{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 2215,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "contrastive_loss": 1.0729,
      "epoch": 0.002257336343115124,
      "grad_norm": 74.65375518798828,
      "learning_rate": 2.0000000000000002e-07,
      "lm_loss": 18.1331,
      "loss": 4.5004,
      "step": 1,
      "text_contrastive_loss": 3.2285,
      "train_positive_log_prob": -273.6295,
      "train_positive_token_accuracy": 0.0003,
      "train_positive_token_prob": 0.001
    },
    {
      "contrastive_loss": 1.1852,
      "epoch": 0.004514672686230248,
      "grad_norm": 88.49497985839844,
      "learning_rate": 4.0000000000000003e-07,
      "lm_loss": 18.3479,
      "loss": 4.7987,
      "step": 2,
      "text_contrastive_loss": 3.5573,
      "train_positive_log_prob": -263.4631,
      "train_positive_token_accuracy": 0.0014,
      "train_positive_token_prob": 0.0009
    },
    {
      "contrastive_loss": 0.7489,
      "epoch": 0.006772009029345372,
      "grad_norm": 67.62373352050781,
      "learning_rate": 6.000000000000001e-07,
      "lm_loss": 17.6708,
      "loss": 4.1388,
      "step": 3,
      "text_contrastive_loss": 3.2457,
      "train_positive_log_prob": -267.5312,
      "train_positive_token_accuracy": 0.0015,
      "train_positive_token_prob": 0.0015
    },
    {
      "contrastive_loss": 0.7492,
      "epoch": 0.009029345372460496,
      "grad_norm": 58.398277282714844,
      "learning_rate": 8.000000000000001e-07,
      "lm_loss": 18.0811,
      "loss": 3.9772,
      "step": 4,
      "text_contrastive_loss": 2.8399,
      "train_positive_log_prob": -267.6794,
      "train_positive_token_accuracy": 0.0005,
      "train_positive_token_prob": 0.001
    },
    {
      "contrastive_loss": 0.8508,
      "epoch": 0.011286681715575621,
      "grad_norm": 61.87724304199219,
      "learning_rate": 1.0000000000000002e-06,
      "lm_loss": 17.6707,
      "loss": 4.1976,
      "step": 5,
      "text_contrastive_loss": 3.1595,
      "train_positive_log_prob": -262.9114,
      "train_positive_token_accuracy": 0.0016,
      "train_positive_token_prob": 0.0015
    },
    {
      "contrastive_loss": 0.9909,
      "epoch": 0.013544018058690745,
      "grad_norm": 58.21712875366211,
      "learning_rate": 1.2000000000000002e-06,
      "lm_loss": 17.7713,
      "loss": 4.2961,
      "step": 6,
      "text_contrastive_loss": 3.0561,
      "train_positive_log_prob": -260.9442,
      "train_positive_token_accuracy": 0.0019,
      "train_positive_token_prob": 0.0013
    },
    {
      "contrastive_loss": 0.8544,
      "epoch": 0.01580135440180587,
      "grad_norm": 57.41749954223633,
      "learning_rate": 1.4000000000000001e-06,
      "lm_loss": 17.6181,
      "loss": 3.7124,
      "step": 7,
      "text_contrastive_loss": 2.1923,
      "train_positive_log_prob": -258.7651,
      "train_positive_token_accuracy": 0.0016,
      "train_positive_token_prob": 0.0013
    },
    {
      "contrastive_loss": 0.7607,
      "epoch": 0.01805869074492099,
      "grad_norm": 53.608184814453125,
      "learning_rate": 1.6000000000000001e-06,
      "lm_loss": 17.6923,
      "loss": 3.8436,
      "step": 8,
      "text_contrastive_loss": 2.6273,
      "train_positive_log_prob": -260.3477,
      "train_positive_token_accuracy": 0.0021,
      "train_positive_token_prob": 0.0021
    },
    {
      "contrastive_loss": 0.841,
      "epoch": 0.020316027088036117,
      "grad_norm": 54.720584869384766,
      "learning_rate": 1.8000000000000001e-06,
      "lm_loss": 17.152,
      "loss": 3.8483,
      "step": 9,
      "text_contrastive_loss": 2.5842,
      "train_positive_log_prob": -257.68,
      "train_positive_token_accuracy": 0.0016,
      "train_positive_token_prob": 0.0016
    },
    {
      "contrastive_loss": 0.8786,
      "epoch": 0.022573363431151242,
      "grad_norm": 49.16249084472656,
      "learning_rate": 2.0000000000000003e-06,
      "lm_loss": 17.1386,
      "loss": 3.6662,
      "step": 10,
      "text_contrastive_loss": 2.1474,
      "train_positive_log_prob": -251.052,
      "train_positive_token_accuracy": 0.0024,
      "train_positive_token_prob": 0.0023
    },
    {
      "contrastive_loss": 0.8012,
      "epoch": 0.024830699774266364,
      "grad_norm": 44.05373764038086,
      "learning_rate": 2.2e-06,
      "lm_loss": 16.7479,
      "loss": 3.3581,
      "step": 11,
      "text_contrastive_loss": 1.7641,
      "train_positive_log_prob": -241.9287,
      "train_positive_token_accuracy": 0.0022,
      "train_positive_token_prob": 0.002
    },
    {
      "contrastive_loss": 0.875,
      "epoch": 0.02708803611738149,
      "grad_norm": 44.777740478515625,
      "learning_rate": 2.4000000000000003e-06,
      "lm_loss": 16.5121,
      "loss": 3.5405,
      "step": 12,
      "text_contrastive_loss": 2.0286,
      "train_positive_log_prob": -246.7872,
      "train_positive_token_accuracy": 0.0026,
      "train_positive_token_prob": 0.0021
    },
    {
      "contrastive_loss": 0.7919,
      "epoch": 0.029345372460496615,
      "grad_norm": 47.89744186401367,
      "learning_rate": 2.6e-06,
      "lm_loss": 16.6649,
      "loss": 3.7886,
      "step": 13,
      "text_contrastive_loss": 2.6604,
      "train_positive_log_prob": -248.2189,
      "train_positive_token_accuracy": 0.0039,
      "train_positive_token_prob": 0.0024
    },
    {
      "contrastive_loss": 0.5726,
      "epoch": 0.03160270880361174,
      "grad_norm": 43.6002082824707,
      "learning_rate": 2.8000000000000003e-06,
      "lm_loss": 16.1374,
      "loss": 3.1962,
      "step": 14,
      "text_contrastive_loss": 2.0196,
      "train_positive_log_prob": -243.8907,
      "train_positive_token_accuracy": 0.0016,
      "train_positive_token_prob": 0.0019
    },
    {
      "contrastive_loss": 0.7991,
      "epoch": 0.033860045146726865,
      "grad_norm": 40.39301681518555,
      "learning_rate": 3e-06,
      "lm_loss": 15.683,
      "loss": 3.4338,
      "step": 15,
      "text_contrastive_loss": 2.1329,
      "train_positive_log_prob": -227.1494,
      "train_positive_token_accuracy": 0.0024,
      "train_positive_token_prob": 0.0019
    },
    {
      "contrastive_loss": 0.8454,
      "epoch": 0.03611738148984198,
      "grad_norm": 38.99760055541992,
      "learning_rate": 3.2000000000000003e-06,
      "lm_loss": 15.3319,
      "loss": 3.4371,
      "step": 16,
      "text_contrastive_loss": 2.1172,
      "train_positive_log_prob": -227.2166,
      "train_positive_token_accuracy": 0.004,
      "train_positive_token_prob": 0.0023
    },
    {
      "contrastive_loss": 0.8265,
      "epoch": 0.03837471783295711,
      "grad_norm": 37.754451751708984,
      "learning_rate": 3.4000000000000005e-06,
      "lm_loss": 14.9289,
      "loss": 3.3596,
      "step": 17,
      "text_contrastive_loss": 2.0804,
      "train_positive_log_prob": -220.669,
      "train_positive_token_accuracy": 0.0032,
      "train_positive_token_prob": 0.0028
    },
    {
      "contrastive_loss": 0.8901,
      "epoch": 0.040632054176072234,
      "grad_norm": 36.27416229248047,
      "learning_rate": 3.6000000000000003e-06,
      "lm_loss": 14.8213,
      "loss": 3.3589,
      "step": 18,
      "text_contrastive_loss": 1.9733,
      "train_positive_log_prob": -225.507,
      "train_positive_token_accuracy": 0.0033,
      "train_positive_token_prob": 0.003
    },
    {
      "contrastive_loss": 0.7757,
      "epoch": 0.04288939051918736,
      "grad_norm": 33.69345474243164,
      "learning_rate": 3.8000000000000005e-06,
      "lm_loss": 14.3138,
      "loss": 3.0299,
      "step": 19,
      "text_contrastive_loss": 1.6456,
      "train_positive_log_prob": -206.9811,
      "train_positive_token_accuracy": 0.0038,
      "train_positive_token_prob": 0.0027
    },
    {
      "contrastive_loss": 0.8273,
      "epoch": 0.045146726862302484,
      "grad_norm": 30.30756187438965,
      "learning_rate": 4.000000000000001e-06,
      "lm_loss": 13.9409,
      "loss": 2.9315,
      "step": 20,
      "text_contrastive_loss": 1.4203,
      "train_positive_log_prob": -204.2179,
      "train_positive_token_accuracy": 0.0053,
      "train_positive_token_prob": 0.0035
    },
    {
      "contrastive_loss": 0.7968,
      "epoch": 0.04740406320541761,
      "grad_norm": 27.533042907714844,
      "learning_rate": 4.2000000000000004e-06,
      "lm_loss": 13.582,
      "loss": 2.8958,
      "step": 21,
      "text_contrastive_loss": 1.4817,
      "train_positive_log_prob": -205.3794,
      "train_positive_token_accuracy": 0.0031,
      "train_positive_token_prob": 0.0028
    },
    {
      "contrastive_loss": 0.8334,
      "epoch": 0.04966139954853273,
      "grad_norm": 27.34656524658203,
      "learning_rate": 4.4e-06,
      "lm_loss": 13.4126,
      "loss": 2.9698,
      "step": 22,
      "text_contrastive_loss": 1.5904,
      "train_positive_log_prob": -195.7889,
      "train_positive_token_accuracy": 0.0075,
      "train_positive_token_prob": 0.0042
    },
    {
      "contrastive_loss": 0.8397,
      "epoch": 0.05191873589164785,
      "grad_norm": 26.550691604614258,
      "learning_rate": 4.600000000000001e-06,
      "lm_loss": 13.0318,
      "loss": 2.8192,
      "step": 23,
      "text_contrastive_loss": 1.3528,
      "train_positive_log_prob": -192.6257,
      "train_positive_token_accuracy": 0.009,
      "train_positive_token_prob": 0.0044
    },
    {
      "contrastive_loss": 0.7736,
      "epoch": 0.05417607223476298,
      "grad_norm": 24.160314559936523,
      "learning_rate": 4.800000000000001e-06,
      "lm_loss": 12.7815,
      "loss": 2.7666,
      "step": 24,
      "text_contrastive_loss": 1.4298,
      "train_positive_log_prob": -187.5415,
      "train_positive_token_accuracy": 0.008,
      "train_positive_token_prob": 0.0044
    },
    {
      "contrastive_loss": 0.7041,
      "epoch": 0.056433408577878104,
      "grad_norm": 23.253559112548828,
      "learning_rate": 5e-06,
      "lm_loss": 12.3029,
      "loss": 2.559,
      "step": 25,
      "text_contrastive_loss": 1.2492,
      "train_positive_log_prob": -182.9034,
      "train_positive_token_accuracy": 0.0126,
      "train_positive_token_prob": 0.006
    },
    {
      "contrastive_loss": 0.7477,
      "epoch": 0.05869074492099323,
      "grad_norm": 22.743453979492188,
      "learning_rate": 5.2e-06,
      "lm_loss": 11.9511,
      "loss": 2.5664,
      "step": 26,
      "text_contrastive_loss": 1.2472,
      "train_positive_log_prob": -179.451,
      "train_positive_token_accuracy": 0.0177,
      "train_positive_token_prob": 0.0089
    },
    {
      "contrastive_loss": 0.8568,
      "epoch": 0.060948081264108354,
      "grad_norm": 22.865564346313477,
      "learning_rate": 5.400000000000001e-06,
      "lm_loss": 11.7099,
      "loss": 2.7531,
      "step": 27,
      "text_contrastive_loss": 1.4506,
      "train_positive_log_prob": -176.3327,
      "train_positive_token_accuracy": 0.0179,
      "train_positive_token_prob": 0.0087
    },
    {
      "contrastive_loss": 0.8545,
      "epoch": 0.06320541760722348,
      "grad_norm": 22.831811904907227,
      "learning_rate": 5.600000000000001e-06,
      "lm_loss": 11.4956,
      "loss": 2.6602,
      "step": 28,
      "text_contrastive_loss": 1.3122,
      "train_positive_log_prob": -170.1002,
      "train_positive_token_accuracy": 0.0251,
      "train_positive_token_prob": 0.0117
    },
    {
      "contrastive_loss": 0.8978,
      "epoch": 0.0654627539503386,
      "grad_norm": 24.4145450592041,
      "learning_rate": 5.8e-06,
      "lm_loss": 11.0337,
      "loss": 2.587,
      "step": 29,
      "text_contrastive_loss": 1.1715,
      "train_positive_log_prob": -158.9954,
      "train_positive_token_accuracy": 0.0339,
      "train_positive_token_prob": 0.0133
    },
    {
      "contrastive_loss": 0.8086,
      "epoch": 0.06772009029345373,
      "grad_norm": 22.760570526123047,
      "learning_rate": 6e-06,
      "lm_loss": 10.7794,
      "loss": 2.4137,
      "step": 30,
      "text_contrastive_loss": 1.0544,
      "train_positive_log_prob": -158.916,
      "train_positive_token_accuracy": 0.0395,
      "train_positive_token_prob": 0.0142
    },
    {
      "contrastive_loss": 0.706,
      "epoch": 0.06997742663656885,
      "grad_norm": 18.718631744384766,
      "learning_rate": 6.200000000000001e-06,
      "lm_loss": 10.463,
      "loss": 2.2648,
      "step": 31,
      "text_contrastive_loss": 1.025,
      "train_positive_log_prob": -155.0657,
      "train_positive_token_accuracy": 0.0464,
      "train_positive_token_prob": 0.0166
    },
    {
      "contrastive_loss": 0.8945,
      "epoch": 0.07223476297968397,
      "grad_norm": 21.011608123779297,
      "learning_rate": 6.4000000000000006e-06,
      "lm_loss": 10.1752,
      "loss": 2.4952,
      "step": 32,
      "text_contrastive_loss": 1.1663,
      "train_positive_log_prob": -148.1847,
      "train_positive_token_accuracy": 0.0488,
      "train_positive_token_prob": 0.0187
    },
    {
      "contrastive_loss": 0.778,
      "epoch": 0.0744920993227991,
      "grad_norm": 20.67786407470703,
      "learning_rate": 6.600000000000001e-06,
      "lm_loss": 10.0549,
      "loss": 2.4966,
      "step": 33,
      "text_contrastive_loss": 1.4263,
      "train_positive_log_prob": -149.3017,
      "train_positive_token_accuracy": 0.0579,
      "train_positive_token_prob": 0.0213
    },
    {
      "contrastive_loss": 0.7866,
      "epoch": 0.07674943566591422,
      "grad_norm": 20.956958770751953,
      "learning_rate": 6.800000000000001e-06,
      "lm_loss": 9.8113,
      "loss": 2.3065,
      "step": 34,
      "text_contrastive_loss": 1.0776,
      "train_positive_log_prob": -145.5954,
      "train_positive_token_accuracy": 0.0663,
      "train_positive_token_prob": 0.0236
    },
    {
      "contrastive_loss": 0.7381,
      "epoch": 0.07900677200902935,
      "grad_norm": 20.436098098754883,
      "learning_rate": 7e-06,
      "lm_loss": 9.712,
      "loss": 2.2814,
      "step": 35,
      "text_contrastive_loss": 1.1442,
      "train_positive_log_prob": -146.7038,
      "train_positive_token_accuracy": 0.0727,
      "train_positive_token_prob": 0.0256
    },
    {
      "contrastive_loss": 0.8602,
      "epoch": 0.08126410835214447,
      "grad_norm": 22.360729217529297,
      "learning_rate": 7.2000000000000005e-06,
      "lm_loss": 9.438,
      "loss": 2.3129,
      "step": 36,
      "text_contrastive_loss": 1.0178,
      "train_positive_log_prob": -143.5192,
      "train_positive_token_accuracy": 0.0753,
      "train_positive_token_prob": 0.0252
    },
    {
      "contrastive_loss": 0.7534,
      "epoch": 0.0835214446952596,
      "grad_norm": 20.295902252197266,
      "learning_rate": 7.4e-06,
      "lm_loss": 9.2734,
      "loss": 2.2474,
      "step": 37,
      "text_contrastive_loss": 1.1334,
      "train_positive_log_prob": -134.3203,
      "train_positive_token_accuracy": 0.0709,
      "train_positive_token_prob": 0.026
    },
    {
      "contrastive_loss": 0.7088,
      "epoch": 0.08577878103837472,
      "grad_norm": 19.866477966308594,
      "learning_rate": 7.600000000000001e-06,
      "lm_loss": 9.163,
      "loss": 2.1854,
      "step": 38,
      "text_contrastive_loss": 1.1206,
      "train_positive_log_prob": -140.95,
      "train_positive_token_accuracy": 0.0759,
      "train_positive_token_prob": 0.0253
    },
    {
      "contrastive_loss": 0.6501,
      "epoch": 0.08803611738148984,
      "grad_norm": 20.59491729736328,
      "learning_rate": 7.800000000000002e-06,
      "lm_loss": 9.0296,
      "loss": 2.0447,
      "step": 39,
      "text_contrastive_loss": 0.9833,
      "train_positive_log_prob": -134.0005,
      "train_positive_token_accuracy": 0.0805,
      "train_positive_token_prob": 0.0268
    },
    {
      "contrastive_loss": 0.8234,
      "epoch": 0.09029345372460497,
      "grad_norm": 21.421241760253906,
      "learning_rate": 8.000000000000001e-06,
      "lm_loss": 8.9082,
      "loss": 2.2105,
      "step": 40,
      "text_contrastive_loss": 0.9924,
      "train_positive_log_prob": -133.1723,
      "train_positive_token_accuracy": 0.0834,
      "train_positive_token_prob": 0.0266
    },
    {
      "contrastive_loss": 0.633,
      "epoch": 0.09255079006772009,
      "grad_norm": 18.046871185302734,
      "learning_rate": 8.2e-06,
      "lm_loss": 8.7144,
      "loss": 2.0768,
      "step": 41,
      "text_contrastive_loss": 1.1447,
      "train_positive_log_prob": -129.3175,
      "train_positive_token_accuracy": 0.0721,
      "train_positive_token_prob": 0.0237
    },
    {
      "contrastive_loss": 0.7048,
      "epoch": 0.09480812641083522,
      "grad_norm": 21.57222557067871,
      "learning_rate": 8.400000000000001e-06,
      "lm_loss": 8.6116,
      "loss": 2.1106,
      "step": 42,
      "text_contrastive_loss": 1.0892,
      "train_positive_log_prob": -128.9057,
      "train_positive_token_accuracy": 0.083,
      "train_positive_token_prob": 0.0233
    },
    {
      "contrastive_loss": 0.7261,
      "epoch": 0.09706546275395034,
      "grad_norm": 23.071239471435547,
      "learning_rate": 8.6e-06,
      "lm_loss": 8.5381,
      "loss": 2.1658,
      "step": 43,
      "text_contrastive_loss": 1.1717,
      "train_positive_log_prob": -124.6441,
      "train_positive_token_accuracy": 0.076,
      "train_positive_token_prob": 0.0226
    },
    {
      "contrastive_loss": 0.7527,
      "epoch": 0.09932279909706546,
      "grad_norm": 22.00257110595703,
      "learning_rate": 8.8e-06,
      "lm_loss": 8.5846,
      "loss": 2.1009,
      "step": 44,
      "text_contrastive_loss": 0.9795,
      "train_positive_log_prob": -125.0441,
      "train_positive_token_accuracy": 0.0711,
      "train_positive_token_prob": 0.0205
    },
    {
      "contrastive_loss": 0.6884,
      "epoch": 0.10158013544018059,
      "grad_norm": 20.627120971679688,
      "learning_rate": 9e-06,
      "lm_loss": 8.2954,
      "loss": 2.0499,
      "step": 45,
      "text_contrastive_loss": 1.064,
      "train_positive_log_prob": -122.1275,
      "train_positive_token_accuracy": 0.0743,
      "train_positive_token_prob": 0.0205
    },
    {
      "contrastive_loss": 0.7249,
      "epoch": 0.1038374717832957,
      "grad_norm": 21.081621170043945,
      "learning_rate": 9.200000000000002e-06,
      "lm_loss": 8.4546,
      "loss": 2.0632,
      "step": 46,
      "text_contrastive_loss": 0.9856,
      "train_positive_log_prob": -127.9356,
      "train_positive_token_accuracy": 0.0687,
      "train_positive_token_prob": 0.0178
    },
    {
      "contrastive_loss": 0.7611,
      "epoch": 0.10609480812641084,
      "grad_norm": 20.1016902923584,
      "learning_rate": 9.4e-06,
      "lm_loss": 8.3393,
      "loss": 2.2063,
      "step": 47,
      "text_contrastive_loss": 1.2226,
      "train_positive_log_prob": -123.2022,
      "train_positive_token_accuracy": 0.0693,
      "train_positive_token_prob": 0.0176
    },
    {
      "contrastive_loss": 0.6507,
      "epoch": 0.10835214446952596,
      "grad_norm": 18.013376235961914,
      "learning_rate": 9.600000000000001e-06,
      "lm_loss": 8.1474,
      "loss": 2.0983,
      "step": 48,
      "text_contrastive_loss": 1.2658,
      "train_positive_log_prob": -117.2156,
      "train_positive_token_accuracy": 0.0728,
      "train_positive_token_prob": 0.0181
    },
    {
      "contrastive_loss": 0.6709,
      "epoch": 0.11060948081264109,
      "grad_norm": 19.485240936279297,
      "learning_rate": 9.800000000000001e-06,
      "lm_loss": 8.0471,
      "loss": 1.9798,
      "step": 49,
      "text_contrastive_loss": 1.0085,
      "train_positive_log_prob": -118.3807,
      "train_positive_token_accuracy": 0.0672,
      "train_positive_token_prob": 0.0169
    },
    {
      "contrastive_loss": 0.718,
      "epoch": 0.11286681715575621,
      "grad_norm": 19.43058967590332,
      "learning_rate": 1e-05,
      "lm_loss": 8.0583,
      "loss": 1.9316,
      "step": 50,
      "text_contrastive_loss": 0.8156,
      "train_positive_log_prob": -121.3418,
      "train_positive_token_accuracy": 0.0672,
      "train_positive_token_prob": 0.0177
    },
    {
      "contrastive_loss": 0.6668,
      "epoch": 0.11512415349887133,
      "grad_norm": 18.925294876098633,
      "learning_rate": 9.999994735903083e-06,
      "lm_loss": 8.1304,
      "loss": 2.0567,
      "step": 51,
      "text_contrastive_loss": 1.1538,
      "train_positive_log_prob": -119.5728,
      "train_positive_token_accuracy": 0.0622,
      "train_positive_token_prob": 0.0156
    },
    {
      "contrastive_loss": 0.7649,
      "epoch": 0.11738148984198646,
      "grad_norm": 22.27410316467285,
      "learning_rate": 9.999978943623417e-06,
      "lm_loss": 7.9671,
      "loss": 2.0171,
      "step": 52,
      "text_contrastive_loss": 0.911,
      "train_positive_log_prob": -116.2374,
      "train_positive_token_accuracy": 0.0651,
      "train_positive_token_prob": 0.016
    },
    {
      "contrastive_loss": 0.6431,
      "epoch": 0.11963882618510158,
      "grad_norm": 18.041515350341797,
      "learning_rate": 9.999952623194252e-06,
      "lm_loss": 7.9681,
      "loss": 1.8983,
      "step": 53,
      "text_contrastive_loss": 0.9168,
      "train_positive_log_prob": -117.9658,
      "train_positive_token_accuracy": 0.0652,
      "train_positive_token_prob": 0.0155
    },
    {
      "contrastive_loss": 0.6277,
      "epoch": 0.12189616252821671,
      "grad_norm": 20.22205352783203,
      "learning_rate": 9.999915774671009e-06,
      "lm_loss": 7.8363,
      "loss": 1.9114,
      "step": 54,
      "text_contrastive_loss": 1.0003,
      "train_positive_log_prob": -114.7927,
      "train_positive_token_accuracy": 0.0659,
      "train_positive_token_prob": 0.016
    },
    {
      "contrastive_loss": 0.7325,
      "epoch": 0.12415349887133183,
      "grad_norm": 23.12201499938965,
      "learning_rate": 9.999868398131282e-06,
      "lm_loss": 7.8999,
      "loss": 2.0765,
      "step": 55,
      "text_contrastive_loss": 1.108,
      "train_positive_log_prob": -119.4265,
      "train_positive_token_accuracy": 0.0669,
      "train_positive_token_prob": 0.0164
    },
    {
      "contrastive_loss": 0.5996,
      "epoch": 0.12641083521444696,
      "grad_norm": 18.512279510498047,
      "learning_rate": 9.999810493674826e-06,
      "lm_loss": 7.8678,
      "loss": 1.8542,
      "step": 56,
      "text_contrastive_loss": 0.9356,
      "train_positive_log_prob": -114.8177,
      "train_positive_token_accuracy": 0.0701,
      "train_positive_token_prob": 0.0164
    },
    {
      "contrastive_loss": 0.7241,
      "epoch": 0.12866817155756208,
      "grad_norm": 20.64830780029297,
      "learning_rate": 9.999742061423567e-06,
      "lm_loss": 7.7702,
      "loss": 1.9937,
      "step": 57,
      "text_contrastive_loss": 0.9852,
      "train_positive_log_prob": -115.5182,
      "train_positive_token_accuracy": 0.0673,
      "train_positive_token_prob": 0.0163
    },
    {
      "contrastive_loss": 0.6388,
      "epoch": 0.1309255079006772,
      "grad_norm": 17.783964157104492,
      "learning_rate": 9.999663101521599e-06,
      "lm_loss": 7.7735,
      "loss": 1.9498,
      "step": 58,
      "text_contrastive_loss": 1.0673,
      "train_positive_log_prob": -114.9075,
      "train_positive_token_accuracy": 0.0632,
      "train_positive_token_prob": 0.0157
    },
    {
      "contrastive_loss": 0.6919,
      "epoch": 0.13318284424379231,
      "grad_norm": 19.88160514831543,
      "learning_rate": 9.999573614135183e-06,
      "lm_loss": 7.7051,
      "loss": 1.9881,
      "step": 59,
      "text_contrastive_loss": 1.0513,
      "train_positive_log_prob": -114.5836,
      "train_positive_token_accuracy": 0.0612,
      "train_positive_token_prob": 0.0156
    },
    {
      "contrastive_loss": 0.6188,
      "epoch": 0.13544018058690746,
      "grad_norm": 18.36577796936035,
      "learning_rate": 9.999473599452746e-06,
      "lm_loss": 7.6918,
      "loss": 1.9356,
      "step": 60,
      "text_contrastive_loss": 1.0952,
      "train_positive_log_prob": -115.5601,
      "train_positive_token_accuracy": 0.0621,
      "train_positive_token_prob": 0.016
    },
    {
      "contrastive_loss": 0.6115,
      "epoch": 0.13769751693002258,
      "grad_norm": 19.831995010375977,
      "learning_rate": 9.999363057684885e-06,
      "lm_loss": 7.6733,
      "loss": 1.8999,
      "step": 61,
      "text_contrastive_loss": 1.0422,
      "train_positive_log_prob": -113.1827,
      "train_positive_token_accuracy": 0.0638,
      "train_positive_token_prob": 0.0165
    },
    {
      "contrastive_loss": 0.6345,
      "epoch": 0.1399548532731377,
      "grad_norm": 18.292102813720703,
      "learning_rate": 9.999241989064358e-06,
      "lm_loss": 7.5556,
      "loss": 1.8898,
      "step": 62,
      "text_contrastive_loss": 0.9995,
      "train_positive_log_prob": -111.4475,
      "train_positive_token_accuracy": 0.0699,
      "train_positive_token_prob": 0.0165
    },
    {
      "contrastive_loss": 0.7301,
      "epoch": 0.14221218961625282,
      "grad_norm": 18.254596710205078,
      "learning_rate": 9.999110393846097e-06,
      "lm_loss": 7.5002,
      "loss": 2.1006,
      "step": 63,
      "text_contrastive_loss": 1.241,
      "train_positive_log_prob": -109.9251,
      "train_positive_token_accuracy": 0.0642,
      "train_positive_token_prob": 0.0157
    },
    {
      "contrastive_loss": 0.6444,
      "epoch": 0.14446952595936793,
      "grad_norm": 17.319774627685547,
      "learning_rate": 9.998968272307187e-06,
      "lm_loss": 7.4982,
      "loss": 1.8944,
      "step": 64,
      "text_contrastive_loss": 1.0002,
      "train_positive_log_prob": -111.0935,
      "train_positive_token_accuracy": 0.0693,
      "train_positive_token_prob": 0.0168
    },
    {
      "contrastive_loss": 0.7149,
      "epoch": 0.14672686230248308,
      "grad_norm": 18.645326614379883,
      "learning_rate": 9.99881562474689e-06,
      "lm_loss": 7.5473,
      "loss": 1.9202,
      "step": 65,
      "text_contrastive_loss": 0.9012,
      "train_positive_log_prob": -111.0243,
      "train_positive_token_accuracy": 0.0685,
      "train_positive_token_prob": 0.0164
    },
    {
      "contrastive_loss": 0.6409,
      "epoch": 0.1489841986455982,
      "grad_norm": 18.37657928466797,
      "learning_rate": 9.998652451486626e-06,
      "lm_loss": 7.6032,
      "loss": 1.8893,
      "step": 66,
      "text_contrastive_loss": 0.9761,
      "train_positive_log_prob": -113.0417,
      "train_positive_token_accuracy": 0.0633,
      "train_positive_token_prob": 0.0162
    },
    {
      "contrastive_loss": 0.7272,
      "epoch": 0.15124153498871332,
      "grad_norm": 18.725929260253906,
      "learning_rate": 9.998478752869976e-06,
      "lm_loss": 7.4061,
      "loss": 1.9524,
      "step": 67,
      "text_contrastive_loss": 0.9692,
      "train_positive_log_prob": -110.0508,
      "train_positive_token_accuracy": 0.0754,
      "train_positive_token_prob": 0.0184
    },
    {
      "contrastive_loss": 0.7003,
      "epoch": 0.15349887133182843,
      "grad_norm": 19.5133113861084,
      "learning_rate": 9.998294529262688e-06,
      "lm_loss": 7.5117,
      "loss": 1.977,
      "step": 68,
      "text_contrastive_loss": 1.0512,
      "train_positive_log_prob": -109.5966,
      "train_positive_token_accuracy": 0.0648,
      "train_positive_token_prob": 0.0175
    },
    {
      "contrastive_loss": 0.7785,
      "epoch": 0.15575620767494355,
      "grad_norm": 20.068553924560547,
      "learning_rate": 9.998099781052673e-06,
      "lm_loss": 7.3481,
      "loss": 2.0972,
      "step": 69,
      "text_contrastive_loss": 1.1678,
      "train_positive_log_prob": -111.4008,
      "train_positive_token_accuracy": 0.0657,
      "train_positive_token_prob": 0.0162
    },
    {
      "contrastive_loss": 0.7231,
      "epoch": 0.1580135440180587,
      "grad_norm": 22.557743072509766,
      "learning_rate": 9.997894508649995e-06,
      "lm_loss": 7.4716,
      "loss": 1.9544,
      "step": 70,
      "text_contrastive_loss": 0.9681,
      "train_positive_log_prob": -109.6256,
      "train_positive_token_accuracy": 0.066,
      "train_positive_token_prob": 0.0163
    },
    {
      "contrastive_loss": 0.5091,
      "epoch": 0.16027088036117382,
      "grad_norm": 15.837409019470215,
      "learning_rate": 9.997678712486889e-06,
      "lm_loss": 7.4683,
      "loss": 1.6324,
      "step": 71,
      "text_contrastive_loss": 0.753,
      "train_positive_log_prob": -104.9369,
      "train_positive_token_accuracy": 0.0659,
      "train_positive_token_prob": 0.0167
    },
    {
      "contrastive_loss": 0.6891,
      "epoch": 0.16252821670428894,
      "grad_norm": 17.99678611755371,
      "learning_rate": 9.99745239301774e-06,
      "lm_loss": 7.3014,
      "loss": 1.9611,
      "step": 72,
      "text_contrastive_loss": 1.0837,
      "train_positive_log_prob": -108.8716,
      "train_positive_token_accuracy": 0.0668,
      "train_positive_token_prob": 0.0174
    },
    {
      "contrastive_loss": 0.7978,
      "epoch": 0.16478555304740405,
      "grad_norm": 18.621131896972656,
      "learning_rate": 9.997215550719097e-06,
      "lm_loss": 7.2734,
      "loss": 2.1024,
      "step": 73,
      "text_contrastive_loss": 1.1546,
      "train_positive_log_prob": -107.9936,
      "train_positive_token_accuracy": 0.0742,
      "train_positive_token_prob": 0.0189
    },
    {
      "contrastive_loss": 0.6444,
      "epoch": 0.1670428893905192,
      "grad_norm": 19.960073471069336,
      "learning_rate": 9.996968186089664e-06,
      "lm_loss": 7.336,
      "loss": 1.8888,
      "step": 74,
      "text_contrastive_loss": 1.0217,
      "train_positive_log_prob": -108.2594,
      "train_positive_token_accuracy": 0.0709,
      "train_positive_token_prob": 0.0172
    },
    {
      "contrastive_loss": 0.8172,
      "epoch": 0.16930022573363432,
      "grad_norm": 19.790021896362305,
      "learning_rate": 9.996710299650302e-06,
      "lm_loss": 7.2372,
      "loss": 2.0423,
      "step": 75,
      "text_contrastive_loss": 1.0028,
      "train_positive_log_prob": -105.7312,
      "train_positive_token_accuracy": 0.0746,
      "train_positive_token_prob": 0.018
    },
    {
      "contrastive_loss": 0.7105,
      "epoch": 0.17155756207674944,
      "grad_norm": 19.25592041015625,
      "learning_rate": 9.996441891944023e-06,
      "lm_loss": 7.2776,
      "loss": 1.9127,
      "step": 76,
      "text_contrastive_loss": 0.9489,
      "train_positive_log_prob": -106.119,
      "train_positive_token_accuracy": 0.0651,
      "train_positive_token_prob": 0.0175
    },
    {
      "contrastive_loss": 0.7012,
      "epoch": 0.17381489841986456,
      "grad_norm": 17.665687561035156,
      "learning_rate": 9.996162963536004e-06,
      "lm_loss": 7.2626,
      "loss": 1.9637,
      "step": 77,
      "text_contrastive_loss": 1.0724,
      "train_positive_log_prob": -106.6461,
      "train_positive_token_accuracy": 0.0724,
      "train_positive_token_prob": 0.0179
    },
    {
      "contrastive_loss": 0.7044,
      "epoch": 0.17607223476297967,
      "grad_norm": 17.54836082458496,
      "learning_rate": 9.995873515013562e-06,
      "lm_loss": 7.1187,
      "loss": 1.9171,
      "step": 78,
      "text_contrastive_loss": 1.0017,
      "train_positive_log_prob": -104.0536,
      "train_positive_token_accuracy": 0.0748,
      "train_positive_token_prob": 0.0188
    },
    {
      "contrastive_loss": 0.6873,
      "epoch": 0.17832957110609482,
      "grad_norm": 18.3485050201416,
      "learning_rate": 9.99557354698617e-06,
      "lm_loss": 7.1606,
      "loss": 1.8431,
      "step": 79,
      "text_contrastive_loss": 0.8796,
      "train_positive_log_prob": -106.0165,
      "train_positive_token_accuracy": 0.0654,
      "train_positive_token_prob": 0.0176
    },
    {
      "contrastive_loss": 0.6392,
      "epoch": 0.18058690744920994,
      "grad_norm": 20.063926696777344,
      "learning_rate": 9.995263060085456e-06,
      "lm_loss": 7.1678,
      "loss": 1.8971,
      "step": 80,
      "text_contrastive_loss": 1.082,
      "train_positive_log_prob": -108.6669,
      "train_positive_token_accuracy": 0.0691,
      "train_positive_token_prob": 0.0189
    },
    {
      "contrastive_loss": 0.7355,
      "epoch": 0.18284424379232506,
      "grad_norm": 20.3895320892334,
      "learning_rate": 9.99494205496519e-06,
      "lm_loss": 7.1325,
      "loss": 2.0276,
      "step": 81,
      "text_contrastive_loss": 1.1576,
      "train_positive_log_prob": -106.0987,
      "train_positive_token_accuracy": 0.068,
      "train_positive_token_prob": 0.0186
    },
    {
      "contrastive_loss": 0.6852,
      "epoch": 0.18510158013544017,
      "grad_norm": 18.73487663269043,
      "learning_rate": 9.994610532301296e-06,
      "lm_loss": 7.1894,
      "loss": 1.9248,
      "step": 82,
      "text_contrastive_loss": 1.0412,
      "train_positive_log_prob": -107.9551,
      "train_positive_token_accuracy": 0.0627,
      "train_positive_token_prob": 0.0176
    },
    {
      "contrastive_loss": 0.6669,
      "epoch": 0.1873589164785553,
      "grad_norm": 18.10122299194336,
      "learning_rate": 9.99426849279184e-06,
      "lm_loss": 7.1451,
      "loss": 1.9421,
      "step": 83,
      "text_contrastive_loss": 1.1214,
      "train_positive_log_prob": -106.0044,
      "train_positive_token_accuracy": 0.0661,
      "train_positive_token_prob": 0.0184
    },
    {
      "contrastive_loss": 0.6057,
      "epoch": 0.18961625282167044,
      "grad_norm": 18.441043853759766,
      "learning_rate": 9.993915937157033e-06,
      "lm_loss": 7.2141,
      "loss": 1.8116,
      "step": 84,
      "text_contrastive_loss": 0.9691,
      "train_positive_log_prob": -105.2496,
      "train_positive_token_accuracy": 0.0718,
      "train_positive_token_prob": 0.0179
    },
    {
      "contrastive_loss": 0.8542,
      "epoch": 0.19187358916478556,
      "grad_norm": 20.69758415222168,
      "learning_rate": 9.99355286613923e-06,
      "lm_loss": 6.9868,
      "loss": 2.2421,
      "step": 85,
      "text_contrastive_loss": 1.3785,
      "train_positive_log_prob": -103.302,
      "train_positive_token_accuracy": 0.0745,
      "train_positive_token_prob": 0.0193
    },
    {
      "contrastive_loss": 0.68,
      "epoch": 0.19413092550790068,
      "grad_norm": 20.38166046142578,
      "learning_rate": 9.993179280502926e-06,
      "lm_loss": 7.0946,
      "loss": 1.8517,
      "step": 86,
      "text_contrastive_loss": 0.9244,
      "train_positive_log_prob": -104.6733,
      "train_positive_token_accuracy": 0.0739,
      "train_positive_token_prob": 0.0194
    },
    {
      "contrastive_loss": 0.7434,
      "epoch": 0.1963882618510158,
      "grad_norm": 20.94504165649414,
      "learning_rate": 9.99279518103476e-06,
      "lm_loss": 7.1518,
      "loss": 1.9784,
      "step": 87,
      "text_contrastive_loss": 1.0396,
      "train_positive_log_prob": -103.5898,
      "train_positive_token_accuracy": 0.0666,
      "train_positive_token_prob": 0.0183
    },
    {
      "contrastive_loss": 0.7131,
      "epoch": 0.1986455981941309,
      "grad_norm": 18.84032440185547,
      "learning_rate": 9.992400568543506e-06,
      "lm_loss": 6.9985,
      "loss": 1.9961,
      "step": 88,
      "text_contrastive_loss": 1.1664,
      "train_positive_log_prob": -104.3755,
      "train_positive_token_accuracy": 0.0817,
      "train_positive_token_prob": 0.0218
    },
    {
      "contrastive_loss": 0.7467,
      "epoch": 0.20090293453724606,
      "grad_norm": 18.880041122436523,
      "learning_rate": 9.991995443860074e-06,
      "lm_loss": 7.1445,
      "loss": 1.9219,
      "step": 89,
      "text_contrastive_loss": 0.9216,
      "train_positive_log_prob": -104.6812,
      "train_positive_token_accuracy": 0.0776,
      "train_positive_token_prob": 0.0204
    },
    {
      "contrastive_loss": 0.7152,
      "epoch": 0.20316027088036118,
      "grad_norm": 18.664297103881836,
      "learning_rate": 9.991579807837511e-06,
      "lm_loss": 7.0983,
      "loss": 1.9426,
      "step": 90,
      "text_contrastive_loss": 1.0351,
      "train_positive_log_prob": -102.8157,
      "train_positive_token_accuracy": 0.072,
      "train_positive_token_prob": 0.0195
    },
    {
      "contrastive_loss": 0.5356,
      "epoch": 0.2054176072234763,
      "grad_norm": 16.25419044494629,
      "learning_rate": 9.991153661350996e-06,
      "lm_loss": 7.1611,
      "loss": 1.6608,
      "step": 91,
      "text_contrastive_loss": 0.818,
      "train_positive_log_prob": -105.3775,
      "train_positive_token_accuracy": 0.0735,
      "train_positive_token_prob": 0.0204
    },
    {
      "contrastive_loss": 0.6375,
      "epoch": 0.2076749435665914,
      "grad_norm": 19.357620239257812,
      "learning_rate": 9.990717005297841e-06,
      "lm_loss": 7.0838,
      "loss": 1.8145,
      "step": 92,
      "text_contrastive_loss": 0.9373,
      "train_positive_log_prob": -101.9163,
      "train_positive_token_accuracy": 0.0722,
      "train_positive_token_prob": 0.02
    },
    {
      "contrastive_loss": 0.6711,
      "epoch": 0.20993227990970656,
      "grad_norm": 16.920473098754883,
      "learning_rate": 9.990269840597484e-06,
      "lm_loss": 6.9962,
      "loss": 1.8757,
      "step": 93,
      "text_contrastive_loss": 1.0101,
      "train_positive_log_prob": -101.5844,
      "train_positive_token_accuracy": 0.0721,
      "train_positive_token_prob": 0.0204
    },
    {
      "contrastive_loss": 0.6933,
      "epoch": 0.21218961625282168,
      "grad_norm": 18.641815185546875,
      "learning_rate": 9.989812168191495e-06,
      "lm_loss": 6.9525,
      "loss": 1.9171,
      "step": 94,
      "text_contrastive_loss": 1.057,
      "train_positive_log_prob": -102.7659,
      "train_positive_token_accuracy": 0.0772,
      "train_positive_token_prob": 0.0212
    },
    {
      "contrastive_loss": 0.7655,
      "epoch": 0.2144469525959368,
      "grad_norm": 18.8684024810791,
      "learning_rate": 9.989343989043563e-06,
      "lm_loss": 6.8671,
      "loss": 1.9745,
      "step": 95,
      "text_contrastive_loss": 1.0446,
      "train_positive_log_prob": -100.0049,
      "train_positive_token_accuracy": 0.0746,
      "train_positive_token_prob": 0.0227
    },
    {
      "contrastive_loss": 0.8027,
      "epoch": 0.21670428893905191,
      "grad_norm": 20.620410919189453,
      "learning_rate": 9.988865304139509e-06,
      "lm_loss": 6.9551,
      "loss": 2.0692,
      "step": 96,
      "text_contrastive_loss": 1.1421,
      "train_positive_log_prob": -104.0833,
      "train_positive_token_accuracy": 0.0715,
      "train_positive_token_prob": 0.0212
    },
    {
      "contrastive_loss": 0.7118,
      "epoch": 0.21896162528216703,
      "grad_norm": 21.403032302856445,
      "learning_rate": 9.988376114487264e-06,
      "lm_loss": 7.0753,
      "loss": 1.9118,
      "step": 97,
      "text_contrastive_loss": 0.9851,
      "train_positive_log_prob": -103.0357,
      "train_positive_token_accuracy": 0.0754,
      "train_positive_token_prob": 0.0212
    },
    {
      "contrastive_loss": 0.5414,
      "epoch": 0.22121896162528218,
      "grad_norm": 14.425271034240723,
      "learning_rate": 9.98787642111689e-06,
      "lm_loss": 6.9234,
      "loss": 1.7147,
      "step": 98,
      "text_contrastive_loss": 0.9619,
      "train_positive_log_prob": -102.0397,
      "train_positive_token_accuracy": 0.0758,
      "train_positive_token_prob": 0.022
    },
    {
      "contrastive_loss": 0.826,
      "epoch": 0.2234762979683973,
      "grad_norm": 19.397075653076172,
      "learning_rate": 9.98736622508056e-06,
      "lm_loss": 6.983,
      "loss": 2.0691,
      "step": 99,
      "text_contrastive_loss": 1.0896,
      "train_positive_log_prob": -105.7801,
      "train_positive_token_accuracy": 0.083,
      "train_positive_token_prob": 0.0229
    },
    {
      "contrastive_loss": 0.822,
      "epoch": 0.22573363431151242,
      "grad_norm": 24.32610321044922,
      "learning_rate": 9.98684552745256e-06,
      "lm_loss": 6.8964,
      "loss": 1.9765,
      "step": 100,
      "text_contrastive_loss": 0.9296,
      "train_positive_log_prob": -104.2558,
      "train_positive_token_accuracy": 0.0736,
      "train_positive_token_prob": 0.0218
    },
    {
      "contrastive_loss": 0.6104,
      "epoch": 0.22799097065462753,
      "grad_norm": 20.24129867553711,
      "learning_rate": 9.986314329329294e-06,
      "lm_loss": 6.8942,
      "loss": 1.8582,
      "step": 101,
      "text_contrastive_loss": 1.1168,
      "train_positive_log_prob": -103.4058,
      "train_positive_token_accuracy": 0.0732,
      "train_positive_token_prob": 0.0212
    },
    {
      "contrastive_loss": 0.5948,
      "epoch": 0.23024830699774265,
      "grad_norm": 21.55072021484375,
      "learning_rate": 9.985772631829272e-06,
      "lm_loss": 6.856,
      "loss": 1.7716,
      "step": 102,
      "text_contrastive_loss": 0.9824,
      "train_positive_log_prob": -101.0808,
      "train_positive_token_accuracy": 0.0763,
      "train_positive_token_prob": 0.0218
    },
    {
      "contrastive_loss": 0.6196,
      "epoch": 0.2325056433408578,
      "grad_norm": 17.736631393432617,
      "learning_rate": 9.985220436093112e-06,
      "lm_loss": 6.7926,
      "loss": 1.8132,
      "step": 103,
      "text_contrastive_loss": 1.0287,
      "train_positive_log_prob": -101.809,
      "train_positive_token_accuracy": 0.0722,
      "train_positive_token_prob": 0.0222
    },
    {
      "contrastive_loss": 0.7161,
      "epoch": 0.23476297968397292,
      "grad_norm": 21.27935218811035,
      "learning_rate": 9.984657743283543e-06,
      "lm_loss": 6.8636,
      "loss": 1.9482,
      "step": 104,
      "text_contrastive_loss": 1.0915,
      "train_positive_log_prob": -100.2482,
      "train_positive_token_accuracy": 0.0687,
      "train_positive_token_prob": 0.0207
    },
    {
      "contrastive_loss": 0.6279,
      "epoch": 0.23702031602708803,
      "grad_norm": 18.59088134765625,
      "learning_rate": 9.984084554585387e-06,
      "lm_loss": 6.7919,
      "loss": 1.8682,
      "step": 105,
      "text_contrastive_loss": 1.1222,
      "train_positive_log_prob": -100.372,
      "train_positive_token_accuracy": 0.0775,
      "train_positive_token_prob": 0.023
    },
    {
      "contrastive_loss": 0.5217,
      "epoch": 0.23927765237020315,
      "grad_norm": 14.964653015136719,
      "learning_rate": 9.983500871205577e-06,
      "lm_loss": 6.8479,
      "loss": 1.6139,
      "step": 106,
      "text_contrastive_loss": 0.8147,
      "train_positive_log_prob": -101.6839,
      "train_positive_token_accuracy": 0.0642,
      "train_positive_token_prob": 0.0196
    },
    {
      "contrastive_loss": 0.7429,
      "epoch": 0.24153498871331827,
      "grad_norm": 21.03154182434082,
      "learning_rate": 9.982906694373136e-06,
      "lm_loss": 6.7614,
      "loss": 1.9369,
      "step": 107,
      "text_contrastive_loss": 1.0357,
      "train_positive_log_prob": -99.8404,
      "train_positive_token_accuracy": 0.0741,
      "train_positive_token_prob": 0.0217
    },
    {
      "contrastive_loss": 0.6753,
      "epoch": 0.24379232505643342,
      "grad_norm": 19.622690200805664,
      "learning_rate": 9.98230202533919e-06,
      "lm_loss": 6.8005,
      "loss": 1.8331,
      "step": 108,
      "text_contrastive_loss": 0.9555,
      "train_positive_log_prob": -98.8159,
      "train_positive_token_accuracy": 0.0737,
      "train_positive_token_prob": 0.0207
    },
    {
      "contrastive_loss": 0.6735,
      "epoch": 0.24604966139954854,
      "grad_norm": 17.51546859741211,
      "learning_rate": 9.98168686537695e-06,
      "lm_loss": 6.8147,
      "loss": 1.9004,
      "step": 109,
      "text_contrastive_loss": 1.0909,
      "train_positive_log_prob": -102.1914,
      "train_positive_token_accuracy": 0.0617,
      "train_positive_token_prob": 0.0203
    },
    {
      "contrastive_loss": 0.5209,
      "epoch": 0.24830699774266365,
      "grad_norm": 20.796781539916992,
      "learning_rate": 9.98106121578172e-06,
      "lm_loss": 6.8527,
      "loss": 1.6981,
      "step": 110,
      "text_contrastive_loss": 0.984,
      "train_positive_log_prob": -99.4987,
      "train_positive_token_accuracy": 0.0678,
      "train_positive_token_prob": 0.0199
    },
    {
      "contrastive_loss": 0.5796,
      "epoch": 0.2505643340857788,
      "grad_norm": 18.110681533813477,
      "learning_rate": 9.980425077870895e-06,
      "lm_loss": 6.7838,
      "loss": 1.7806,
      "step": 111,
      "text_contrastive_loss": 1.0451,
      "train_positive_log_prob": -99.5029,
      "train_positive_token_accuracy": 0.0671,
      "train_positive_token_prob": 0.0197
    },
    {
      "contrastive_loss": 0.6582,
      "epoch": 0.2528216704288939,
      "grad_norm": 18.792470932006836,
      "learning_rate": 9.979778452983949e-06,
      "lm_loss": 6.7502,
      "loss": 1.8141,
      "step": 112,
      "text_contrastive_loss": 0.9618,
      "train_positive_log_prob": -97.1377,
      "train_positive_token_accuracy": 0.0725,
      "train_positive_token_prob": 0.0195
    },
    {
      "contrastive_loss": 0.6102,
      "epoch": 0.255079006772009,
      "grad_norm": 17.86269760131836,
      "learning_rate": 9.979121342482442e-06,
      "lm_loss": 6.7461,
      "loss": 1.7826,
      "step": 113,
      "text_contrastive_loss": 0.9956,
      "train_positive_log_prob": -100.1923,
      "train_positive_token_accuracy": 0.0655,
      "train_positive_token_prob": 0.0206
    },
    {
      "contrastive_loss": 0.7073,
      "epoch": 0.25733634311512416,
      "grad_norm": 20.309282302856445,
      "learning_rate": 9.978453747750012e-06,
      "lm_loss": 6.7841,
      "loss": 1.8668,
      "step": 114,
      "text_contrastive_loss": 0.9621,
      "train_positive_log_prob": -103.271,
      "train_positive_token_accuracy": 0.0677,
      "train_positive_token_prob": 0.0204
    },
    {
      "contrastive_loss": 0.584,
      "epoch": 0.2595936794582393,
      "grad_norm": 15.47262191772461,
      "learning_rate": 9.977775670192373e-06,
      "lm_loss": 6.7531,
      "loss": 1.7331,
      "step": 115,
      "text_contrastive_loss": 0.9477,
      "train_positive_log_prob": -102.5584,
      "train_positive_token_accuracy": 0.0756,
      "train_positive_token_prob": 0.0216
    },
    {
      "contrastive_loss": 0.7128,
      "epoch": 0.2618510158013544,
      "grad_norm": 21.398481369018555,
      "learning_rate": 9.977087111237307e-06,
      "lm_loss": 6.7714,
      "loss": 1.859,
      "step": 116,
      "text_contrastive_loss": 0.9381,
      "train_positive_log_prob": -99.1344,
      "train_positive_token_accuracy": 0.0755,
      "train_positive_token_prob": 0.0211
    },
    {
      "contrastive_loss": 0.7079,
      "epoch": 0.26410835214446954,
      "grad_norm": 20.449430465698242,
      "learning_rate": 9.976388072334674e-06,
      "lm_loss": 6.7875,
      "loss": 1.8355,
      "step": 117,
      "text_contrastive_loss": 0.8977,
      "train_positive_log_prob": -102.1042,
      "train_positive_token_accuracy": 0.0776,
      "train_positive_token_prob": 0.0217
    },
    {
      "contrastive_loss": 0.6718,
      "epoch": 0.26636568848758463,
      "grad_norm": 21.4807071685791,
      "learning_rate": 9.975678554956397e-06,
      "lm_loss": 6.707,
      "loss": 1.857,
      "step": 118,
      "text_contrastive_loss": 1.0288,
      "train_positive_log_prob": -99.3987,
      "train_positive_token_accuracy": 0.0743,
      "train_positive_token_prob": 0.0227
    },
    {
      "contrastive_loss": 0.6264,
      "epoch": 0.2686230248306998,
      "grad_norm": 19.080724716186523,
      "learning_rate": 9.974958560596464e-06,
      "lm_loss": 6.5665,
      "loss": 1.7351,
      "step": 119,
      "text_contrastive_loss": 0.9041,
      "train_positive_log_prob": -96.908,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.0234
    },
    {
      "contrastive_loss": 0.623,
      "epoch": 0.2708803611738149,
      "grad_norm": 17.793132781982422,
      "learning_rate": 9.97422809077092e-06,
      "lm_loss": 6.7075,
      "loss": 1.8309,
      "step": 120,
      "text_contrastive_loss": 1.0742,
      "train_positive_log_prob": -97.9892,
      "train_positive_token_accuracy": 0.0711,
      "train_positive_token_prob": 0.0212
    },
    {
      "contrastive_loss": 0.7049,
      "epoch": 0.27313769751693,
      "grad_norm": 20.13200569152832,
      "learning_rate": 9.973487147017874e-06,
      "lm_loss": 6.6935,
      "loss": 1.9066,
      "step": 121,
      "text_contrastive_loss": 1.0646,
      "train_positive_log_prob": -99.6141,
      "train_positive_token_accuracy": 0.0667,
      "train_positive_token_prob": 0.0219
    },
    {
      "contrastive_loss": 0.6301,
      "epoch": 0.27539503386004516,
      "grad_norm": 20.30483055114746,
      "learning_rate": 9.972735730897484e-06,
      "lm_loss": 6.7637,
      "loss": 1.7692,
      "step": 122,
      "text_contrastive_loss": 0.9255,
      "train_positive_log_prob": -99.4747,
      "train_positive_token_accuracy": 0.0696,
      "train_positive_token_prob": 0.0219
    },
    {
      "contrastive_loss": 0.6715,
      "epoch": 0.27765237020316025,
      "grad_norm": 18.955608367919922,
      "learning_rate": 9.97197384399196e-06,
      "lm_loss": 6.659,
      "loss": 1.859,
      "step": 123,
      "text_contrastive_loss": 1.0431,
      "train_positive_log_prob": -99.4913,
      "train_positive_token_accuracy": 0.0729,
      "train_positive_token_prob": 0.0228
    },
    {
      "contrastive_loss": 0.5571,
      "epoch": 0.2799097065462754,
      "grad_norm": 18.156646728515625,
      "learning_rate": 9.971201487905563e-06,
      "lm_loss": 6.7414,
      "loss": 1.6083,
      "step": 124,
      "text_contrastive_loss": 0.7542,
      "train_positive_log_prob": -100.8597,
      "train_positive_token_accuracy": 0.0697,
      "train_positive_token_prob": 0.0211
    },
    {
      "contrastive_loss": 0.5687,
      "epoch": 0.28216704288939054,
      "grad_norm": 16.17840003967285,
      "learning_rate": 9.970418664264596e-06,
      "lm_loss": 6.7946,
      "loss": 1.7881,
      "step": 125,
      "text_contrastive_loss": 1.0799,
      "train_positive_log_prob": -100.4893,
      "train_positive_token_accuracy": 0.0684,
      "train_positive_token_prob": 0.022
    },
    {
      "contrastive_loss": 0.7128,
      "epoch": 0.28442437923250563,
      "grad_norm": 20.05196189880371,
      "learning_rate": 9.969625374717401e-06,
      "lm_loss": 6.723,
      "loss": 1.9051,
      "step": 126,
      "text_contrastive_loss": 1.04,
      "train_positive_log_prob": -98.9507,
      "train_positive_token_accuracy": 0.0693,
      "train_positive_token_prob": 0.0208
    },
    {
      "contrastive_loss": 0.6544,
      "epoch": 0.2866817155756208,
      "grad_norm": 20.08206558227539,
      "learning_rate": 9.96882162093436e-06,
      "lm_loss": 6.6549,
      "loss": 1.7353,
      "step": 127,
      "text_contrastive_loss": 0.8308,
      "train_positive_log_prob": -99.0368,
      "train_positive_token_accuracy": 0.0735,
      "train_positive_token_prob": 0.0229
    },
    {
      "contrastive_loss": 0.6874,
      "epoch": 0.28893905191873587,
      "grad_norm": 19.576711654663086,
      "learning_rate": 9.968007404607887e-06,
      "lm_loss": 6.6307,
      "loss": 1.8066,
      "step": 128,
      "text_contrastive_loss": 0.9122,
      "train_positive_log_prob": -99.4894,
      "train_positive_token_accuracy": 0.0695,
      "train_positive_token_prob": 0.0223
    },
    {
      "contrastive_loss": 0.5249,
      "epoch": 0.291196388261851,
      "grad_norm": 16.32411766052246,
      "learning_rate": 9.96718272745243e-06,
      "lm_loss": 6.572,
      "loss": 1.5725,
      "step": 129,
      "text_contrastive_loss": 0.7809,
      "train_positive_log_prob": -94.3693,
      "train_positive_token_accuracy": 0.0721,
      "train_positive_token_prob": 0.0223
    },
    {
      "contrastive_loss": 0.603,
      "epoch": 0.29345372460496616,
      "grad_norm": 19.72479248046875,
      "learning_rate": 9.966347591204459e-06,
      "lm_loss": 6.4371,
      "loss": 1.6972,
      "step": 130,
      "text_contrastive_loss": 0.901,
      "train_positive_log_prob": -93.3102,
      "train_positive_token_accuracy": 0.0741,
      "train_positive_token_prob": 0.023
    },
    {
      "contrastive_loss": 0.4895,
      "epoch": 0.29571106094808125,
      "grad_norm": 14.938592910766602,
      "learning_rate": 9.96550199762247e-06,
      "lm_loss": 6.7028,
      "loss": 1.6372,
      "step": 131,
      "text_contrastive_loss": 0.9549,
      "train_positive_log_prob": -99.5458,
      "train_positive_token_accuracy": 0.071,
      "train_positive_token_prob": 0.0219
    },
    {
      "contrastive_loss": 0.6468,
      "epoch": 0.2979683972911964,
      "grad_norm": 18.830659866333008,
      "learning_rate": 9.964645948486978e-06,
      "lm_loss": 6.5754,
      "loss": 1.8032,
      "step": 132,
      "text_contrastive_loss": 0.9978,
      "train_positive_log_prob": -96.2181,
      "train_positive_token_accuracy": 0.0713,
      "train_positive_token_prob": 0.0222
    },
    {
      "contrastive_loss": 0.7476,
      "epoch": 0.3002257336343115,
      "grad_norm": 19.2735538482666,
      "learning_rate": 9.963779445600512e-06,
      "lm_loss": 6.7231,
      "loss": 1.9824,
      "step": 133,
      "text_contrastive_loss": 1.1248,
      "train_positive_log_prob": -102.0271,
      "train_positive_token_accuracy": 0.0764,
      "train_positive_token_prob": 0.0223
    },
    {
      "contrastive_loss": 0.5515,
      "epoch": 0.30248306997742663,
      "grad_norm": 17.205963134765625,
      "learning_rate": 9.962902490787616e-06,
      "lm_loss": 6.5939,
      "loss": 1.7153,
      "step": 134,
      "text_contrastive_loss": 1.0088,
      "train_positive_log_prob": -96.5667,
      "train_positive_token_accuracy": 0.0686,
      "train_positive_token_prob": 0.0224
    },
    {
      "contrastive_loss": 0.4688,
      "epoch": 0.3047404063205418,
      "grad_norm": 16.025466918945312,
      "learning_rate": 9.962015085894838e-06,
      "lm_loss": 6.5824,
      "loss": 1.5918,
      "step": 135,
      "text_contrastive_loss": 0.9294,
      "train_positive_log_prob": -96.3242,
      "train_positive_token_accuracy": 0.0718,
      "train_positive_token_prob": 0.023
    },
    {
      "contrastive_loss": 0.6728,
      "epoch": 0.30699774266365687,
      "grad_norm": 21.674942016601562,
      "learning_rate": 9.961117232790734e-06,
      "lm_loss": 6.5266,
      "loss": 1.8339,
      "step": 136,
      "text_contrastive_loss": 1.0169,
      "train_positive_log_prob": -96.6234,
      "train_positive_token_accuracy": 0.0734,
      "train_positive_token_prob": 0.0229
    },
    {
      "contrastive_loss": 0.7276,
      "epoch": 0.309255079006772,
      "grad_norm": 17.904706954956055,
      "learning_rate": 9.960208933365857e-06,
      "lm_loss": 6.4983,
      "loss": 1.88,
      "step": 137,
      "text_contrastive_loss": 1.0052,
      "train_positive_log_prob": -96.9173,
      "train_positive_token_accuracy": 0.0712,
      "train_positive_token_prob": 0.0231
    },
    {
      "contrastive_loss": 0.5621,
      "epoch": 0.3115124153498871,
      "grad_norm": 17.837366104125977,
      "learning_rate": 9.959290189532757e-06,
      "lm_loss": 6.6279,
      "loss": 1.7579,
      "step": 138,
      "text_contrastive_loss": 1.066,
      "train_positive_log_prob": -98.5638,
      "train_positive_token_accuracy": 0.0617,
      "train_positive_token_prob": 0.021
    },
    {
      "contrastive_loss": 0.7193,
      "epoch": 0.31376975169300225,
      "grad_norm": 18.712196350097656,
      "learning_rate": 9.958361003225979e-06,
      "lm_loss": 6.5824,
      "loss": 1.8925,
      "step": 139,
      "text_contrastive_loss": 1.0299,
      "train_positive_log_prob": -98.5002,
      "train_positive_token_accuracy": 0.0762,
      "train_positive_token_prob": 0.023
    },
    {
      "contrastive_loss": 0.7363,
      "epoch": 0.3160270880361174,
      "grad_norm": 18.103960037231445,
      "learning_rate": 9.957421376402053e-06,
      "lm_loss": 6.5925,
      "loss": 1.8567,
      "step": 140,
      "text_contrastive_loss": 0.9223,
      "train_positive_log_prob": -97.703,
      "train_positive_token_accuracy": 0.0738,
      "train_positive_token_prob": 0.0235
    },
    {
      "contrastive_loss": 0.6386,
      "epoch": 0.3182844243792325,
      "grad_norm": 18.02215576171875,
      "learning_rate": 9.956471311039491e-06,
      "lm_loss": 6.5221,
      "loss": 1.7777,
      "step": 141,
      "text_contrastive_loss": 0.9737,
      "train_positive_log_prob": -96.9393,
      "train_positive_token_accuracy": 0.0799,
      "train_positive_token_prob": 0.0252
    },
    {
      "contrastive_loss": 0.6095,
      "epoch": 0.32054176072234764,
      "grad_norm": 17.972013473510742,
      "learning_rate": 9.95551080913879e-06,
      "lm_loss": 6.5494,
      "loss": 1.7287,
      "step": 142,
      "text_contrastive_loss": 0.9285,
      "train_positive_log_prob": -97.9343,
      "train_positive_token_accuracy": 0.0656,
      "train_positive_token_prob": 0.0225
    },
    {
      "contrastive_loss": 0.6191,
      "epoch": 0.3227990970654628,
      "grad_norm": 16.62250518798828,
      "learning_rate": 9.954539872722417e-06,
      "lm_loss": 6.5831,
      "loss": 1.7514,
      "step": 143,
      "text_contrastive_loss": 0.9479,
      "train_positive_log_prob": -99.0788,
      "train_positive_token_accuracy": 0.076,
      "train_positive_token_prob": 0.023
    },
    {
      "contrastive_loss": 0.6935,
      "epoch": 0.32505643340857787,
      "grad_norm": 18.340702056884766,
      "learning_rate": 9.953558503834819e-06,
      "lm_loss": 6.574,
      "loss": 1.8538,
      "step": 144,
      "text_contrastive_loss": 1.0058,
      "train_positive_log_prob": -99.1228,
      "train_positive_token_accuracy": 0.0806,
      "train_positive_token_prob": 0.0243
    },
    {
      "contrastive_loss": 0.7105,
      "epoch": 0.327313769751693,
      "grad_norm": 20.077089309692383,
      "learning_rate": 9.9525667045424e-06,
      "lm_loss": 6.4291,
      "loss": 1.8989,
      "step": 145,
      "text_contrastive_loss": 1.0909,
      "train_positive_log_prob": -96.109,
      "train_positive_token_accuracy": 0.076,
      "train_positive_token_prob": 0.0242
    },
    {
      "contrastive_loss": 0.5658,
      "epoch": 0.3295711060948081,
      "grad_norm": 17.12446403503418,
      "learning_rate": 9.951564476933534e-06,
      "lm_loss": 6.6439,
      "loss": 1.6466,
      "step": 146,
      "text_contrastive_loss": 0.8327,
      "train_positive_log_prob": -100.1287,
      "train_positive_token_accuracy": 0.0734,
      "train_positive_token_prob": 0.0227
    },
    {
      "contrastive_loss": 0.5777,
      "epoch": 0.33182844243792325,
      "grad_norm": 18.200572967529297,
      "learning_rate": 9.950551823118544e-06,
      "lm_loss": 6.4616,
      "loss": 1.7331,
      "step": 147,
      "text_contrastive_loss": 1.0184,
      "train_positive_log_prob": -96.6485,
      "train_positive_token_accuracy": 0.0721,
      "train_positive_token_prob": 0.0227
    },
    {
      "contrastive_loss": 0.7021,
      "epoch": 0.3340857787810384,
      "grad_norm": 16.875293731689453,
      "learning_rate": 9.949528745229721e-06,
      "lm_loss": 6.5708,
      "loss": 1.8853,
      "step": 148,
      "text_contrastive_loss": 1.0521,
      "train_positive_log_prob": -99.7688,
      "train_positive_token_accuracy": 0.0707,
      "train_positive_token_prob": 0.0229
    },
    {
      "contrastive_loss": 0.6348,
      "epoch": 0.3363431151241535,
      "grad_norm": 19.054262161254883,
      "learning_rate": 9.948495245421294e-06,
      "lm_loss": 6.4714,
      "loss": 1.7528,
      "step": 149,
      "text_contrastive_loss": 0.9417,
      "train_positive_log_prob": -95.0963,
      "train_positive_token_accuracy": 0.0792,
      "train_positive_token_prob": 0.024
    },
    {
      "contrastive_loss": 0.6802,
      "epoch": 0.33860045146726864,
      "grad_norm": 17.193363189697266,
      "learning_rate": 9.94745132586944e-06,
      "lm_loss": 6.5273,
      "loss": 1.7791,
      "step": 150,
      "text_contrastive_loss": 0.8923,
      "train_positive_log_prob": -99.2347,
      "train_positive_token_accuracy": 0.0681,
      "train_positive_token_prob": 0.0226
    },
    {
      "contrastive_loss": 0.6213,
      "epoch": 0.34085778781038373,
      "grad_norm": 18.566396713256836,
      "learning_rate": 9.946396988772275e-06,
      "lm_loss": 6.6128,
      "loss": 1.7275,
      "step": 151,
      "text_contrastive_loss": 0.89,
      "train_positive_log_prob": -96.8675,
      "train_positive_token_accuracy": 0.0624,
      "train_positive_token_prob": 0.0219
    },
    {
      "contrastive_loss": 0.5195,
      "epoch": 0.3431151241534989,
      "grad_norm": 19.143999099731445,
      "learning_rate": 9.945332236349857e-06,
      "lm_loss": 6.5027,
      "loss": 1.5934,
      "step": 152,
      "text_contrastive_loss": 0.8473,
      "train_positive_log_prob": -98.3293,
      "train_positive_token_accuracy": 0.0765,
      "train_positive_token_prob": 0.0241
    },
    {
      "contrastive_loss": 0.7038,
      "epoch": 0.345372460496614,
      "grad_norm": 19.382593154907227,
      "learning_rate": 9.944257070844165e-06,
      "lm_loss": 6.5004,
      "loss": 1.833,
      "step": 153,
      "text_contrastive_loss": 0.9582,
      "train_positive_log_prob": -95.6292,
      "train_positive_token_accuracy": 0.0733,
      "train_positive_token_prob": 0.0228
    },
    {
      "contrastive_loss": 0.5754,
      "epoch": 0.3476297968397291,
      "grad_norm": 17.497941970825195,
      "learning_rate": 9.943171494519111e-06,
      "lm_loss": 6.4694,
      "loss": 1.7106,
      "step": 154,
      "text_contrastive_loss": 0.9764,
      "train_positive_log_prob": -94.1347,
      "train_positive_token_accuracy": 0.0685,
      "train_positive_token_prob": 0.0227
    },
    {
      "contrastive_loss": 0.578,
      "epoch": 0.34988713318284426,
      "grad_norm": 19.888530731201172,
      "learning_rate": 9.942075509660527e-06,
      "lm_loss": 6.3295,
      "loss": 1.6232,
      "step": 155,
      "text_contrastive_loss": 0.8245,
      "train_positive_log_prob": -92.8392,
      "train_positive_token_accuracy": 0.0754,
      "train_positive_token_prob": 0.0241
    },
    {
      "contrastive_loss": 0.705,
      "epoch": 0.35214446952595935,
      "grad_norm": 18.749338150024414,
      "learning_rate": 9.94096911857616e-06,
      "lm_loss": 6.3607,
      "loss": 1.8029,
      "step": 156,
      "text_contrastive_loss": 0.9237,
      "train_positive_log_prob": -96.1845,
      "train_positive_token_accuracy": 0.077,
      "train_positive_token_prob": 0.0238
    },
    {
      "contrastive_loss": 0.582,
      "epoch": 0.3544018058690745,
      "grad_norm": 19.937015533447266,
      "learning_rate": 9.939852323595671e-06,
      "lm_loss": 6.3937,
      "loss": 1.6613,
      "step": 157,
      "text_contrastive_loss": 0.8797,
      "train_positive_log_prob": -94.1576,
      "train_positive_token_accuracy": 0.0814,
      "train_positive_token_prob": 0.0243
    },
    {
      "contrastive_loss": 0.5762,
      "epoch": 0.35665914221218964,
      "grad_norm": 18.210784912109375,
      "learning_rate": 9.938725127070628e-06,
      "lm_loss": 6.3963,
      "loss": 1.7224,
      "step": 158,
      "text_contrastive_loss": 1.0131,
      "train_positive_log_prob": -97.7453,
      "train_positive_token_accuracy": 0.0749,
      "train_positive_token_prob": 0.0243
    },
    {
      "contrastive_loss": 0.5866,
      "epoch": 0.35891647855530473,
      "grad_norm": 17.754047393798828,
      "learning_rate": 9.937587531374497e-06,
      "lm_loss": 6.4707,
      "loss": 1.7096,
      "step": 159,
      "text_contrastive_loss": 0.9519,
      "train_positive_log_prob": -95.9966,
      "train_positive_token_accuracy": 0.0729,
      "train_positive_token_prob": 0.0242
    },
    {
      "contrastive_loss": 0.6803,
      "epoch": 0.3611738148984199,
      "grad_norm": 20.1284236907959,
      "learning_rate": 9.936439538902644e-06,
      "lm_loss": 6.3979,
      "loss": 1.7264,
      "step": 160,
      "text_contrastive_loss": 0.8128,
      "train_positive_log_prob": -94.7203,
      "train_positive_token_accuracy": 0.0765,
      "train_positive_token_prob": 0.0239
    },
    {
      "contrastive_loss": 0.6147,
      "epoch": 0.36343115124153497,
      "grad_norm": 18.447181701660156,
      "learning_rate": 9.935281152072329e-06,
      "lm_loss": 6.3058,
      "loss": 1.7724,
      "step": 161,
      "text_contrastive_loss": 1.0541,
      "train_positive_log_prob": -92.9408,
      "train_positive_token_accuracy": 0.0753,
      "train_positive_token_prob": 0.0248
    },
    {
      "contrastive_loss": 0.6066,
      "epoch": 0.3656884875846501,
      "grad_norm": 16.914417266845703,
      "learning_rate": 9.934112373322695e-06,
      "lm_loss": 6.3964,
      "loss": 1.7119,
      "step": 162,
      "text_contrastive_loss": 0.9314,
      "train_positive_log_prob": -94.6963,
      "train_positive_token_accuracy": 0.0734,
      "train_positive_token_prob": 0.024
    },
    {
      "contrastive_loss": 0.5901,
      "epoch": 0.36794582392776526,
      "grad_norm": 17.683534622192383,
      "learning_rate": 9.932933205114766e-06,
      "lm_loss": 6.3766,
      "loss": 1.6508,
      "step": 163,
      "text_contrastive_loss": 0.846,
      "train_positive_log_prob": -95.7006,
      "train_positive_token_accuracy": 0.0825,
      "train_positive_token_prob": 0.0246
    },
    {
      "contrastive_loss": 0.5814,
      "epoch": 0.37020316027088035,
      "grad_norm": 17.29725456237793,
      "learning_rate": 9.931743649931446e-06,
      "lm_loss": 6.5045,
      "loss": 1.653,
      "step": 164,
      "text_contrastive_loss": 0.8424,
      "train_positive_log_prob": -96.6768,
      "train_positive_token_accuracy": 0.0715,
      "train_positive_token_prob": 0.0226
    },
    {
      "contrastive_loss": 0.5415,
      "epoch": 0.3724604966139955,
      "grad_norm": 17.485736846923828,
      "learning_rate": 9.93054371027751e-06,
      "lm_loss": 6.3399,
      "loss": 1.6119,
      "step": 165,
      "text_contrastive_loss": 0.8727,
      "train_positive_log_prob": -90.8089,
      "train_positive_token_accuracy": 0.0804,
      "train_positive_token_prob": 0.025
    },
    {
      "contrastive_loss": 0.5808,
      "epoch": 0.3747178329571106,
      "grad_norm": 15.62707233428955,
      "learning_rate": 9.929333388679593e-06,
      "lm_loss": 6.3657,
      "loss": 1.6355,
      "step": 166,
      "text_contrastive_loss": 0.8362,
      "train_positive_log_prob": -92.1776,
      "train_positive_token_accuracy": 0.082,
      "train_positive_token_prob": 0.0239
    },
    {
      "contrastive_loss": 0.6004,
      "epoch": 0.37697516930022573,
      "grad_norm": 18.117691040039062,
      "learning_rate": 9.928112687686197e-06,
      "lm_loss": 6.3642,
      "loss": 1.6828,
      "step": 167,
      "text_contrastive_loss": 0.8919,
      "train_positive_log_prob": -91.6888,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.0243
    },
    {
      "contrastive_loss": 0.5473,
      "epoch": 0.3792325056433409,
      "grad_norm": 15.383481979370117,
      "learning_rate": 9.92688160986768e-06,
      "lm_loss": 6.3471,
      "loss": 1.6303,
      "step": 168,
      "text_contrastive_loss": 0.8967,
      "train_positive_log_prob": -94.2657,
      "train_positive_token_accuracy": 0.0689,
      "train_positive_token_prob": 0.0237
    },
    {
      "contrastive_loss": 0.6605,
      "epoch": 0.38148984198645597,
      "grad_norm": 20.276687622070312,
      "learning_rate": 9.925640157816246e-06,
      "lm_loss": 6.4145,
      "loss": 1.7597,
      "step": 169,
      "text_contrastive_loss": 0.9155,
      "train_positive_log_prob": -93.8055,
      "train_positive_token_accuracy": 0.0743,
      "train_positive_token_prob": 0.023
    },
    {
      "contrastive_loss": 0.6517,
      "epoch": 0.3837471783295711,
      "grad_norm": 18.185070037841797,
      "learning_rate": 9.924388334145943e-06,
      "lm_loss": 6.4825,
      "loss": 1.7708,
      "step": 170,
      "text_contrastive_loss": 0.9418,
      "train_positive_log_prob": -96.2035,
      "train_positive_token_accuracy": 0.0719,
      "train_positive_token_prob": 0.0226
    },
    {
      "contrastive_loss": 0.5431,
      "epoch": 0.3860045146726862,
      "grad_norm": 19.181861877441406,
      "learning_rate": 9.92312614149266e-06,
      "lm_loss": 6.3756,
      "loss": 1.6018,
      "step": 171,
      "text_contrastive_loss": 0.8423,
      "train_positive_log_prob": -93.6186,
      "train_positive_token_accuracy": 0.0774,
      "train_positive_token_prob": 0.0238
    },
    {
      "contrastive_loss": 0.5546,
      "epoch": 0.38826185101580135,
      "grad_norm": 16.07084083557129,
      "learning_rate": 9.92185358251412e-06,
      "lm_loss": 6.2966,
      "loss": 1.6335,
      "step": 172,
      "text_contrastive_loss": 0.8986,
      "train_positive_log_prob": -91.7924,
      "train_positive_token_accuracy": 0.0756,
      "train_positive_token_prob": 0.0237
    },
    {
      "contrastive_loss": 0.6237,
      "epoch": 0.3905191873589165,
      "grad_norm": 18.402652740478516,
      "learning_rate": 9.92057065988987e-06,
      "lm_loss": 6.3649,
      "loss": 1.6979,
      "step": 173,
      "text_contrastive_loss": 0.8754,
      "train_positive_log_prob": -94.9973,
      "train_positive_token_accuracy": 0.0699,
      "train_positive_token_prob": 0.0234
    },
    {
      "contrastive_loss": 0.5243,
      "epoch": 0.3927765237020316,
      "grad_norm": 16.405948638916016,
      "learning_rate": 9.919277376321284e-06,
      "lm_loss": 6.2933,
      "loss": 1.643,
      "step": 174,
      "text_contrastive_loss": 0.9787,
      "train_positive_log_prob": -93.1227,
      "train_positive_token_accuracy": 0.0644,
      "train_positive_token_prob": 0.023
    },
    {
      "contrastive_loss": 0.645,
      "epoch": 0.39503386004514673,
      "grad_norm": 17.82147789001465,
      "learning_rate": 9.917973734531549e-06,
      "lm_loss": 6.3834,
      "loss": 1.7463,
      "step": 175,
      "text_contrastive_loss": 0.9259,
      "train_positive_log_prob": -93.3786,
      "train_positive_token_accuracy": 0.0676,
      "train_positive_token_prob": 0.0233
    },
    {
      "contrastive_loss": 0.5635,
      "epoch": 0.3972911963882618,
      "grad_norm": 16.221330642700195,
      "learning_rate": 9.916659737265664e-06,
      "lm_loss": 6.3933,
      "loss": 1.6698,
      "step": 176,
      "text_contrastive_loss": 0.9339,
      "train_positive_log_prob": -96.3004,
      "train_positive_token_accuracy": 0.0752,
      "train_positive_token_prob": 0.0243
    },
    {
      "contrastive_loss": 0.5209,
      "epoch": 0.39954853273137697,
      "grad_norm": 16.04338836669922,
      "learning_rate": 9.915335387290432e-06,
      "lm_loss": 6.3647,
      "loss": 1.5876,
      "step": 177,
      "text_contrastive_loss": 0.8605,
      "train_positive_log_prob": -95.7909,
      "train_positive_token_accuracy": 0.0737,
      "train_positive_token_prob": 0.0242
    },
    {
      "contrastive_loss": 0.6082,
      "epoch": 0.4018058690744921,
      "grad_norm": 19.90717315673828,
      "learning_rate": 9.914000687394457e-06,
      "lm_loss": 6.5428,
      "loss": 1.7621,
      "step": 178,
      "text_contrastive_loss": 0.9992,
      "train_positive_log_prob": -99.5424,
      "train_positive_token_accuracy": 0.0724,
      "train_positive_token_prob": 0.0235
    },
    {
      "contrastive_loss": 0.6049,
      "epoch": 0.4040632054176072,
      "grad_norm": 15.602974891662598,
      "learning_rate": 9.912655640388134e-06,
      "lm_loss": 6.2935,
      "loss": 1.644,
      "step": 179,
      "text_contrastive_loss": 0.8196,
      "train_positive_log_prob": -94.5021,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.0254
    },
    {
      "contrastive_loss": 0.6542,
      "epoch": 0.40632054176072235,
      "grad_norm": 20.750207901000977,
      "learning_rate": 9.911300249103646e-06,
      "lm_loss": 6.3268,
      "loss": 1.7814,
      "step": 180,
      "text_contrastive_loss": 0.989,
      "train_positive_log_prob": -94.0351,
      "train_positive_token_accuracy": 0.077,
      "train_positive_token_prob": 0.0248
    },
    {
      "contrastive_loss": 0.6459,
      "epoch": 0.40857787810383744,
      "grad_norm": 17.191402435302734,
      "learning_rate": 9.909934516394957e-06,
      "lm_loss": 6.2459,
      "loss": 1.7891,
      "step": 181,
      "text_contrastive_loss": 1.0371,
      "train_positive_log_prob": -92.1278,
      "train_positive_token_accuracy": 0.0747,
      "train_positive_token_prob": 0.0253
    },
    {
      "contrastive_loss": 0.6721,
      "epoch": 0.4108352144469526,
      "grad_norm": 16.632160186767578,
      "learning_rate": 9.908558445137807e-06,
      "lm_loss": 6.3747,
      "loss": 1.7507,
      "step": 182,
      "text_contrastive_loss": 0.8822,
      "train_positive_log_prob": -93.629,
      "train_positive_token_accuracy": 0.0766,
      "train_positive_token_prob": 0.0253
    },
    {
      "contrastive_loss": 0.5982,
      "epoch": 0.41309255079006774,
      "grad_norm": 17.62737274169922,
      "learning_rate": 9.907172038229706e-06,
      "lm_loss": 6.3935,
      "loss": 1.6699,
      "step": 183,
      "text_contrastive_loss": 0.8648,
      "train_positive_log_prob": -93.3324,
      "train_positive_token_accuracy": 0.0709,
      "train_positive_token_prob": 0.0245
    },
    {
      "contrastive_loss": 0.6416,
      "epoch": 0.4153498871331828,
      "grad_norm": 17.71466636657715,
      "learning_rate": 9.905775298589923e-06,
      "lm_loss": 6.353,
      "loss": 1.7138,
      "step": 184,
      "text_contrastive_loss": 0.8738,
      "train_positive_log_prob": -92.9928,
      "train_positive_token_accuracy": 0.0707,
      "train_positive_token_prob": 0.0244
    },
    {
      "contrastive_loss": 0.6686,
      "epoch": 0.417607223476298,
      "grad_norm": 16.830604553222656,
      "learning_rate": 9.904368229159494e-06,
      "lm_loss": 6.2932,
      "loss": 1.7439,
      "step": 185,
      "text_contrastive_loss": 0.8918,
      "train_positive_log_prob": -93.4648,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.0259
    },
    {
      "contrastive_loss": 0.5155,
      "epoch": 0.4198645598194131,
      "grad_norm": 17.492033004760742,
      "learning_rate": 9.90295083290119e-06,
      "lm_loss": 6.3031,
      "loss": 1.6393,
      "step": 186,
      "text_contrastive_loss": 0.9871,
      "train_positive_log_prob": -93.1214,
      "train_positive_token_accuracy": 0.0748,
      "train_positive_token_prob": 0.0242
    },
    {
      "contrastive_loss": 0.6918,
      "epoch": 0.4221218961625282,
      "grad_norm": 17.344141006469727,
      "learning_rate": 9.901523112799543e-06,
      "lm_loss": 6.2857,
      "loss": 1.7474,
      "step": 187,
      "text_contrastive_loss": 0.8539,
      "train_positive_log_prob": -93.4994,
      "train_positive_token_accuracy": 0.0714,
      "train_positive_token_prob": 0.0236
    },
    {
      "contrastive_loss": 0.5633,
      "epoch": 0.42437923250564336,
      "grad_norm": 17.498371124267578,
      "learning_rate": 9.90008507186081e-06,
      "lm_loss": 6.2037,
      "loss": 1.6121,
      "step": 188,
      "text_contrastive_loss": 0.8569,
      "train_positive_log_prob": -91.7199,
      "train_positive_token_accuracy": 0.07,
      "train_positive_token_prob": 0.0244
    },
    {
      "contrastive_loss": 0.5294,
      "epoch": 0.42663656884875845,
      "grad_norm": 16.007226943969727,
      "learning_rate": 9.898636713112992e-06,
      "lm_loss": 6.3962,
      "loss": 1.6096,
      "step": 189,
      "text_contrastive_loss": 0.8811,
      "train_positive_log_prob": -93.7179,
      "train_positive_token_accuracy": 0.0757,
      "train_positive_token_prob": 0.0241
    },
    {
      "contrastive_loss": 0.6992,
      "epoch": 0.4288939051918736,
      "grad_norm": 19.596721649169922,
      "learning_rate": 9.897178039605803e-06,
      "lm_loss": 6.2761,
      "loss": 1.8828,
      "step": 190,
      "text_contrastive_loss": 1.1119,
      "train_positive_log_prob": -90.9802,
      "train_positive_token_accuracy": 0.0757,
      "train_positive_token_prob": 0.0249
    },
    {
      "contrastive_loss": 0.5659,
      "epoch": 0.43115124153498874,
      "grad_norm": 17.168418884277344,
      "learning_rate": 9.895709054410686e-06,
      "lm_loss": 6.3628,
      "loss": 1.6115,
      "step": 191,
      "text_contrastive_loss": 0.8187,
      "train_positive_log_prob": -97.2303,
      "train_positive_token_accuracy": 0.08,
      "train_positive_token_prob": 0.0249
    },
    {
      "contrastive_loss": 0.6841,
      "epoch": 0.43340857787810383,
      "grad_norm": 20.078487396240234,
      "learning_rate": 9.894229760620793e-06,
      "lm_loss": 6.2957,
      "loss": 1.762,
      "step": 192,
      "text_contrastive_loss": 0.8967,
      "train_positive_log_prob": -93.2803,
      "train_positive_token_accuracy": 0.0712,
      "train_positive_token_prob": 0.0238
    },
    {
      "contrastive_loss": 0.5674,
      "epoch": 0.435665914221219,
      "grad_norm": 16.747314453125,
      "learning_rate": 9.892740161350981e-06,
      "lm_loss": 6.3721,
      "loss": 1.6006,
      "step": 193,
      "text_contrastive_loss": 0.792,
      "train_positive_log_prob": -94.6121,
      "train_positive_token_accuracy": 0.0776,
      "train_positive_token_prob": 0.024
    },
    {
      "contrastive_loss": 0.5034,
      "epoch": 0.43792325056433407,
      "grad_norm": 16.926101684570312,
      "learning_rate": 9.891240259737809e-06,
      "lm_loss": 6.4089,
      "loss": 1.5184,
      "step": 194,
      "text_contrastive_loss": 0.7484,
      "train_positive_log_prob": -95.6889,
      "train_positive_token_accuracy": 0.0759,
      "train_positive_token_prob": 0.0232
    },
    {
      "contrastive_loss": 0.6301,
      "epoch": 0.4401805869074492,
      "grad_norm": 16.07118797302246,
      "learning_rate": 9.889730058939529e-06,
      "lm_loss": 6.3629,
      "loss": 1.7402,
      "step": 195,
      "text_contrastive_loss": 0.9476,
      "train_positive_log_prob": -95.1156,
      "train_positive_token_accuracy": 0.0679,
      "train_positive_token_prob": 0.0234
    },
    {
      "contrastive_loss": 0.6138,
      "epoch": 0.44243792325056436,
      "grad_norm": 18.82052230834961,
      "learning_rate": 9.888209562136074e-06,
      "lm_loss": 6.2125,
      "loss": 1.6899,
      "step": 196,
      "text_contrastive_loss": 0.9098,
      "train_positive_log_prob": -92.8738,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.025
    },
    {
      "contrastive_loss": 0.5423,
      "epoch": 0.44469525959367945,
      "grad_norm": 15.113842010498047,
      "learning_rate": 9.886678772529069e-06,
      "lm_loss": 6.1721,
      "loss": 1.6142,
      "step": 197,
      "text_contrastive_loss": 0.9094,
      "train_positive_log_prob": -92.171,
      "train_positive_token_accuracy": 0.0743,
      "train_positive_token_prob": 0.0243
    },
    {
      "contrastive_loss": 0.5986,
      "epoch": 0.4469525959367946,
      "grad_norm": 16.718223571777344,
      "learning_rate": 9.885137693341795e-06,
      "lm_loss": 6.3469,
      "loss": 1.7091,
      "step": 198,
      "text_contrastive_loss": 0.9516,
      "train_positive_log_prob": -94.9289,
      "train_positive_token_accuracy": 0.07,
      "train_positive_token_prob": 0.0241
    },
    {
      "contrastive_loss": 0.607,
      "epoch": 0.4492099322799097,
      "grad_norm": 18.055437088012695,
      "learning_rate": 9.883586327819214e-06,
      "lm_loss": 6.2876,
      "loss": 1.7736,
      "step": 199,
      "text_contrastive_loss": 1.0758,
      "train_positive_log_prob": -93.7248,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.0247
    },
    {
      "contrastive_loss": 0.6925,
      "epoch": 0.45146726862302483,
      "grad_norm": 17.790586471557617,
      "learning_rate": 9.88202467922794e-06,
      "lm_loss": 6.3828,
      "loss": 1.7746,
      "step": 200,
      "text_contrastive_loss": 0.8877,
      "train_positive_log_prob": -93.5735,
      "train_positive_token_accuracy": 0.0807,
      "train_positive_token_prob": 0.0247
    },
    {
      "contrastive_loss": 0.5999,
      "epoch": 0.45372460496614,
      "grad_norm": 17.330663681030273,
      "learning_rate": 9.880452750856239e-06,
      "lm_loss": 6.3503,
      "loss": 1.7844,
      "step": 201,
      "text_contrastive_loss": 1.099,
      "train_positive_log_prob": -95.7016,
      "train_positive_token_accuracy": 0.0726,
      "train_positive_token_prob": 0.0244
    },
    {
      "contrastive_loss": 0.5973,
      "epoch": 0.45598194130925507,
      "grad_norm": 18.505414962768555,
      "learning_rate": 9.878870546014025e-06,
      "lm_loss": 6.3365,
      "loss": 1.671,
      "step": 202,
      "text_contrastive_loss": 0.88,
      "train_positive_log_prob": -93.5629,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.025
    },
    {
      "contrastive_loss": 0.6106,
      "epoch": 0.4582392776523702,
      "grad_norm": 18.446922302246094,
      "learning_rate": 9.877278068032852e-06,
      "lm_loss": 6.3287,
      "loss": 1.6306,
      "step": 203,
      "text_contrastive_loss": 0.7743,
      "train_positive_log_prob": -94.0187,
      "train_positive_token_accuracy": 0.0755,
      "train_positive_token_prob": 0.0255
    },
    {
      "contrastive_loss": 0.5605,
      "epoch": 0.4604966139954853,
      "grad_norm": 17.675628662109375,
      "learning_rate": 9.875675320265903e-06,
      "lm_loss": 6.3369,
      "loss": 1.6047,
      "step": 204,
      "text_contrastive_loss": 0.821,
      "train_positive_log_prob": -95.2803,
      "train_positive_token_accuracy": 0.0686,
      "train_positive_token_prob": 0.0246
    },
    {
      "contrastive_loss": 0.5259,
      "epoch": 0.46275395033860045,
      "grad_norm": 16.840137481689453,
      "learning_rate": 9.874062306087983e-06,
      "lm_loss": 6.1988,
      "loss": 1.5437,
      "step": 205,
      "text_contrastive_loss": 0.7958,
      "train_positive_log_prob": -91.2163,
      "train_positive_token_accuracy": 0.0815,
      "train_positive_token_prob": 0.0262
    },
    {
      "contrastive_loss": 0.6749,
      "epoch": 0.4650112866817156,
      "grad_norm": 19.432479858398438,
      "learning_rate": 9.872439028895518e-06,
      "lm_loss": 6.1775,
      "loss": 1.7606,
      "step": 206,
      "text_contrastive_loss": 0.9359,
      "train_positive_log_prob": -90.0577,
      "train_positive_token_accuracy": 0.0721,
      "train_positive_token_prob": 0.0254
    },
    {
      "contrastive_loss": 0.5187,
      "epoch": 0.4672686230248307,
      "grad_norm": 17.64603614807129,
      "learning_rate": 9.870805492106546e-06,
      "lm_loss": 6.2887,
      "loss": 1.5682,
      "step": 207,
      "text_contrastive_loss": 0.8411,
      "train_positive_log_prob": -92.9095,
      "train_positive_token_accuracy": 0.0751,
      "train_positive_token_prob": 0.0248
    },
    {
      "contrastive_loss": 0.6564,
      "epoch": 0.46952595936794583,
      "grad_norm": 17.92214012145996,
      "learning_rate": 9.869161699160704e-06,
      "lm_loss": 6.1896,
      "loss": 1.7195,
      "step": 208,
      "text_contrastive_loss": 0.8883,
      "train_positive_log_prob": -90.647,
      "train_positive_token_accuracy": 0.0816,
      "train_positive_token_prob": 0.0267
    },
    {
      "contrastive_loss": 0.5406,
      "epoch": 0.4717832957110609,
      "grad_norm": 15.97896671295166,
      "learning_rate": 9.867507653519225e-06,
      "lm_loss": 6.222,
      "loss": 1.5611,
      "step": 209,
      "text_contrastive_loss": 0.7966,
      "train_positive_log_prob": -91.7062,
      "train_positive_token_accuracy": 0.0822,
      "train_positive_token_prob": 0.0267
    },
    {
      "contrastive_loss": 0.5576,
      "epoch": 0.47404063205417607,
      "grad_norm": 17.953554153442383,
      "learning_rate": 9.865843358664933e-06,
      "lm_loss": 6.311,
      "loss": 1.5939,
      "step": 210,
      "text_contrastive_loss": 0.8105,
      "train_positive_log_prob": -93.7491,
      "train_positive_token_accuracy": 0.0739,
      "train_positive_token_prob": 0.025
    },
    {
      "contrastive_loss": 0.7181,
      "epoch": 0.4762979683972912,
      "grad_norm": 19.0045223236084,
      "learning_rate": 9.86416881810223e-06,
      "lm_loss": 6.3363,
      "loss": 1.8132,
      "step": 211,
      "text_contrastive_loss": 0.923,
      "train_positive_log_prob": -93.8539,
      "train_positive_token_accuracy": 0.0707,
      "train_positive_token_prob": 0.0261
    },
    {
      "contrastive_loss": 0.4872,
      "epoch": 0.4785553047404063,
      "grad_norm": 16.929540634155273,
      "learning_rate": 9.862484035357095e-06,
      "lm_loss": 6.2512,
      "loss": 1.5339,
      "step": 212,
      "text_contrastive_loss": 0.8432,
      "train_positive_log_prob": -92.4767,
      "train_positive_token_accuracy": 0.0803,
      "train_positive_token_prob": 0.0261
    },
    {
      "contrastive_loss": 0.6875,
      "epoch": 0.48081264108352145,
      "grad_norm": 19.721738815307617,
      "learning_rate": 9.860789013977074e-06,
      "lm_loss": 6.1811,
      "loss": 1.8533,
      "step": 213,
      "text_contrastive_loss": 1.0952,
      "train_positive_log_prob": -93.8491,
      "train_positive_token_accuracy": 0.0841,
      "train_positive_token_prob": 0.0278
    },
    {
      "contrastive_loss": 0.5638,
      "epoch": 0.48306997742663654,
      "grad_norm": 16.685081481933594,
      "learning_rate": 9.859083757531265e-06,
      "lm_loss": 6.2524,
      "loss": 1.6364,
      "step": 214,
      "text_contrastive_loss": 0.8948,
      "train_positive_log_prob": -91.1474,
      "train_positive_token_accuracy": 0.0721,
      "train_positive_token_prob": 0.0259
    },
    {
      "contrastive_loss": 0.6151,
      "epoch": 0.4853273137697517,
      "grad_norm": 18.529338836669922,
      "learning_rate": 9.857368269610325e-06,
      "lm_loss": 6.1617,
      "loss": 1.6349,
      "step": 215,
      "text_contrastive_loss": 0.8073,
      "train_positive_log_prob": -92.3024,
      "train_positive_token_accuracy": 0.0787,
      "train_positive_token_prob": 0.0279
    },
    {
      "contrastive_loss": 0.6331,
      "epoch": 0.48758465011286684,
      "grad_norm": 16.865758895874023,
      "learning_rate": 9.85564255382645e-06,
      "lm_loss": 6.2704,
      "loss": 1.6857,
      "step": 216,
      "text_contrastive_loss": 0.8511,
      "train_positive_log_prob": -93.0253,
      "train_positive_token_accuracy": 0.074,
      "train_positive_token_prob": 0.0257
    },
    {
      "contrastive_loss": 0.5651,
      "epoch": 0.4898419864559819,
      "grad_norm": 16.25713539123535,
      "learning_rate": 9.853906613813378e-06,
      "lm_loss": 6.2145,
      "loss": 1.6537,
      "step": 217,
      "text_contrastive_loss": 0.9342,
      "train_positive_log_prob": -89.8655,
      "train_positive_token_accuracy": 0.0702,
      "train_positive_token_prob": 0.0264
    },
    {
      "contrastive_loss": 0.653,
      "epoch": 0.49209932279909707,
      "grad_norm": 18.443195343017578,
      "learning_rate": 9.852160453226367e-06,
      "lm_loss": 6.1917,
      "loss": 1.7427,
      "step": 218,
      "text_contrastive_loss": 0.9409,
      "train_positive_log_prob": -92.3912,
      "train_positive_token_accuracy": 0.0728,
      "train_positive_token_prob": 0.0262
    },
    {
      "contrastive_loss": 0.6124,
      "epoch": 0.49435665914221216,
      "grad_norm": 19.45539093017578,
      "learning_rate": 9.850404075742204e-06,
      "lm_loss": 6.3033,
      "loss": 1.7178,
      "step": 219,
      "text_contrastive_loss": 0.9501,
      "train_positive_log_prob": -90.8647,
      "train_positive_token_accuracy": 0.0715,
      "train_positive_token_prob": 0.0251
    },
    {
      "contrastive_loss": 0.6076,
      "epoch": 0.4966139954853273,
      "grad_norm": 17.693368911743164,
      "learning_rate": 9.848637485059183e-06,
      "lm_loss": 6.0934,
      "loss": 1.6177,
      "step": 220,
      "text_contrastive_loss": 0.8015,
      "train_positive_log_prob": -88.1391,
      "train_positive_token_accuracy": 0.0818,
      "train_positive_token_prob": 0.0275
    },
    {
      "contrastive_loss": 0.5889,
      "epoch": 0.49887133182844245,
      "grad_norm": 17.95065689086914,
      "learning_rate": 9.846860684897107e-06,
      "lm_loss": 6.2495,
      "loss": 1.6106,
      "step": 221,
      "text_contrastive_loss": 0.7935,
      "train_positive_log_prob": -93.0622,
      "train_positive_token_accuracy": 0.0745,
      "train_positive_token_prob": 0.0261
    },
    {
      "contrastive_loss": 0.5507,
      "epoch": 0.5011286681715575,
      "grad_norm": 17.545827865600586,
      "learning_rate": 9.845073678997275e-06,
      "lm_loss": 6.2366,
      "loss": 1.5471,
      "step": 222,
      "text_contrastive_loss": 0.7453,
      "train_positive_log_prob": -95.0166,
      "train_positive_token_accuracy": 0.0754,
      "train_positive_token_prob": 0.0261
    },
    {
      "contrastive_loss": 0.4272,
      "epoch": 0.5033860045146726,
      "grad_norm": 14.557065963745117,
      "learning_rate": 9.843276471122473e-06,
      "lm_loss": 6.3137,
      "loss": 1.5591,
      "step": 223,
      "text_contrastive_loss": 1.0011,
      "train_positive_log_prob": -94.4131,
      "train_positive_token_accuracy": 0.0713,
      "train_positive_token_prob": 0.0244
    },
    {
      "contrastive_loss": 0.5857,
      "epoch": 0.5056433408577878,
      "grad_norm": 16.679834365844727,
      "learning_rate": 9.84146906505698e-06,
      "lm_loss": 6.019,
      "loss": 1.5642,
      "step": 224,
      "text_contrastive_loss": 0.7532,
      "train_positive_log_prob": -88.7104,
      "train_positive_token_accuracy": 0.0774,
      "train_positive_token_prob": 0.0264
    },
    {
      "contrastive_loss": 0.602,
      "epoch": 0.5079006772009029,
      "grad_norm": 17.186115264892578,
      "learning_rate": 9.83965146460653e-06,
      "lm_loss": 6.2476,
      "loss": 1.7351,
      "step": 225,
      "text_contrastive_loss": 1.0167,
      "train_positive_log_prob": -93.3259,
      "train_positive_token_accuracy": 0.0719,
      "train_positive_token_prob": 0.0254
    },
    {
      "contrastive_loss": 0.5749,
      "epoch": 0.510158013544018,
      "grad_norm": 17.85243797302246,
      "learning_rate": 9.83782367359834e-06,
      "lm_loss": 6.2224,
      "loss": 1.6575,
      "step": 226,
      "text_contrastive_loss": 0.9207,
      "train_positive_log_prob": -91.0266,
      "train_positive_token_accuracy": 0.0678,
      "train_positive_token_prob": 0.0241
    },
    {
      "contrastive_loss": 0.5526,
      "epoch": 0.5124153498871332,
      "grad_norm": 16.159278869628906,
      "learning_rate": 9.835985695881076e-06,
      "lm_loss": 6.2173,
      "loss": 1.596,
      "step": 227,
      "text_contrastive_loss": 0.8432,
      "train_positive_log_prob": -90.9736,
      "train_positive_token_accuracy": 0.0747,
      "train_positive_token_prob": 0.0256
    },
    {
      "contrastive_loss": 0.4397,
      "epoch": 0.5146726862302483,
      "grad_norm": 15.639877319335938,
      "learning_rate": 9.834137535324852e-06,
      "lm_loss": 6.2489,
      "loss": 1.4753,
      "step": 228,
      "text_contrastive_loss": 0.8213,
      "train_positive_log_prob": -92.6611,
      "train_positive_token_accuracy": 0.0772,
      "train_positive_token_prob": 0.0253
    },
    {
      "contrastive_loss": 0.4949,
      "epoch": 0.5169300225733634,
      "grad_norm": 15.5403413772583,
      "learning_rate": 9.83227919582123e-06,
      "lm_loss": 6.1217,
      "loss": 1.5301,
      "step": 229,
      "text_contrastive_loss": 0.8462,
      "train_positive_log_prob": -91.2759,
      "train_positive_token_accuracy": 0.0715,
      "train_positive_token_prob": 0.0244
    },
    {
      "contrastive_loss": 0.4991,
      "epoch": 0.5191873589164786,
      "grad_norm": 14.614914894104004,
      "learning_rate": 9.830410681283203e-06,
      "lm_loss": 6.2602,
      "loss": 1.5834,
      "step": 230,
      "text_contrastive_loss": 0.9167,
      "train_positive_log_prob": -91.4541,
      "train_positive_token_accuracy": 0.0738,
      "train_positive_token_prob": 0.0249
    },
    {
      "contrastive_loss": 0.6095,
      "epoch": 0.5214446952595937,
      "grad_norm": 20.50278091430664,
      "learning_rate": 9.828531995645183e-06,
      "lm_loss": 6.2406,
      "loss": 1.6358,
      "step": 231,
      "text_contrastive_loss": 0.8044,
      "train_positive_log_prob": -91.4917,
      "train_positive_token_accuracy": 0.069,
      "train_positive_token_prob": 0.0248
    },
    {
      "contrastive_loss": 0.6543,
      "epoch": 0.5237020316027088,
      "grad_norm": 18.58837127685547,
      "learning_rate": 9.826643142863006e-06,
      "lm_loss": 6.2583,
      "loss": 1.8025,
      "step": 232,
      "text_contrastive_loss": 1.0446,
      "train_positive_log_prob": -92.2403,
      "train_positive_token_accuracy": 0.084,
      "train_positive_token_prob": 0.0255
    },
    {
      "contrastive_loss": 0.5722,
      "epoch": 0.5259593679458239,
      "grad_norm": 17.41712188720703,
      "learning_rate": 9.824744126913914e-06,
      "lm_loss": 6.171,
      "loss": 1.6904,
      "step": 233,
      "text_contrastive_loss": 1.0021,
      "train_positive_log_prob": -87.9141,
      "train_positive_token_accuracy": 0.0749,
      "train_positive_token_prob": 0.0254
    },
    {
      "contrastive_loss": 0.6116,
      "epoch": 0.5282167042889391,
      "grad_norm": 16.216840744018555,
      "learning_rate": 9.822834951796547e-06,
      "lm_loss": 6.0939,
      "loss": 1.642,
      "step": 234,
      "text_contrastive_loss": 0.842,
      "train_positive_log_prob": -88.2419,
      "train_positive_token_accuracy": 0.079,
      "train_positive_token_prob": 0.0256
    },
    {
      "contrastive_loss": 0.549,
      "epoch": 0.5304740406320542,
      "grad_norm": 17.31159782409668,
      "learning_rate": 9.820915621530939e-06,
      "lm_loss": 6.2401,
      "loss": 1.5438,
      "step": 235,
      "text_contrastive_loss": 0.7415,
      "train_positive_log_prob": -90.8246,
      "train_positive_token_accuracy": 0.0714,
      "train_positive_token_prob": 0.0249
    },
    {
      "contrastive_loss": 0.6312,
      "epoch": 0.5327313769751693,
      "grad_norm": 16.976350784301758,
      "learning_rate": 9.818986140158507e-06,
      "lm_loss": 6.1592,
      "loss": 1.6354,
      "step": 236,
      "text_contrastive_loss": 0.7766,
      "train_positive_log_prob": -90.1244,
      "train_positive_token_accuracy": 0.0707,
      "train_positive_token_prob": 0.0245
    },
    {
      "contrastive_loss": 0.5212,
      "epoch": 0.5349887133182845,
      "grad_norm": 15.452056884765625,
      "learning_rate": 9.817046511742042e-06,
      "lm_loss": 6.1457,
      "loss": 1.4973,
      "step": 237,
      "text_contrastive_loss": 0.7231,
      "train_positive_log_prob": -88.4154,
      "train_positive_token_accuracy": 0.0695,
      "train_positive_token_prob": 0.024
    },
    {
      "contrastive_loss": 0.601,
      "epoch": 0.5372460496613995,
      "grad_norm": 16.932661056518555,
      "learning_rate": 9.815096740365698e-06,
      "lm_loss": 6.1435,
      "loss": 1.6303,
      "step": 238,
      "text_contrastive_loss": 0.8299,
      "train_positive_log_prob": -90.5476,
      "train_positive_token_accuracy": 0.0739,
      "train_positive_token_prob": 0.0255
    },
    {
      "contrastive_loss": 0.5377,
      "epoch": 0.5395033860045146,
      "grad_norm": 14.377546310424805,
      "learning_rate": 9.81313683013499e-06,
      "lm_loss": 6.2176,
      "loss": 1.6099,
      "step": 239,
      "text_contrastive_loss": 0.9008,
      "train_positive_log_prob": -90.4694,
      "train_positive_token_accuracy": 0.0738,
      "train_positive_token_prob": 0.0251
    },
    {
      "contrastive_loss": 0.6264,
      "epoch": 0.5417607223476298,
      "grad_norm": 17.758268356323242,
      "learning_rate": 9.811166785176785e-06,
      "lm_loss": 6.1442,
      "loss": 1.7602,
      "step": 240,
      "text_contrastive_loss": 1.0388,
      "train_positive_log_prob": -89.7901,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.0243
    },
    {
      "contrastive_loss": 0.5626,
      "epoch": 0.5440180586907449,
      "grad_norm": 14.423271179199219,
      "learning_rate": 9.809186609639281e-06,
      "lm_loss": 6.1467,
      "loss": 1.4961,
      "step": 241,
      "text_contrastive_loss": 0.6376,
      "train_positive_log_prob": -90.6907,
      "train_positive_token_accuracy": 0.0768,
      "train_positive_token_prob": 0.0271
    },
    {
      "contrastive_loss": 0.6618,
      "epoch": 0.54627539503386,
      "grad_norm": 18.368301391601562,
      "learning_rate": 9.807196307692015e-06,
      "lm_loss": 6.092,
      "loss": 1.7655,
      "step": 242,
      "text_contrastive_loss": 0.9889,
      "train_positive_log_prob": -89.1187,
      "train_positive_token_accuracy": 0.075,
      "train_positive_token_prob": 0.0265
    },
    {
      "contrastive_loss": 0.663,
      "epoch": 0.5485327313769752,
      "grad_norm": 19.776853561401367,
      "learning_rate": 9.805195883525844e-06,
      "lm_loss": 6.246,
      "loss": 1.7617,
      "step": 243,
      "text_contrastive_loss": 0.9481,
      "train_positive_log_prob": -91.979,
      "train_positive_token_accuracy": 0.078,
      "train_positive_token_prob": 0.0257
    },
    {
      "contrastive_loss": 0.5471,
      "epoch": 0.5507900677200903,
      "grad_norm": 15.385061264038086,
      "learning_rate": 9.803185341352936e-06,
      "lm_loss": 6.0717,
      "loss": 1.5648,
      "step": 244,
      "text_contrastive_loss": 0.8211,
      "train_positive_log_prob": -90.7888,
      "train_positive_token_accuracy": 0.0776,
      "train_positive_token_prob": 0.0262
    },
    {
      "contrastive_loss": 0.6105,
      "epoch": 0.5530474040632054,
      "grad_norm": 19.231651306152344,
      "learning_rate": 9.80116468540677e-06,
      "lm_loss": 6.1306,
      "loss": 1.6456,
      "step": 245,
      "text_contrastive_loss": 0.8441,
      "train_positive_log_prob": -90.2334,
      "train_positive_token_accuracy": 0.0717,
      "train_positive_token_prob": 0.0262
    },
    {
      "contrastive_loss": 0.5546,
      "epoch": 0.5553047404063205,
      "grad_norm": 17.634424209594727,
      "learning_rate": 9.799133919942117e-06,
      "lm_loss": 6.2802,
      "loss": 1.574,
      "step": 246,
      "text_contrastive_loss": 0.7826,
      "train_positive_log_prob": -93.5424,
      "train_positive_token_accuracy": 0.0803,
      "train_positive_token_prob": 0.0257
    },
    {
      "contrastive_loss": 0.483,
      "epoch": 0.5575620767494357,
      "grad_norm": 17.656702041625977,
      "learning_rate": 9.797093049235034e-06,
      "lm_loss": 6.3099,
      "loss": 1.5714,
      "step": 247,
      "text_contrastive_loss": 0.9149,
      "train_positive_log_prob": -95.4399,
      "train_positive_token_accuracy": 0.0767,
      "train_positive_token_prob": 0.0254
    },
    {
      "contrastive_loss": 0.6128,
      "epoch": 0.5598194130925508,
      "grad_norm": 18.952295303344727,
      "learning_rate": 9.795042077582856e-06,
      "lm_loss": 6.1993,
      "loss": 1.6999,
      "step": 248,
      "text_contrastive_loss": 0.9345,
      "train_positive_log_prob": -90.1788,
      "train_positive_token_accuracy": 0.073,
      "train_positive_token_prob": 0.0262
    },
    {
      "contrastive_loss": 0.5971,
      "epoch": 0.5620767494356659,
      "grad_norm": 16.89877700805664,
      "learning_rate": 9.792981009304192e-06,
      "lm_loss": 6.1549,
      "loss": 1.6729,
      "step": 249,
      "text_contrastive_loss": 0.9206,
      "train_positive_log_prob": -88.6223,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.0272
    },
    {
      "contrastive_loss": 0.6054,
      "epoch": 0.5643340857787811,
      "grad_norm": 19.338794708251953,
      "learning_rate": 9.790909848738907e-06,
      "lm_loss": 6.1171,
      "loss": 1.74,
      "step": 250,
      "text_contrastive_loss": 1.0457,
      "train_positive_log_prob": -90.9953,
      "train_positive_token_accuracy": 0.0767,
      "train_positive_token_prob": 0.0273
    },
    {
      "contrastive_loss": 0.6219,
      "epoch": 0.5665914221218962,
      "grad_norm": 18.177352905273438,
      "learning_rate": 9.788828600248114e-06,
      "lm_loss": 6.0648,
      "loss": 1.6477,
      "step": 251,
      "text_contrastive_loss": 0.8387,
      "train_positive_log_prob": -89.4804,
      "train_positive_token_accuracy": 0.0839,
      "train_positive_token_prob": 0.0283
    },
    {
      "contrastive_loss": 0.5716,
      "epoch": 0.5688487584650113,
      "grad_norm": 16.08686637878418,
      "learning_rate": 9.786737268214172e-06,
      "lm_loss": 6.1455,
      "loss": 1.6651,
      "step": 252,
      "text_contrastive_loss": 0.9579,
      "train_positive_log_prob": -91.418,
      "train_positive_token_accuracy": 0.0827,
      "train_positive_token_prob": 0.0276
    },
    {
      "contrastive_loss": 0.5646,
      "epoch": 0.5711060948081265,
      "grad_norm": 15.260428428649902,
      "learning_rate": 9.784635857040672e-06,
      "lm_loss": 6.0727,
      "loss": 1.6243,
      "step": 253,
      "text_contrastive_loss": 0.9048,
      "train_positive_log_prob": -88.6699,
      "train_positive_token_accuracy": 0.0725,
      "train_positive_token_prob": 0.0265
    },
    {
      "contrastive_loss": 0.6481,
      "epoch": 0.5733634311512416,
      "grad_norm": 17.00851058959961,
      "learning_rate": 9.782524371152425e-06,
      "lm_loss": 6.1111,
      "loss": 1.7243,
      "step": 254,
      "text_contrastive_loss": 0.9302,
      "train_positive_log_prob": -89.567,
      "train_positive_token_accuracy": 0.0754,
      "train_positive_token_prob": 0.0263
    },
    {
      "contrastive_loss": 0.6073,
      "epoch": 0.5756207674943566,
      "grad_norm": 18.875898361206055,
      "learning_rate": 9.780402814995458e-06,
      "lm_loss": 6.1165,
      "loss": 1.6024,
      "step": 255,
      "text_contrastive_loss": 0.7667,
      "train_positive_log_prob": -93.2308,
      "train_positive_token_accuracy": 0.072,
      "train_positive_token_prob": 0.0265
    },
    {
      "contrastive_loss": 0.505,
      "epoch": 0.5778781038374717,
      "grad_norm": 14.658570289611816,
      "learning_rate": 9.778271193037003e-06,
      "lm_loss": 6.0952,
      "loss": 1.6059,
      "step": 256,
      "text_contrastive_loss": 0.9828,
      "train_positive_log_prob": -91.6629,
      "train_positive_token_accuracy": 0.0743,
      "train_positive_token_prob": 0.027
    },
    {
      "contrastive_loss": 0.5923,
      "epoch": 0.5801354401805869,
      "grad_norm": 15.614409446716309,
      "learning_rate": 9.776129509765487e-06,
      "lm_loss": 6.1658,
      "loss": 1.6292,
      "step": 257,
      "text_contrastive_loss": 0.8406,
      "train_positive_log_prob": -92.0102,
      "train_positive_token_accuracy": 0.0809,
      "train_positive_token_prob": 0.0273
    },
    {
      "contrastive_loss": 0.7068,
      "epoch": 0.582392776523702,
      "grad_norm": 18.77219581604004,
      "learning_rate": 9.773977769690517e-06,
      "lm_loss": 6.0314,
      "loss": 1.7734,
      "step": 258,
      "text_contrastive_loss": 0.9269,
      "train_positive_log_prob": -88.1188,
      "train_positive_token_accuracy": 0.0786,
      "train_positive_token_prob": 0.0262
    },
    {
      "contrastive_loss": 0.6451,
      "epoch": 0.5846501128668171,
      "grad_norm": 16.3891658782959,
      "learning_rate": 9.771815977342882e-06,
      "lm_loss": 6.1078,
      "loss": 1.7128,
      "step": 259,
      "text_contrastive_loss": 0.9138,
      "train_positive_log_prob": -89.9655,
      "train_positive_token_accuracy": 0.0748,
      "train_positive_token_prob": 0.0262
    },
    {
      "contrastive_loss": 0.5838,
      "epoch": 0.5869074492099323,
      "grad_norm": 19.314834594726562,
      "learning_rate": 9.76964413727454e-06,
      "lm_loss": 6.1401,
      "loss": 1.6666,
      "step": 260,
      "text_contrastive_loss": 0.9377,
      "train_positive_log_prob": -93.8053,
      "train_positive_token_accuracy": 0.0816,
      "train_positive_token_prob": 0.0273
    },
    {
      "contrastive_loss": 0.7249,
      "epoch": 0.5891647855530474,
      "grad_norm": 20.733478546142578,
      "learning_rate": 9.767462254058593e-06,
      "lm_loss": 6.0405,
      "loss": 1.8595,
      "step": 261,
      "text_contrastive_loss": 1.0611,
      "train_positive_log_prob": -87.4025,
      "train_positive_token_accuracy": 0.0724,
      "train_positive_token_prob": 0.0259
    },
    {
      "contrastive_loss": 0.521,
      "epoch": 0.5914221218961625,
      "grad_norm": 15.232973098754883,
      "learning_rate": 9.765270332289307e-06,
      "lm_loss": 6.1078,
      "loss": 1.4998,
      "step": 262,
      "text_contrastive_loss": 0.736,
      "train_positive_log_prob": -91.445,
      "train_positive_token_accuracy": 0.0764,
      "train_positive_token_prob": 0.0263
    },
    {
      "contrastive_loss": 0.7181,
      "epoch": 0.5936794582392777,
      "grad_norm": 17.208232879638672,
      "learning_rate": 9.763068376582075e-06,
      "lm_loss": 6.1621,
      "loss": 1.9106,
      "step": 263,
      "text_contrastive_loss": 1.1527,
      "train_positive_log_prob": -91.2295,
      "train_positive_token_accuracy": 0.0834,
      "train_positive_token_prob": 0.0279
    },
    {
      "contrastive_loss": 0.4552,
      "epoch": 0.5959367945823928,
      "grad_norm": 14.609410285949707,
      "learning_rate": 9.76085639157342e-06,
      "lm_loss": 6.0441,
      "loss": 1.4435,
      "step": 264,
      "text_contrastive_loss": 0.7678,
      "train_positive_log_prob": -90.1864,
      "train_positive_token_accuracy": 0.0791,
      "train_positive_token_prob": 0.0279
    },
    {
      "contrastive_loss": 0.5018,
      "epoch": 0.5981941309255079,
      "grad_norm": 16.79071617126465,
      "learning_rate": 9.758634381920982e-06,
      "lm_loss": 6.1323,
      "loss": 1.5243,
      "step": 265,
      "text_contrastive_loss": 0.8185,
      "train_positive_log_prob": -91.6752,
      "train_positive_token_accuracy": 0.0755,
      "train_positive_token_prob": 0.0265
    },
    {
      "contrastive_loss": 0.6111,
      "epoch": 0.600451467268623,
      "grad_norm": 17.34381866455078,
      "learning_rate": 9.756402352303513e-06,
      "lm_loss": 6.2033,
      "loss": 1.7488,
      "step": 266,
      "text_contrastive_loss": 1.0348,
      "train_positive_log_prob": -93.725,
      "train_positive_token_accuracy": 0.0789,
      "train_positive_token_prob": 0.0266
    },
    {
      "contrastive_loss": 0.5962,
      "epoch": 0.6027088036117382,
      "grad_norm": 16.367290496826172,
      "learning_rate": 9.754160307420858e-06,
      "lm_loss": 6.1638,
      "loss": 1.6855,
      "step": 267,
      "text_contrastive_loss": 0.946,
      "train_positive_log_prob": -90.6276,
      "train_positive_token_accuracy": 0.0776,
      "train_positive_token_prob": 0.0266
    },
    {
      "contrastive_loss": 0.6552,
      "epoch": 0.6049661399548533,
      "grad_norm": 18.06829261779785,
      "learning_rate": 9.751908251993956e-06,
      "lm_loss": 6.1996,
      "loss": 1.7855,
      "step": 268,
      "text_contrastive_loss": 1.0206,
      "train_positive_log_prob": -91.3954,
      "train_positive_token_accuracy": 0.0753,
      "train_positive_token_prob": 0.0264
    },
    {
      "contrastive_loss": 0.5985,
      "epoch": 0.6072234762979684,
      "grad_norm": 18.242778778076172,
      "learning_rate": 9.749646190764823e-06,
      "lm_loss": 6.1059,
      "loss": 1.7115,
      "step": 269,
      "text_contrastive_loss": 1.0049,
      "train_positive_log_prob": -90.995,
      "train_positive_token_accuracy": 0.0794,
      "train_positive_token_prob": 0.0273
    },
    {
      "contrastive_loss": 0.6986,
      "epoch": 0.6094808126410836,
      "grad_norm": 19.880321502685547,
      "learning_rate": 9.747374128496541e-06,
      "lm_loss": 6.1151,
      "loss": 1.8399,
      "step": 270,
      "text_contrastive_loss": 1.0596,
      "train_positive_log_prob": -92.3053,
      "train_positive_token_accuracy": 0.0815,
      "train_positive_token_prob": 0.027
    },
    {
      "contrastive_loss": 0.4825,
      "epoch": 0.6117381489841986,
      "grad_norm": 13.749444007873535,
      "learning_rate": 9.745092069973254e-06,
      "lm_loss": 6.0851,
      "loss": 1.5329,
      "step": 271,
      "text_contrastive_loss": 0.8839,
      "train_positive_log_prob": -90.4711,
      "train_positive_token_accuracy": 0.0751,
      "train_positive_token_prob": 0.0266
    },
    {
      "contrastive_loss": 0.4315,
      "epoch": 0.6139954853273137,
      "grad_norm": 15.846484184265137,
      "learning_rate": 9.74280002000015e-06,
      "lm_loss": 6.0431,
      "loss": 1.4397,
      "step": 272,
      "text_contrastive_loss": 0.8078,
      "train_positive_log_prob": -88.2874,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.0271
    },
    {
      "contrastive_loss": 0.6614,
      "epoch": 0.6162528216704289,
      "grad_norm": 17.792285919189453,
      "learning_rate": 9.74049798340346e-06,
      "lm_loss": 6.0391,
      "loss": 1.7364,
      "step": 273,
      "text_contrastive_loss": 0.9422,
      "train_positive_log_prob": -91.1986,
      "train_positive_token_accuracy": 0.0823,
      "train_positive_token_prob": 0.0278
    },
    {
      "contrastive_loss": 0.5268,
      "epoch": 0.618510158013544,
      "grad_norm": 15.569907188415527,
      "learning_rate": 9.738185965030444e-06,
      "lm_loss": 6.1219,
      "loss": 1.6263,
      "step": 274,
      "text_contrastive_loss": 0.9747,
      "train_positive_log_prob": -91.7319,
      "train_positive_token_accuracy": 0.0756,
      "train_positive_token_prob": 0.0256
    },
    {
      "contrastive_loss": 0.5985,
      "epoch": 0.6207674943566591,
      "grad_norm": 20.254619598388672,
      "learning_rate": 9.735863969749373e-06,
      "lm_loss": 6.0733,
      "loss": 1.6646,
      "step": 275,
      "text_contrastive_loss": 0.9175,
      "train_positive_log_prob": -93.1891,
      "train_positive_token_accuracy": 0.0774,
      "train_positive_token_prob": 0.0278
    },
    {
      "contrastive_loss": 0.526,
      "epoch": 0.6230248306997742,
      "grad_norm": 17.3663272857666,
      "learning_rate": 9.733532002449533e-06,
      "lm_loss": 6.0813,
      "loss": 1.4958,
      "step": 276,
      "text_contrastive_loss": 0.7232,
      "train_positive_log_prob": -91.3601,
      "train_positive_token_accuracy": 0.0827,
      "train_positive_token_prob": 0.0277
    },
    {
      "contrastive_loss": 0.7095,
      "epoch": 0.6252821670428894,
      "grad_norm": 18.9420223236084,
      "learning_rate": 9.731190068041205e-06,
      "lm_loss": 5.9635,
      "loss": 1.7941,
      "step": 277,
      "text_contrastive_loss": 0.9764,
      "train_positive_log_prob": -88.9673,
      "train_positive_token_accuracy": 0.0835,
      "train_positive_token_prob": 0.0269
    },
    {
      "contrastive_loss": 0.6617,
      "epoch": 0.6275395033860045,
      "grad_norm": 18.74312400817871,
      "learning_rate": 9.728838171455655e-06,
      "lm_loss": 6.1322,
      "loss": 1.7378,
      "step": 278,
      "text_contrastive_loss": 0.9256,
      "train_positive_log_prob": -92.0784,
      "train_positive_token_accuracy": 0.0739,
      "train_positive_token_prob": 0.0245
    },
    {
      "contrastive_loss": 0.4955,
      "epoch": 0.6297968397291196,
      "grad_norm": 16.41619873046875,
      "learning_rate": 9.72647631764513e-06,
      "lm_loss": 6.094,
      "loss": 1.4873,
      "step": 279,
      "text_contrastive_loss": 0.7647,
      "train_positive_log_prob": -88.8882,
      "train_positive_token_accuracy": 0.0728,
      "train_positive_token_prob": 0.0257
    },
    {
      "contrastive_loss": 0.5533,
      "epoch": 0.6320541760722348,
      "grad_norm": 16.351163864135742,
      "learning_rate": 9.724104511582838e-06,
      "lm_loss": 6.0338,
      "loss": 1.5977,
      "step": 280,
      "text_contrastive_loss": 0.882,
      "train_positive_log_prob": -89.4713,
      "train_positive_token_accuracy": 0.0735,
      "train_positive_token_prob": 0.0252
    },
    {
      "contrastive_loss": 0.683,
      "epoch": 0.6343115124153499,
      "grad_norm": 19.409170150756836,
      "learning_rate": 9.721722758262948e-06,
      "lm_loss": 6.127,
      "loss": 1.7248,
      "step": 281,
      "text_contrastive_loss": 0.8581,
      "train_positive_log_prob": -91.4759,
      "train_positive_token_accuracy": 0.0767,
      "train_positive_token_prob": 0.0251
    },
    {
      "contrastive_loss": 0.5802,
      "epoch": 0.636568848758465,
      "grad_norm": 17.677600860595703,
      "learning_rate": 9.719331062700572e-06,
      "lm_loss": 6.1104,
      "loss": 1.5941,
      "step": 282,
      "text_contrastive_loss": 0.8057,
      "train_positive_log_prob": -90.5352,
      "train_positive_token_accuracy": 0.0699,
      "train_positive_token_prob": 0.0262
    },
    {
      "contrastive_loss": 0.6348,
      "epoch": 0.6388261851015802,
      "grad_norm": 17.54817771911621,
      "learning_rate": 9.716929429931757e-06,
      "lm_loss": 6.0538,
      "loss": 1.7448,
      "step": 283,
      "text_contrastive_loss": 1.0093,
      "train_positive_log_prob": -89.4645,
      "train_positive_token_accuracy": 0.0764,
      "train_positive_token_prob": 0.0249
    },
    {
      "contrastive_loss": 0.6313,
      "epoch": 0.6410835214446953,
      "grad_norm": 18.95325469970703,
      "learning_rate": 9.714517865013473e-06,
      "lm_loss": 6.1678,
      "loss": 1.6922,
      "step": 284,
      "text_contrastive_loss": 0.8882,
      "train_positive_log_prob": -91.0495,
      "train_positive_token_accuracy": 0.0728,
      "train_positive_token_prob": 0.0257
    },
    {
      "contrastive_loss": 0.6045,
      "epoch": 0.6433408577878104,
      "grad_norm": 18.337867736816406,
      "learning_rate": 9.712096373023603e-06,
      "lm_loss": 6.1029,
      "loss": 1.6479,
      "step": 285,
      "text_contrastive_loss": 0.8663,
      "train_positive_log_prob": -94.3844,
      "train_positive_token_accuracy": 0.0783,
      "train_positive_token_prob": 0.0263
    },
    {
      "contrastive_loss": 0.735,
      "epoch": 0.6455981941309256,
      "grad_norm": 19.584972381591797,
      "learning_rate": 9.70966495906094e-06,
      "lm_loss": 6.0201,
      "loss": 1.9011,
      "step": 286,
      "text_contrastive_loss": 1.1282,
      "train_positive_log_prob": -89.2677,
      "train_positive_token_accuracy": 0.0774,
      "train_positive_token_prob": 0.0271
    },
    {
      "contrastive_loss": 0.5304,
      "epoch": 0.6478555304740407,
      "grad_norm": 17.35251235961914,
      "learning_rate": 9.707223628245157e-06,
      "lm_loss": 5.9482,
      "loss": 1.5545,
      "step": 287,
      "text_contrastive_loss": 0.8585,
      "train_positive_log_prob": -88.387,
      "train_positive_token_accuracy": 0.087,
      "train_positive_token_prob": 0.028
    },
    {
      "contrastive_loss": 0.5458,
      "epoch": 0.6501128668171557,
      "grad_norm": 18.532630920410156,
      "learning_rate": 9.70477238571682e-06,
      "lm_loss": 6.1462,
      "loss": 1.6045,
      "step": 288,
      "text_contrastive_loss": 0.8882,
      "train_positive_log_prob": -91.0414,
      "train_positive_token_accuracy": 0.0746,
      "train_positive_token_prob": 0.0249
    },
    {
      "contrastive_loss": 0.7276,
      "epoch": 0.6523702031602708,
      "grad_norm": 18.21759605407715,
      "learning_rate": 9.702311236637357e-06,
      "lm_loss": 6.0582,
      "loss": 1.814,
      "step": 289,
      "text_contrastive_loss": 0.961,
      "train_positive_log_prob": -90.4924,
      "train_positive_token_accuracy": 0.08,
      "train_positive_token_prob": 0.0268
    },
    {
      "contrastive_loss": 0.6422,
      "epoch": 0.654627539503386,
      "grad_norm": 17.986358642578125,
      "learning_rate": 9.699840186189061e-06,
      "lm_loss": 6.0396,
      "loss": 1.6586,
      "step": 290,
      "text_contrastive_loss": 0.8247,
      "train_positive_log_prob": -91.0663,
      "train_positive_token_accuracy": 0.0746,
      "train_positive_token_prob": 0.0265
    },
    {
      "contrastive_loss": 0.5277,
      "epoch": 0.6568848758465011,
      "grad_norm": 16.52487564086914,
      "learning_rate": 9.697359239575069e-06,
      "lm_loss": 6.1162,
      "loss": 1.6542,
      "step": 291,
      "text_contrastive_loss": 1.0297,
      "train_positive_log_prob": -93.8718,
      "train_positive_token_accuracy": 0.0807,
      "train_positive_token_prob": 0.0275
    },
    {
      "contrastive_loss": 0.5414,
      "epoch": 0.6591422121896162,
      "grad_norm": 16.79463005065918,
      "learning_rate": 9.694868402019362e-06,
      "lm_loss": 6.0678,
      "loss": 1.647,
      "step": 292,
      "text_contrastive_loss": 0.9975,
      "train_positive_log_prob": -89.9989,
      "train_positive_token_accuracy": 0.0745,
      "train_positive_token_prob": 0.0255
    },
    {
      "contrastive_loss": 0.7381,
      "epoch": 0.6613995485327314,
      "grad_norm": 16.14636993408203,
      "learning_rate": 9.69236767876674e-06,
      "lm_loss": 5.9773,
      "loss": 1.8722,
      "step": 293,
      "text_contrastive_loss": 1.0728,
      "train_positive_log_prob": -89.5195,
      "train_positive_token_accuracy": 0.0715,
      "train_positive_token_prob": 0.0256
    },
    {
      "contrastive_loss": 0.5277,
      "epoch": 0.6636568848758465,
      "grad_norm": 14.656967163085938,
      "learning_rate": 9.689857075082828e-06,
      "lm_loss": 6.0224,
      "loss": 1.6384,
      "step": 294,
      "text_contrastive_loss": 1.0168,
      "train_positive_log_prob": -90.3835,
      "train_positive_token_accuracy": 0.0794,
      "train_positive_token_prob": 0.0278
    },
    {
      "contrastive_loss": 0.7446,
      "epoch": 0.6659142212189616,
      "grad_norm": 17.57710075378418,
      "learning_rate": 9.687336596254045e-06,
      "lm_loss": 6.1203,
      "loss": 1.8206,
      "step": 295,
      "text_contrastive_loss": 0.928,
      "train_positive_log_prob": -88.887,
      "train_positive_token_accuracy": 0.075,
      "train_positive_token_prob": 0.0263
    },
    {
      "contrastive_loss": 0.6219,
      "epoch": 0.6681715575620768,
      "grad_norm": 16.115657806396484,
      "learning_rate": 9.68480624758761e-06,
      "lm_loss": 6.0482,
      "loss": 1.6568,
      "step": 296,
      "text_contrastive_loss": 0.8601,
      "train_positive_log_prob": -87.8189,
      "train_positive_token_accuracy": 0.0815,
      "train_positive_token_prob": 0.0276
    },
    {
      "contrastive_loss": 0.5262,
      "epoch": 0.6704288939051919,
      "grad_norm": 16.029603958129883,
      "learning_rate": 9.682266034411527e-06,
      "lm_loss": 6.0701,
      "loss": 1.5084,
      "step": 297,
      "text_contrastive_loss": 0.7502,
      "train_positive_log_prob": -89.1563,
      "train_positive_token_accuracy": 0.0721,
      "train_positive_token_prob": 0.0258
    },
    {
      "contrastive_loss": 0.5522,
      "epoch": 0.672686230248307,
      "grad_norm": 16.295137405395508,
      "learning_rate": 9.679715962074566e-06,
      "lm_loss": 6.0496,
      "loss": 1.6329,
      "step": 298,
      "text_contrastive_loss": 0.9515,
      "train_positive_log_prob": -88.3593,
      "train_positive_token_accuracy": 0.0717,
      "train_positive_token_prob": 0.0272
    },
    {
      "contrastive_loss": 0.5422,
      "epoch": 0.6749435665914221,
      "grad_norm": 15.412567138671875,
      "learning_rate": 9.677156035946253e-06,
      "lm_loss": 5.9921,
      "loss": 1.6354,
      "step": 299,
      "text_contrastive_loss": 0.9878,
      "train_positive_log_prob": -88.6905,
      "train_positive_token_accuracy": 0.075,
      "train_positive_token_prob": 0.027
    },
    {
      "contrastive_loss": 0.5602,
      "epoch": 0.6772009029345373,
      "grad_norm": 16.239044189453125,
      "learning_rate": 9.674586261416874e-06,
      "lm_loss": 6.0446,
      "loss": 1.5681,
      "step": 300,
      "text_contrastive_loss": 0.8069,
      "train_positive_log_prob": -88.4974,
      "train_positive_token_accuracy": 0.0782,
      "train_positive_token_prob": 0.0268
    },
    {
      "contrastive_loss": 0.6695,
      "epoch": 0.6794582392776524,
      "grad_norm": 17.917858123779297,
      "learning_rate": 9.672006643897444e-06,
      "lm_loss": 6.0657,
      "loss": 1.8314,
      "step": 301,
      "text_contrastive_loss": 1.1107,
      "train_positive_log_prob": -89.0193,
      "train_positive_token_accuracy": 0.0745,
      "train_positive_token_prob": 0.0261
    },
    {
      "contrastive_loss": 0.5875,
      "epoch": 0.6817155756207675,
      "grad_norm": 15.977974891662598,
      "learning_rate": 9.669417188819704e-06,
      "lm_loss": 6.1059,
      "loss": 1.6711,
      "step": 302,
      "text_contrastive_loss": 0.9461,
      "train_positive_log_prob": -91.2699,
      "train_positive_token_accuracy": 0.076,
      "train_positive_token_prob": 0.0269
    },
    {
      "contrastive_loss": 0.5363,
      "epoch": 0.6839729119638827,
      "grad_norm": 16.47319221496582,
      "learning_rate": 9.666817901636115e-06,
      "lm_loss": 6.0319,
      "loss": 1.6197,
      "step": 303,
      "text_contrastive_loss": 0.9603,
      "train_positive_log_prob": -88.4529,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0276
    },
    {
      "contrastive_loss": 0.5232,
      "epoch": 0.6862302483069977,
      "grad_norm": 16.018712997436523,
      "learning_rate": 9.664208787819833e-06,
      "lm_loss": 6.0542,
      "loss": 1.5921,
      "step": 304,
      "text_contrastive_loss": 0.9269,
      "train_positive_log_prob": -88.3527,
      "train_positive_token_accuracy": 0.0771,
      "train_positive_token_prob": 0.0257
    },
    {
      "contrastive_loss": 0.5353,
      "epoch": 0.6884875846501128,
      "grad_norm": 14.660124778747559,
      "learning_rate": 9.66158985286471e-06,
      "lm_loss": 5.9663,
      "loss": 1.563,
      "step": 305,
      "text_contrastive_loss": 0.8621,
      "train_positive_log_prob": -87.9518,
      "train_positive_token_accuracy": 0.0816,
      "train_positive_token_prob": 0.028
    },
    {
      "contrastive_loss": 0.5071,
      "epoch": 0.690744920993228,
      "grad_norm": 15.332354545593262,
      "learning_rate": 9.658961102285276e-06,
      "lm_loss": 5.9832,
      "loss": 1.5337,
      "step": 306,
      "text_contrastive_loss": 0.8565,
      "train_positive_log_prob": -87.9938,
      "train_positive_token_accuracy": 0.0826,
      "train_positive_token_prob": 0.027
    },
    {
      "contrastive_loss": 0.6551,
      "epoch": 0.6930022573363431,
      "grad_norm": 15.486719131469727,
      "learning_rate": 9.656322541616734e-06,
      "lm_loss": 6.0244,
      "loss": 1.6758,
      "step": 307,
      "text_contrastive_loss": 0.8365,
      "train_positive_log_prob": -91.1411,
      "train_positive_token_accuracy": 0.0736,
      "train_positive_token_prob": 0.0268
    },
    {
      "contrastive_loss": 0.6736,
      "epoch": 0.6952595936794582,
      "grad_norm": 20.303863525390625,
      "learning_rate": 9.653674176414936e-06,
      "lm_loss": 6.0129,
      "loss": 1.7671,
      "step": 308,
      "text_contrastive_loss": 0.9845,
      "train_positive_log_prob": -88.1039,
      "train_positive_token_accuracy": 0.0744,
      "train_positive_token_prob": 0.0269
    },
    {
      "contrastive_loss": 0.5507,
      "epoch": 0.6975169300225733,
      "grad_norm": 15.779223442077637,
      "learning_rate": 9.651016012256382e-06,
      "lm_loss": 5.9777,
      "loss": 1.5803,
      "step": 309,
      "text_contrastive_loss": 0.8637,
      "train_positive_log_prob": -86.7909,
      "train_positive_token_accuracy": 0.0812,
      "train_positive_token_prob": 0.0275
    },
    {
      "contrastive_loss": 0.6486,
      "epoch": 0.6997742663656885,
      "grad_norm": 16.559284210205078,
      "learning_rate": 9.648348054738208e-06,
      "lm_loss": 6.1013,
      "loss": 1.8002,
      "step": 310,
      "text_contrastive_loss": 1.0828,
      "train_positive_log_prob": -88.926,
      "train_positive_token_accuracy": 0.0708,
      "train_positive_token_prob": 0.0259
    },
    {
      "contrastive_loss": 0.4758,
      "epoch": 0.7020316027088036,
      "grad_norm": 15.927628517150879,
      "learning_rate": 9.64567030947817e-06,
      "lm_loss": 6.0964,
      "loss": 1.4774,
      "step": 311,
      "text_contrastive_loss": 0.784,
      "train_positive_log_prob": -90.2778,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.027
    },
    {
      "contrastive_loss": 0.5897,
      "epoch": 0.7042889390519187,
      "grad_norm": 16.94893455505371,
      "learning_rate": 9.642982782114628e-06,
      "lm_loss": 5.9471,
      "loss": 1.6554,
      "step": 312,
      "text_contrastive_loss": 0.9419,
      "train_positive_log_prob": -86.047,
      "train_positive_token_accuracy": 0.0815,
      "train_positive_token_prob": 0.0269
    },
    {
      "contrastive_loss": 0.6094,
      "epoch": 0.7065462753950339,
      "grad_norm": 16.972312927246094,
      "learning_rate": 9.640285478306546e-06,
      "lm_loss": 5.9373,
      "loss": 1.7118,
      "step": 313,
      "text_contrastive_loss": 1.0174,
      "train_positive_log_prob": -86.931,
      "train_positive_token_accuracy": 0.0763,
      "train_positive_token_prob": 0.0272
    },
    {
      "contrastive_loss": 0.4797,
      "epoch": 0.708803611738149,
      "grad_norm": 15.580543518066406,
      "learning_rate": 9.63757840373347e-06,
      "lm_loss": 5.9767,
      "loss": 1.4474,
      "step": 314,
      "text_contrastive_loss": 0.7402,
      "train_positive_log_prob": -86.7767,
      "train_positive_token_accuracy": 0.0807,
      "train_positive_token_prob": 0.0285
    },
    {
      "contrastive_loss": 0.5058,
      "epoch": 0.7110609480812641,
      "grad_norm": 17.959449768066406,
      "learning_rate": 9.634861564095525e-06,
      "lm_loss": 6.1116,
      "loss": 1.5566,
      "step": 315,
      "text_contrastive_loss": 0.8794,
      "train_positive_log_prob": -87.5941,
      "train_positive_token_accuracy": 0.0856,
      "train_positive_token_prob": 0.0269
    },
    {
      "contrastive_loss": 0.6315,
      "epoch": 0.7133182844243793,
      "grad_norm": 18.653593063354492,
      "learning_rate": 9.632134965113389e-06,
      "lm_loss": 5.8974,
      "loss": 1.8327,
      "step": 316,
      "text_contrastive_loss": 1.2229,
      "train_positive_log_prob": -88.2529,
      "train_positive_token_accuracy": 0.082,
      "train_positive_token_prob": 0.0275
    },
    {
      "contrastive_loss": 0.6441,
      "epoch": 0.7155756207674944,
      "grad_norm": 19.774742126464844,
      "learning_rate": 9.629398612528299e-06,
      "lm_loss": 6.1244,
      "loss": 1.7442,
      "step": 317,
      "text_contrastive_loss": 0.9754,
      "train_positive_log_prob": -93.2565,
      "train_positive_token_accuracy": 0.0759,
      "train_positive_token_prob": 0.027
    },
    {
      "contrastive_loss": 0.5449,
      "epoch": 0.7178329571106095,
      "grad_norm": 15.776238441467285,
      "learning_rate": 9.626652512102021e-06,
      "lm_loss": 6.0455,
      "loss": 1.6211,
      "step": 318,
      "text_contrastive_loss": 0.9434,
      "train_positive_log_prob": -89.6836,
      "train_positive_token_accuracy": 0.079,
      "train_positive_token_prob": 0.0271
    },
    {
      "contrastive_loss": 0.5327,
      "epoch": 0.7200902934537246,
      "grad_norm": 15.428654670715332,
      "learning_rate": 9.623896669616855e-06,
      "lm_loss": 5.9935,
      "loss": 1.595,
      "step": 319,
      "text_contrastive_loss": 0.926,
      "train_positive_log_prob": -87.5854,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.0268
    },
    {
      "contrastive_loss": 0.5538,
      "epoch": 0.7223476297968398,
      "grad_norm": 15.515487670898438,
      "learning_rate": 9.621131090875603e-06,
      "lm_loss": 5.9955,
      "loss": 1.5415,
      "step": 320,
      "text_contrastive_loss": 0.7762,
      "train_positive_log_prob": -87.0324,
      "train_positive_token_accuracy": 0.0762,
      "train_positive_token_prob": 0.0271
    },
    {
      "contrastive_loss": 0.7214,
      "epoch": 0.7246049661399548,
      "grad_norm": 16.490665435791016,
      "learning_rate": 9.618355781701584e-06,
      "lm_loss": 5.9352,
      "loss": 1.719,
      "step": 321,
      "text_contrastive_loss": 0.8082,
      "train_positive_log_prob": -87.7036,
      "train_positive_token_accuracy": 0.0656,
      "train_positive_token_prob": 0.0261
    },
    {
      "contrastive_loss": 0.6069,
      "epoch": 0.7268623024830699,
      "grad_norm": 16.947256088256836,
      "learning_rate": 9.61557074793859e-06,
      "lm_loss": 6.0106,
      "loss": 1.7047,
      "step": 322,
      "text_contrastive_loss": 0.9934,
      "train_positive_log_prob": -87.4139,
      "train_positive_token_accuracy": 0.0851,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.7181,
      "epoch": 0.7291196388261851,
      "grad_norm": 17.712127685546875,
      "learning_rate": 9.612775995450896e-06,
      "lm_loss": 6.1349,
      "loss": 1.8107,
      "step": 323,
      "text_contrastive_loss": 0.9581,
      "train_positive_log_prob": -91.6418,
      "train_positive_token_accuracy": 0.0779,
      "train_positive_token_prob": 0.0269
    },
    {
      "contrastive_loss": 0.5646,
      "epoch": 0.7313769751693002,
      "grad_norm": 17.118139266967773,
      "learning_rate": 9.609971530123243e-06,
      "lm_loss": 5.973,
      "loss": 1.5571,
      "step": 324,
      "text_contrastive_loss": 0.7903,
      "train_positive_log_prob": -88.3355,
      "train_positive_token_accuracy": 0.0848,
      "train_positive_token_prob": 0.0276
    },
    {
      "contrastive_loss": 0.5351,
      "epoch": 0.7336343115124153,
      "grad_norm": 18.682863235473633,
      "learning_rate": 9.607157357860823e-06,
      "lm_loss": 6.1304,
      "loss": 1.6209,
      "step": 325,
      "text_contrastive_loss": 0.9455,
      "train_positive_log_prob": -91.7654,
      "train_positive_token_accuracy": 0.0812,
      "train_positive_token_prob": 0.0273
    },
    {
      "contrastive_loss": 0.614,
      "epoch": 0.7358916478555305,
      "grad_norm": 20.833709716796875,
      "learning_rate": 9.604333484589266e-06,
      "lm_loss": 5.931,
      "loss": 1.6868,
      "step": 326,
      "text_contrastive_loss": 0.9595,
      "train_positive_log_prob": -88.2185,
      "train_positive_token_accuracy": 0.0772,
      "train_positive_token_prob": 0.0274
    },
    {
      "contrastive_loss": 0.5644,
      "epoch": 0.7381489841986456,
      "grad_norm": 17.01936912536621,
      "learning_rate": 9.601499916254626e-06,
      "lm_loss": 6.0155,
      "loss": 1.6586,
      "step": 327,
      "text_contrastive_loss": 0.9854,
      "train_positive_log_prob": -89.2936,
      "train_positive_token_accuracy": 0.0768,
      "train_positive_token_prob": 0.028
    },
    {
      "contrastive_loss": 0.5838,
      "epoch": 0.7404063205417607,
      "grad_norm": 16.37017822265625,
      "learning_rate": 9.598656658823378e-06,
      "lm_loss": 6.0186,
      "loss": 1.6197,
      "step": 328,
      "text_contrastive_loss": 0.8681,
      "train_positive_log_prob": -87.4516,
      "train_positive_token_accuracy": 0.0777,
      "train_positive_token_prob": 0.0274
    },
    {
      "contrastive_loss": 0.4752,
      "epoch": 0.7426636568848759,
      "grad_norm": 13.509170532226562,
      "learning_rate": 9.595803718282391e-06,
      "lm_loss": 6.0354,
      "loss": 1.462,
      "step": 329,
      "text_contrastive_loss": 0.7667,
      "train_positive_log_prob": -90.4357,
      "train_positive_token_accuracy": 0.0696,
      "train_positive_token_prob": 0.0261
    },
    {
      "contrastive_loss": 0.5827,
      "epoch": 0.744920993227991,
      "grad_norm": 15.919811248779297,
      "learning_rate": 9.59294110063893e-06,
      "lm_loss": 5.9231,
      "loss": 1.57,
      "step": 330,
      "text_contrastive_loss": 0.79,
      "train_positive_log_prob": -88.336,
      "train_positive_token_accuracy": 0.0705,
      "train_positive_token_prob": 0.0271
    },
    {
      "contrastive_loss": 0.6355,
      "epoch": 0.7471783295711061,
      "grad_norm": 15.94876766204834,
      "learning_rate": 9.590068811920637e-06,
      "lm_loss": 6.018,
      "loss": 1.7106,
      "step": 331,
      "text_contrastive_loss": 0.9466,
      "train_positive_log_prob": -87.8264,
      "train_positive_token_accuracy": 0.0774,
      "train_positive_token_prob": 0.0281
    },
    {
      "contrastive_loss": 0.4436,
      "epoch": 0.7494356659142212,
      "grad_norm": 12.93568229675293,
      "learning_rate": 9.587186858175507e-06,
      "lm_loss": 5.9285,
      "loss": 1.4444,
      "step": 332,
      "text_contrastive_loss": 0.8159,
      "train_positive_log_prob": -87.7702,
      "train_positive_token_accuracy": 0.0755,
      "train_positive_token_prob": 0.0273
    },
    {
      "contrastive_loss": 0.5154,
      "epoch": 0.7516930022573364,
      "grad_norm": 13.687461853027344,
      "learning_rate": 9.584295245471898e-06,
      "lm_loss": 6.047,
      "loss": 1.5184,
      "step": 333,
      "text_contrastive_loss": 0.7965,
      "train_positive_log_prob": -89.6451,
      "train_positive_token_accuracy": 0.0748,
      "train_positive_token_prob": 0.0269
    },
    {
      "contrastive_loss": 0.6214,
      "epoch": 0.7539503386004515,
      "grad_norm": 16.401796340942383,
      "learning_rate": 9.581393979898502e-06,
      "lm_loss": 6.08,
      "loss": 1.6637,
      "step": 334,
      "text_contrastive_loss": 0.8686,
      "train_positive_log_prob": -90.8207,
      "train_positive_token_accuracy": 0.0724,
      "train_positive_token_prob": 0.0262
    },
    {
      "contrastive_loss": 0.5947,
      "epoch": 0.7562076749435666,
      "grad_norm": 16.64836311340332,
      "learning_rate": 9.578483067564335e-06,
      "lm_loss": 5.9331,
      "loss": 1.6623,
      "step": 335,
      "text_contrastive_loss": 0.9486,
      "train_positive_log_prob": -88.5124,
      "train_positive_token_accuracy": 0.0783,
      "train_positive_token_prob": 0.0264
    },
    {
      "contrastive_loss": 0.5481,
      "epoch": 0.7584650112866818,
      "grad_norm": 15.983649253845215,
      "learning_rate": 9.575562514598727e-06,
      "lm_loss": 6.044,
      "loss": 1.5863,
      "step": 336,
      "text_contrastive_loss": 0.8677,
      "train_positive_log_prob": -90.2519,
      "train_positive_token_accuracy": 0.0772,
      "train_positive_token_prob": 0.0287
    },
    {
      "contrastive_loss": 0.5021,
      "epoch": 0.7607223476297968,
      "grad_norm": 16.104372024536133,
      "learning_rate": 9.572632327151309e-06,
      "lm_loss": 5.9966,
      "loss": 1.4857,
      "step": 337,
      "text_contrastive_loss": 0.7679,
      "train_positive_log_prob": -87.5348,
      "train_positive_token_accuracy": 0.0805,
      "train_positive_token_prob": 0.0267
    },
    {
      "contrastive_loss": 0.5592,
      "epoch": 0.7629796839729119,
      "grad_norm": 15.639152526855469,
      "learning_rate": 9.569692511391995e-06,
      "lm_loss": 6.0392,
      "loss": 1.6813,
      "step": 338,
      "text_contrastive_loss": 1.0363,
      "train_positive_log_prob": -89.8587,
      "train_positive_token_accuracy": 0.0738,
      "train_positive_token_prob": 0.0263
    },
    {
      "contrastive_loss": 0.6336,
      "epoch": 0.7652370203160271,
      "grad_norm": 18.634450912475586,
      "learning_rate": 9.566743073510976e-06,
      "lm_loss": 5.8507,
      "loss": 1.6356,
      "step": 339,
      "text_contrastive_loss": 0.8337,
      "train_positive_log_prob": -84.6849,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0273
    },
    {
      "contrastive_loss": 0.55,
      "epoch": 0.7674943566591422,
      "grad_norm": 14.68127727508545,
      "learning_rate": 9.563784019718704e-06,
      "lm_loss": 5.979,
      "loss": 1.543,
      "step": 340,
      "text_contrastive_loss": 0.7901,
      "train_positive_log_prob": -87.0723,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.028
    },
    {
      "contrastive_loss": 0.6238,
      "epoch": 0.7697516930022573,
      "grad_norm": 16.066402435302734,
      "learning_rate": 9.560815356245875e-06,
      "lm_loss": 5.8746,
      "loss": 1.6467,
      "step": 341,
      "text_contrastive_loss": 0.8709,
      "train_positive_log_prob": -86.8341,
      "train_positive_token_accuracy": 0.0848,
      "train_positive_token_prob": 0.0288
    },
    {
      "contrastive_loss": 0.5541,
      "epoch": 0.7720090293453724,
      "grad_norm": 16.494718551635742,
      "learning_rate": 9.557837089343424e-06,
      "lm_loss": 6.0958,
      "loss": 1.5481,
      "step": 342,
      "text_contrastive_loss": 0.7688,
      "train_positive_log_prob": -92.3658,
      "train_positive_token_accuracy": 0.0817,
      "train_positive_token_prob": 0.0269
    },
    {
      "contrastive_loss": 0.6157,
      "epoch": 0.7742663656884876,
      "grad_norm": 16.388378143310547,
      "learning_rate": 9.554849225282503e-06,
      "lm_loss": 6.1625,
      "loss": 1.6874,
      "step": 343,
      "text_contrastive_loss": 0.9109,
      "train_positive_log_prob": -92.0531,
      "train_positive_token_accuracy": 0.0764,
      "train_positive_token_prob": 0.027
    },
    {
      "contrastive_loss": 0.6033,
      "epoch": 0.7765237020316027,
      "grad_norm": 15.661701202392578,
      "learning_rate": 9.551851770354477e-06,
      "lm_loss": 6.1156,
      "loss": 1.6969,
      "step": 344,
      "text_contrastive_loss": 0.964,
      "train_positive_log_prob": -89.2016,
      "train_positive_token_accuracy": 0.0715,
      "train_positive_token_prob": 0.0257
    },
    {
      "contrastive_loss": 0.6114,
      "epoch": 0.7787810383747178,
      "grad_norm": 19.551776885986328,
      "learning_rate": 9.548844730870903e-06,
      "lm_loss": 6.0139,
      "loss": 1.7044,
      "step": 345,
      "text_contrastive_loss": 0.9831,
      "train_positive_log_prob": -89.7135,
      "train_positive_token_accuracy": 0.0702,
      "train_positive_token_prob": 0.0265
    },
    {
      "contrastive_loss": 0.6824,
      "epoch": 0.781038374717833,
      "grad_norm": 18.047565460205078,
      "learning_rate": 9.545828113163516e-06,
      "lm_loss": 6.0352,
      "loss": 1.7321,
      "step": 346,
      "text_contrastive_loss": 0.8924,
      "train_positive_log_prob": -88.4301,
      "train_positive_token_accuracy": 0.0733,
      "train_positive_token_prob": 0.0272
    },
    {
      "contrastive_loss": 0.5448,
      "epoch": 0.7832957110609481,
      "grad_norm": 16.720369338989258,
      "learning_rate": 9.542801923584228e-06,
      "lm_loss": 5.9956,
      "loss": 1.5758,
      "step": 347,
      "text_contrastive_loss": 0.8628,
      "train_positive_log_prob": -88.8306,
      "train_positive_token_accuracy": 0.0799,
      "train_positive_token_prob": 0.0276
    },
    {
      "contrastive_loss": 0.6701,
      "epoch": 0.7855530474040632,
      "grad_norm": 16.758535385131836,
      "learning_rate": 9.5397661685051e-06,
      "lm_loss": 5.9736,
      "loss": 1.7021,
      "step": 348,
      "text_contrastive_loss": 0.8693,
      "train_positive_log_prob": -86.5943,
      "train_positive_token_accuracy": 0.0798,
      "train_positive_token_prob": 0.0271
    },
    {
      "contrastive_loss": 0.6332,
      "epoch": 0.7878103837471784,
      "grad_norm": 16.43297576904297,
      "learning_rate": 9.536720854318333e-06,
      "lm_loss": 6.0055,
      "loss": 1.7895,
      "step": 349,
      "text_contrastive_loss": 1.1115,
      "train_positive_log_prob": -86.9145,
      "train_positive_token_accuracy": 0.0745,
      "train_positive_token_prob": 0.026
    },
    {
      "contrastive_loss": 0.571,
      "epoch": 0.7900677200902935,
      "grad_norm": 16.152088165283203,
      "learning_rate": 9.533665987436262e-06,
      "lm_loss": 6.0885,
      "loss": 1.634,
      "step": 350,
      "text_contrastive_loss": 0.9083,
      "train_positive_log_prob": -91.7586,
      "train_positive_token_accuracy": 0.0741,
      "train_positive_token_prob": 0.0257
    },
    {
      "contrastive_loss": 0.4889,
      "epoch": 0.7923250564334086,
      "grad_norm": 14.552695274353027,
      "learning_rate": 9.530601574291331e-06,
      "lm_loss": 5.9661,
      "loss": 1.4834,
      "step": 351,
      "text_contrastive_loss": 0.7957,
      "train_positive_log_prob": -88.2302,
      "train_positive_token_accuracy": 0.0681,
      "train_positive_token_prob": 0.0253
    },
    {
      "contrastive_loss": 0.5776,
      "epoch": 0.7945823927765236,
      "grad_norm": 16.552778244018555,
      "learning_rate": 9.527527621336087e-06,
      "lm_loss": 6.033,
      "loss": 1.6147,
      "step": 352,
      "text_contrastive_loss": 0.8675,
      "train_positive_log_prob": -90.2359,
      "train_positive_token_accuracy": 0.0786,
      "train_positive_token_prob": 0.0267
    },
    {
      "contrastive_loss": 0.5818,
      "epoch": 0.7968397291196389,
      "grad_norm": 14.326408386230469,
      "learning_rate": 9.524444135043168e-06,
      "lm_loss": 5.9204,
      "loss": 1.6186,
      "step": 353,
      "text_contrastive_loss": 0.8896,
      "train_positive_log_prob": -86.4452,
      "train_positive_token_accuracy": 0.0784,
      "train_positive_token_prob": 0.0279
    },
    {
      "contrastive_loss": 0.5184,
      "epoch": 0.7990970654627539,
      "grad_norm": 15.130330085754395,
      "learning_rate": 9.521351121905278e-06,
      "lm_loss": 5.9565,
      "loss": 1.5699,
      "step": 354,
      "text_contrastive_loss": 0.9116,
      "train_positive_log_prob": -87.9751,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0269
    },
    {
      "contrastive_loss": 0.4493,
      "epoch": 0.801354401805869,
      "grad_norm": 17.21162223815918,
      "learning_rate": 9.518248588435185e-06,
      "lm_loss": 6.1181,
      "loss": 1.4731,
      "step": 355,
      "text_contrastive_loss": 0.8239,
      "train_positive_log_prob": -90.6486,
      "train_positive_token_accuracy": 0.0707,
      "train_positive_token_prob": 0.0263
    },
    {
      "contrastive_loss": 0.6361,
      "epoch": 0.8036117381489842,
      "grad_norm": 16.533164978027344,
      "learning_rate": 9.515136541165708e-06,
      "lm_loss": 5.8704,
      "loss": 1.6609,
      "step": 356,
      "text_contrastive_loss": 0.8754,
      "train_positive_log_prob": -85.9868,
      "train_positive_token_accuracy": 0.0795,
      "train_positive_token_prob": 0.0267
    },
    {
      "contrastive_loss": 0.5051,
      "epoch": 0.8058690744920993,
      "grad_norm": 14.658692359924316,
      "learning_rate": 9.512014986649691e-06,
      "lm_loss": 5.9747,
      "loss": 1.5898,
      "step": 357,
      "text_contrastive_loss": 0.9744,
      "train_positive_log_prob": -89.1526,
      "train_positive_token_accuracy": 0.0717,
      "train_positive_token_prob": 0.0265
    },
    {
      "contrastive_loss": 0.4978,
      "epoch": 0.8081264108352144,
      "grad_norm": 17.369964599609375,
      "learning_rate": 9.50888393146e-06,
      "lm_loss": 5.9468,
      "loss": 1.5709,
      "step": 358,
      "text_contrastive_loss": 0.9568,
      "train_positive_log_prob": -85.2053,
      "train_positive_token_accuracy": 0.0722,
      "train_positive_token_prob": 0.0271
    },
    {
      "contrastive_loss": 0.572,
      "epoch": 0.8103837471783296,
      "grad_norm": 16.059120178222656,
      "learning_rate": 9.50574338218951e-06,
      "lm_loss": 5.9644,
      "loss": 1.6031,
      "step": 359,
      "text_contrastive_loss": 0.8693,
      "train_positive_log_prob": -87.2093,
      "train_positive_token_accuracy": 0.0737,
      "train_positive_token_prob": 0.0274
    },
    {
      "contrastive_loss": 0.6199,
      "epoch": 0.8126410835214447,
      "grad_norm": 16.445810317993164,
      "learning_rate": 9.502593345451078e-06,
      "lm_loss": 5.9581,
      "loss": 1.6576,
      "step": 360,
      "text_contrastive_loss": 0.8838,
      "train_positive_log_prob": -88.2559,
      "train_positive_token_accuracy": 0.0783,
      "train_positive_token_prob": 0.0274
    },
    {
      "contrastive_loss": 0.4576,
      "epoch": 0.8148984198645598,
      "grad_norm": 14.760526657104492,
      "learning_rate": 9.499433827877547e-06,
      "lm_loss": 5.9476,
      "loss": 1.5293,
      "step": 361,
      "text_contrastive_loss": 0.9539,
      "train_positive_log_prob": -84.9882,
      "train_positive_token_accuracy": 0.0785,
      "train_positive_token_prob": 0.0278
    },
    {
      "contrastive_loss": 0.4922,
      "epoch": 0.8171557562076749,
      "grad_norm": 16.34233856201172,
      "learning_rate": 9.49626483612172e-06,
      "lm_loss": 5.8911,
      "loss": 1.4942,
      "step": 362,
      "text_contrastive_loss": 0.8257,
      "train_positive_log_prob": -86.315,
      "train_positive_token_accuracy": 0.0778,
      "train_positive_token_prob": 0.0278
    },
    {
      "contrastive_loss": 0.5968,
      "epoch": 0.8194130925507901,
      "grad_norm": 18.121335983276367,
      "learning_rate": 9.493086376856346e-06,
      "lm_loss": 5.9198,
      "loss": 1.7071,
      "step": 363,
      "text_contrastive_loss": 1.0368,
      "train_positive_log_prob": -87.2706,
      "train_positive_token_accuracy": 0.0755,
      "train_positive_token_prob": 0.0272
    },
    {
      "contrastive_loss": 0.5812,
      "epoch": 0.8216704288939052,
      "grad_norm": 17.509723663330078,
      "learning_rate": 9.489898456774116e-06,
      "lm_loss": 5.9005,
      "loss": 1.6531,
      "step": 364,
      "text_contrastive_loss": 0.9639,
      "train_positive_log_prob": -86.044,
      "train_positive_token_accuracy": 0.0723,
      "train_positive_token_prob": 0.027
    },
    {
      "contrastive_loss": 0.5617,
      "epoch": 0.8239277652370203,
      "grad_norm": 15.32075023651123,
      "learning_rate": 9.486701082587635e-06,
      "lm_loss": 5.9199,
      "loss": 1.5451,
      "step": 365,
      "text_contrastive_loss": 0.7828,
      "train_positive_log_prob": -86.8076,
      "train_positive_token_accuracy": 0.0743,
      "train_positive_token_prob": 0.027
    },
    {
      "contrastive_loss": 0.5565,
      "epoch": 0.8261851015801355,
      "grad_norm": 18.541889190673828,
      "learning_rate": 9.483494261029418e-06,
      "lm_loss": 5.8448,
      "loss": 1.545,
      "step": 366,
      "text_contrastive_loss": 0.8081,
      "train_positive_log_prob": -85.7326,
      "train_positive_token_accuracy": 0.0754,
      "train_positive_token_prob": 0.0286
    },
    {
      "contrastive_loss": 0.665,
      "epoch": 0.8284424379232506,
      "grad_norm": 17.35003662109375,
      "learning_rate": 9.480277998851875e-06,
      "lm_loss": 5.9089,
      "loss": 1.7371,
      "step": 367,
      "text_contrastive_loss": 0.9623,
      "train_positive_log_prob": -87.941,
      "train_positive_token_accuracy": 0.0748,
      "train_positive_token_prob": 0.0285
    },
    {
      "contrastive_loss": 0.5716,
      "epoch": 0.8306997742663657,
      "grad_norm": 16.31109046936035,
      "learning_rate": 9.47705230282729e-06,
      "lm_loss": 5.887,
      "loss": 1.6179,
      "step": 368,
      "text_contrastive_loss": 0.9152,
      "train_positive_log_prob": -87.7974,
      "train_positive_token_accuracy": 0.0699,
      "train_positive_token_prob": 0.0265
    },
    {
      "contrastive_loss": 0.5609,
      "epoch": 0.8329571106094809,
      "grad_norm": 15.051897048950195,
      "learning_rate": 9.473817179747815e-06,
      "lm_loss": 5.9945,
      "loss": 1.5513,
      "step": 369,
      "text_contrastive_loss": 0.782,
      "train_positive_log_prob": -88.5668,
      "train_positive_token_accuracy": 0.064,
      "train_positive_token_prob": 0.0261
    },
    {
      "contrastive_loss": 0.7267,
      "epoch": 0.835214446952596,
      "grad_norm": 17.49429702758789,
      "learning_rate": 9.470572636425451e-06,
      "lm_loss": 6.0031,
      "loss": 1.8776,
      "step": 370,
      "text_contrastive_loss": 1.1012,
      "train_positive_log_prob": -89.394,
      "train_positive_token_accuracy": 0.0832,
      "train_positive_token_prob": 0.028
    },
    {
      "contrastive_loss": 0.5684,
      "epoch": 0.837471783295711,
      "grad_norm": 17.741493225097656,
      "learning_rate": 9.467318679692031e-06,
      "lm_loss": 5.854,
      "loss": 1.6599,
      "step": 371,
      "text_contrastive_loss": 1.0122,
      "train_positive_log_prob": -87.42,
      "train_positive_token_accuracy": 0.0772,
      "train_positive_token_prob": 0.0282
    },
    {
      "contrastive_loss": 0.4824,
      "epoch": 0.8397291196388262,
      "grad_norm": 15.962331771850586,
      "learning_rate": 9.464055316399217e-06,
      "lm_loss": 5.934,
      "loss": 1.489,
      "step": 372,
      "text_contrastive_loss": 0.8265,
      "train_positive_log_prob": -88.3151,
      "train_positive_token_accuracy": 0.079,
      "train_positive_token_prob": 0.0274
    },
    {
      "contrastive_loss": 0.5473,
      "epoch": 0.8419864559819413,
      "grad_norm": 16.54502296447754,
      "learning_rate": 9.46078255341847e-06,
      "lm_loss": 5.919,
      "loss": 1.5184,
      "step": 373,
      "text_contrastive_loss": 0.7583,
      "train_positive_log_prob": -86.8454,
      "train_positive_token_accuracy": 0.0892,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.5836,
      "epoch": 0.8442437923250564,
      "grad_norm": 15.227729797363281,
      "learning_rate": 9.457500397641049e-06,
      "lm_loss": 5.9626,
      "loss": 1.6041,
      "step": 374,
      "text_contrastive_loss": 0.8483,
      "train_positive_log_prob": -90.3974,
      "train_positive_token_accuracy": 0.0794,
      "train_positive_token_prob": 0.0283
    },
    {
      "contrastive_loss": 0.5116,
      "epoch": 0.8465011286681715,
      "grad_norm": 16.31110191345215,
      "learning_rate": 9.454208855977986e-06,
      "lm_loss": 5.8935,
      "loss": 1.5011,
      "step": 375,
      "text_contrastive_loss": 0.8003,
      "train_positive_log_prob": -86.6806,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.0287
    },
    {
      "contrastive_loss": 0.476,
      "epoch": 0.8487584650112867,
      "grad_norm": 16.87764549255371,
      "learning_rate": 9.450907935360081e-06,
      "lm_loss": 5.7913,
      "loss": 1.5026,
      "step": 376,
      "text_contrastive_loss": 0.8949,
      "train_positive_log_prob": -85.0093,
      "train_positive_token_accuracy": 0.0772,
      "train_positive_token_prob": 0.0278
    },
    {
      "contrastive_loss": 0.6421,
      "epoch": 0.8510158013544018,
      "grad_norm": 16.616607666015625,
      "learning_rate": 9.447597642737878e-06,
      "lm_loss": 6.0081,
      "loss": 1.7121,
      "step": 377,
      "text_contrastive_loss": 0.9384,
      "train_positive_log_prob": -90.167,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.0274
    },
    {
      "contrastive_loss": 0.5528,
      "epoch": 0.8532731376975169,
      "grad_norm": 17.461790084838867,
      "learning_rate": 9.44427798508166e-06,
      "lm_loss": 5.9656,
      "loss": 1.5774,
      "step": 378,
      "text_contrastive_loss": 0.8561,
      "train_positive_log_prob": -86.3173,
      "train_positive_token_accuracy": 0.0775,
      "train_positive_token_prob": 0.0269
    },
    {
      "contrastive_loss": 0.5893,
      "epoch": 0.8555304740406321,
      "grad_norm": 16.50901985168457,
      "learning_rate": 9.440948969381425e-06,
      "lm_loss": 5.8756,
      "loss": 1.6351,
      "step": 379,
      "text_contrastive_loss": 0.9163,
      "train_positive_log_prob": -87.2872,
      "train_positive_token_accuracy": 0.086,
      "train_positive_token_prob": 0.0289
    },
    {
      "contrastive_loss": 0.5734,
      "epoch": 0.8577878103837472,
      "grad_norm": 15.030078887939453,
      "learning_rate": 9.437610602646878e-06,
      "lm_loss": 5.9427,
      "loss": 1.565,
      "step": 380,
      "text_contrastive_loss": 0.7946,
      "train_positive_log_prob": -88.9976,
      "train_positive_token_accuracy": 0.0769,
      "train_positive_token_prob": 0.0272
    },
    {
      "contrastive_loss": 0.5823,
      "epoch": 0.8600451467268623,
      "grad_norm": 14.964461326599121,
      "learning_rate": 9.434262891907413e-06,
      "lm_loss": 5.9993,
      "loss": 1.5932,
      "step": 381,
      "text_contrastive_loss": 0.8221,
      "train_positive_log_prob": -88.2041,
      "train_positive_token_accuracy": 0.0731,
      "train_positive_token_prob": 0.0265
    },
    {
      "contrastive_loss": 0.5695,
      "epoch": 0.8623024830699775,
      "grad_norm": 16.77033042907715,
      "learning_rate": 9.430905844212102e-06,
      "lm_loss": 5.9179,
      "loss": 1.6473,
      "step": 382,
      "text_contrastive_loss": 0.9719,
      "train_positive_log_prob": -86.2016,
      "train_positive_token_accuracy": 0.0737,
      "train_positive_token_prob": 0.0268
    },
    {
      "contrastive_loss": 0.6778,
      "epoch": 0.8645598194130926,
      "grad_norm": 19.320667266845703,
      "learning_rate": 9.427539466629672e-06,
      "lm_loss": 5.8167,
      "loss": 1.6839,
      "step": 383,
      "text_contrastive_loss": 0.8489,
      "train_positive_log_prob": -86.042,
      "train_positive_token_accuracy": 0.0832,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.4745,
      "epoch": 0.8668171557562077,
      "grad_norm": 15.715731620788574,
      "learning_rate": 9.424163766248499e-06,
      "lm_loss": 5.9824,
      "loss": 1.4254,
      "step": 384,
      "text_contrastive_loss": 0.7053,
      "train_positive_log_prob": -89.9199,
      "train_positive_token_accuracy": 0.085,
      "train_positive_token_prob": 0.0269
    },
    {
      "contrastive_loss": 0.6555,
      "epoch": 0.8690744920993227,
      "grad_norm": 17.485872268676758,
      "learning_rate": 9.420778750176588e-06,
      "lm_loss": 5.8326,
      "loss": 1.6989,
      "step": 385,
      "text_contrastive_loss": 0.9202,
      "train_positive_log_prob": -85.9371,
      "train_positive_token_accuracy": 0.0769,
      "train_positive_token_prob": 0.0261
    },
    {
      "contrastive_loss": 0.6277,
      "epoch": 0.871331828442438,
      "grad_norm": 17.302583694458008,
      "learning_rate": 9.41738442554156e-06,
      "lm_loss": 5.941,
      "loss": 1.6261,
      "step": 386,
      "text_contrastive_loss": 0.8084,
      "train_positive_log_prob": -89.8349,
      "train_positive_token_accuracy": 0.0785,
      "train_positive_token_prob": 0.0276
    },
    {
      "contrastive_loss": 0.6555,
      "epoch": 0.873589164785553,
      "grad_norm": 16.820449829101562,
      "learning_rate": 9.41398079949064e-06,
      "lm_loss": 5.7898,
      "loss": 1.6802,
      "step": 387,
      "text_contrastive_loss": 0.8915,
      "train_positive_log_prob": -83.5024,
      "train_positive_token_accuracy": 0.0783,
      "train_positive_token_prob": 0.0273
    },
    {
      "contrastive_loss": 0.6577,
      "epoch": 0.8758465011286681,
      "grad_norm": 16.29709243774414,
      "learning_rate": 9.41056787919063e-06,
      "lm_loss": 5.9397,
      "loss": 1.6741,
      "step": 388,
      "text_contrastive_loss": 0.8449,
      "train_positive_log_prob": -88.4692,
      "train_positive_token_accuracy": 0.074,
      "train_positive_token_prob": 0.0272
    },
    {
      "contrastive_loss": 0.5398,
      "epoch": 0.8781038374717833,
      "grad_norm": 14.739457130432129,
      "learning_rate": 9.407145671827909e-06,
      "lm_loss": 5.8599,
      "loss": 1.5127,
      "step": 389,
      "text_contrastive_loss": 0.7739,
      "train_positive_log_prob": -87.0304,
      "train_positive_token_accuracy": 0.0818,
      "train_positive_token_prob": 0.0283
    },
    {
      "contrastive_loss": 0.512,
      "epoch": 0.8803611738148984,
      "grad_norm": 16.041366577148438,
      "learning_rate": 9.403714184608411e-06,
      "lm_loss": 5.8536,
      "loss": 1.5133,
      "step": 390,
      "text_contrastive_loss": 0.8318,
      "train_positive_log_prob": -87.7598,
      "train_positive_token_accuracy": 0.0844,
      "train_positive_token_prob": 0.0282
    },
    {
      "contrastive_loss": 0.6058,
      "epoch": 0.8826185101580135,
      "grad_norm": 15.920665740966797,
      "learning_rate": 9.400273424757607e-06,
      "lm_loss": 5.7786,
      "loss": 1.6172,
      "step": 391,
      "text_contrastive_loss": 0.867,
      "train_positive_log_prob": -86.4933,
      "train_positive_token_accuracy": 0.0864,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.552,
      "epoch": 0.8848758465011287,
      "grad_norm": 15.671189308166504,
      "learning_rate": 9.396823399520495e-06,
      "lm_loss": 5.9443,
      "loss": 1.6113,
      "step": 392,
      "text_contrastive_loss": 0.9297,
      "train_positive_log_prob": -89.9789,
      "train_positive_token_accuracy": 0.08,
      "train_positive_token_prob": 0.0267
    },
    {
      "contrastive_loss": 0.6316,
      "epoch": 0.8871331828442438,
      "grad_norm": 18.04585075378418,
      "learning_rate": 9.393364116161582e-06,
      "lm_loss": 5.981,
      "loss": 1.7371,
      "step": 393,
      "text_contrastive_loss": 1.0147,
      "train_positive_log_prob": -87.4024,
      "train_positive_token_accuracy": 0.0692,
      "train_positive_token_prob": 0.0264
    },
    {
      "contrastive_loss": 0.5855,
      "epoch": 0.8893905191873589,
      "grad_norm": 16.53270149230957,
      "learning_rate": 9.38989558196487e-06,
      "lm_loss": 5.8239,
      "loss": 1.645,
      "step": 394,
      "text_contrastive_loss": 0.9542,
      "train_positive_log_prob": -85.8799,
      "train_positive_token_accuracy": 0.0787,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.6909,
      "epoch": 0.891647855530474,
      "grad_norm": 18.113792419433594,
      "learning_rate": 9.386417804233836e-06,
      "lm_loss": 5.9493,
      "loss": 1.6901,
      "step": 395,
      "text_contrastive_loss": 0.8086,
      "train_positive_log_prob": -88.4249,
      "train_positive_token_accuracy": 0.0757,
      "train_positive_token_prob": 0.0268
    },
    {
      "contrastive_loss": 0.6344,
      "epoch": 0.8939051918735892,
      "grad_norm": 16.652870178222656,
      "learning_rate": 9.382930790291426e-06,
      "lm_loss": 5.9354,
      "loss": 1.6869,
      "step": 396,
      "text_contrastive_loss": 0.9179,
      "train_positive_log_prob": -88.9619,
      "train_positive_token_accuracy": 0.0758,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.4617,
      "epoch": 0.8961625282167043,
      "grad_norm": 12.797913551330566,
      "learning_rate": 9.37943454748003e-06,
      "lm_loss": 6.0363,
      "loss": 1.4401,
      "step": 397,
      "text_contrastive_loss": 0.7496,
      "train_positive_log_prob": -89.4817,
      "train_positive_token_accuracy": 0.0746,
      "train_positive_token_prob": 0.0267
    },
    {
      "contrastive_loss": 0.5744,
      "epoch": 0.8984198645598194,
      "grad_norm": 16.787803649902344,
      "learning_rate": 9.375929083161475e-06,
      "lm_loss": 5.9057,
      "loss": 1.6225,
      "step": 398,
      "text_contrastive_loss": 0.9151,
      "train_positive_log_prob": -86.6751,
      "train_positive_token_accuracy": 0.0719,
      "train_positive_token_prob": 0.0268
    },
    {
      "contrastive_loss": 0.5262,
      "epoch": 0.9006772009029346,
      "grad_norm": 16.495084762573242,
      "learning_rate": 9.372414404717001e-06,
      "lm_loss": 5.7736,
      "loss": 1.5032,
      "step": 399,
      "text_contrastive_loss": 0.7992,
      "train_positive_log_prob": -85.8852,
      "train_positive_token_accuracy": 0.0764,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.5468,
      "epoch": 0.9029345372460497,
      "grad_norm": 16.830516815185547,
      "learning_rate": 9.36889051954725e-06,
      "lm_loss": 5.884,
      "loss": 1.5455,
      "step": 400,
      "text_contrastive_loss": 0.8206,
      "train_positive_log_prob": -85.6623,
      "train_positive_token_accuracy": 0.0778,
      "train_positive_token_prob": 0.0278
    },
    {
      "contrastive_loss": 0.6198,
      "epoch": 0.9051918735891648,
      "grad_norm": 17.817441940307617,
      "learning_rate": 9.365357435072255e-06,
      "lm_loss": 5.8643,
      "loss": 1.6149,
      "step": 401,
      "text_contrastive_loss": 0.8172,
      "train_positive_log_prob": -88.1916,
      "train_positive_token_accuracy": 0.086,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.5756,
      "epoch": 0.90744920993228,
      "grad_norm": 16.032751083374023,
      "learning_rate": 9.361815158731413e-06,
      "lm_loss": 5.9592,
      "loss": 1.6194,
      "step": 402,
      "text_contrastive_loss": 0.8957,
      "train_positive_log_prob": -88.2699,
      "train_positive_token_accuracy": 0.0752,
      "train_positive_token_prob": 0.0276
    },
    {
      "contrastive_loss": 0.6894,
      "epoch": 0.909706546275395,
      "grad_norm": 20.829038619995117,
      "learning_rate": 9.358263697983479e-06,
      "lm_loss": 5.8097,
      "loss": 1.7285,
      "step": 403,
      "text_contrastive_loss": 0.9163,
      "train_positive_log_prob": -85.4369,
      "train_positive_token_accuracy": 0.0765,
      "train_positive_token_prob": 0.0276
    },
    {
      "contrastive_loss": 0.5692,
      "epoch": 0.9119638826185101,
      "grad_norm": 17.53789520263672,
      "learning_rate": 9.354703060306546e-06,
      "lm_loss": 5.8393,
      "loss": 1.6301,
      "step": 404,
      "text_contrastive_loss": 0.954,
      "train_positive_log_prob": -84.8315,
      "train_positive_token_accuracy": 0.0742,
      "train_positive_token_prob": 0.027
    },
    {
      "contrastive_loss": 0.5173,
      "epoch": 0.9142212189616253,
      "grad_norm": 18.327421188354492,
      "learning_rate": 9.351133253198027e-06,
      "lm_loss": 5.8648,
      "loss": 1.5791,
      "step": 405,
      "text_contrastive_loss": 0.9506,
      "train_positive_log_prob": -87.812,
      "train_positive_token_accuracy": 0.0811,
      "train_positive_token_prob": 0.0276
    },
    {
      "contrastive_loss": 0.5595,
      "epoch": 0.9164785553047404,
      "grad_norm": 15.92544937133789,
      "learning_rate": 9.347554284174654e-06,
      "lm_loss": 5.8902,
      "loss": 1.6165,
      "step": 406,
      "text_contrastive_loss": 0.9359,
      "train_positive_log_prob": -86.93,
      "train_positive_token_accuracy": 0.0789,
      "train_positive_token_prob": 0.0271
    },
    {
      "contrastive_loss": 0.6278,
      "epoch": 0.9187358916478555,
      "grad_norm": 17.234172821044922,
      "learning_rate": 9.343966160772438e-06,
      "lm_loss": 5.89,
      "loss": 1.6909,
      "step": 407,
      "text_contrastive_loss": 0.9483,
      "train_positive_log_prob": -85.5923,
      "train_positive_token_accuracy": 0.0761,
      "train_positive_token_prob": 0.0271
    },
    {
      "contrastive_loss": 0.5346,
      "epoch": 0.9209932279909706,
      "grad_norm": 14.652546882629395,
      "learning_rate": 9.340368890546672e-06,
      "lm_loss": 5.9793,
      "loss": 1.5382,
      "step": 408,
      "text_contrastive_loss": 0.8115,
      "train_positive_log_prob": -89.6913,
      "train_positive_token_accuracy": 0.0745,
      "train_positive_token_prob": 0.0263
    },
    {
      "contrastive_loss": 0.4637,
      "epoch": 0.9232505643340858,
      "grad_norm": 14.887948989868164,
      "learning_rate": 9.336762481071906e-06,
      "lm_loss": 5.9219,
      "loss": 1.4281,
      "step": 409,
      "text_contrastive_loss": 0.7445,
      "train_positive_log_prob": -87.2568,
      "train_positive_token_accuracy": 0.0811,
      "train_positive_token_prob": 0.0272
    },
    {
      "contrastive_loss": 0.6062,
      "epoch": 0.9255079006772009,
      "grad_norm": 16.95090103149414,
      "learning_rate": 9.333146939941938e-06,
      "lm_loss": 6.0113,
      "loss": 1.6833,
      "step": 410,
      "text_contrastive_loss": 0.952,
      "train_positive_log_prob": -89.957,
      "train_positive_token_accuracy": 0.0807,
      "train_positive_token_prob": 0.0262
    },
    {
      "contrastive_loss": 0.598,
      "epoch": 0.927765237020316,
      "grad_norm": 14.304362297058105,
      "learning_rate": 9.329522274769791e-06,
      "lm_loss": 5.8227,
      "loss": 1.5791,
      "step": 411,
      "text_contrastive_loss": 0.7976,
      "train_positive_log_prob": -83.8585,
      "train_positive_token_accuracy": 0.0759,
      "train_positive_token_prob": 0.0276
    },
    {
      "contrastive_loss": 0.631,
      "epoch": 0.9300225733634312,
      "grad_norm": 17.765522003173828,
      "learning_rate": 9.325888493187699e-06,
      "lm_loss": 6.051,
      "loss": 1.72,
      "step": 412,
      "text_contrastive_loss": 0.9678,
      "train_positive_log_prob": -91.0685,
      "train_positive_token_accuracy": 0.0755,
      "train_positive_token_prob": 0.0265
    },
    {
      "contrastive_loss": 0.577,
      "epoch": 0.9322799097065463,
      "grad_norm": 17.135438919067383,
      "learning_rate": 9.322245602847094e-06,
      "lm_loss": 5.9059,
      "loss": 1.6223,
      "step": 413,
      "text_contrastive_loss": 0.9095,
      "train_positive_log_prob": -87.7117,
      "train_positive_token_accuracy": 0.075,
      "train_positive_token_prob": 0.0275
    },
    {
      "contrastive_loss": 0.5003,
      "epoch": 0.9345372460496614,
      "grad_norm": 14.984756469726562,
      "learning_rate": 9.31859361141859e-06,
      "lm_loss": 5.8585,
      "loss": 1.4603,
      "step": 414,
      "text_contrastive_loss": 0.7482,
      "train_positive_log_prob": -82.9794,
      "train_positive_token_accuracy": 0.0794,
      "train_positive_token_prob": 0.0281
    },
    {
      "contrastive_loss": 0.5006,
      "epoch": 0.9367945823927766,
      "grad_norm": 15.582765579223633,
      "learning_rate": 9.314932526591956e-06,
      "lm_loss": 5.9528,
      "loss": 1.4835,
      "step": 415,
      "text_contrastive_loss": 0.7753,
      "train_positive_log_prob": -89.5734,
      "train_positive_token_accuracy": 0.0732,
      "train_positive_token_prob": 0.0268
    },
    {
      "contrastive_loss": 0.6021,
      "epoch": 0.9390519187358917,
      "grad_norm": 16.197175979614258,
      "learning_rate": 9.311262356076118e-06,
      "lm_loss": 5.779,
      "loss": 1.6585,
      "step": 416,
      "text_contrastive_loss": 0.9572,
      "train_positive_log_prob": -84.5415,
      "train_positive_token_accuracy": 0.0758,
      "train_positive_token_prob": 0.0272
    },
    {
      "contrastive_loss": 0.5616,
      "epoch": 0.9413092550790068,
      "grad_norm": 17.086196899414062,
      "learning_rate": 9.30758310759913e-06,
      "lm_loss": 5.899,
      "loss": 1.5774,
      "step": 417,
      "text_contrastive_loss": 0.8519,
      "train_positive_log_prob": -89.5021,
      "train_positive_token_accuracy": 0.0772,
      "train_positive_token_prob": 0.0279
    },
    {
      "contrastive_loss": 0.5974,
      "epoch": 0.9435665914221218,
      "grad_norm": 17.816383361816406,
      "learning_rate": 9.303894788908158e-06,
      "lm_loss": 5.8604,
      "loss": 1.6101,
      "step": 418,
      "text_contrastive_loss": 0.8533,
      "train_positive_log_prob": -86.6261,
      "train_positive_token_accuracy": 0.0748,
      "train_positive_token_prob": 0.027
    },
    {
      "contrastive_loss": 0.59,
      "epoch": 0.945823927765237,
      "grad_norm": 17.613948822021484,
      "learning_rate": 9.300197407769472e-06,
      "lm_loss": 5.7805,
      "loss": 1.6857,
      "step": 419,
      "text_contrastive_loss": 1.0353,
      "train_positive_log_prob": -84.2001,
      "train_positive_token_accuracy": 0.0764,
      "train_positive_token_prob": 0.0287
    },
    {
      "contrastive_loss": 0.4702,
      "epoch": 0.9480812641083521,
      "grad_norm": 16.38786506652832,
      "learning_rate": 9.296490971968416e-06,
      "lm_loss": 5.8173,
      "loss": 1.4483,
      "step": 420,
      "text_contrastive_loss": 0.7927,
      "train_positive_log_prob": -84.5788,
      "train_positive_token_accuracy": 0.0795,
      "train_positive_token_prob": 0.0281
    },
    {
      "contrastive_loss": 0.5483,
      "epoch": 0.9503386004514672,
      "grad_norm": 18.509519577026367,
      "learning_rate": 9.292775489309409e-06,
      "lm_loss": 5.8759,
      "loss": 1.5136,
      "step": 421,
      "text_contrastive_loss": 0.7555,
      "train_positive_log_prob": -88.1345,
      "train_positive_token_accuracy": 0.074,
      "train_positive_token_prob": 0.028
    },
    {
      "contrastive_loss": 0.4699,
      "epoch": 0.9525959367945824,
      "grad_norm": 16.658477783203125,
      "learning_rate": 9.289050967615914e-06,
      "lm_loss": 5.9344,
      "loss": 1.4823,
      "step": 422,
      "text_contrastive_loss": 0.838,
      "train_positive_log_prob": -89.5952,
      "train_positive_token_accuracy": 0.0766,
      "train_positive_token_prob": 0.0278
    },
    {
      "contrastive_loss": 0.5532,
      "epoch": 0.9548532731376975,
      "grad_norm": 15.291207313537598,
      "learning_rate": 9.285317414730427e-06,
      "lm_loss": 6.0415,
      "loss": 1.6304,
      "step": 423,
      "text_contrastive_loss": 0.9461,
      "train_positive_log_prob": -90.5035,
      "train_positive_token_accuracy": 0.073,
      "train_positive_token_prob": 0.0259
    },
    {
      "contrastive_loss": 0.5333,
      "epoch": 0.9571106094808126,
      "grad_norm": 15.894474029541016,
      "learning_rate": 9.281574838514464e-06,
      "lm_loss": 5.9181,
      "loss": 1.5743,
      "step": 424,
      "text_contrastive_loss": 0.8984,
      "train_positive_log_prob": -89.7682,
      "train_positive_token_accuracy": 0.0873,
      "train_positive_token_prob": 0.0287
    },
    {
      "contrastive_loss": 0.5708,
      "epoch": 0.9593679458239278,
      "grad_norm": 16.08817481994629,
      "learning_rate": 9.277823246848537e-06,
      "lm_loss": 6.0141,
      "loss": 1.6473,
      "step": 425,
      "text_contrastive_loss": 0.9503,
      "train_positive_log_prob": -86.6867,
      "train_positive_token_accuracy": 0.0808,
      "train_positive_token_prob": 0.028
    },
    {
      "contrastive_loss": 0.612,
      "epoch": 0.9616252821670429,
      "grad_norm": 15.471580505371094,
      "learning_rate": 9.274062647632144e-06,
      "lm_loss": 5.8701,
      "loss": 1.6208,
      "step": 426,
      "text_contrastive_loss": 0.8436,
      "train_positive_log_prob": -87.7061,
      "train_positive_token_accuracy": 0.0805,
      "train_positive_token_prob": 0.0282
    },
    {
      "contrastive_loss": 0.6127,
      "epoch": 0.963882618510158,
      "grad_norm": 17.222087860107422,
      "learning_rate": 9.270293048783747e-06,
      "lm_loss": 5.9028,
      "loss": 1.6201,
      "step": 427,
      "text_contrastive_loss": 0.8343,
      "train_positive_log_prob": -86.9072,
      "train_positive_token_accuracy": 0.074,
      "train_positive_token_prob": 0.0275
    },
    {
      "contrastive_loss": 0.5228,
      "epoch": 0.9661399548532731,
      "grad_norm": 17.653852462768555,
      "learning_rate": 9.266514458240762e-06,
      "lm_loss": 5.8615,
      "loss": 1.4908,
      "step": 428,
      "text_contrastive_loss": 0.7636,
      "train_positive_log_prob": -86.3916,
      "train_positive_token_accuracy": 0.075,
      "train_positive_token_prob": 0.0279
    },
    {
      "contrastive_loss": 0.5396,
      "epoch": 0.9683972911963883,
      "grad_norm": 16.23486328125,
      "learning_rate": 9.262726883959535e-06,
      "lm_loss": 5.8543,
      "loss": 1.5029,
      "step": 429,
      "text_contrastive_loss": 0.7557,
      "train_positive_log_prob": -88.9117,
      "train_positive_token_accuracy": 0.0777,
      "train_positive_token_prob": 0.0286
    },
    {
      "contrastive_loss": 0.5503,
      "epoch": 0.9706546275395034,
      "grad_norm": 15.685914039611816,
      "learning_rate": 9.258930333915325e-06,
      "lm_loss": 5.8241,
      "loss": 1.5681,
      "step": 430,
      "text_contrastive_loss": 0.8707,
      "train_positive_log_prob": -86.4459,
      "train_positive_token_accuracy": 0.0861,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.4516,
      "epoch": 0.9729119638826185,
      "grad_norm": 15.291813850402832,
      "learning_rate": 9.2551248161023e-06,
      "lm_loss": 5.8405,
      "loss": 1.4498,
      "step": 431,
      "text_contrastive_loss": 0.8281,
      "train_positive_log_prob": -86.9725,
      "train_positive_token_accuracy": 0.075,
      "train_positive_token_prob": 0.0286
    },
    {
      "contrastive_loss": 0.5189,
      "epoch": 0.9751693002257337,
      "grad_norm": 14.862266540527344,
      "learning_rate": 9.251310338533504e-06,
      "lm_loss": 5.83,
      "loss": 1.5722,
      "step": 432,
      "text_contrastive_loss": 0.9407,
      "train_positive_log_prob": -85.3116,
      "train_positive_token_accuracy": 0.0726,
      "train_positive_token_prob": 0.0267
    },
    {
      "contrastive_loss": 0.5968,
      "epoch": 0.9774266365688488,
      "grad_norm": 16.92352867126465,
      "learning_rate": 9.247486909240849e-06,
      "lm_loss": 6.0227,
      "loss": 1.6701,
      "step": 433,
      "text_contrastive_loss": 0.942,
      "train_positive_log_prob": -86.9785,
      "train_positive_token_accuracy": 0.079,
      "train_positive_token_prob": 0.0273
    },
    {
      "contrastive_loss": 0.6644,
      "epoch": 0.9796839729119639,
      "grad_norm": 19.324182510375977,
      "learning_rate": 9.243654536275095e-06,
      "lm_loss": 5.8235,
      "loss": 1.655,
      "step": 434,
      "text_contrastive_loss": 0.8166,
      "train_positive_log_prob": -87.3978,
      "train_positive_token_accuracy": 0.0898,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.6476,
      "epoch": 0.981941309255079,
      "grad_norm": 16.82012176513672,
      "learning_rate": 9.23981322770584e-06,
      "lm_loss": 5.7523,
      "loss": 1.6639,
      "step": 435,
      "text_contrastive_loss": 0.882,
      "train_positive_log_prob": -85.5649,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0287
    },
    {
      "contrastive_loss": 0.6013,
      "epoch": 0.9841986455981941,
      "grad_norm": 16.209348678588867,
      "learning_rate": 9.235962991621484e-06,
      "lm_loss": 5.7624,
      "loss": 1.7297,
      "step": 436,
      "text_contrastive_loss": 1.1045,
      "train_positive_log_prob": -86.0311,
      "train_positive_token_accuracy": 0.0924,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.6078,
      "epoch": 0.9864559819413092,
      "grad_norm": 15.903900146484375,
      "learning_rate": 9.232103836129239e-06,
      "lm_loss": 5.9091,
      "loss": 1.6515,
      "step": 437,
      "text_contrastive_loss": 0.9056,
      "train_positive_log_prob": -89.1635,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.0289
    },
    {
      "contrastive_loss": 0.5193,
      "epoch": 0.9887133182844243,
      "grad_norm": 14.519559860229492,
      "learning_rate": 9.22823576935509e-06,
      "lm_loss": 5.864,
      "loss": 1.5648,
      "step": 438,
      "text_contrastive_loss": 0.9181,
      "train_positive_log_prob": -87.7092,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.0279
    },
    {
      "contrastive_loss": 0.6269,
      "epoch": 0.9909706546275395,
      "grad_norm": 16.698381423950195,
      "learning_rate": 9.224358799443791e-06,
      "lm_loss": 5.8001,
      "loss": 1.6402,
      "step": 439,
      "text_contrastive_loss": 0.8666,
      "train_positive_log_prob": -84.8054,
      "train_positive_token_accuracy": 0.0801,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.5207,
      "epoch": 0.9932279909706546,
      "grad_norm": 14.701775550842285,
      "learning_rate": 9.220472934558838e-06,
      "lm_loss": 5.937,
      "loss": 1.5043,
      "step": 440,
      "text_contrastive_loss": 0.7798,
      "train_positive_log_prob": -88.6172,
      "train_positive_token_accuracy": 0.0701,
      "train_positive_token_prob": 0.0278
    },
    {
      "contrastive_loss": 0.5213,
      "epoch": 0.9954853273137697,
      "grad_norm": 15.258160591125488,
      "learning_rate": 9.216578182882459e-06,
      "lm_loss": 5.7891,
      "loss": 1.5187,
      "step": 441,
      "text_contrastive_loss": 0.8369,
      "train_positive_log_prob": -86.4508,
      "train_positive_token_accuracy": 0.0779,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.6021,
      "epoch": 0.9977426636568849,
      "grad_norm": 16.111305236816406,
      "learning_rate": 9.212674552615594e-06,
      "lm_loss": 5.9081,
      "loss": 1.6279,
      "step": 442,
      "text_contrastive_loss": 0.8701,
      "train_positive_log_prob": -86.1487,
      "train_positive_token_accuracy": 0.0721,
      "train_positive_token_prob": 0.0283
    },
    {
      "contrastive_loss": 0.4096,
      "epoch": 1.0,
      "grad_norm": 20.45907974243164,
      "learning_rate": 9.208762051977879e-06,
      "lm_loss": 5.918,
      "loss": 1.349,
      "step": 443,
      "text_contrastive_loss": 0.6952,
      "train_positive_log_prob": -89.6557,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.0273
    },
    {
      "contrastive_loss": 0.4947,
      "epoch": 1.002257336343115,
      "grad_norm": 12.947877883911133,
      "learning_rate": 9.204840689207626e-06,
      "lm_loss": 5.8088,
      "loss": 1.5178,
      "step": 444,
      "text_contrastive_loss": 0.8845,
      "train_positive_log_prob": -84.2974,
      "train_positive_token_accuracy": 0.074,
      "train_positive_token_prob": 0.0277
    },
    {
      "contrastive_loss": 0.6478,
      "epoch": 1.0045146726862302,
      "grad_norm": 14.91596794128418,
      "learning_rate": 9.20091047256181e-06,
      "lm_loss": 5.7999,
      "loss": 1.7457,
      "step": 445,
      "text_contrastive_loss": 1.0357,
      "train_positive_log_prob": -86.7551,
      "train_positive_token_accuracy": 0.0765,
      "train_positive_token_prob": 0.0269
    },
    {
      "contrastive_loss": 0.5065,
      "epoch": 1.0067720090293453,
      "grad_norm": 14.268144607543945,
      "learning_rate": 9.196971410316047e-06,
      "lm_loss": 5.8751,
      "loss": 1.5296,
      "step": 446,
      "text_contrastive_loss": 0.871,
      "train_positive_log_prob": -88.0539,
      "train_positive_token_accuracy": 0.0831,
      "train_positive_token_prob": 0.0279
    },
    {
      "contrastive_loss": 0.4398,
      "epoch": 1.0090293453724606,
      "grad_norm": 12.608469009399414,
      "learning_rate": 9.193023510764578e-06,
      "lm_loss": 5.793,
      "loss": 1.396,
      "step": 447,
      "text_contrastive_loss": 0.7539,
      "train_positive_log_prob": -85.8542,
      "train_positive_token_accuracy": 0.0799,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.5217,
      "epoch": 1.0112866817155757,
      "grad_norm": 15.451228141784668,
      "learning_rate": 9.189066782220253e-06,
      "lm_loss": 5.9104,
      "loss": 1.5425,
      "step": 448,
      "text_contrastive_loss": 0.8596,
      "train_positive_log_prob": -86.4638,
      "train_positive_token_accuracy": 0.0769,
      "train_positive_token_prob": 0.0278
    },
    {
      "contrastive_loss": 0.5485,
      "epoch": 1.0135440180586908,
      "grad_norm": 15.606856346130371,
      "learning_rate": 9.185101233014516e-06,
      "lm_loss": 5.9077,
      "loss": 1.6274,
      "step": 449,
      "text_contrastive_loss": 0.9763,
      "train_positive_log_prob": -88.8938,
      "train_positive_token_accuracy": 0.0846,
      "train_positive_token_prob": 0.0278
    },
    {
      "contrastive_loss": 0.519,
      "epoch": 1.0158013544018059,
      "grad_norm": 13.44371223449707,
      "learning_rate": 9.181126871497378e-06,
      "lm_loss": 5.8679,
      "loss": 1.5475,
      "step": 450,
      "text_contrastive_loss": 0.8833,
      "train_positive_log_prob": -88.6146,
      "train_positive_token_accuracy": 0.0706,
      "train_positive_token_prob": 0.0268
    },
    {
      "contrastive_loss": 0.4338,
      "epoch": 1.018058690744921,
      "grad_norm": 12.92390251159668,
      "learning_rate": 9.177143706037411e-06,
      "lm_loss": 5.8065,
      "loss": 1.3952,
      "step": 451,
      "text_contrastive_loss": 0.7617,
      "train_positive_log_prob": -86.5998,
      "train_positive_token_accuracy": 0.0822,
      "train_positive_token_prob": 0.0278
    },
    {
      "contrastive_loss": 0.4333,
      "epoch": 1.020316027088036,
      "grad_norm": 14.111056327819824,
      "learning_rate": 9.173151745021722e-06,
      "lm_loss": 5.7691,
      "loss": 1.4061,
      "step": 452,
      "text_contrastive_loss": 0.7918,
      "train_positive_log_prob": -84.734,
      "train_positive_token_accuracy": 0.083,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.3544,
      "epoch": 1.0225733634311513,
      "grad_norm": 12.541257858276367,
      "learning_rate": 9.169150996855939e-06,
      "lm_loss": 5.8573,
      "loss": 1.3147,
      "step": 453,
      "text_contrastive_loss": 0.7492,
      "train_positive_log_prob": -87.4012,
      "train_positive_token_accuracy": 0.0804,
      "train_positive_token_prob": 0.027
    },
    {
      "contrastive_loss": 0.5402,
      "epoch": 1.0248306997742664,
      "grad_norm": 14.790188789367676,
      "learning_rate": 9.16514146996419e-06,
      "lm_loss": 5.8104,
      "loss": 1.5138,
      "step": 454,
      "text_contrastive_loss": 0.7851,
      "train_positive_log_prob": -87.9943,
      "train_positive_token_accuracy": 0.0779,
      "train_positive_token_prob": 0.027
    },
    {
      "contrastive_loss": 0.4695,
      "epoch": 1.0270880361173815,
      "grad_norm": 15.407713890075684,
      "learning_rate": 9.161123172789091e-06,
      "lm_loss": 5.758,
      "loss": 1.4119,
      "step": 455,
      "text_contrastive_loss": 0.7332,
      "train_positive_log_prob": -85.8075,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.0287
    },
    {
      "contrastive_loss": 0.5372,
      "epoch": 1.0293453724604966,
      "grad_norm": 16.16050910949707,
      "learning_rate": 9.157096113791727e-06,
      "lm_loss": 5.8291,
      "loss": 1.5989,
      "step": 456,
      "text_contrastive_loss": 0.9578,
      "train_positive_log_prob": -85.2038,
      "train_positive_token_accuracy": 0.0786,
      "train_positive_token_prob": 0.0279
    },
    {
      "contrastive_loss": 0.496,
      "epoch": 1.0316027088036117,
      "grad_norm": 14.388401985168457,
      "learning_rate": 9.153060301451629e-06,
      "lm_loss": 5.9564,
      "loss": 1.5376,
      "step": 457,
      "text_contrastive_loss": 0.8919,
      "train_positive_log_prob": -86.6673,
      "train_positive_token_accuracy": 0.0765,
      "train_positive_token_prob": 0.0272
    },
    {
      "contrastive_loss": 0.5267,
      "epoch": 1.0338600451467268,
      "grad_norm": 16.01051139831543,
      "learning_rate": 9.149015744266759e-06,
      "lm_loss": 5.729,
      "loss": 1.6183,
      "step": 458,
      "text_contrastive_loss": 1.0374,
      "train_positive_log_prob": -85.9048,
      "train_positive_token_accuracy": 0.08,
      "train_positive_token_prob": 0.028
    },
    {
      "contrastive_loss": 0.4585,
      "epoch": 1.036117381489842,
      "grad_norm": 17.170934677124023,
      "learning_rate": 9.144962450753491e-06,
      "lm_loss": 5.9998,
      "loss": 1.4977,
      "step": 459,
      "text_contrastive_loss": 0.8784,
      "train_positive_log_prob": -88.2386,
      "train_positive_token_accuracy": 0.0749,
      "train_positive_token_prob": 0.0276
    },
    {
      "contrastive_loss": 0.5279,
      "epoch": 1.0383747178329572,
      "grad_norm": 15.702168464660645,
      "learning_rate": 9.140900429446601e-06,
      "lm_loss": 5.8052,
      "loss": 1.5142,
      "step": 460,
      "text_contrastive_loss": 0.8115,
      "train_positive_log_prob": -83.5433,
      "train_positive_token_accuracy": 0.0771,
      "train_positive_token_prob": 0.0279
    },
    {
      "contrastive_loss": 0.564,
      "epoch": 1.0406320541760723,
      "grad_norm": 13.882556915283203,
      "learning_rate": 9.136829688899236e-06,
      "lm_loss": 5.8419,
      "loss": 1.6218,
      "step": 461,
      "text_contrastive_loss": 0.9472,
      "train_positive_log_prob": -86.0309,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.027
    },
    {
      "contrastive_loss": 0.396,
      "epoch": 1.0428893905191874,
      "grad_norm": 13.365018844604492,
      "learning_rate": 9.132750237682907e-06,
      "lm_loss": 5.8004,
      "loss": 1.3386,
      "step": 462,
      "text_contrastive_loss": 0.7252,
      "train_positive_log_prob": -86.9392,
      "train_positive_token_accuracy": 0.087,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.5425,
      "epoch": 1.0451467268623025,
      "grad_norm": 16.31186294555664,
      "learning_rate": 9.128662084387462e-06,
      "lm_loss": 5.8817,
      "loss": 1.524,
      "step": 463,
      "text_contrastive_loss": 0.7866,
      "train_positive_log_prob": -85.8379,
      "train_positive_token_accuracy": 0.0782,
      "train_positive_token_prob": 0.0276
    },
    {
      "contrastive_loss": 0.582,
      "epoch": 1.0474040632054176,
      "grad_norm": 14.941473960876465,
      "learning_rate": 9.12456523762108e-06,
      "lm_loss": 5.7214,
      "loss": 1.6567,
      "step": 464,
      "text_contrastive_loss": 1.0052,
      "train_positive_log_prob": -82.0933,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.4046,
      "epoch": 1.0496613995485327,
      "grad_norm": 12.706572532653809,
      "learning_rate": 9.120459706010233e-06,
      "lm_loss": 5.8326,
      "loss": 1.3838,
      "step": 465,
      "text_contrastive_loss": 0.7918,
      "train_positive_log_prob": -86.5063,
      "train_positive_token_accuracy": 0.0764,
      "train_positive_token_prob": 0.0289
    },
    {
      "contrastive_loss": 0.3926,
      "epoch": 1.0519187358916477,
      "grad_norm": 15.198698997497559,
      "learning_rate": 9.116345498199693e-06,
      "lm_loss": 5.7933,
      "loss": 1.3493,
      "step": 466,
      "text_contrastive_loss": 0.7547,
      "train_positive_log_prob": -84.9983,
      "train_positive_token_accuracy": 0.0873,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.4502,
      "epoch": 1.054176072234763,
      "grad_norm": 13.700481414794922,
      "learning_rate": 9.112222622852494e-06,
      "lm_loss": 5.805,
      "loss": 1.4557,
      "step": 467,
      "text_contrastive_loss": 0.8499,
      "train_positive_log_prob": -85.4643,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0285
    },
    {
      "contrastive_loss": 0.4237,
      "epoch": 1.0564334085778782,
      "grad_norm": 13.614365577697754,
      "learning_rate": 9.108091088649922e-06,
      "lm_loss": 5.7999,
      "loss": 1.4297,
      "step": 468,
      "text_contrastive_loss": 0.852,
      "train_positive_log_prob": -86.592,
      "train_positive_token_accuracy": 0.0761,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.5719,
      "epoch": 1.0586907449209932,
      "grad_norm": 16.305103302001953,
      "learning_rate": 9.103950904291496e-06,
      "lm_loss": 5.6919,
      "loss": 1.5975,
      "step": 469,
      "text_contrastive_loss": 0.913,
      "train_positive_log_prob": -84.6675,
      "train_positive_token_accuracy": 0.0775,
      "train_positive_token_prob": 0.0283
    },
    {
      "contrastive_loss": 0.4606,
      "epoch": 1.0609480812641083,
      "grad_norm": 13.533616065979004,
      "learning_rate": 9.099802078494947e-06,
      "lm_loss": 5.6829,
      "loss": 1.479,
      "step": 470,
      "text_contrastive_loss": 0.9002,
      "train_positive_log_prob": -82.9966,
      "train_positive_token_accuracy": 0.0792,
      "train_positive_token_prob": 0.0287
    },
    {
      "contrastive_loss": 0.5167,
      "epoch": 1.0632054176072234,
      "grad_norm": 14.10770320892334,
      "learning_rate": 9.095644619996206e-06,
      "lm_loss": 5.9261,
      "loss": 1.6337,
      "step": 471,
      "text_contrastive_loss": 1.0488,
      "train_positive_log_prob": -87.8771,
      "train_positive_token_accuracy": 0.0685,
      "train_positive_token_prob": 0.0262
    },
    {
      "contrastive_loss": 0.4674,
      "epoch": 1.0654627539503385,
      "grad_norm": 14.255465507507324,
      "learning_rate": 9.09147853754938e-06,
      "lm_loss": 5.81,
      "loss": 1.4521,
      "step": 472,
      "text_contrastive_loss": 0.8076,
      "train_positive_log_prob": -85.3749,
      "train_positive_token_accuracy": 0.0803,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.5032,
      "epoch": 1.0677200902934538,
      "grad_norm": 15.463451385498047,
      "learning_rate": 9.087303839926727e-06,
      "lm_loss": 5.8396,
      "loss": 1.5136,
      "step": 473,
      "text_contrastive_loss": 0.8528,
      "train_positive_log_prob": -86.4972,
      "train_positive_token_accuracy": 0.0754,
      "train_positive_token_prob": 0.028
    },
    {
      "contrastive_loss": 0.521,
      "epoch": 1.069977426636569,
      "grad_norm": 14.949661254882812,
      "learning_rate": 9.08312053591866e-06,
      "lm_loss": 5.8701,
      "loss": 1.5653,
      "step": 474,
      "text_contrastive_loss": 0.9145,
      "train_positive_log_prob": -85.7861,
      "train_positive_token_accuracy": 0.073,
      "train_positive_token_prob": 0.027
    },
    {
      "contrastive_loss": 0.5747,
      "epoch": 1.072234762979684,
      "grad_norm": 14.694694519042969,
      "learning_rate": 9.0789286343337e-06,
      "lm_loss": 5.8051,
      "loss": 1.5711,
      "step": 475,
      "text_contrastive_loss": 0.8318,
      "train_positive_log_prob": -85.7406,
      "train_positive_token_accuracy": 0.0759,
      "train_positive_token_prob": 0.0277
    },
    {
      "contrastive_loss": 0.4812,
      "epoch": 1.074492099322799,
      "grad_norm": 16.750852584838867,
      "learning_rate": 9.07472814399848e-06,
      "lm_loss": 5.7319,
      "loss": 1.5312,
      "step": 476,
      "text_contrastive_loss": 0.9536,
      "train_positive_log_prob": -86.2696,
      "train_positive_token_accuracy": 0.0831,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.5897,
      "epoch": 1.0767494356659142,
      "grad_norm": 15.837838172912598,
      "learning_rate": 9.070519073757717e-06,
      "lm_loss": 5.704,
      "loss": 1.5723,
      "step": 477,
      "text_contrastive_loss": 0.8244,
      "train_positive_log_prob": -84.8902,
      "train_positive_token_accuracy": 0.0837,
      "train_positive_token_prob": 0.0291
    },
    {
      "contrastive_loss": 0.567,
      "epoch": 1.0790067720090293,
      "grad_norm": 14.569334030151367,
      "learning_rate": 9.06630143247419e-06,
      "lm_loss": 5.8504,
      "loss": 1.6406,
      "step": 478,
      "text_contrastive_loss": 0.9772,
      "train_positive_log_prob": -83.4529,
      "train_positive_token_accuracy": 0.0761,
      "train_positive_token_prob": 0.0278
    },
    {
      "contrastive_loss": 0.4571,
      "epoch": 1.0812641083521444,
      "grad_norm": 14.058526992797852,
      "learning_rate": 9.062075229028728e-06,
      "lm_loss": 5.8129,
      "loss": 1.4873,
      "step": 479,
      "text_contrastive_loss": 0.8978,
      "train_positive_log_prob": -86.3999,
      "train_positive_token_accuracy": 0.0767,
      "train_positive_token_prob": 0.0278
    },
    {
      "contrastive_loss": 0.5346,
      "epoch": 1.0835214446952597,
      "grad_norm": 14.769458770751953,
      "learning_rate": 9.057840472320192e-06,
      "lm_loss": 5.7671,
      "loss": 1.477,
      "step": 480,
      "text_contrastive_loss": 0.7315,
      "train_positive_log_prob": -85.201,
      "train_positive_token_accuracy": 0.0799,
      "train_positive_token_prob": 0.0274
    },
    {
      "contrastive_loss": 0.4574,
      "epoch": 1.0857787810383748,
      "grad_norm": 13.187503814697266,
      "learning_rate": 9.053597171265447e-06,
      "lm_loss": 5.8272,
      "loss": 1.4742,
      "step": 481,
      "text_contrastive_loss": 0.8683,
      "train_positive_log_prob": -88.2313,
      "train_positive_token_accuracy": 0.0826,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.447,
      "epoch": 1.0880361173814899,
      "grad_norm": 14.901713371276855,
      "learning_rate": 9.04934533479935e-06,
      "lm_loss": 5.8659,
      "loss": 1.4509,
      "step": 482,
      "text_contrastive_loss": 0.8345,
      "train_positive_log_prob": -86.9744,
      "train_positive_token_accuracy": 0.0811,
      "train_positive_token_prob": 0.0271
    },
    {
      "contrastive_loss": 0.4869,
      "epoch": 1.090293453724605,
      "grad_norm": 16.12311553955078,
      "learning_rate": 9.045084971874738e-06,
      "lm_loss": 5.8612,
      "loss": 1.5182,
      "step": 483,
      "text_contrastive_loss": 0.8905,
      "train_positive_log_prob": -89.0412,
      "train_positive_token_accuracy": 0.0825,
      "train_positive_token_prob": 0.028
    },
    {
      "contrastive_loss": 0.4039,
      "epoch": 1.09255079006772,
      "grad_norm": 15.610627174377441,
      "learning_rate": 9.040816091462393e-06,
      "lm_loss": 5.8886,
      "loss": 1.394,
      "step": 484,
      "text_contrastive_loss": 0.8024,
      "train_positive_log_prob": -87.6831,
      "train_positive_token_accuracy": 0.086,
      "train_positive_token_prob": 0.0289
    },
    {
      "contrastive_loss": 0.5351,
      "epoch": 1.0948081264108351,
      "grad_norm": 14.818351745605469,
      "learning_rate": 9.036538702551037e-06,
      "lm_loss": 5.7065,
      "loss": 1.5589,
      "step": 485,
      "text_contrastive_loss": 0.9065,
      "train_positive_log_prob": -82.4999,
      "train_positive_token_accuracy": 0.0851,
      "train_positive_token_prob": 0.0285
    },
    {
      "contrastive_loss": 0.4522,
      "epoch": 1.0970654627539504,
      "grad_norm": 13.92127513885498,
      "learning_rate": 9.032252814147302e-06,
      "lm_loss": 5.8093,
      "loss": 1.4679,
      "step": 486,
      "text_contrastive_loss": 0.8695,
      "train_positive_log_prob": -85.7787,
      "train_positive_token_accuracy": 0.0757,
      "train_positive_token_prob": 0.0279
    },
    {
      "contrastive_loss": 0.4749,
      "epoch": 1.0993227990970655,
      "grad_norm": 15.42867374420166,
      "learning_rate": 9.027958435275726e-06,
      "lm_loss": 5.8436,
      "loss": 1.4605,
      "step": 487,
      "text_contrastive_loss": 0.8026,
      "train_positive_log_prob": -87.102,
      "train_positive_token_accuracy": 0.0765,
      "train_positive_token_prob": 0.0275
    },
    {
      "contrastive_loss": 0.4627,
      "epoch": 1.1015801354401806,
      "grad_norm": 14.040528297424316,
      "learning_rate": 9.023655574978716e-06,
      "lm_loss": 5.7913,
      "loss": 1.5035,
      "step": 488,
      "text_contrastive_loss": 0.9234,
      "train_positive_log_prob": -88.9274,
      "train_positive_token_accuracy": 0.0697,
      "train_positive_token_prob": 0.0279
    },
    {
      "contrastive_loss": 0.3857,
      "epoch": 1.1038374717832957,
      "grad_norm": 13.681938171386719,
      "learning_rate": 9.019344242316542e-06,
      "lm_loss": 5.7986,
      "loss": 1.408,
      "step": 489,
      "text_contrastive_loss": 0.8849,
      "train_positive_log_prob": -86.6199,
      "train_positive_token_accuracy": 0.0779,
      "train_positive_token_prob": 0.0281
    },
    {
      "contrastive_loss": 0.4946,
      "epoch": 1.1060948081264108,
      "grad_norm": 16.23743438720703,
      "learning_rate": 9.015024446367315e-06,
      "lm_loss": 5.7592,
      "loss": 1.4949,
      "step": 490,
      "text_contrastive_loss": 0.8488,
      "train_positive_log_prob": -84.2068,
      "train_positive_token_accuracy": 0.0836,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.4335,
      "epoch": 1.108352144469526,
      "grad_norm": 15.248078346252441,
      "learning_rate": 9.010696196226963e-06,
      "lm_loss": 5.7904,
      "loss": 1.5199,
      "step": 491,
      "text_contrastive_loss": 1.0148,
      "train_positive_log_prob": -82.6063,
      "train_positive_token_accuracy": 0.0764,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.4875,
      "epoch": 1.110609480812641,
      "grad_norm": 15.537084579467773,
      "learning_rate": 9.00635950100922e-06,
      "lm_loss": 5.7493,
      "loss": 1.5141,
      "step": 492,
      "text_contrastive_loss": 0.9034,
      "train_positive_log_prob": -88.9566,
      "train_positive_token_accuracy": 0.079,
      "train_positive_token_prob": 0.0288
    },
    {
      "contrastive_loss": 0.5147,
      "epoch": 1.1128668171557563,
      "grad_norm": 15.767317771911621,
      "learning_rate": 9.002014369845592e-06,
      "lm_loss": 5.8588,
      "loss": 1.5701,
      "step": 493,
      "text_contrastive_loss": 0.9391,
      "train_positive_log_prob": -86.4833,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.0288
    },
    {
      "contrastive_loss": 0.5019,
      "epoch": 1.1151241534988714,
      "grad_norm": 15.360732078552246,
      "learning_rate": 8.997660811885367e-06,
      "lm_loss": 5.9284,
      "loss": 1.5486,
      "step": 494,
      "text_contrastive_loss": 0.9077,
      "train_positive_log_prob": -84.8963,
      "train_positive_token_accuracy": 0.0747,
      "train_positive_token_prob": 0.0277
    },
    {
      "contrastive_loss": 0.5222,
      "epoch": 1.1173814898419865,
      "grad_norm": 14.968240737915039,
      "learning_rate": 8.993298836295556e-06,
      "lm_loss": 5.8004,
      "loss": 1.5103,
      "step": 495,
      "text_contrastive_loss": 0.8162,
      "train_positive_log_prob": -84.1571,
      "train_positive_token_accuracy": 0.0732,
      "train_positive_token_prob": 0.0283
    },
    {
      "contrastive_loss": 0.4355,
      "epoch": 1.1196388261851016,
      "grad_norm": 14.490927696228027,
      "learning_rate": 8.988928452260909e-06,
      "lm_loss": 5.8659,
      "loss": 1.3729,
      "step": 496,
      "text_contrastive_loss": 0.7017,
      "train_positive_log_prob": -88.259,
      "train_positive_token_accuracy": 0.0792,
      "train_positive_token_prob": 0.0289
    },
    {
      "contrastive_loss": 0.4653,
      "epoch": 1.1218961625282167,
      "grad_norm": 15.169868469238281,
      "learning_rate": 8.984549668983875e-06,
      "lm_loss": 5.8415,
      "loss": 1.4662,
      "step": 497,
      "text_contrastive_loss": 0.8333,
      "train_positive_log_prob": -87.096,
      "train_positive_token_accuracy": 0.0825,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.4849,
      "epoch": 1.1241534988713318,
      "grad_norm": 16.12611961364746,
      "learning_rate": 8.980162495684587e-06,
      "lm_loss": 5.6943,
      "loss": 1.4348,
      "step": 498,
      "text_contrastive_loss": 0.761,
      "train_positive_log_prob": -84.3932,
      "train_positive_token_accuracy": 0.0817,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.4739,
      "epoch": 1.1264108352144468,
      "grad_norm": 14.112802505493164,
      "learning_rate": 8.975766941600852e-06,
      "lm_loss": 5.6894,
      "loss": 1.4551,
      "step": 499,
      "text_contrastive_loss": 0.8245,
      "train_positive_log_prob": -81.3625,
      "train_positive_token_accuracy": 0.0817,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.5062,
      "epoch": 1.1286681715575622,
      "grad_norm": 15.006860733032227,
      "learning_rate": 8.971363015988115e-06,
      "lm_loss": 5.7202,
      "loss": 1.4517,
      "step": 500,
      "text_contrastive_loss": 0.7469,
      "train_positive_log_prob": -86.2701,
      "train_positive_token_accuracy": 0.079,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.4546,
      "epoch": 1.1309255079006773,
      "grad_norm": 14.931099891662598,
      "learning_rate": 8.966950728119453e-06,
      "lm_loss": 5.7483,
      "loss": 1.4561,
      "step": 501,
      "text_contrastive_loss": 0.8532,
      "train_positive_log_prob": -86.7438,
      "train_positive_token_accuracy": 0.0779,
      "train_positive_token_prob": 0.0279
    },
    {
      "contrastive_loss": 0.5577,
      "epoch": 1.1331828442437923,
      "grad_norm": 14.916293144226074,
      "learning_rate": 8.962530087285552e-06,
      "lm_loss": 5.7719,
      "loss": 1.5366,
      "step": 502,
      "text_contrastive_loss": 0.8034,
      "train_positive_log_prob": -84.6355,
      "train_positive_token_accuracy": 0.077,
      "train_positive_token_prob": 0.0285
    },
    {
      "contrastive_loss": 0.3864,
      "epoch": 1.1354401805869074,
      "grad_norm": 12.708629608154297,
      "learning_rate": 8.958101102794686e-06,
      "lm_loss": 5.7789,
      "loss": 1.4053,
      "step": 503,
      "text_contrastive_loss": 0.8821,
      "train_positive_log_prob": -86.2771,
      "train_positive_token_accuracy": 0.0759,
      "train_positive_token_prob": 0.0291
    },
    {
      "contrastive_loss": 0.5313,
      "epoch": 1.1376975169300225,
      "grad_norm": 15.195137023925781,
      "learning_rate": 8.953663783972692e-06,
      "lm_loss": 5.752,
      "loss": 1.5259,
      "step": 504,
      "text_contrastive_loss": 0.8388,
      "train_positive_log_prob": -86.0143,
      "train_positive_token_accuracy": 0.0789,
      "train_positive_token_prob": 0.0279
    },
    {
      "contrastive_loss": 0.3863,
      "epoch": 1.1399548532731376,
      "grad_norm": 14.520092964172363,
      "learning_rate": 8.949218140162965e-06,
      "lm_loss": 5.733,
      "loss": 1.4052,
      "step": 505,
      "text_contrastive_loss": 0.8912,
      "train_positive_log_prob": -84.1584,
      "train_positive_token_accuracy": 0.0766,
      "train_positive_token_prob": 0.0277
    },
    {
      "contrastive_loss": 0.4617,
      "epoch": 1.1422121896162527,
      "grad_norm": 14.958356857299805,
      "learning_rate": 8.944764180726423e-06,
      "lm_loss": 5.7915,
      "loss": 1.5506,
      "step": 506,
      "text_contrastive_loss": 1.0196,
      "train_positive_log_prob": -85.0416,
      "train_positive_token_accuracy": 0.0758,
      "train_positive_token_prob": 0.0281
    },
    {
      "contrastive_loss": 0.6412,
      "epoch": 1.144469525959368,
      "grad_norm": 17.7445068359375,
      "learning_rate": 8.940301915041496e-06,
      "lm_loss": 5.7276,
      "loss": 1.6771,
      "step": 507,
      "text_contrastive_loss": 0.9263,
      "train_positive_log_prob": -84.1906,
      "train_positive_token_accuracy": 0.0874,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.5004,
      "epoch": 1.146726862302483,
      "grad_norm": 14.941315650939941,
      "learning_rate": 8.935831352504103e-06,
      "lm_loss": 5.7347,
      "loss": 1.4447,
      "step": 508,
      "text_contrastive_loss": 0.7416,
      "train_positive_log_prob": -83.6701,
      "train_positive_token_accuracy": 0.0782,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.5204,
      "epoch": 1.1489841986455982,
      "grad_norm": 17.664854049682617,
      "learning_rate": 8.931352502527633e-06,
      "lm_loss": 5.7146,
      "loss": 1.5246,
      "step": 509,
      "text_contrastive_loss": 0.8656,
      "train_positive_log_prob": -82.9064,
      "train_positive_token_accuracy": 0.0781,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.548,
      "epoch": 1.1512415349887133,
      "grad_norm": 17.793493270874023,
      "learning_rate": 8.926865374542928e-06,
      "lm_loss": 5.7054,
      "loss": 1.6607,
      "step": 510,
      "text_contrastive_loss": 1.0843,
      "train_positive_log_prob": -82.4141,
      "train_positive_token_accuracy": 0.0768,
      "train_positive_token_prob": 0.0287
    },
    {
      "contrastive_loss": 0.5057,
      "epoch": 1.1534988713318284,
      "grad_norm": 16.280363082885742,
      "learning_rate": 8.922369977998257e-06,
      "lm_loss": 5.8599,
      "loss": 1.5999,
      "step": 511,
      "text_contrastive_loss": 1.0164,
      "train_positive_log_prob": -88.3531,
      "train_positive_token_accuracy": 0.0803,
      "train_positive_token_prob": 0.0287
    },
    {
      "contrastive_loss": 0.4861,
      "epoch": 1.1557562076749435,
      "grad_norm": 17.49152183532715,
      "learning_rate": 8.917866322359303e-06,
      "lm_loss": 5.8364,
      "loss": 1.5015,
      "step": 512,
      "text_contrastive_loss": 0.8635,
      "train_positive_log_prob": -87.2909,
      "train_positive_token_accuracy": 0.0799,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.4601,
      "epoch": 1.1580135440180588,
      "grad_norm": 13.432955741882324,
      "learning_rate": 8.913354417109136e-06,
      "lm_loss": 5.7393,
      "loss": 1.3987,
      "step": 513,
      "text_contrastive_loss": 0.7295,
      "train_positive_log_prob": -83.783,
      "train_positive_token_accuracy": 0.0789,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.4495,
      "epoch": 1.1602708803611739,
      "grad_norm": 14.593243598937988,
      "learning_rate": 8.908834271748202e-06,
      "lm_loss": 5.7306,
      "loss": 1.4098,
      "step": 514,
      "text_contrastive_loss": 0.7744,
      "train_positive_log_prob": -85.9168,
      "train_positive_token_accuracy": 0.0795,
      "train_positive_token_prob": 0.0286
    },
    {
      "contrastive_loss": 0.4307,
      "epoch": 1.162528216704289,
      "grad_norm": 14.839759826660156,
      "learning_rate": 8.904305895794292e-06,
      "lm_loss": 5.7948,
      "loss": 1.4511,
      "step": 515,
      "text_contrastive_loss": 0.8819,
      "train_positive_log_prob": -86.7461,
      "train_positive_token_accuracy": 0.0861,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.6057,
      "epoch": 1.164785553047404,
      "grad_norm": 15.436018943786621,
      "learning_rate": 8.899769298782528e-06,
      "lm_loss": 5.8003,
      "loss": 1.6617,
      "step": 516,
      "text_contrastive_loss": 0.9521,
      "train_positive_log_prob": -86.7558,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.5433,
      "epoch": 1.1670428893905191,
      "grad_norm": 16.551664352416992,
      "learning_rate": 8.895224490265346e-06,
      "lm_loss": 5.6868,
      "loss": 1.5589,
      "step": 517,
      "text_contrastive_loss": 0.8938,
      "train_positive_log_prob": -82.3282,
      "train_positive_token_accuracy": 0.0801,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.463,
      "epoch": 1.1693002257336342,
      "grad_norm": 13.993252754211426,
      "learning_rate": 8.890671479812472e-06,
      "lm_loss": 5.7785,
      "loss": 1.3917,
      "step": 518,
      "text_contrastive_loss": 0.7018,
      "train_positive_log_prob": -84.2373,
      "train_positive_token_accuracy": 0.0756,
      "train_positive_token_prob": 0.0286
    },
    {
      "contrastive_loss": 0.5316,
      "epoch": 1.1715575620767495,
      "grad_norm": 15.224872589111328,
      "learning_rate": 8.886110277010902e-06,
      "lm_loss": 5.7554,
      "loss": 1.5065,
      "step": 519,
      "text_contrastive_loss": 0.7989,
      "train_positive_log_prob": -86.3766,
      "train_positive_token_accuracy": 0.0737,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.5913,
      "epoch": 1.1738148984198646,
      "grad_norm": 15.95876693725586,
      "learning_rate": 8.88154089146488e-06,
      "lm_loss": 5.6424,
      "loss": 1.606,
      "step": 520,
      "text_contrastive_loss": 0.9009,
      "train_positive_log_prob": -80.9136,
      "train_positive_token_accuracy": 0.0877,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.4302,
      "epoch": 1.1760722347629797,
      "grad_norm": 12.770486831665039,
      "learning_rate": 8.876963332795881e-06,
      "lm_loss": 5.7211,
      "loss": 1.4309,
      "step": 521,
      "text_contrastive_loss": 0.8572,
      "train_positive_log_prob": -85.3662,
      "train_positive_token_accuracy": 0.0832,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.5878,
      "epoch": 1.1783295711060948,
      "grad_norm": 15.525639533996582,
      "learning_rate": 8.87237761064259e-06,
      "lm_loss": 5.6982,
      "loss": 1.5839,
      "step": 522,
      "text_contrastive_loss": 0.8525,
      "train_positive_log_prob": -83.6426,
      "train_positive_token_accuracy": 0.0875,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.4203,
      "epoch": 1.18058690744921,
      "grad_norm": 14.036768913269043,
      "learning_rate": 8.867783734660883e-06,
      "lm_loss": 5.7352,
      "loss": 1.4365,
      "step": 523,
      "text_contrastive_loss": 0.8852,
      "train_positive_log_prob": -82.3299,
      "train_positive_token_accuracy": 0.0819,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.4719,
      "epoch": 1.182844243792325,
      "grad_norm": 14.465782165527344,
      "learning_rate": 8.8631817145238e-06,
      "lm_loss": 5.8459,
      "loss": 1.516,
      "step": 524,
      "text_contrastive_loss": 0.9191,
      "train_positive_log_prob": -87.0917,
      "train_positive_token_accuracy": 0.0829,
      "train_positive_token_prob": 0.0286
    },
    {
      "contrastive_loss": 0.4598,
      "epoch": 1.18510158013544,
      "grad_norm": 14.271872520446777,
      "learning_rate": 8.858571559921539e-06,
      "lm_loss": 5.6123,
      "loss": 1.4508,
      "step": 525,
      "text_contrastive_loss": 0.8597,
      "train_positive_log_prob": -83.13,
      "train_positive_token_accuracy": 0.0833,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.5269,
      "epoch": 1.1873589164785554,
      "grad_norm": 14.845014572143555,
      "learning_rate": 8.853953280561412e-06,
      "lm_loss": 5.6389,
      "loss": 1.5117,
      "step": 526,
      "text_contrastive_loss": 0.8418,
      "train_positive_log_prob": -81.1695,
      "train_positive_token_accuracy": 0.0768,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.3811,
      "epoch": 1.1896162528216705,
      "grad_norm": 12.324769973754883,
      "learning_rate": 8.849326886167854e-06,
      "lm_loss": 5.7842,
      "loss": 1.3254,
      "step": 527,
      "text_contrastive_loss": 0.7317,
      "train_positive_log_prob": -85.9033,
      "train_positive_token_accuracy": 0.0757,
      "train_positive_token_prob": 0.0289
    },
    {
      "contrastive_loss": 0.4383,
      "epoch": 1.1918735891647856,
      "grad_norm": 14.406002044677734,
      "learning_rate": 8.844692386482379e-06,
      "lm_loss": 5.6966,
      "loss": 1.3858,
      "step": 528,
      "text_contrastive_loss": 0.7558,
      "train_positive_log_prob": -84.6946,
      "train_positive_token_accuracy": 0.0838,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.3975,
      "epoch": 1.1941309255079007,
      "grad_norm": 12.731202125549316,
      "learning_rate": 8.840049791263567e-06,
      "lm_loss": 5.7486,
      "loss": 1.3348,
      "step": 529,
      "text_contrastive_loss": 0.7249,
      "train_positive_log_prob": -84.5876,
      "train_positive_token_accuracy": 0.0836,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.4899,
      "epoch": 1.1963882618510158,
      "grad_norm": 15.5960054397583,
      "learning_rate": 8.835399110287046e-06,
      "lm_loss": 5.729,
      "loss": 1.5556,
      "step": 530,
      "text_contrastive_loss": 0.9857,
      "train_positive_log_prob": -84.167,
      "train_positive_token_accuracy": 0.0766,
      "train_positive_token_prob": 0.0283
    },
    {
      "contrastive_loss": 0.5163,
      "epoch": 1.1986455981941309,
      "grad_norm": 14.73003101348877,
      "learning_rate": 8.830740353345475e-06,
      "lm_loss": 5.6938,
      "loss": 1.5633,
      "step": 531,
      "text_contrastive_loss": 0.9551,
      "train_positive_log_prob": -83.7345,
      "train_positive_token_accuracy": 0.0866,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.4693,
      "epoch": 1.200902934537246,
      "grad_norm": 14.445740699768066,
      "learning_rate": 8.826073530248508e-06,
      "lm_loss": 5.6517,
      "loss": 1.3906,
      "step": 532,
      "text_contrastive_loss": 0.7122,
      "train_positive_log_prob": -82.456,
      "train_positive_token_accuracy": 0.076,
      "train_positive_token_prob": 0.0291
    },
    {
      "contrastive_loss": 0.3856,
      "epoch": 1.2031602708803613,
      "grad_norm": 13.313594818115234,
      "learning_rate": 8.82139865082279e-06,
      "lm_loss": 5.7347,
      "loss": 1.3076,
      "step": 533,
      "text_contrastive_loss": 0.6969,
      "train_positive_log_prob": -85.2587,
      "train_positive_token_accuracy": 0.0799,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.5691,
      "epoch": 1.2054176072234764,
      "grad_norm": 15.315177917480469,
      "learning_rate": 8.81671572491193e-06,
      "lm_loss": 5.866,
      "loss": 1.6133,
      "step": 534,
      "text_contrastive_loss": 0.9153,
      "train_positive_log_prob": -86.1575,
      "train_positive_token_accuracy": 0.0769,
      "train_positive_token_prob": 0.028
    },
    {
      "contrastive_loss": 0.5431,
      "epoch": 1.2076749435665914,
      "grad_norm": 14.558280944824219,
      "learning_rate": 8.812024762376477e-06,
      "lm_loss": 5.7293,
      "loss": 1.5564,
      "step": 535,
      "text_contrastive_loss": 0.8807,
      "train_positive_log_prob": -82.6475,
      "train_positive_token_accuracy": 0.0777,
      "train_positive_token_prob": 0.0287
    },
    {
      "contrastive_loss": 0.4477,
      "epoch": 1.2099322799097065,
      "grad_norm": 17.240751266479492,
      "learning_rate": 8.807325773093904e-06,
      "lm_loss": 5.7842,
      "loss": 1.5066,
      "step": 536,
      "text_contrastive_loss": 0.9611,
      "train_positive_log_prob": -84.7264,
      "train_positive_token_accuracy": 0.0701,
      "train_positive_token_prob": 0.027
    },
    {
      "contrastive_loss": 0.5589,
      "epoch": 1.2121896162528216,
      "grad_norm": 15.851217269897461,
      "learning_rate": 8.802618766958586e-06,
      "lm_loss": 5.7564,
      "loss": 1.5424,
      "step": 537,
      "text_contrastive_loss": 0.8157,
      "train_positive_log_prob": -86.8839,
      "train_positive_token_accuracy": 0.0719,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.5343,
      "epoch": 1.2144469525959367,
      "grad_norm": 15.090217590332031,
      "learning_rate": 8.797903753881775e-06,
      "lm_loss": 5.7031,
      "loss": 1.5509,
      "step": 538,
      "text_contrastive_loss": 0.8924,
      "train_positive_log_prob": -84.3439,
      "train_positive_token_accuracy": 0.0784,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.4792,
      "epoch": 1.2167042889390518,
      "grad_norm": 14.871121406555176,
      "learning_rate": 8.793180743791587e-06,
      "lm_loss": 5.6515,
      "loss": 1.4767,
      "step": 539,
      "text_contrastive_loss": 0.8647,
      "train_positive_log_prob": -84.1551,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.4826,
      "epoch": 1.2189616252821671,
      "grad_norm": 14.766619682312012,
      "learning_rate": 8.788449746632976e-06,
      "lm_loss": 5.7669,
      "loss": 1.5008,
      "step": 540,
      "text_contrastive_loss": 0.8829,
      "train_positive_log_prob": -85.2913,
      "train_positive_token_accuracy": 0.0726,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.4565,
      "epoch": 1.2212189616252822,
      "grad_norm": 13.520822525024414,
      "learning_rate": 8.78371077236771e-06,
      "lm_loss": 5.7719,
      "loss": 1.456,
      "step": 541,
      "text_contrastive_loss": 0.8445,
      "train_positive_log_prob": -83.2678,
      "train_positive_token_accuracy": 0.0747,
      "train_positive_token_prob": 0.028
    },
    {
      "contrastive_loss": 0.4993,
      "epoch": 1.2234762979683973,
      "grad_norm": 15.36131763458252,
      "learning_rate": 8.778963830974362e-06,
      "lm_loss": 5.8409,
      "loss": 1.6026,
      "step": 542,
      "text_contrastive_loss": 1.0384,
      "train_positive_log_prob": -86.1752,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.0281
    },
    {
      "contrastive_loss": 0.3354,
      "epoch": 1.2257336343115124,
      "grad_norm": 11.211355209350586,
      "learning_rate": 8.77420893244827e-06,
      "lm_loss": 5.7047,
      "loss": 1.2846,
      "step": 543,
      "text_contrastive_loss": 0.7573,
      "train_positive_log_prob": -84.77,
      "train_positive_token_accuracy": 0.0783,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.4381,
      "epoch": 1.2279909706546275,
      "grad_norm": 15.713208198547363,
      "learning_rate": 8.769446086801536e-06,
      "lm_loss": 5.6901,
      "loss": 1.4123,
      "step": 544,
      "text_contrastive_loss": 0.8102,
      "train_positive_log_prob": -85.6427,
      "train_positive_token_accuracy": 0.0838,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.4511,
      "epoch": 1.2302483069977426,
      "grad_norm": 15.119928359985352,
      "learning_rate": 8.764675304062992e-06,
      "lm_loss": 5.8012,
      "loss": 1.4833,
      "step": 545,
      "text_contrastive_loss": 0.9043,
      "train_positive_log_prob": -86.3176,
      "train_positive_token_accuracy": 0.0732,
      "train_positive_token_prob": 0.0283
    },
    {
      "contrastive_loss": 0.5804,
      "epoch": 1.2325056433408579,
      "grad_norm": 17.2160587310791,
      "learning_rate": 8.759896594278183e-06,
      "lm_loss": 5.7205,
      "loss": 1.6213,
      "step": 546,
      "text_contrastive_loss": 0.9377,
      "train_positive_log_prob": -83.6166,
      "train_positive_token_accuracy": 0.0852,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.4843,
      "epoch": 1.234762979683973,
      "grad_norm": 14.953021049499512,
      "learning_rate": 8.755109967509345e-06,
      "lm_loss": 5.7142,
      "loss": 1.4419,
      "step": 547,
      "text_contrastive_loss": 0.7725,
      "train_positive_log_prob": -83.6403,
      "train_positive_token_accuracy": 0.0761,
      "train_positive_token_prob": 0.028
    },
    {
      "contrastive_loss": 0.5046,
      "epoch": 1.237020316027088,
      "grad_norm": 15.522333145141602,
      "learning_rate": 8.750315433835387e-06,
      "lm_loss": 5.6737,
      "loss": 1.5456,
      "step": 548,
      "text_contrastive_loss": 0.9472,
      "train_positive_log_prob": -83.5554,
      "train_positive_token_accuracy": 0.0798,
      "train_positive_token_prob": 0.0285
    },
    {
      "contrastive_loss": 0.5071,
      "epoch": 1.2392776523702032,
      "grad_norm": 14.70351505279541,
      "learning_rate": 8.745513003351862e-06,
      "lm_loss": 5.6814,
      "loss": 1.5055,
      "step": 549,
      "text_contrastive_loss": 0.8606,
      "train_positive_log_prob": -85.7084,
      "train_positive_token_accuracy": 0.0834,
      "train_positive_token_prob": 0.0291
    },
    {
      "contrastive_loss": 0.5358,
      "epoch": 1.2415349887133182,
      "grad_norm": 16.600210189819336,
      "learning_rate": 8.740702686170955e-06,
      "lm_loss": 5.814,
      "loss": 1.6439,
      "step": 550,
      "text_contrastive_loss": 1.0533,
      "train_positive_log_prob": -86.0512,
      "train_positive_token_accuracy": 0.0839,
      "train_positive_token_prob": 0.0277
    },
    {
      "contrastive_loss": 0.3952,
      "epoch": 1.2437923250564333,
      "grad_norm": 15.454834938049316,
      "learning_rate": 8.735884492421457e-06,
      "lm_loss": 5.6809,
      "loss": 1.3336,
      "step": 551,
      "text_contrastive_loss": 0.7406,
      "train_positive_log_prob": -83.021,
      "train_positive_token_accuracy": 0.0823,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.5029,
      "epoch": 1.2460496613995486,
      "grad_norm": 16.872909545898438,
      "learning_rate": 8.731058432248743e-06,
      "lm_loss": 5.7728,
      "loss": 1.606,
      "step": 552,
      "text_contrastive_loss": 1.0516,
      "train_positive_log_prob": -84.6123,
      "train_positive_token_accuracy": 0.0736,
      "train_positive_token_prob": 0.0272
    },
    {
      "contrastive_loss": 0.4669,
      "epoch": 1.2483069977426637,
      "grad_norm": 15.279902458190918,
      "learning_rate": 8.726224515814752e-06,
      "lm_loss": 5.7958,
      "loss": 1.5404,
      "step": 553,
      "text_contrastive_loss": 0.9878,
      "train_positive_log_prob": -86.8693,
      "train_positive_token_accuracy": 0.0766,
      "train_positive_token_prob": 0.0275
    },
    {
      "contrastive_loss": 0.5938,
      "epoch": 1.2505643340857788,
      "grad_norm": 16.064992904663086,
      "learning_rate": 8.721382753297967e-06,
      "lm_loss": 5.7911,
      "loss": 1.5923,
      "step": 554,
      "text_contrastive_loss": 0.8389,
      "train_positive_log_prob": -84.9457,
      "train_positive_token_accuracy": 0.086,
      "train_positive_token_prob": 0.0281
    },
    {
      "contrastive_loss": 0.4106,
      "epoch": 1.252821670428894,
      "grad_norm": 15.163758277893066,
      "learning_rate": 8.71653315489339e-06,
      "lm_loss": 5.7514,
      "loss": 1.4164,
      "step": 555,
      "text_contrastive_loss": 0.8613,
      "train_positive_log_prob": -86.3594,
      "train_positive_token_accuracy": 0.0801,
      "train_positive_token_prob": 0.0279
    },
    {
      "contrastive_loss": 0.3873,
      "epoch": 1.255079006772009,
      "grad_norm": 13.098235130310059,
      "learning_rate": 8.711675730812522e-06,
      "lm_loss": 5.731,
      "loss": 1.3723,
      "step": 556,
      "text_contrastive_loss": 0.8238,
      "train_positive_log_prob": -84.1503,
      "train_positive_token_accuracy": 0.0809,
      "train_positive_token_prob": 0.0286
    },
    {
      "contrastive_loss": 0.5018,
      "epoch": 1.257336343115124,
      "grad_norm": 15.086501121520996,
      "learning_rate": 8.706810491283346e-06,
      "lm_loss": 5.7284,
      "loss": 1.4601,
      "step": 557,
      "text_contrastive_loss": 0.7709,
      "train_positive_log_prob": -83.2167,
      "train_positive_token_accuracy": 0.0764,
      "train_positive_token_prob": 0.0282
    },
    {
      "contrastive_loss": 0.5713,
      "epoch": 1.2595936794582392,
      "grad_norm": 17.556636810302734,
      "learning_rate": 8.701937446550298e-06,
      "lm_loss": 5.6779,
      "loss": 1.6101,
      "step": 558,
      "text_contrastive_loss": 0.942,
      "train_positive_log_prob": -84.7448,
      "train_positive_token_accuracy": 0.0835,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.4792,
      "epoch": 1.2618510158013545,
      "grad_norm": 13.67333698272705,
      "learning_rate": 8.69705660687425e-06,
      "lm_loss": 5.7951,
      "loss": 1.4587,
      "step": 559,
      "text_contrastive_loss": 0.7999,
      "train_positive_log_prob": -86.564,
      "train_positive_token_accuracy": 0.0819,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.6232,
      "epoch": 1.2641083521444696,
      "grad_norm": 16.826778411865234,
      "learning_rate": 8.692167982532487e-06,
      "lm_loss": 5.7352,
      "loss": 1.7336,
      "step": 560,
      "text_contrastive_loss": 1.0739,
      "train_positive_log_prob": -84.1038,
      "train_positive_token_accuracy": 0.0804,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.4988,
      "epoch": 1.2663656884875847,
      "grad_norm": 15.485930442810059,
      "learning_rate": 8.687271583818687e-06,
      "lm_loss": 5.6988,
      "loss": 1.5127,
      "step": 561,
      "text_contrastive_loss": 0.888,
      "train_positive_log_prob": -84.5729,
      "train_positive_token_accuracy": 0.0779,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.4706,
      "epoch": 1.2686230248306998,
      "grad_norm": 14.440827369689941,
      "learning_rate": 8.682367421042895e-06,
      "lm_loss": 5.7956,
      "loss": 1.4354,
      "step": 562,
      "text_contrastive_loss": 0.7705,
      "train_positive_log_prob": -87.1599,
      "train_positive_token_accuracy": 0.0743,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.6045,
      "epoch": 1.2708803611738149,
      "grad_norm": 16.22441864013672,
      "learning_rate": 8.677455504531507e-06,
      "lm_loss": 5.8231,
      "loss": 1.7022,
      "step": 563,
      "text_contrastive_loss": 1.0307,
      "train_positive_log_prob": -87.6188,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.4107,
      "epoch": 1.27313769751693,
      "grad_norm": 14.138707160949707,
      "learning_rate": 8.672535844627243e-06,
      "lm_loss": 5.8323,
      "loss": 1.446,
      "step": 564,
      "text_contrastive_loss": 0.904,
      "train_positive_log_prob": -89.0805,
      "train_positive_token_accuracy": 0.077,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.5039,
      "epoch": 1.275395033860045,
      "grad_norm": 15.108689308166504,
      "learning_rate": 8.667608451689135e-06,
      "lm_loss": 5.7063,
      "loss": 1.4567,
      "step": 565,
      "text_contrastive_loss": 0.7643,
      "train_positive_log_prob": -84.7062,
      "train_positive_token_accuracy": 0.0811,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.5203,
      "epoch": 1.2776523702031604,
      "grad_norm": 14.707185745239258,
      "learning_rate": 8.662673336092487e-06,
      "lm_loss": 5.8072,
      "loss": 1.5847,
      "step": 566,
      "text_contrastive_loss": 0.9673,
      "train_positive_log_prob": -86.0674,
      "train_positive_token_accuracy": 0.0783,
      "train_positive_token_prob": 0.0288
    },
    {
      "contrastive_loss": 0.4584,
      "epoch": 1.2799097065462754,
      "grad_norm": 14.290581703186035,
      "learning_rate": 8.657730508228874e-06,
      "lm_loss": 5.8627,
      "loss": 1.4925,
      "step": 567,
      "text_contrastive_loss": 0.8957,
      "train_positive_log_prob": -87.209,
      "train_positive_token_accuracy": 0.0725,
      "train_positive_token_prob": 0.0278
    },
    {
      "contrastive_loss": 0.5026,
      "epoch": 1.2821670428893905,
      "grad_norm": 15.093138694763184,
      "learning_rate": 8.652779978506103e-06,
      "lm_loss": 5.8232,
      "loss": 1.577,
      "step": 568,
      "text_contrastive_loss": 0.9841,
      "train_positive_log_prob": -87.1005,
      "train_positive_token_accuracy": 0.0739,
      "train_positive_token_prob": 0.0288
    },
    {
      "contrastive_loss": 0.4428,
      "epoch": 1.2844243792325056,
      "grad_norm": 14.957511901855469,
      "learning_rate": 8.647821757348202e-06,
      "lm_loss": 5.8571,
      "loss": 1.3795,
      "step": 569,
      "text_contrastive_loss": 0.7018,
      "train_positive_log_prob": -88.7241,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.5481,
      "epoch": 1.2866817155756207,
      "grad_norm": 13.965394020080566,
      "learning_rate": 8.642855855195394e-06,
      "lm_loss": 5.7554,
      "loss": 1.5387,
      "step": 570,
      "text_contrastive_loss": 0.83,
      "train_positive_log_prob": -85.7922,
      "train_positive_token_accuracy": 0.0755,
      "train_positive_token_prob": 0.0281
    },
    {
      "contrastive_loss": 0.4729,
      "epoch": 1.2889390519187358,
      "grad_norm": 13.30693244934082,
      "learning_rate": 8.637882282504075e-06,
      "lm_loss": 5.8883,
      "loss": 1.4739,
      "step": 571,
      "text_contrastive_loss": 0.8244,
      "train_positive_log_prob": -89.8679,
      "train_positive_token_accuracy": 0.0806,
      "train_positive_token_prob": 0.0279
    },
    {
      "contrastive_loss": 0.5084,
      "epoch": 1.291196388261851,
      "grad_norm": 15.2211332321167,
      "learning_rate": 8.632901049746793e-06,
      "lm_loss": 5.7314,
      "loss": 1.6142,
      "step": 572,
      "text_contrastive_loss": 1.0654,
      "train_positive_log_prob": -84.6486,
      "train_positive_token_accuracy": 0.0767,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.5172,
      "epoch": 1.2934537246049662,
      "grad_norm": 16.37986183166504,
      "learning_rate": 8.627912167412222e-06,
      "lm_loss": 5.6582,
      "loss": 1.5207,
      "step": 573,
      "text_contrastive_loss": 0.8753,
      "train_positive_log_prob": -84.4503,
      "train_positive_token_accuracy": 0.0845,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.4995,
      "epoch": 1.2957110609480813,
      "grad_norm": 16.388891220092773,
      "learning_rate": 8.622915646005152e-06,
      "lm_loss": 5.7194,
      "loss": 1.5174,
      "step": 574,
      "text_contrastive_loss": 0.8919,
      "train_positive_log_prob": -85.2107,
      "train_positive_token_accuracy": 0.0823,
      "train_positive_token_prob": 0.0285
    },
    {
      "contrastive_loss": 0.414,
      "epoch": 1.2979683972911964,
      "grad_norm": 14.741597175598145,
      "learning_rate": 8.617911496046446e-06,
      "lm_loss": 5.7209,
      "loss": 1.3457,
      "step": 575,
      "text_contrastive_loss": 0.7192,
      "train_positive_log_prob": -85.7448,
      "train_positive_token_accuracy": 0.0753,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.4004,
      "epoch": 1.3002257336343115,
      "grad_norm": 13.857163429260254,
      "learning_rate": 8.612899728073039e-06,
      "lm_loss": 5.818,
      "loss": 1.3764,
      "step": 576,
      "text_contrastive_loss": 0.7884,
      "train_positive_log_prob": -88.0437,
      "train_positive_token_accuracy": 0.0811,
      "train_positive_token_prob": 0.0282
    },
    {
      "contrastive_loss": 0.458,
      "epoch": 1.3024830699774266,
      "grad_norm": 14.838714599609375,
      "learning_rate": 8.607880352637905e-06,
      "lm_loss": 5.7136,
      "loss": 1.3895,
      "step": 577,
      "text_contrastive_loss": 0.7202,
      "train_positive_log_prob": -82.3788,
      "train_positive_token_accuracy": 0.087,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.467,
      "epoch": 1.304740406320542,
      "grad_norm": 16.11862564086914,
      "learning_rate": 8.602853380310033e-06,
      "lm_loss": 5.7059,
      "loss": 1.4292,
      "step": 578,
      "text_contrastive_loss": 0.7831,
      "train_positive_log_prob": -83.1163,
      "train_positive_token_accuracy": 0.0834,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.5316,
      "epoch": 1.3069977426636568,
      "grad_norm": 16.140466690063477,
      "learning_rate": 8.59781882167441e-06,
      "lm_loss": 5.7495,
      "loss": 1.5035,
      "step": 579,
      "text_contrastive_loss": 0.7939,
      "train_positive_log_prob": -85.2785,
      "train_positive_token_accuracy": 0.083,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.4612,
      "epoch": 1.309255079006772,
      "grad_norm": 14.843207359313965,
      "learning_rate": 8.592776687332003e-06,
      "lm_loss": 5.8023,
      "loss": 1.5381,
      "step": 580,
      "text_contrastive_loss": 0.9934,
      "train_positive_log_prob": -85.1521,
      "train_positive_token_accuracy": 0.0825,
      "train_positive_token_prob": 0.028
    },
    {
      "contrastive_loss": 0.4867,
      "epoch": 1.3115124153498872,
      "grad_norm": 14.938104629516602,
      "learning_rate": 8.58772698789972e-06,
      "lm_loss": 5.7041,
      "loss": 1.4913,
      "step": 581,
      "text_contrastive_loss": 0.8684,
      "train_positive_log_prob": -83.6713,
      "train_positive_token_accuracy": 0.0858,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.4529,
      "epoch": 1.3137697516930023,
      "grad_norm": 14.714336395263672,
      "learning_rate": 8.582669734010407e-06,
      "lm_loss": 5.6696,
      "loss": 1.4519,
      "step": 582,
      "text_contrastive_loss": 0.8641,
      "train_positive_log_prob": -83.7359,
      "train_positive_token_accuracy": 0.0786,
      "train_positive_token_prob": 0.0285
    },
    {
      "contrastive_loss": 0.4953,
      "epoch": 1.3160270880361173,
      "grad_norm": 14.955601692199707,
      "learning_rate": 8.577604936312813e-06,
      "lm_loss": 5.6612,
      "loss": 1.4908,
      "step": 583,
      "text_contrastive_loss": 0.8588,
      "train_positive_log_prob": -82.7071,
      "train_positive_token_accuracy": 0.0759,
      "train_positive_token_prob": 0.028
    },
    {
      "contrastive_loss": 0.5167,
      "epoch": 1.3182844243792324,
      "grad_norm": 16.038482666015625,
      "learning_rate": 8.572532605471572e-06,
      "lm_loss": 5.759,
      "loss": 1.5626,
      "step": 584,
      "text_contrastive_loss": 0.94,
      "train_positive_log_prob": -85.9355,
      "train_positive_token_accuracy": 0.0728,
      "train_positive_token_prob": 0.0285
    },
    {
      "contrastive_loss": 0.5387,
      "epoch": 1.3205417607223477,
      "grad_norm": 14.877217292785645,
      "learning_rate": 8.567452752167183e-06,
      "lm_loss": 5.6663,
      "loss": 1.5811,
      "step": 585,
      "text_contrastive_loss": 0.9515,
      "train_positive_log_prob": -83.557,
      "train_positive_token_accuracy": 0.0808,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.4369,
      "epoch": 1.3227990970654628,
      "grad_norm": 14.755962371826172,
      "learning_rate": 8.562365387095977e-06,
      "lm_loss": 5.6585,
      "loss": 1.4837,
      "step": 586,
      "text_contrastive_loss": 0.9619,
      "train_positive_log_prob": -82.978,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.0291
    },
    {
      "contrastive_loss": 0.4224,
      "epoch": 1.325056433408578,
      "grad_norm": 14.569847106933594,
      "learning_rate": 8.557270520970111e-06,
      "lm_loss": 5.6073,
      "loss": 1.3786,
      "step": 587,
      "text_contrastive_loss": 0.7909,
      "train_positive_log_prob": -84.6998,
      "train_positive_token_accuracy": 0.0786,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.4036,
      "epoch": 1.327313769751693,
      "grad_norm": 13.44758129119873,
      "learning_rate": 8.552168164517532e-06,
      "lm_loss": 5.6238,
      "loss": 1.3288,
      "step": 588,
      "text_contrastive_loss": 0.7257,
      "train_positive_log_prob": -84.845,
      "train_positive_token_accuracy": 0.0803,
      "train_positive_token_prob": 0.0291
    },
    {
      "contrastive_loss": 0.5184,
      "epoch": 1.329571106094808,
      "grad_norm": 13.879511833190918,
      "learning_rate": 8.547058328481959e-06,
      "lm_loss": 5.739,
      "loss": 1.5182,
      "step": 589,
      "text_contrastive_loss": 0.8516,
      "train_positive_log_prob": -83.8685,
      "train_positive_token_accuracy": 0.0695,
      "train_positive_token_prob": 0.0276
    },
    {
      "contrastive_loss": 0.4487,
      "epoch": 1.3318284424379232,
      "grad_norm": 13.161958694458008,
      "learning_rate": 8.54194102362286e-06,
      "lm_loss": 5.6756,
      "loss": 1.4122,
      "step": 590,
      "text_contrastive_loss": 0.7918,
      "train_positive_log_prob": -84.561,
      "train_positive_token_accuracy": 0.0758,
      "train_positive_token_prob": 0.0277
    },
    {
      "contrastive_loss": 0.5622,
      "epoch": 1.3340857787810383,
      "grad_norm": 15.899773597717285,
      "learning_rate": 8.536816260715433e-06,
      "lm_loss": 5.819,
      "loss": 1.6092,
      "step": 591,
      "text_contrastive_loss": 0.9302,
      "train_positive_log_prob": -86.9451,
      "train_positive_token_accuracy": 0.0719,
      "train_positive_token_prob": 0.0272
    },
    {
      "contrastive_loss": 0.5415,
      "epoch": 1.3363431151241536,
      "grad_norm": 14.699732780456543,
      "learning_rate": 8.531684050550575e-06,
      "lm_loss": 5.6373,
      "loss": 1.574,
      "step": 592,
      "text_contrastive_loss": 0.9375,
      "train_positive_log_prob": -82.4896,
      "train_positive_token_accuracy": 0.0785,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.436,
      "epoch": 1.3386004514672687,
      "grad_norm": 13.371824264526367,
      "learning_rate": 8.526544403934868e-06,
      "lm_loss": 5.7423,
      "loss": 1.4383,
      "step": 593,
      "text_contrastive_loss": 0.8561,
      "train_positive_log_prob": -85.6432,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.0286
    },
    {
      "contrastive_loss": 0.4243,
      "epoch": 1.3408577878103838,
      "grad_norm": 12.91390323638916,
      "learning_rate": 8.521397331690551e-06,
      "lm_loss": 5.6372,
      "loss": 1.3804,
      "step": 594,
      "text_contrastive_loss": 0.7848,
      "train_positive_log_prob": -83.7875,
      "train_positive_token_accuracy": 0.0762,
      "train_positive_token_prob": 0.0282
    },
    {
      "contrastive_loss": 0.4223,
      "epoch": 1.3431151241534989,
      "grad_norm": 14.641847610473633,
      "learning_rate": 8.516242844655498e-06,
      "lm_loss": 5.7631,
      "loss": 1.368,
      "step": 595,
      "text_contrastive_loss": 0.7388,
      "train_positive_log_prob": -84.9133,
      "train_positive_token_accuracy": 0.0758,
      "train_positive_token_prob": 0.028
    },
    {
      "contrastive_loss": 0.6231,
      "epoch": 1.345372460496614,
      "grad_norm": 18.55071258544922,
      "learning_rate": 8.5110809536832e-06,
      "lm_loss": 5.8372,
      "loss": 1.6618,
      "step": 596,
      "text_contrastive_loss": 0.9099,
      "train_positive_log_prob": -86.3248,
      "train_positive_token_accuracy": 0.0711,
      "train_positive_token_prob": 0.0271
    },
    {
      "contrastive_loss": 0.5289,
      "epoch": 1.347629796839729,
      "grad_norm": 17.197635650634766,
      "learning_rate": 8.50591166964273e-06,
      "lm_loss": 5.804,
      "loss": 1.5507,
      "step": 597,
      "text_contrastive_loss": 0.8828,
      "train_positive_log_prob": -86.7005,
      "train_positive_token_accuracy": 0.0764,
      "train_positive_token_prob": 0.0281
    },
    {
      "contrastive_loss": 0.5315,
      "epoch": 1.3498871331828441,
      "grad_norm": 12.700179100036621,
      "learning_rate": 8.500735003418734e-06,
      "lm_loss": 5.6742,
      "loss": 1.5912,
      "step": 598,
      "text_contrastive_loss": 0.9845,
      "train_positive_log_prob": -82.3024,
      "train_positive_token_accuracy": 0.0819,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.5325,
      "epoch": 1.3521444695259595,
      "grad_norm": 15.083609580993652,
      "learning_rate": 8.495550965911403e-06,
      "lm_loss": 5.6869,
      "loss": 1.6384,
      "step": 599,
      "text_contrastive_loss": 1.0745,
      "train_positive_log_prob": -85.2988,
      "train_positive_token_accuracy": 0.0784,
      "train_positive_token_prob": 0.0287
    },
    {
      "contrastive_loss": 0.5557,
      "epoch": 1.3544018058690745,
      "grad_norm": 16.89891815185547,
      "learning_rate": 8.490359568036446e-06,
      "lm_loss": 5.7464,
      "loss": 1.6062,
      "step": 600,
      "text_contrastive_loss": 0.9517,
      "train_positive_log_prob": -86.6905,
      "train_positive_token_accuracy": 0.0746,
      "train_positive_token_prob": 0.0282
    },
    {
      "contrastive_loss": 0.5086,
      "epoch": 1.3566591422121896,
      "grad_norm": 14.070127487182617,
      "learning_rate": 8.485160820725073e-06,
      "lm_loss": 5.834,
      "loss": 1.4916,
      "step": 601,
      "text_contrastive_loss": 0.7992,
      "train_positive_log_prob": -88.8102,
      "train_positive_token_accuracy": 0.0819,
      "train_positive_token_prob": 0.0289
    },
    {
      "contrastive_loss": 0.492,
      "epoch": 1.3589164785553047,
      "grad_norm": 15.26427936553955,
      "learning_rate": 8.479954734923967e-06,
      "lm_loss": 5.6653,
      "loss": 1.458,
      "step": 602,
      "text_contrastive_loss": 0.799,
      "train_positive_log_prob": -83.984,
      "train_positive_token_accuracy": 0.087,
      "train_positive_token_prob": 0.0291
    },
    {
      "contrastive_loss": 0.4101,
      "epoch": 1.3611738148984198,
      "grad_norm": 13.670388221740723,
      "learning_rate": 8.474741321595263e-06,
      "lm_loss": 5.7464,
      "loss": 1.3817,
      "step": 603,
      "text_contrastive_loss": 0.7938,
      "train_positive_log_prob": -84.6457,
      "train_positive_token_accuracy": 0.0814,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.5669,
      "epoch": 1.363431151241535,
      "grad_norm": 16.399883270263672,
      "learning_rate": 8.46952059171653e-06,
      "lm_loss": 5.7989,
      "loss": 1.5579,
      "step": 604,
      "text_contrastive_loss": 0.8222,
      "train_positive_log_prob": -87.2108,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.5052,
      "epoch": 1.36568848758465,
      "grad_norm": 14.55126667022705,
      "learning_rate": 8.464292556280734e-06,
      "lm_loss": 5.6811,
      "loss": 1.4282,
      "step": 605,
      "text_contrastive_loss": 0.7097,
      "train_positive_log_prob": -83.2894,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.4126,
      "epoch": 1.3679458239277653,
      "grad_norm": 13.797871589660645,
      "learning_rate": 8.459057226296232e-06,
      "lm_loss": 5.6525,
      "loss": 1.3848,
      "step": 606,
      "text_contrastive_loss": 0.8138,
      "train_positive_log_prob": -83.4893,
      "train_positive_token_accuracy": 0.0801,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.4463,
      "epoch": 1.3702031602708804,
      "grad_norm": 14.512334823608398,
      "learning_rate": 8.453814612786736e-06,
      "lm_loss": 5.7283,
      "loss": 1.5064,
      "step": 607,
      "text_contrastive_loss": 0.9745,
      "train_positive_log_prob": -86.4649,
      "train_positive_token_accuracy": 0.0763,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.4708,
      "epoch": 1.3724604966139955,
      "grad_norm": 15.86633586883545,
      "learning_rate": 8.4485647267913e-06,
      "lm_loss": 5.7088,
      "loss": 1.4663,
      "step": 608,
      "text_contrastive_loss": 0.8493,
      "train_positive_log_prob": -85.651,
      "train_positive_token_accuracy": 0.0771,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.4696,
      "epoch": 1.3747178329571106,
      "grad_norm": 14.49863338470459,
      "learning_rate": 8.443307579364282e-06,
      "lm_loss": 5.7355,
      "loss": 1.4756,
      "step": 609,
      "text_contrastive_loss": 0.8649,
      "train_positive_log_prob": -86.3931,
      "train_positive_token_accuracy": 0.077,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.5206,
      "epoch": 1.3769751693002257,
      "grad_norm": 17.275390625,
      "learning_rate": 8.43804318157534e-06,
      "lm_loss": 5.742,
      "loss": 1.5092,
      "step": 610,
      "text_contrastive_loss": 0.8289,
      "train_positive_log_prob": -84.9649,
      "train_positive_token_accuracy": 0.0697,
      "train_positive_token_prob": 0.0277
    },
    {
      "contrastive_loss": 0.5528,
      "epoch": 1.379232505643341,
      "grad_norm": 16.464248657226562,
      "learning_rate": 8.432771544509395e-06,
      "lm_loss": 5.8354,
      "loss": 1.6296,
      "step": 611,
      "text_contrastive_loss": 0.9863,
      "train_positive_log_prob": -89.8798,
      "train_positive_token_accuracy": 0.0751,
      "train_positive_token_prob": 0.0281
    },
    {
      "contrastive_loss": 0.431,
      "epoch": 1.3814898419864559,
      "grad_norm": 16.348220825195312,
      "learning_rate": 8.427492679266605e-06,
      "lm_loss": 5.7165,
      "loss": 1.4121,
      "step": 612,
      "text_contrastive_loss": 0.8189,
      "train_positive_log_prob": -86.3244,
      "train_positive_token_accuracy": 0.0792,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.5288,
      "epoch": 1.3837471783295712,
      "grad_norm": 14.306358337402344,
      "learning_rate": 8.422206596962357e-06,
      "lm_loss": 5.653,
      "loss": 1.4818,
      "step": 613,
      "text_contrastive_loss": 0.7755,
      "train_positive_log_prob": -80.8444,
      "train_positive_token_accuracy": 0.0759,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.6173,
      "epoch": 1.3860045146726863,
      "grad_norm": 17.739561080932617,
      "learning_rate": 8.416913308727229e-06,
      "lm_loss": 5.6375,
      "loss": 1.5781,
      "step": 614,
      "text_contrastive_loss": 0.794,
      "train_positive_log_prob": -83.7935,
      "train_positive_token_accuracy": 0.0883,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.4479,
      "epoch": 1.3882618510158014,
      "grad_norm": 15.596267700195312,
      "learning_rate": 8.411612825706976e-06,
      "lm_loss": 5.6922,
      "loss": 1.4017,
      "step": 615,
      "text_contrastive_loss": 0.7692,
      "train_positive_log_prob": -84.7406,
      "train_positive_token_accuracy": 0.0848,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.5118,
      "epoch": 1.3905191873589164,
      "grad_norm": 14.741986274719238,
      "learning_rate": 8.4063051590625e-06,
      "lm_loss": 5.6216,
      "loss": 1.53,
      "step": 616,
      "text_contrastive_loss": 0.9121,
      "train_positive_log_prob": -84.3669,
      "train_positive_token_accuracy": 0.0809,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.3836,
      "epoch": 1.3927765237020315,
      "grad_norm": 13.403180122375488,
      "learning_rate": 8.400990319969829e-06,
      "lm_loss": 5.609,
      "loss": 1.3193,
      "step": 617,
      "text_contrastive_loss": 0.7495,
      "train_positive_log_prob": -81.4905,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.5574,
      "epoch": 1.3950338600451468,
      "grad_norm": 14.65643310546875,
      "learning_rate": 8.395668319620092e-06,
      "lm_loss": 5.6239,
      "loss": 1.5532,
      "step": 618,
      "text_contrastive_loss": 0.8669,
      "train_positive_log_prob": -82.0509,
      "train_positive_token_accuracy": 0.0776,
      "train_positive_token_prob": 0.0286
    },
    {
      "contrastive_loss": 0.4803,
      "epoch": 1.3972911963882617,
      "grad_norm": 15.410263061523438,
      "learning_rate": 8.390339169219504e-06,
      "lm_loss": 5.6461,
      "loss": 1.4675,
      "step": 619,
      "text_contrastive_loss": 0.8452,
      "train_positive_log_prob": -84.9576,
      "train_positive_token_accuracy": 0.0854,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.5147,
      "epoch": 1.399548532731377,
      "grad_norm": 15.741019248962402,
      "learning_rate": 8.385002879989328e-06,
      "lm_loss": 5.7148,
      "loss": 1.5343,
      "step": 620,
      "text_contrastive_loss": 0.8963,
      "train_positive_log_prob": -83.5083,
      "train_positive_token_accuracy": 0.0756,
      "train_positive_token_prob": 0.0288
    },
    {
      "contrastive_loss": 0.4724,
      "epoch": 1.4018058690744921,
      "grad_norm": 16.445375442504883,
      "learning_rate": 8.37965946316586e-06,
      "lm_loss": 5.7327,
      "loss": 1.4951,
      "step": 621,
      "text_contrastive_loss": 0.899,
      "train_positive_log_prob": -83.3037,
      "train_positive_token_accuracy": 0.0753,
      "train_positive_token_prob": 0.0288
    },
    {
      "contrastive_loss": 0.385,
      "epoch": 1.4040632054176072,
      "grad_norm": 13.678441047668457,
      "learning_rate": 8.37430893000041e-06,
      "lm_loss": 5.6494,
      "loss": 1.3745,
      "step": 622,
      "text_contrastive_loss": 0.8491,
      "train_positive_log_prob": -83.9212,
      "train_positive_token_accuracy": 0.0826,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.5289,
      "epoch": 1.4063205417607223,
      "grad_norm": 16.341611862182617,
      "learning_rate": 8.368951291759264e-06,
      "lm_loss": 5.6466,
      "loss": 1.5547,
      "step": 623,
      "text_contrastive_loss": 0.9223,
      "train_positive_log_prob": -83.3343,
      "train_positive_token_accuracy": 0.0807,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.6226,
      "epoch": 1.4085778781038374,
      "grad_norm": 16.759748458862305,
      "learning_rate": 8.363586559723675e-06,
      "lm_loss": 5.6907,
      "loss": 1.6319,
      "step": 624,
      "text_contrastive_loss": 0.8803,
      "train_positive_log_prob": -83.9181,
      "train_positive_token_accuracy": 0.0646,
      "train_positive_token_prob": 0.0273
    },
    {
      "contrastive_loss": 0.4707,
      "epoch": 1.4108352144469527,
      "grad_norm": 15.580821990966797,
      "learning_rate": 8.35821474518983e-06,
      "lm_loss": 5.6648,
      "loss": 1.4457,
      "step": 625,
      "text_contrastive_loss": 0.817,
      "train_positive_log_prob": -83.7316,
      "train_positive_token_accuracy": 0.084,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.4396,
      "epoch": 1.4130925507900678,
      "grad_norm": 13.662397384643555,
      "learning_rate": 8.352835859468829e-06,
      "lm_loss": 5.6666,
      "loss": 1.4652,
      "step": 626,
      "text_contrastive_loss": 0.9179,
      "train_positive_log_prob": -82.5654,
      "train_positive_token_accuracy": 0.0791,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.471,
      "epoch": 1.4153498871331829,
      "grad_norm": 15.736557960510254,
      "learning_rate": 8.347449913886662e-06,
      "lm_loss": 5.5869,
      "loss": 1.5001,
      "step": 627,
      "text_contrastive_loss": 0.941,
      "train_positive_log_prob": -82.6043,
      "train_positive_token_accuracy": 0.0806,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.4888,
      "epoch": 1.417607223476298,
      "grad_norm": 12.857428550720215,
      "learning_rate": 8.34205691978419e-06,
      "lm_loss": 5.6094,
      "loss": 1.5109,
      "step": 628,
      "text_contrastive_loss": 0.9223,
      "train_positive_log_prob": -84.2689,
      "train_positive_token_accuracy": 0.0777,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.4558,
      "epoch": 1.419864559819413,
      "grad_norm": 15.458051681518555,
      "learning_rate": 8.336656888517103e-06,
      "lm_loss": 5.6711,
      "loss": 1.4628,
      "step": 629,
      "text_contrastive_loss": 0.8797,
      "train_positive_log_prob": -84.7394,
      "train_positive_token_accuracy": 0.0758,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.4538,
      "epoch": 1.4221218961625282,
      "grad_norm": 13.901751518249512,
      "learning_rate": 8.331249831455921e-06,
      "lm_loss": 5.7476,
      "loss": 1.4583,
      "step": 630,
      "text_contrastive_loss": 0.8595,
      "train_positive_log_prob": -84.4216,
      "train_positive_token_accuracy": 0.0795,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.3812,
      "epoch": 1.4243792325056432,
      "grad_norm": 11.676010131835938,
      "learning_rate": 8.325835759985951e-06,
      "lm_loss": 5.7537,
      "loss": 1.3213,
      "step": 631,
      "text_contrastive_loss": 0.7296,
      "train_positive_log_prob": -85.0485,
      "train_positive_token_accuracy": 0.0801,
      "train_positive_token_prob": 0.0289
    },
    {
      "contrastive_loss": 0.4459,
      "epoch": 1.4266365688487586,
      "grad_norm": 14.951569557189941,
      "learning_rate": 8.320414685507272e-06,
      "lm_loss": 5.7978,
      "loss": 1.3755,
      "step": 632,
      "text_contrastive_loss": 0.6995,
      "train_positive_log_prob": -88.2355,
      "train_positive_token_accuracy": 0.0834,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.5285,
      "epoch": 1.4288939051918736,
      "grad_norm": 15.13301944732666,
      "learning_rate": 8.31498661943471e-06,
      "lm_loss": 5.7731,
      "loss": 1.5599,
      "step": 633,
      "text_contrastive_loss": 0.9081,
      "train_positive_log_prob": -86.3685,
      "train_positive_token_accuracy": 0.0708,
      "train_positive_token_prob": 0.0281
    },
    {
      "contrastive_loss": 0.4824,
      "epoch": 1.4311512415349887,
      "grad_norm": 14.152031898498535,
      "learning_rate": 8.309551573197809e-06,
      "lm_loss": 5.7039,
      "loss": 1.4794,
      "step": 634,
      "text_contrastive_loss": 0.8533,
      "train_positive_log_prob": -85.8893,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.3754,
      "epoch": 1.4334085778781038,
      "grad_norm": 13.575399398803711,
      "learning_rate": 8.304109558240817e-06,
      "lm_loss": 5.7387,
      "loss": 1.2938,
      "step": 635,
      "text_contrastive_loss": 0.6891,
      "train_positive_log_prob": -85.2221,
      "train_positive_token_accuracy": 0.0757,
      "train_positive_token_prob": 0.0283
    },
    {
      "contrastive_loss": 0.4933,
      "epoch": 1.435665914221219,
      "grad_norm": 14.869497299194336,
      "learning_rate": 8.298660586022646e-06,
      "lm_loss": 5.8215,
      "loss": 1.5802,
      "step": 636,
      "text_contrastive_loss": 1.0095,
      "train_positive_log_prob": -85.527,
      "train_positive_token_accuracy": 0.0776,
      "train_positive_token_prob": 0.0285
    },
    {
      "contrastive_loss": 0.5576,
      "epoch": 1.437923250564334,
      "grad_norm": 18.26089859008789,
      "learning_rate": 8.293204668016867e-06,
      "lm_loss": 5.6906,
      "loss": 1.618,
      "step": 637,
      "text_contrastive_loss": 0.9827,
      "train_positive_log_prob": -86.403,
      "train_positive_token_accuracy": 0.0849,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.5191,
      "epoch": 1.440180586907449,
      "grad_norm": 15.64136028289795,
      "learning_rate": 8.287741815711674e-06,
      "lm_loss": 5.7477,
      "loss": 1.5562,
      "step": 638,
      "text_contrastive_loss": 0.9246,
      "train_positive_log_prob": -85.3564,
      "train_positive_token_accuracy": 0.0807,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.5803,
      "epoch": 1.4424379232505644,
      "grad_norm": 15.560050010681152,
      "learning_rate": 8.282272040609855e-06,
      "lm_loss": 5.7404,
      "loss": 1.5859,
      "step": 639,
      "text_contrastive_loss": 0.8631,
      "train_positive_log_prob": -85.7271,
      "train_positive_token_accuracy": 0.0772,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.5483,
      "epoch": 1.4446952595936795,
      "grad_norm": 15.672173500061035,
      "learning_rate": 8.276795354228785e-06,
      "lm_loss": 5.6833,
      "loss": 1.5249,
      "step": 640,
      "text_contrastive_loss": 0.8164,
      "train_positive_log_prob": -84.3885,
      "train_positive_token_accuracy": 0.0829,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3893,
      "epoch": 1.4469525959367946,
      "grad_norm": 12.000456809997559,
      "learning_rate": 8.271311768100386e-06,
      "lm_loss": 5.5456,
      "loss": 1.3374,
      "step": 641,
      "text_contrastive_loss": 0.7871,
      "train_positive_log_prob": -82.8222,
      "train_positive_token_accuracy": 0.0787,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.5464,
      "epoch": 1.4492099322799097,
      "grad_norm": 15.725830078125,
      "learning_rate": 8.26582129377111e-06,
      "lm_loss": 5.6961,
      "loss": 1.5786,
      "step": 642,
      "text_contrastive_loss": 0.9251,
      "train_positive_log_prob": -83.1883,
      "train_positive_token_accuracy": 0.0706,
      "train_positive_token_prob": 0.0285
    },
    {
      "contrastive_loss": 0.3856,
      "epoch": 1.4514672686230248,
      "grad_norm": 14.19225788116455,
      "learning_rate": 8.26032394280191e-06,
      "lm_loss": 5.7577,
      "loss": 1.357,
      "step": 643,
      "text_contrastive_loss": 0.7912,
      "train_positive_log_prob": -85.9421,
      "train_positive_token_accuracy": 0.0772,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.583,
      "epoch": 1.45372460496614,
      "grad_norm": 14.988269805908203,
      "learning_rate": 8.254819726768224e-06,
      "lm_loss": 5.658,
      "loss": 1.6074,
      "step": 644,
      "text_contrastive_loss": 0.9173,
      "train_positive_log_prob": -84.5007,
      "train_positive_token_accuracy": 0.0751,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.3982,
      "epoch": 1.455981941309255,
      "grad_norm": 14.792583465576172,
      "learning_rate": 8.249308657259943e-06,
      "lm_loss": 5.6463,
      "loss": 1.3618,
      "step": 645,
      "text_contrastive_loss": 0.798,
      "train_positive_log_prob": -84.7384,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.4313,
      "epoch": 1.4582392776523703,
      "grad_norm": 13.597331047058105,
      "learning_rate": 8.243790745881389e-06,
      "lm_loss": 5.6629,
      "loss": 1.4241,
      "step": 646,
      "text_contrastive_loss": 0.8531,
      "train_positive_log_prob": -83.5331,
      "train_positive_token_accuracy": 0.0784,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.5069,
      "epoch": 1.4604966139954854,
      "grad_norm": 15.051453590393066,
      "learning_rate": 8.238266004251284e-06,
      "lm_loss": 5.7407,
      "loss": 1.4731,
      "step": 647,
      "text_contrastive_loss": 0.7842,
      "train_positive_log_prob": -85.7958,
      "train_positive_token_accuracy": 0.0766,
      "train_positive_token_prob": 0.0288
    },
    {
      "contrastive_loss": 0.466,
      "epoch": 1.4627539503386005,
      "grad_norm": 13.057682037353516,
      "learning_rate": 8.232734444002748e-06,
      "lm_loss": 5.7397,
      "loss": 1.446,
      "step": 648,
      "text_contrastive_loss": 0.812,
      "train_positive_log_prob": -85.3991,
      "train_positive_token_accuracy": 0.0851,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.4597,
      "epoch": 1.4650112866817155,
      "grad_norm": 14.503249168395996,
      "learning_rate": 8.22719607678324e-06,
      "lm_loss": 5.713,
      "loss": 1.4275,
      "step": 649,
      "text_contrastive_loss": 0.793,
      "train_positive_log_prob": -84.7791,
      "train_positive_token_accuracy": 0.0787,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.4196,
      "epoch": 1.4672686230248306,
      "grad_norm": 12.781978607177734,
      "learning_rate": 8.221650914254566e-06,
      "lm_loss": 5.7031,
      "loss": 1.3435,
      "step": 650,
      "text_contrastive_loss": 0.7072,
      "train_positive_log_prob": -83.7659,
      "train_positive_token_accuracy": 0.0801,
      "train_positive_token_prob": 0.0289
    },
    {
      "contrastive_loss": 0.3606,
      "epoch": 1.469525959367946,
      "grad_norm": 12.280264854431152,
      "learning_rate": 8.216098968092833e-06,
      "lm_loss": 5.652,
      "loss": 1.2835,
      "step": 651,
      "text_contrastive_loss": 0.7153,
      "train_positive_log_prob": -83.9411,
      "train_positive_token_accuracy": 0.0834,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3972,
      "epoch": 1.4717832957110608,
      "grad_norm": 15.713282585144043,
      "learning_rate": 8.210540249988435e-06,
      "lm_loss": 5.6435,
      "loss": 1.2746,
      "step": 652,
      "text_contrastive_loss": 0.6261,
      "train_positive_log_prob": -84.1212,
      "train_positive_token_accuracy": 0.0888,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.4787,
      "epoch": 1.4740406320541761,
      "grad_norm": 15.67564582824707,
      "learning_rate": 8.204974771646023e-06,
      "lm_loss": 5.6599,
      "loss": 1.453,
      "step": 653,
      "text_contrastive_loss": 0.8167,
      "train_positive_log_prob": -83.7458,
      "train_positive_token_accuracy": 0.0845,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.5185,
      "epoch": 1.4762979683972912,
      "grad_norm": 15.40292739868164,
      "learning_rate": 8.199402544784485e-06,
      "lm_loss": 5.6211,
      "loss": 1.6099,
      "step": 654,
      "text_contrastive_loss": 1.0584,
      "train_positive_log_prob": -82.9765,
      "train_positive_token_accuracy": 0.082,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.4908,
      "epoch": 1.4785553047404063,
      "grad_norm": 15.722895622253418,
      "learning_rate": 8.193823581136919e-06,
      "lm_loss": 5.7332,
      "loss": 1.559,
      "step": 655,
      "text_contrastive_loss": 0.9897,
      "train_positive_log_prob": -84.6532,
      "train_positive_token_accuracy": 0.0762,
      "train_positive_token_prob": 0.028
    },
    {
      "contrastive_loss": 0.4544,
      "epoch": 1.4808126410835214,
      "grad_norm": 14.068140029907227,
      "learning_rate": 8.188237892450603e-06,
      "lm_loss": 5.782,
      "loss": 1.4121,
      "step": 656,
      "text_contrastive_loss": 0.759,
      "train_positive_log_prob": -85.3028,
      "train_positive_token_accuracy": 0.0702,
      "train_positive_token_prob": 0.0276
    },
    {
      "contrastive_loss": 0.5044,
      "epoch": 1.4830699774266365,
      "grad_norm": 16.4137020111084,
      "learning_rate": 8.182645490486986e-06,
      "lm_loss": 5.6261,
      "loss": 1.4556,
      "step": 657,
      "text_contrastive_loss": 0.7772,
      "train_positive_log_prob": -83.9315,
      "train_positive_token_accuracy": 0.0757,
      "train_positive_token_prob": 0.0291
    },
    {
      "contrastive_loss": 0.4391,
      "epoch": 1.4853273137697518,
      "grad_norm": 14.37668514251709,
      "learning_rate": 8.177046387021641e-06,
      "lm_loss": 5.6918,
      "loss": 1.4325,
      "step": 658,
      "text_contrastive_loss": 0.8484,
      "train_positive_log_prob": -84.3999,
      "train_positive_token_accuracy": 0.0759,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.5499,
      "epoch": 1.487584650112867,
      "grad_norm": 16.686614990234375,
      "learning_rate": 8.17144059384426e-06,
      "lm_loss": 5.6319,
      "loss": 1.5167,
      "step": 659,
      "text_contrastive_loss": 0.8072,
      "train_positive_log_prob": -84.5463,
      "train_positive_token_accuracy": 0.0801,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.4126,
      "epoch": 1.489841986455982,
      "grad_norm": 12.877717971801758,
      "learning_rate": 8.165828122758615e-06,
      "lm_loss": 5.7432,
      "loss": 1.3588,
      "step": 660,
      "text_contrastive_loss": 0.7439,
      "train_positive_log_prob": -87.4941,
      "train_positive_token_accuracy": 0.0813,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.6014,
      "epoch": 1.492099322799097,
      "grad_norm": 17.75623321533203,
      "learning_rate": 8.160208985582547e-06,
      "lm_loss": 5.6466,
      "loss": 1.6029,
      "step": 661,
      "text_contrastive_loss": 0.8738,
      "train_positive_log_prob": -82.8446,
      "train_positive_token_accuracy": 0.0716,
      "train_positive_token_prob": 0.0291
    },
    {
      "contrastive_loss": 0.3884,
      "epoch": 1.4943566591422122,
      "grad_norm": 13.010163307189941,
      "learning_rate": 8.154583194147929e-06,
      "lm_loss": 5.6319,
      "loss": 1.2631,
      "step": 662,
      "text_contrastive_loss": 0.623,
      "train_positive_log_prob": -80.9603,
      "train_positive_token_accuracy": 0.0799,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.4159,
      "epoch": 1.4966139954853273,
      "grad_norm": 14.1793851852417,
      "learning_rate": 8.148950760300642e-06,
      "lm_loss": 5.7532,
      "loss": 1.3051,
      "step": 663,
      "text_contrastive_loss": 0.6278,
      "train_positive_log_prob": -86.3863,
      "train_positive_token_accuracy": 0.0799,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.5176,
      "epoch": 1.4988713318284423,
      "grad_norm": 16.44997215270996,
      "learning_rate": 8.14331169590056e-06,
      "lm_loss": 5.7242,
      "loss": 1.4979,
      "step": 664,
      "text_contrastive_loss": 0.8158,
      "train_positive_log_prob": -85.5964,
      "train_positive_token_accuracy": 0.0778,
      "train_positive_token_prob": 0.0286
    },
    {
      "contrastive_loss": 0.3672,
      "epoch": 1.5011286681715577,
      "grad_norm": 14.054045677185059,
      "learning_rate": 8.137666012821514e-06,
      "lm_loss": 5.7167,
      "loss": 1.3639,
      "step": 665,
      "text_contrastive_loss": 0.8502,
      "train_positive_log_prob": -83.6982,
      "train_positive_token_accuracy": 0.0686,
      "train_positive_token_prob": 0.0281
    },
    {
      "contrastive_loss": 0.5241,
      "epoch": 1.5033860045146725,
      "grad_norm": 15.702958106994629,
      "learning_rate": 8.132013722951275e-06,
      "lm_loss": 5.6649,
      "loss": 1.4999,
      "step": 666,
      "text_contrastive_loss": 0.8186,
      "train_positive_log_prob": -83.3398,
      "train_positive_token_accuracy": 0.086,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.4516,
      "epoch": 1.5056433408577878,
      "grad_norm": 16.065961837768555,
      "learning_rate": 8.12635483819152e-06,
      "lm_loss": 5.8018,
      "loss": 1.412,
      "step": 667,
      "text_contrastive_loss": 0.7605,
      "train_positive_log_prob": -85.0356,
      "train_positive_token_accuracy": 0.0776,
      "train_positive_token_prob": 0.0278
    },
    {
      "contrastive_loss": 0.4771,
      "epoch": 1.507900677200903,
      "grad_norm": 16.118410110473633,
      "learning_rate": 8.12068937045782e-06,
      "lm_loss": 5.592,
      "loss": 1.4338,
      "step": 668,
      "text_contrastive_loss": 0.7951,
      "train_positive_log_prob": -83.6645,
      "train_positive_token_accuracy": 0.0799,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.4767,
      "epoch": 1.510158013544018,
      "grad_norm": 14.855873107910156,
      "learning_rate": 8.115017331679602e-06,
      "lm_loss": 5.7021,
      "loss": 1.466,
      "step": 669,
      "text_contrastive_loss": 0.8381,
      "train_positive_log_prob": -84.6433,
      "train_positive_token_accuracy": 0.08,
      "train_positive_token_prob": 0.0289
    },
    {
      "contrastive_loss": 0.5037,
      "epoch": 1.5124153498871333,
      "grad_norm": 16.366140365600586,
      "learning_rate": 8.109338733800132e-06,
      "lm_loss": 5.6545,
      "loss": 1.5321,
      "step": 670,
      "text_contrastive_loss": 0.926,
      "train_positive_log_prob": -83.7774,
      "train_positive_token_accuracy": 0.0783,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.4368,
      "epoch": 1.5146726862302482,
      "grad_norm": 14.199953079223633,
      "learning_rate": 8.103653588776483e-06,
      "lm_loss": 5.573,
      "loss": 1.3232,
      "step": 671,
      "text_contrastive_loss": 0.6582,
      "train_positive_log_prob": -80.8225,
      "train_positive_token_accuracy": 0.0835,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3749,
      "epoch": 1.5169300225733635,
      "grad_norm": 14.447689056396484,
      "learning_rate": 8.09796190857952e-06,
      "lm_loss": 5.6185,
      "loss": 1.3952,
      "step": 672,
      "text_contrastive_loss": 0.917,
      "train_positive_log_prob": -81.0253,
      "train_positive_token_accuracy": 0.0712,
      "train_positive_token_prob": 0.0288
    },
    {
      "contrastive_loss": 0.4518,
      "epoch": 1.5191873589164786,
      "grad_norm": 13.073445320129395,
      "learning_rate": 8.09226370519386e-06,
      "lm_loss": 5.5978,
      "loss": 1.3589,
      "step": 673,
      "text_contrastive_loss": 0.6947,
      "train_positive_log_prob": -85.5665,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.4737,
      "epoch": 1.5214446952595937,
      "grad_norm": 15.316915512084961,
      "learning_rate": 8.08655899061787e-06,
      "lm_loss": 5.8387,
      "loss": 1.5077,
      "step": 674,
      "text_contrastive_loss": 0.9004,
      "train_positive_log_prob": -87.4698,
      "train_positive_token_accuracy": 0.0821,
      "train_positive_token_prob": 0.0275
    },
    {
      "contrastive_loss": 0.551,
      "epoch": 1.5237020316027088,
      "grad_norm": 15.95355224609375,
      "learning_rate": 8.080847776863609e-06,
      "lm_loss": 5.5784,
      "loss": 1.5882,
      "step": 675,
      "text_contrastive_loss": 0.9586,
      "train_positive_log_prob": -82.7395,
      "train_positive_token_accuracy": 0.0777,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.4212,
      "epoch": 1.5259593679458239,
      "grad_norm": 12.847784996032715,
      "learning_rate": 8.075130075956836e-06,
      "lm_loss": 5.6926,
      "loss": 1.3971,
      "step": 676,
      "text_contrastive_loss": 0.8133,
      "train_positive_log_prob": -84.1003,
      "train_positive_token_accuracy": 0.0791,
      "train_positive_token_prob": 0.0286
    },
    {
      "contrastive_loss": 0.4108,
      "epoch": 1.5282167042889392,
      "grad_norm": 12.959222793579102,
      "learning_rate": 8.069405899936961e-06,
      "lm_loss": 5.6467,
      "loss": 1.3945,
      "step": 677,
      "text_contrastive_loss": 0.838,
      "train_positive_log_prob": -84.1897,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.4695,
      "epoch": 1.530474040632054,
      "grad_norm": 14.164462089538574,
      "learning_rate": 8.06367526085703e-06,
      "lm_loss": 5.6998,
      "loss": 1.4137,
      "step": 678,
      "text_contrastive_loss": 0.7485,
      "train_positive_log_prob": -83.6287,
      "train_positive_token_accuracy": 0.0748,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.4206,
      "epoch": 1.5327313769751694,
      "grad_norm": 13.234776496887207,
      "learning_rate": 8.057938170783704e-06,
      "lm_loss": 5.6494,
      "loss": 1.3521,
      "step": 679,
      "text_contrastive_loss": 0.7333,
      "train_positive_log_prob": -83.8824,
      "train_positive_token_accuracy": 0.0758,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.4758,
      "epoch": 1.5349887133182845,
      "grad_norm": 14.323420524597168,
      "learning_rate": 8.052194641797217e-06,
      "lm_loss": 5.8014,
      "loss": 1.4548,
      "step": 680,
      "text_contrastive_loss": 0.7976,
      "train_positive_log_prob": -85.9592,
      "train_positive_token_accuracy": 0.0794,
      "train_positive_token_prob": 0.0283
    },
    {
      "contrastive_loss": 0.4938,
      "epoch": 1.5372460496613995,
      "grad_norm": 16.471073150634766,
      "learning_rate": 8.046444685991369e-06,
      "lm_loss": 5.6084,
      "loss": 1.4596,
      "step": 681,
      "text_contrastive_loss": 0.8099,
      "train_positive_log_prob": -83.9705,
      "train_positive_token_accuracy": 0.0806,
      "train_positive_token_prob": 0.0291
    },
    {
      "contrastive_loss": 0.4681,
      "epoch": 1.5395033860045146,
      "grad_norm": 17.488418579101562,
      "learning_rate": 8.040688315473489e-06,
      "lm_loss": 5.5904,
      "loss": 1.4683,
      "step": 682,
      "text_contrastive_loss": 0.8823,
      "train_positive_log_prob": -84.3386,
      "train_positive_token_accuracy": 0.079,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.5798,
      "epoch": 1.5417607223476297,
      "grad_norm": 16.649585723876953,
      "learning_rate": 8.034925542364412e-06,
      "lm_loss": 5.7301,
      "loss": 1.589,
      "step": 683,
      "text_contrastive_loss": 0.8725,
      "train_positive_log_prob": -85.1206,
      "train_positive_token_accuracy": 0.0686,
      "train_positive_token_prob": 0.0275
    },
    {
      "contrastive_loss": 0.4264,
      "epoch": 1.544018058690745,
      "grad_norm": 13.816652297973633,
      "learning_rate": 8.029156378798459e-06,
      "lm_loss": 5.7822,
      "loss": 1.4185,
      "step": 684,
      "text_contrastive_loss": 0.8277,
      "train_positive_log_prob": -85.9443,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0285
    },
    {
      "contrastive_loss": 0.527,
      "epoch": 1.54627539503386,
      "grad_norm": 14.386276245117188,
      "learning_rate": 8.023380836923404e-06,
      "lm_loss": 5.7393,
      "loss": 1.54,
      "step": 685,
      "text_contrastive_loss": 0.8782,
      "train_positive_log_prob": -84.275,
      "train_positive_token_accuracy": 0.0782,
      "train_positive_token_prob": 0.0288
    },
    {
      "contrastive_loss": 0.477,
      "epoch": 1.5485327313769752,
      "grad_norm": 14.044321060180664,
      "learning_rate": 8.017598928900452e-06,
      "lm_loss": 5.5485,
      "loss": 1.4258,
      "step": 686,
      "text_contrastive_loss": 0.7879,
      "train_positive_log_prob": -81.8228,
      "train_positive_token_accuracy": 0.0766,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.5478,
      "epoch": 1.5507900677200903,
      "grad_norm": 14.924182891845703,
      "learning_rate": 8.011810666904212e-06,
      "lm_loss": 5.7297,
      "loss": 1.5407,
      "step": 687,
      "text_contrastive_loss": 0.8399,
      "train_positive_log_prob": -85.5175,
      "train_positive_token_accuracy": 0.0741,
      "train_positive_token_prob": 0.0278
    },
    {
      "contrastive_loss": 0.4488,
      "epoch": 1.5530474040632054,
      "grad_norm": 15.37421703338623,
      "learning_rate": 8.006016063122672e-06,
      "lm_loss": 5.6321,
      "loss": 1.4148,
      "step": 688,
      "text_contrastive_loss": 0.8057,
      "train_positive_log_prob": -81.6189,
      "train_positive_token_accuracy": 0.0733,
      "train_positive_token_prob": 0.0282
    },
    {
      "contrastive_loss": 0.5096,
      "epoch": 1.5553047404063205,
      "grad_norm": 16.845478057861328,
      "learning_rate": 8.000215129757178e-06,
      "lm_loss": 5.6407,
      "loss": 1.5149,
      "step": 689,
      "text_contrastive_loss": 0.8824,
      "train_positive_log_prob": -83.489,
      "train_positive_token_accuracy": 0.0794,
      "train_positive_token_prob": 0.0287
    },
    {
      "contrastive_loss": 0.3692,
      "epoch": 1.5575620767494356,
      "grad_norm": 13.815431594848633,
      "learning_rate": 7.994407879022397e-06,
      "lm_loss": 5.7419,
      "loss": 1.3044,
      "step": 690,
      "text_contrastive_loss": 0.7221,
      "train_positive_log_prob": -84.4463,
      "train_positive_token_accuracy": 0.0821,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.5475,
      "epoch": 1.559819413092551,
      "grad_norm": 15.667496681213379,
      "learning_rate": 7.9885943231463e-06,
      "lm_loss": 5.7556,
      "loss": 1.6112,
      "step": 691,
      "text_contrastive_loss": 0.9763,
      "train_positive_log_prob": -84.1954,
      "train_positive_token_accuracy": 0.0764,
      "train_positive_token_prob": 0.0275
    },
    {
      "contrastive_loss": 0.4485,
      "epoch": 1.5620767494356658,
      "grad_norm": 13.958014488220215,
      "learning_rate": 7.98277447437014e-06,
      "lm_loss": 5.6992,
      "loss": 1.4239,
      "step": 692,
      "text_contrastive_loss": 0.8111,
      "train_positive_log_prob": -87.2531,
      "train_positive_token_accuracy": 0.0819,
      "train_positive_token_prob": 0.0289
    },
    {
      "contrastive_loss": 0.5523,
      "epoch": 1.564334085778781,
      "grad_norm": 15.36923885345459,
      "learning_rate": 7.976948344948412e-06,
      "lm_loss": 5.775,
      "loss": 1.5736,
      "step": 693,
      "text_contrastive_loss": 0.8876,
      "train_positive_log_prob": -85.3385,
      "train_positive_token_accuracy": 0.079,
      "train_positive_token_prob": 0.0279
    },
    {
      "contrastive_loss": 0.5094,
      "epoch": 1.5665914221218962,
      "grad_norm": 13.96995735168457,
      "learning_rate": 7.971115947148842e-06,
      "lm_loss": 5.5939,
      "loss": 1.4705,
      "step": 694,
      "text_contrastive_loss": 0.8034,
      "train_positive_log_prob": -82.3289,
      "train_positive_token_accuracy": 0.0756,
      "train_positive_token_prob": 0.0275
    },
    {
      "contrastive_loss": 0.5592,
      "epoch": 1.5688487584650113,
      "grad_norm": 16.41417121887207,
      "learning_rate": 7.965277293252354e-06,
      "lm_loss": 5.5482,
      "loss": 1.5618,
      "step": 695,
      "text_contrastive_loss": 0.8955,
      "train_positive_log_prob": -82.7689,
      "train_positive_token_accuracy": 0.0712,
      "train_positive_token_prob": 0.0285
    },
    {
      "contrastive_loss": 0.5122,
      "epoch": 1.5711060948081266,
      "grad_norm": 15.281829833984375,
      "learning_rate": 7.95943239555304e-06,
      "lm_loss": 5.6175,
      "loss": 1.4699,
      "step": 696,
      "text_contrastive_loss": 0.7918,
      "train_positive_log_prob": -81.5889,
      "train_positive_token_accuracy": 0.0734,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.4928,
      "epoch": 1.5733634311512414,
      "grad_norm": 16.319522857666016,
      "learning_rate": 7.953581266358148e-06,
      "lm_loss": 5.6711,
      "loss": 1.4936,
      "step": 697,
      "text_contrastive_loss": 0.8676,
      "train_positive_log_prob": -84.4906,
      "train_positive_token_accuracy": 0.0732,
      "train_positive_token_prob": 0.0279
    },
    {
      "contrastive_loss": 0.4439,
      "epoch": 1.5756207674943568,
      "grad_norm": 16.381437301635742,
      "learning_rate": 7.94772391798804e-06,
      "lm_loss": 5.5834,
      "loss": 1.2828,
      "step": 698,
      "text_contrastive_loss": 0.5611,
      "train_positive_log_prob": -82.8392,
      "train_positive_token_accuracy": 0.0748,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.528,
      "epoch": 1.5778781038374716,
      "grad_norm": 15.742697715759277,
      "learning_rate": 7.941860362776176e-06,
      "lm_loss": 5.5959,
      "loss": 1.534,
      "step": 699,
      "text_contrastive_loss": 0.8929,
      "train_positive_log_prob": -78.8482,
      "train_positive_token_accuracy": 0.0743,
      "train_positive_token_prob": 0.0279
    },
    {
      "contrastive_loss": 0.4935,
      "epoch": 1.580135440180587,
      "grad_norm": 14.445799827575684,
      "learning_rate": 7.935990613069087e-06,
      "lm_loss": 5.6173,
      "loss": 1.4992,
      "step": 700,
      "text_contrastive_loss": 0.888,
      "train_positive_log_prob": -81.5845,
      "train_positive_token_accuracy": 0.0791,
      "train_positive_token_prob": 0.0291
    },
    {
      "contrastive_loss": 0.4314,
      "epoch": 1.582392776523702,
      "grad_norm": 14.703787803649902,
      "learning_rate": 7.930114681226341e-06,
      "lm_loss": 5.7236,
      "loss": 1.3482,
      "step": 701,
      "text_contrastive_loss": 0.6888,
      "train_positive_log_prob": -82.8079,
      "train_positive_token_accuracy": 0.078,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.4213,
      "epoch": 1.5846501128668171,
      "grad_norm": 16.77519416809082,
      "learning_rate": 7.924232579620533e-06,
      "lm_loss": 5.6974,
      "loss": 1.4344,
      "step": 702,
      "text_contrastive_loss": 0.8867,
      "train_positive_log_prob": -83.9277,
      "train_positive_token_accuracy": 0.0721,
      "train_positive_token_prob": 0.0283
    },
    {
      "contrastive_loss": 0.3852,
      "epoch": 1.5869074492099324,
      "grad_norm": 13.184041023254395,
      "learning_rate": 7.91834432063724e-06,
      "lm_loss": 5.5581,
      "loss": 1.293,
      "step": 703,
      "text_contrastive_loss": 0.704,
      "train_positive_log_prob": -82.5643,
      "train_positive_token_accuracy": 0.0784,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.4773,
      "epoch": 1.5891647855530473,
      "grad_norm": 14.658939361572266,
      "learning_rate": 7.912449916675008e-06,
      "lm_loss": 5.5706,
      "loss": 1.4164,
      "step": 704,
      "text_contrastive_loss": 0.764,
      "train_positive_log_prob": -83.3206,
      "train_positive_token_accuracy": 0.0755,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.5394,
      "epoch": 1.5914221218961626,
      "grad_norm": 15.919829368591309,
      "learning_rate": 7.90654938014533e-06,
      "lm_loss": 5.5969,
      "loss": 1.5109,
      "step": 705,
      "text_contrastive_loss": 0.8237,
      "train_positive_log_prob": -82.9038,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.4613,
      "epoch": 1.5936794582392777,
      "grad_norm": 14.476689338684082,
      "learning_rate": 7.900642723472596e-06,
      "lm_loss": 5.8075,
      "loss": 1.4326,
      "step": 706,
      "text_contrastive_loss": 0.7811,
      "train_positive_log_prob": -88.1512,
      "train_positive_token_accuracy": 0.0772,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.4404,
      "epoch": 1.5959367945823928,
      "grad_norm": 16.509441375732422,
      "learning_rate": 7.894729959094097e-06,
      "lm_loss": 5.7106,
      "loss": 1.4022,
      "step": 707,
      "text_contrastive_loss": 0.7815,
      "train_positive_log_prob": -83.4062,
      "train_positive_token_accuracy": 0.0752,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.5398,
      "epoch": 1.5981941309255079,
      "grad_norm": 15.218833923339844,
      "learning_rate": 7.888811099459974e-06,
      "lm_loss": 5.695,
      "loss": 1.5809,
      "step": 708,
      "text_contrastive_loss": 0.9432,
      "train_positive_log_prob": -85.3343,
      "train_positive_token_accuracy": 0.086,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.4894,
      "epoch": 1.600451467268623,
      "grad_norm": 17.24501609802246,
      "learning_rate": 7.882886157033209e-06,
      "lm_loss": 5.6748,
      "loss": 1.4233,
      "step": 709,
      "text_contrastive_loss": 0.7329,
      "train_positive_log_prob": -84.2801,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.4836,
      "epoch": 1.6027088036117383,
      "grad_norm": 15.260598182678223,
      "learning_rate": 7.876955144289594e-06,
      "lm_loss": 5.6995,
      "loss": 1.4934,
      "step": 710,
      "text_contrastive_loss": 0.8796,
      "train_positive_log_prob": -84.0514,
      "train_positive_token_accuracy": 0.0774,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.5571,
      "epoch": 1.6049661399548532,
      "grad_norm": 17.923580169677734,
      "learning_rate": 7.871018073717693e-06,
      "lm_loss": 5.6762,
      "loss": 1.5727,
      "step": 711,
      "text_contrastive_loss": 0.8961,
      "train_positive_log_prob": -83.0791,
      "train_positive_token_accuracy": 0.0787,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.5156,
      "epoch": 1.6072234762979685,
      "grad_norm": 16.633056640625,
      "learning_rate": 7.865074957818839e-06,
      "lm_loss": 5.8299,
      "loss": 1.4893,
      "step": 712,
      "text_contrastive_loss": 0.7814,
      "train_positive_log_prob": -86.8327,
      "train_positive_token_accuracy": 0.081,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.4718,
      "epoch": 1.6094808126410836,
      "grad_norm": 15.569839477539062,
      "learning_rate": 7.859125809107082e-06,
      "lm_loss": 5.6042,
      "loss": 1.4569,
      "step": 713,
      "text_contrastive_loss": 0.8494,
      "train_positive_log_prob": -85.0284,
      "train_positive_token_accuracy": 0.0785,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.5376,
      "epoch": 1.6117381489841986,
      "grad_norm": 17.901193618774414,
      "learning_rate": 7.853170640109182e-06,
      "lm_loss": 5.7166,
      "loss": 1.5815,
      "step": 714,
      "text_contrastive_loss": 0.9443,
      "train_positive_log_prob": -86.239,
      "train_positive_token_accuracy": 0.0785,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.4856,
      "epoch": 1.6139954853273137,
      "grad_norm": 14.7576322555542,
      "learning_rate": 7.847209463364574e-06,
      "lm_loss": 5.5901,
      "loss": 1.4724,
      "step": 715,
      "text_contrastive_loss": 0.8555,
      "train_positive_log_prob": -80.5992,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.512,
      "epoch": 1.6162528216704288,
      "grad_norm": 15.334814071655273,
      "learning_rate": 7.841242291425342e-06,
      "lm_loss": 5.5756,
      "loss": 1.5143,
      "step": 716,
      "text_contrastive_loss": 0.8895,
      "train_positive_log_prob": -81.8691,
      "train_positive_token_accuracy": 0.075,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.5843,
      "epoch": 1.6185101580135441,
      "grad_norm": 17.95580291748047,
      "learning_rate": 7.835269136856194e-06,
      "lm_loss": 5.6804,
      "loss": 1.6047,
      "step": 717,
      "text_contrastive_loss": 0.9047,
      "train_positive_log_prob": -82.1908,
      "train_positive_token_accuracy": 0.0745,
      "train_positive_token_prob": 0.0291
    },
    {
      "contrastive_loss": 0.4107,
      "epoch": 1.620767494356659,
      "grad_norm": 15.698619842529297,
      "learning_rate": 7.829290012234438e-06,
      "lm_loss": 5.5673,
      "loss": 1.332,
      "step": 718,
      "text_contrastive_loss": 0.7291,
      "train_positive_log_prob": -81.011,
      "train_positive_token_accuracy": 0.0795,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.4688,
      "epoch": 1.6230248306997743,
      "grad_norm": 16.278276443481445,
      "learning_rate": 7.823304930149949e-06,
      "lm_loss": 5.4734,
      "loss": 1.4136,
      "step": 719,
      "text_contrastive_loss": 0.795,
      "train_positive_log_prob": -79.1723,
      "train_positive_token_accuracy": 0.0789,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.4137,
      "epoch": 1.6252821670428894,
      "grad_norm": 15.17823314666748,
      "learning_rate": 7.817313903205148e-06,
      "lm_loss": 5.6698,
      "loss": 1.3413,
      "step": 720,
      "text_contrastive_loss": 0.7212,
      "train_positive_log_prob": -85.289,
      "train_positive_token_accuracy": 0.0823,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.443,
      "epoch": 1.6275395033860045,
      "grad_norm": 16.49557876586914,
      "learning_rate": 7.811316944014974e-06,
      "lm_loss": 5.6835,
      "loss": 1.4397,
      "step": 721,
      "text_contrastive_loss": 0.8566,
      "train_positive_log_prob": -83.278,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.0291
    },
    {
      "contrastive_loss": 0.4804,
      "epoch": 1.6297968397291196,
      "grad_norm": 16.320707321166992,
      "learning_rate": 7.805314065206857e-06,
      "lm_loss": 5.578,
      "loss": 1.4139,
      "step": 722,
      "text_contrastive_loss": 0.7515,
      "train_positive_log_prob": -81.5975,
      "train_positive_token_accuracy": 0.0804,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.4617,
      "epoch": 1.6320541760722347,
      "grad_norm": 16.48221206665039,
      "learning_rate": 7.799305279420691e-06,
      "lm_loss": 5.6306,
      "loss": 1.4554,
      "step": 723,
      "text_contrastive_loss": 0.8613,
      "train_positive_log_prob": -82.1051,
      "train_positive_token_accuracy": 0.075,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.4122,
      "epoch": 1.63431151241535,
      "grad_norm": 14.674779891967773,
      "learning_rate": 7.793290599308807e-06,
      "lm_loss": 5.6626,
      "loss": 1.4416,
      "step": 724,
      "text_contrastive_loss": 0.9262,
      "train_positive_log_prob": -83.4107,
      "train_positive_token_accuracy": 0.0719,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.5598,
      "epoch": 1.6365688487584649,
      "grad_norm": 19.37317657470703,
      "learning_rate": 7.78727003753595e-06,
      "lm_loss": 5.586,
      "loss": 1.5882,
      "step": 725,
      "text_contrastive_loss": 0.9396,
      "train_positive_log_prob": -80.9306,
      "train_positive_token_accuracy": 0.0817,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.4673,
      "epoch": 1.6388261851015802,
      "grad_norm": 16.270069122314453,
      "learning_rate": 7.78124360677925e-06,
      "lm_loss": 5.5472,
      "loss": 1.4907,
      "step": 726,
      "text_contrastive_loss": 0.9374,
      "train_positive_log_prob": -83.2954,
      "train_positive_token_accuracy": 0.0827,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.4777,
      "epoch": 1.6410835214446953,
      "grad_norm": 15.725870132446289,
      "learning_rate": 7.775211319728191e-06,
      "lm_loss": 5.6535,
      "loss": 1.4712,
      "step": 727,
      "text_contrastive_loss": 0.8564,
      "train_positive_log_prob": -83.6781,
      "train_positive_token_accuracy": 0.0726,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.4651,
      "epoch": 1.6433408577878104,
      "grad_norm": 14.012334823608398,
      "learning_rate": 7.769173189084589e-06,
      "lm_loss": 5.6447,
      "loss": 1.4904,
      "step": 728,
      "text_contrastive_loss": 0.9217,
      "train_positive_log_prob": -83.985,
      "train_positive_token_accuracy": 0.0816,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.5654,
      "epoch": 1.6455981941309257,
      "grad_norm": 16.746809005737305,
      "learning_rate": 7.763129227562568e-06,
      "lm_loss": 5.6947,
      "loss": 1.5699,
      "step": 729,
      "text_contrastive_loss": 0.8701,
      "train_positive_log_prob": -84.9315,
      "train_positive_token_accuracy": 0.077,
      "train_positive_token_prob": 0.0287
    },
    {
      "contrastive_loss": 0.4177,
      "epoch": 1.6478555304740405,
      "grad_norm": 11.539710998535156,
      "learning_rate": 7.757079447888529e-06,
      "lm_loss": 5.6066,
      "loss": 1.3854,
      "step": 730,
      "text_contrastive_loss": 0.8141,
      "train_positive_log_prob": -83.3069,
      "train_positive_token_accuracy": 0.076,
      "train_positive_token_prob": 0.0288
    },
    {
      "contrastive_loss": 0.4228,
      "epoch": 1.6501128668171559,
      "grad_norm": 15.049748420715332,
      "learning_rate": 7.75102386280112e-06,
      "lm_loss": 5.57,
      "loss": 1.421,
      "step": 731,
      "text_contrastive_loss": 0.8824,
      "train_positive_log_prob": -82.678,
      "train_positive_token_accuracy": 0.0774,
      "train_positive_token_prob": 0.0286
    },
    {
      "contrastive_loss": 0.4932,
      "epoch": 1.6523702031602707,
      "grad_norm": 15.766206741333008,
      "learning_rate": 7.744962485051217e-06,
      "lm_loss": 5.6068,
      "loss": 1.4178,
      "step": 732,
      "text_contrastive_loss": 0.728,
      "train_positive_log_prob": -82.5477,
      "train_positive_token_accuracy": 0.0785,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.5588,
      "epoch": 1.654627539503386,
      "grad_norm": 14.77518367767334,
      "learning_rate": 7.738895327401891e-06,
      "lm_loss": 5.6551,
      "loss": 1.5478,
      "step": 733,
      "text_contrastive_loss": 0.8469,
      "train_positive_log_prob": -84.3434,
      "train_positive_token_accuracy": 0.0765,
      "train_positive_token_prob": 0.0288
    },
    {
      "contrastive_loss": 0.4651,
      "epoch": 1.6568848758465011,
      "grad_norm": 16.423664093017578,
      "learning_rate": 7.732822402628385e-06,
      "lm_loss": 5.5931,
      "loss": 1.4284,
      "step": 734,
      "text_contrastive_loss": 0.8079,
      "train_positive_log_prob": -81.4274,
      "train_positive_token_accuracy": 0.0856,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.5366,
      "epoch": 1.6591422121896162,
      "grad_norm": 14.504514694213867,
      "learning_rate": 7.726743723518087e-06,
      "lm_loss": 5.5241,
      "loss": 1.4753,
      "step": 735,
      "text_contrastive_loss": 0.7726,
      "train_positive_log_prob": -82.1937,
      "train_positive_token_accuracy": 0.0856,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.5463,
      "epoch": 1.6613995485327315,
      "grad_norm": 16.042978286743164,
      "learning_rate": 7.720659302870496e-06,
      "lm_loss": 5.6109,
      "loss": 1.561,
      "step": 736,
      "text_contrastive_loss": 0.9071,
      "train_positive_log_prob": -82.9788,
      "train_positive_token_accuracy": 0.0851,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.5426,
      "epoch": 1.6636568848758464,
      "grad_norm": 15.57763957977295,
      "learning_rate": 7.714569153497204e-06,
      "lm_loss": 5.7038,
      "loss": 1.5502,
      "step": 737,
      "text_contrastive_loss": 0.8744,
      "train_positive_log_prob": -83.8156,
      "train_positive_token_accuracy": 0.076,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.3659,
      "epoch": 1.6659142212189617,
      "grad_norm": 13.961685180664062,
      "learning_rate": 7.708473288221868e-06,
      "lm_loss": 5.7913,
      "loss": 1.3805,
      "step": 738,
      "text_contrastive_loss": 0.8711,
      "train_positive_log_prob": -85.8103,
      "train_positive_token_accuracy": 0.0714,
      "train_positive_token_prob": 0.0277
    },
    {
      "contrastive_loss": 0.4062,
      "epoch": 1.6681715575620768,
      "grad_norm": 14.212257385253906,
      "learning_rate": 7.702371719880178e-06,
      "lm_loss": 5.7684,
      "loss": 1.3621,
      "step": 739,
      "text_contrastive_loss": 0.7582,
      "train_positive_log_prob": -84.7057,
      "train_positive_token_accuracy": 0.0803,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.5704,
      "epoch": 1.670428893905192,
      "grad_norm": 17.218656539916992,
      "learning_rate": 7.696264461319831e-06,
      "lm_loss": 5.5532,
      "loss": 1.5963,
      "step": 740,
      "text_contrastive_loss": 0.9412,
      "train_positive_log_prob": -80.7605,
      "train_positive_token_accuracy": 0.0741,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.4466,
      "epoch": 1.672686230248307,
      "grad_norm": 14.9937744140625,
      "learning_rate": 7.69015152540051e-06,
      "lm_loss": 5.5905,
      "loss": 1.4574,
      "step": 741,
      "text_contrastive_loss": 0.9035,
      "train_positive_log_prob": -82.3747,
      "train_positive_token_accuracy": 0.0835,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.4056,
      "epoch": 1.674943566591422,
      "grad_norm": 14.841171264648438,
      "learning_rate": 7.684032924993845e-06,
      "lm_loss": 5.6774,
      "loss": 1.3664,
      "step": 742,
      "text_contrastive_loss": 0.7862,
      "train_positive_log_prob": -83.3485,
      "train_positive_token_accuracy": 0.0745,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.3857,
      "epoch": 1.6772009029345374,
      "grad_norm": 13.606510162353516,
      "learning_rate": 7.677908672983404e-06,
      "lm_loss": 5.6338,
      "loss": 1.3532,
      "step": 743,
      "text_contrastive_loss": 0.8083,
      "train_positive_log_prob": -83.7588,
      "train_positive_token_accuracy": 0.0772,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.4237,
      "epoch": 1.6794582392776523,
      "grad_norm": 14.244503021240234,
      "learning_rate": 7.671778782264647e-06,
      "lm_loss": 5.6147,
      "loss": 1.3599,
      "step": 744,
      "text_contrastive_loss": 0.7493,
      "train_positive_log_prob": -81.7582,
      "train_positive_token_accuracy": 0.0799,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.535,
      "epoch": 1.6817155756207676,
      "grad_norm": 14.576723098754883,
      "learning_rate": 7.66564326574491e-06,
      "lm_loss": 5.7495,
      "loss": 1.5361,
      "step": 745,
      "text_contrastive_loss": 0.8524,
      "train_positive_log_prob": -84.6237,
      "train_positive_token_accuracy": 0.074,
      "train_positive_token_prob": 0.0286
    },
    {
      "contrastive_loss": 0.4851,
      "epoch": 1.6839729119638827,
      "grad_norm": 14.613896369934082,
      "learning_rate": 7.65950213634337e-06,
      "lm_loss": 5.713,
      "loss": 1.5287,
      "step": 746,
      "text_contrastive_loss": 0.9446,
      "train_positive_log_prob": -84.5161,
      "train_positive_token_accuracy": 0.076,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.3686,
      "epoch": 1.6862302483069977,
      "grad_norm": 15.065150260925293,
      "learning_rate": 7.653355406991034e-06,
      "lm_loss": 5.5927,
      "loss": 1.3653,
      "step": 747,
      "text_contrastive_loss": 0.8748,
      "train_positive_log_prob": -83.303,
      "train_positive_token_accuracy": 0.0787,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.4851,
      "epoch": 1.6884875846501128,
      "grad_norm": 17.209030151367188,
      "learning_rate": 7.64720309063069e-06,
      "lm_loss": 5.6565,
      "loss": 1.4543,
      "step": 748,
      "text_contrastive_loss": 0.8071,
      "train_positive_log_prob": -83.6971,
      "train_positive_token_accuracy": 0.0784,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.4535,
      "epoch": 1.690744920993228,
      "grad_norm": 12.747918128967285,
      "learning_rate": 7.641045200216896e-06,
      "lm_loss": 5.5735,
      "loss": 1.4081,
      "step": 749,
      "text_contrastive_loss": 0.7944,
      "train_positive_log_prob": -83.2573,
      "train_positive_token_accuracy": 0.0782,
      "train_positive_token_prob": 0.0282
    },
    {
      "contrastive_loss": 0.4117,
      "epoch": 1.6930022573363432,
      "grad_norm": 13.951990127563477,
      "learning_rate": 7.634881748715941e-06,
      "lm_loss": 5.8124,
      "loss": 1.3534,
      "step": 750,
      "text_contrastive_loss": 0.7211,
      "train_positive_log_prob": -85.0944,
      "train_positive_token_accuracy": 0.072,
      "train_positive_token_prob": 0.0283
    },
    {
      "contrastive_loss": 0.5052,
      "epoch": 1.695259593679458,
      "grad_norm": 16.40280532836914,
      "learning_rate": 7.628712749105831e-06,
      "lm_loss": 5.6827,
      "loss": 1.5382,
      "step": 751,
      "text_contrastive_loss": 0.9294,
      "train_positive_log_prob": -83.285,
      "train_positive_token_accuracy": 0.0712,
      "train_positive_token_prob": 0.0276
    },
    {
      "contrastive_loss": 0.3899,
      "epoch": 1.6975169300225734,
      "grad_norm": 13.680509567260742,
      "learning_rate": 7.622538214376248e-06,
      "lm_loss": 5.613,
      "loss": 1.2772,
      "step": 752,
      "text_contrastive_loss": 0.6521,
      "train_positive_log_prob": -83.7103,
      "train_positive_token_accuracy": 0.0767,
      "train_positive_token_prob": 0.0281
    },
    {
      "contrastive_loss": 0.4275,
      "epoch": 1.6997742663656885,
      "grad_norm": 14.89928150177002,
      "learning_rate": 7.616358157528535e-06,
      "lm_loss": 5.5666,
      "loss": 1.3869,
      "step": 753,
      "text_contrastive_loss": 0.8054,
      "train_positive_log_prob": -81.299,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.4492,
      "epoch": 1.7020316027088036,
      "grad_norm": 16.91277503967285,
      "learning_rate": 7.610172591575656e-06,
      "lm_loss": 5.7597,
      "loss": 1.392,
      "step": 754,
      "text_contrastive_loss": 0.7337,
      "train_positive_log_prob": -84.726,
      "train_positive_token_accuracy": 0.0757,
      "train_positive_token_prob": 0.0274
    },
    {
      "contrastive_loss": 0.4825,
      "epoch": 1.7042889390519187,
      "grad_norm": 15.468907356262207,
      "learning_rate": 7.60398152954218e-06,
      "lm_loss": 5.6553,
      "loss": 1.3819,
      "step": 755,
      "text_contrastive_loss": 0.6679,
      "train_positive_log_prob": -83.44,
      "train_positive_token_accuracy": 0.0694,
      "train_positive_token_prob": 0.0277
    },
    {
      "contrastive_loss": 0.4433,
      "epoch": 1.7065462753950338,
      "grad_norm": 14.060961723327637,
      "learning_rate": 7.597784984464248e-06,
      "lm_loss": 5.6453,
      "loss": 1.4166,
      "step": 756,
      "text_contrastive_loss": 0.8175,
      "train_positive_log_prob": -84.5482,
      "train_positive_token_accuracy": 0.0775,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.5329,
      "epoch": 1.708803611738149,
      "grad_norm": 15.737454414367676,
      "learning_rate": 7.5915829693895435e-06,
      "lm_loss": 5.6437,
      "loss": 1.5176,
      "step": 757,
      "text_contrastive_loss": 0.8408,
      "train_positive_log_prob": -82.3182,
      "train_positive_token_accuracy": 0.0766,
      "train_positive_token_prob": 0.0277
    },
    {
      "contrastive_loss": 0.5537,
      "epoch": 1.711060948081264,
      "grad_norm": 15.397076606750488,
      "learning_rate": 7.585375497377271e-06,
      "lm_loss": 5.6778,
      "loss": 1.6266,
      "step": 758,
      "text_contrastive_loss": 1.0103,
      "train_positive_log_prob": -83.88,
      "train_positive_token_accuracy": 0.0738,
      "train_positive_token_prob": 0.0273
    },
    {
      "contrastive_loss": 0.455,
      "epoch": 1.7133182844243793,
      "grad_norm": 14.482380867004395,
      "learning_rate": 7.579162581498125e-06,
      "lm_loss": 5.6564,
      "loss": 1.461,
      "step": 759,
      "text_contrastive_loss": 0.8807,
      "train_positive_log_prob": -83.6313,
      "train_positive_token_accuracy": 0.0721,
      "train_positive_token_prob": 0.0285
    },
    {
      "contrastive_loss": 0.5666,
      "epoch": 1.7155756207674944,
      "grad_norm": 16.678985595703125,
      "learning_rate": 7.572944234834261e-06,
      "lm_loss": 5.5595,
      "loss": 1.5437,
      "step": 760,
      "text_contrastive_loss": 0.8424,
      "train_positive_log_prob": -81.9402,
      "train_positive_token_accuracy": 0.0742,
      "train_positive_token_prob": 0.0279
    },
    {
      "contrastive_loss": 0.5585,
      "epoch": 1.7178329571106095,
      "grad_norm": 14.428262710571289,
      "learning_rate": 7.5667204704792706e-06,
      "lm_loss": 5.5821,
      "loss": 1.5035,
      "step": 761,
      "text_contrastive_loss": 0.7737,
      "train_positive_log_prob": -83.0824,
      "train_positive_token_accuracy": 0.0814,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.5673,
      "epoch": 1.7200902934537246,
      "grad_norm": 15.505300521850586,
      "learning_rate": 7.5604913015381535e-06,
      "lm_loss": 5.585,
      "loss": 1.6184,
      "step": 762,
      "text_contrastive_loss": 0.9851,
      "train_positive_log_prob": -82.4666,
      "train_positive_token_accuracy": 0.0714,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.4963,
      "epoch": 1.7223476297968396,
      "grad_norm": 14.312376976013184,
      "learning_rate": 7.554256741127291e-06,
      "lm_loss": 5.6173,
      "loss": 1.4956,
      "step": 763,
      "text_contrastive_loss": 0.8751,
      "train_positive_log_prob": -83.3581,
      "train_positive_token_accuracy": 0.0813,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.4345,
      "epoch": 1.724604966139955,
      "grad_norm": 14.42605209350586,
      "learning_rate": 7.548016802374412e-06,
      "lm_loss": 5.5431,
      "loss": 1.4118,
      "step": 764,
      "text_contrastive_loss": 0.8461,
      "train_positive_log_prob": -79.7033,
      "train_positive_token_accuracy": 0.0742,
      "train_positive_token_prob": 0.0279
    },
    {
      "contrastive_loss": 0.5725,
      "epoch": 1.7268623024830698,
      "grad_norm": 15.596787452697754,
      "learning_rate": 7.541771498418575e-06,
      "lm_loss": 5.714,
      "loss": 1.5954,
      "step": 765,
      "text_contrastive_loss": 0.9031,
      "train_positive_log_prob": -84.3706,
      "train_positive_token_accuracy": 0.0749,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.4699,
      "epoch": 1.7291196388261851,
      "grad_norm": 14.061925888061523,
      "learning_rate": 7.535520842410136e-06,
      "lm_loss": 5.5871,
      "loss": 1.4654,
      "step": 766,
      "text_contrastive_loss": 0.8736,
      "train_positive_log_prob": -81.562,
      "train_positive_token_accuracy": 0.0736,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.4901,
      "epoch": 1.7313769751693002,
      "grad_norm": 17.323240280151367,
      "learning_rate": 7.529264847510715e-06,
      "lm_loss": 5.6402,
      "loss": 1.4201,
      "step": 767,
      "text_contrastive_loss": 0.7319,
      "train_positive_log_prob": -84.026,
      "train_positive_token_accuracy": 0.0813,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.5083,
      "epoch": 1.7336343115124153,
      "grad_norm": 14.857467651367188,
      "learning_rate": 7.52300352689318e-06,
      "lm_loss": 5.6372,
      "loss": 1.5528,
      "step": 768,
      "text_contrastive_loss": 0.9616,
      "train_positive_log_prob": -84.6456,
      "train_positive_token_accuracy": 0.0744,
      "train_positive_token_prob": 0.0289
    },
    {
      "contrastive_loss": 0.497,
      "epoch": 1.7358916478555306,
      "grad_norm": 16.907846450805664,
      "learning_rate": 7.516736893741611e-06,
      "lm_loss": 5.5738,
      "loss": 1.4939,
      "step": 769,
      "text_contrastive_loss": 0.8791,
      "train_positive_log_prob": -83.9804,
      "train_positive_token_accuracy": 0.0767,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.3947,
      "epoch": 1.7381489841986455,
      "grad_norm": 12.01798152923584,
      "learning_rate": 7.510464961251271e-06,
      "lm_loss": 5.6794,
      "loss": 1.2747,
      "step": 770,
      "text_contrastive_loss": 0.6241,
      "train_positive_log_prob": -84.013,
      "train_positive_token_accuracy": 0.0737,
      "train_positive_token_prob": 0.0289
    },
    {
      "contrastive_loss": 0.4547,
      "epoch": 1.7404063205417608,
      "grad_norm": 14.21453857421875,
      "learning_rate": 7.5041877426285856e-06,
      "lm_loss": 5.5642,
      "loss": 1.4135,
      "step": 771,
      "text_contrastive_loss": 0.8047,
      "train_positive_log_prob": -83.0057,
      "train_positive_token_accuracy": 0.0799,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.5835,
      "epoch": 1.742663656884876,
      "grad_norm": 16.025463104248047,
      "learning_rate": 7.49790525109111e-06,
      "lm_loss": 5.5922,
      "loss": 1.6391,
      "step": 772,
      "text_contrastive_loss": 0.9926,
      "train_positive_log_prob": -83.6232,
      "train_positive_token_accuracy": 0.081,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3763,
      "epoch": 1.744920993227991,
      "grad_norm": 12.589165687561035,
      "learning_rate": 7.491617499867502e-06,
      "lm_loss": 5.4859,
      "loss": 1.2875,
      "step": 773,
      "text_contrastive_loss": 0.7251,
      "train_positive_log_prob": -81.8992,
      "train_positive_token_accuracy": 0.0827,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3712,
      "epoch": 1.747178329571106,
      "grad_norm": 12.508708953857422,
      "learning_rate": 7.485324502197494e-06,
      "lm_loss": 5.6359,
      "loss": 1.2815,
      "step": 774,
      "text_contrastive_loss": 0.6935,
      "train_positive_log_prob": -84.3179,
      "train_positive_token_accuracy": 0.0856,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3641,
      "epoch": 1.7494356659142212,
      "grad_norm": 13.209696769714355,
      "learning_rate": 7.479026271331864e-06,
      "lm_loss": 5.5879,
      "loss": 1.2816,
      "step": 775,
      "text_contrastive_loss": 0.7173,
      "train_positive_log_prob": -81.8326,
      "train_positive_token_accuracy": 0.0843,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.546,
      "epoch": 1.7516930022573365,
      "grad_norm": 15.404215812683105,
      "learning_rate": 7.472722820532414e-06,
      "lm_loss": 5.5922,
      "loss": 1.5356,
      "step": 776,
      "text_contrastive_loss": 0.8608,
      "train_positive_log_prob": -82.8356,
      "train_positive_token_accuracy": 0.0823,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.6744,
      "epoch": 1.7539503386004514,
      "grad_norm": 18.27569007873535,
      "learning_rate": 7.466414163071934e-06,
      "lm_loss": 5.5766,
      "loss": 1.6637,
      "step": 777,
      "text_contrastive_loss": 0.8633,
      "train_positive_log_prob": -82.8614,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.5493,
      "epoch": 1.7562076749435667,
      "grad_norm": 18.299306869506836,
      "learning_rate": 7.460100312234176e-06,
      "lm_loss": 5.6565,
      "loss": 1.5948,
      "step": 778,
      "text_contrastive_loss": 0.9596,
      "train_positive_log_prob": -84.4055,
      "train_positive_token_accuracy": 0.073,
      "train_positive_token_prob": 0.0282
    },
    {
      "contrastive_loss": 0.4558,
      "epoch": 1.7584650112866818,
      "grad_norm": 14.698132514953613,
      "learning_rate": 7.453781281313831e-06,
      "lm_loss": 5.5634,
      "loss": 1.3756,
      "step": 779,
      "text_contrastive_loss": 0.7269,
      "train_positive_log_prob": -81.147,
      "train_positive_token_accuracy": 0.0852,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.5006,
      "epoch": 1.7607223476297968,
      "grad_norm": 14.811718940734863,
      "learning_rate": 7.447457083616494e-06,
      "lm_loss": 5.5326,
      "loss": 1.4775,
      "step": 780,
      "text_contrastive_loss": 0.8473,
      "train_positive_log_prob": -82.9922,
      "train_positive_token_accuracy": 0.0831,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.4054,
      "epoch": 1.762979683972912,
      "grad_norm": 15.256193161010742,
      "learning_rate": 7.441127732458642e-06,
      "lm_loss": 5.5708,
      "loss": 1.379,
      "step": 781,
      "text_contrastive_loss": 0.8329,
      "train_positive_log_prob": -83.2141,
      "train_positive_token_accuracy": 0.0795,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.4579,
      "epoch": 1.765237020316027,
      "grad_norm": 15.612141609191895,
      "learning_rate": 7.434793241167601e-06,
      "lm_loss": 5.4878,
      "loss": 1.5079,
      "step": 782,
      "text_contrastive_loss": 1.0025,
      "train_positive_log_prob": -80.8577,
      "train_positive_token_accuracy": 0.0885,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.5026,
      "epoch": 1.7674943566591423,
      "grad_norm": 16.013240814208984,
      "learning_rate": 7.428453623081522e-06,
      "lm_loss": 5.5057,
      "loss": 1.5084,
      "step": 783,
      "text_contrastive_loss": 0.9105,
      "train_positive_log_prob": -79.1509,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.5301,
      "epoch": 1.7697516930022572,
      "grad_norm": 14.345297813415527,
      "learning_rate": 7.422108891549349e-06,
      "lm_loss": 5.5626,
      "loss": 1.5358,
      "step": 784,
      "text_contrastive_loss": 0.8988,
      "train_positive_log_prob": -83.0716,
      "train_positive_token_accuracy": 0.0842,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.4109,
      "epoch": 1.7720090293453725,
      "grad_norm": 14.167484283447266,
      "learning_rate": 7.415759059930799e-06,
      "lm_loss": 5.5699,
      "loss": 1.323,
      "step": 785,
      "text_contrastive_loss": 0.7102,
      "train_positive_log_prob": -80.807,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.4247,
      "epoch": 1.7742663656884876,
      "grad_norm": 13.521806716918945,
      "learning_rate": 7.409404141596319e-06,
      "lm_loss": 5.6025,
      "loss": 1.3949,
      "step": 786,
      "text_contrastive_loss": 0.8197,
      "train_positive_log_prob": -81.7631,
      "train_positive_token_accuracy": 0.0736,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.4207,
      "epoch": 1.7765237020316027,
      "grad_norm": 15.362735748291016,
      "learning_rate": 7.403044149927074e-06,
      "lm_loss": 5.4708,
      "loss": 1.3121,
      "step": 787,
      "text_contrastive_loss": 0.6886,
      "train_positive_log_prob": -78.3631,
      "train_positive_token_accuracy": 0.0725,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.4382,
      "epoch": 1.7787810383747178,
      "grad_norm": 13.427018165588379,
      "learning_rate": 7.396679098314908e-06,
      "lm_loss": 5.5537,
      "loss": 1.3769,
      "step": 788,
      "text_contrastive_loss": 0.7667,
      "train_positive_log_prob": -82.6966,
      "train_positive_token_accuracy": 0.0805,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.5195,
      "epoch": 1.7810383747178329,
      "grad_norm": 14.191274642944336,
      "learning_rate": 7.390309000162321e-06,
      "lm_loss": 5.6429,
      "loss": 1.5185,
      "step": 789,
      "text_contrastive_loss": 0.8693,
      "train_positive_log_prob": -84.3314,
      "train_positive_token_accuracy": 0.0729,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.4409,
      "epoch": 1.7832957110609482,
      "grad_norm": 15.111398696899414,
      "learning_rate": 7.383933868882438e-06,
      "lm_loss": 5.6692,
      "loss": 1.4199,
      "step": 790,
      "text_contrastive_loss": 0.8241,
      "train_positive_log_prob": -84.2383,
      "train_positive_token_accuracy": 0.0757,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.405,
      "epoch": 1.785553047404063,
      "grad_norm": 12.903446197509766,
      "learning_rate": 7.377553717898983e-06,
      "lm_loss": 5.5162,
      "loss": 1.317,
      "step": 791,
      "text_contrastive_loss": 0.7208,
      "train_positive_log_prob": -80.6727,
      "train_positive_token_accuracy": 0.0847,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3485,
      "epoch": 1.7878103837471784,
      "grad_norm": 12.686439514160156,
      "learning_rate": 7.37116856064625e-06,
      "lm_loss": 5.676,
      "loss": 1.313,
      "step": 792,
      "text_contrastive_loss": 0.7937,
      "train_positive_log_prob": -83.7908,
      "train_positive_token_accuracy": 0.0807,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.4713,
      "epoch": 1.7900677200902935,
      "grad_norm": 14.899544715881348,
      "learning_rate": 7.364778410569071e-06,
      "lm_loss": 5.6578,
      "loss": 1.4345,
      "step": 793,
      "text_contrastive_loss": 0.7949,
      "train_positive_log_prob": -85.3269,
      "train_positive_token_accuracy": 0.079,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.4433,
      "epoch": 1.7923250564334086,
      "grad_norm": 15.099254608154297,
      "learning_rate": 7.358383281122797e-06,
      "lm_loss": 5.6283,
      "loss": 1.4054,
      "step": 794,
      "text_contrastive_loss": 0.7986,
      "train_positive_log_prob": -81.9379,
      "train_positive_token_accuracy": 0.0794,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.5166,
      "epoch": 1.7945823927765236,
      "grad_norm": 12.828969955444336,
      "learning_rate": 7.351983185773259e-06,
      "lm_loss": 5.5089,
      "loss": 1.5173,
      "step": 795,
      "text_contrastive_loss": 0.8996,
      "train_positive_log_prob": -81.3185,
      "train_positive_token_accuracy": 0.0831,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.5595,
      "epoch": 1.7968397291196387,
      "grad_norm": 17.080780029296875,
      "learning_rate": 7.345578137996745e-06,
      "lm_loss": 5.5889,
      "loss": 1.5754,
      "step": 796,
      "text_contrastive_loss": 0.9141,
      "train_positive_log_prob": -82.4799,
      "train_positive_token_accuracy": 0.0834,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.5049,
      "epoch": 1.799097065462754,
      "grad_norm": 15.56982135772705,
      "learning_rate": 7.339168151279974e-06,
      "lm_loss": 5.6143,
      "loss": 1.5121,
      "step": 797,
      "text_contrastive_loss": 0.8917,
      "train_positive_log_prob": -84.216,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.4489,
      "epoch": 1.801354401805869,
      "grad_norm": 15.048324584960938,
      "learning_rate": 7.332753239120061e-06,
      "lm_loss": 5.6385,
      "loss": 1.4327,
      "step": 798,
      "text_contrastive_loss": 0.8399,
      "train_positive_log_prob": -85.109,
      "train_positive_token_accuracy": 0.0805,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.473,
      "epoch": 1.8036117381489842,
      "grad_norm": 14.806596755981445,
      "learning_rate": 7.326333415024494e-06,
      "lm_loss": 5.6085,
      "loss": 1.4863,
      "step": 799,
      "text_contrastive_loss": 0.9048,
      "train_positive_log_prob": -83.4921,
      "train_positive_token_accuracy": 0.0787,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3592,
      "epoch": 1.8058690744920993,
      "grad_norm": 12.606931686401367,
      "learning_rate": 7.319908692511103e-06,
      "lm_loss": 5.6913,
      "loss": 1.3472,
      "step": 800,
      "text_contrastive_loss": 0.8378,
      "train_positive_log_prob": -87.3034,
      "train_positive_token_accuracy": 0.0779,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.4869,
      "epoch": 1.8081264108352144,
      "grad_norm": 14.769548416137695,
      "learning_rate": 7.313479085108033e-06,
      "lm_loss": 5.5339,
      "loss": 1.552,
      "step": 801,
      "text_contrastive_loss": 1.0234,
      "train_positive_log_prob": -81.4544,
      "train_positive_token_accuracy": 0.0799,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.5564,
      "epoch": 1.8103837471783297,
      "grad_norm": 14.806395530700684,
      "learning_rate": 7.307044606353715e-06,
      "lm_loss": 5.5761,
      "loss": 1.5849,
      "step": 802,
      "text_contrastive_loss": 0.9418,
      "train_positive_log_prob": -81.4017,
      "train_positive_token_accuracy": 0.0805,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.598,
      "epoch": 1.8126410835214446,
      "grad_norm": 16.165496826171875,
      "learning_rate": 7.300605269796839e-06,
      "lm_loss": 5.6073,
      "loss": 1.626,
      "step": 803,
      "text_contrastive_loss": 0.9346,
      "train_positive_log_prob": -83.693,
      "train_positive_token_accuracy": 0.0764,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.4908,
      "epoch": 1.81489841986456,
      "grad_norm": 18.01096534729004,
      "learning_rate": 7.2941610889963164e-06,
      "lm_loss": 5.5842,
      "loss": 1.5407,
      "step": 804,
      "text_contrastive_loss": 0.9828,
      "train_positive_log_prob": -83.4605,
      "train_positive_token_accuracy": 0.0748,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.5103,
      "epoch": 1.8171557562076748,
      "grad_norm": 15.611786842346191,
      "learning_rate": 7.2877120775212685e-06,
      "lm_loss": 5.5133,
      "loss": 1.4826,
      "step": 805,
      "text_contrastive_loss": 0.8421,
      "train_positive_log_prob": -80.532,
      "train_positive_token_accuracy": 0.0818,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.4798,
      "epoch": 1.81941309255079,
      "grad_norm": 13.229434967041016,
      "learning_rate": 7.2812582489509844e-06,
      "lm_loss": 5.5485,
      "loss": 1.4481,
      "step": 806,
      "text_contrastive_loss": 0.8268,
      "train_positive_log_prob": -81.2095,
      "train_positive_token_accuracy": 0.0771,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.468,
      "epoch": 1.8216704288939052,
      "grad_norm": 15.610215187072754,
      "learning_rate": 7.2747996168748915e-06,
      "lm_loss": 5.7483,
      "loss": 1.4664,
      "step": 807,
      "text_contrastive_loss": 0.8473,
      "train_positive_log_prob": -87.0547,
      "train_positive_token_accuracy": 0.0748,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.3884,
      "epoch": 1.8239277652370203,
      "grad_norm": 13.929288864135742,
      "learning_rate": 7.26833619489254e-06,
      "lm_loss": 5.569,
      "loss": 1.2913,
      "step": 808,
      "text_contrastive_loss": 0.6921,
      "train_positive_log_prob": -82.9014,
      "train_positive_token_accuracy": 0.09,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.5016,
      "epoch": 1.8261851015801356,
      "grad_norm": 18.220733642578125,
      "learning_rate": 7.261867996613559e-06,
      "lm_loss": 5.6519,
      "loss": 1.6004,
      "step": 809,
      "text_contrastive_loss": 1.0672,
      "train_positive_log_prob": -85.3075,
      "train_positive_token_accuracy": 0.0735,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.3633,
      "epoch": 1.8284424379232505,
      "grad_norm": 12.820158004760742,
      "learning_rate": 7.255395035657639e-06,
      "lm_loss": 5.6493,
      "loss": 1.3667,
      "step": 810,
      "text_contrastive_loss": 0.8769,
      "train_positive_log_prob": -82.6875,
      "train_positive_token_accuracy": 0.0811,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.3875,
      "epoch": 1.8306997742663658,
      "grad_norm": 14.121953964233398,
      "learning_rate": 7.2489173256544975e-06,
      "lm_loss": 5.5581,
      "loss": 1.3184,
      "step": 811,
      "text_contrastive_loss": 0.7502,
      "train_positive_log_prob": -83.6329,
      "train_positive_token_accuracy": 0.0776,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.4855,
      "epoch": 1.8329571106094809,
      "grad_norm": 14.91685676574707,
      "learning_rate": 7.242434880243851e-06,
      "lm_loss": 5.5543,
      "loss": 1.4778,
      "step": 812,
      "text_contrastive_loss": 0.8737,
      "train_positive_log_prob": -84.0102,
      "train_positive_token_accuracy": 0.0826,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.4148,
      "epoch": 1.835214446952596,
      "grad_norm": 13.67586898803711,
      "learning_rate": 7.235947713075389e-06,
      "lm_loss": 5.6382,
      "loss": 1.3529,
      "step": 813,
      "text_contrastive_loss": 0.7487,
      "train_positive_log_prob": -83.2335,
      "train_positive_token_accuracy": 0.0717,
      "train_positive_token_prob": 0.0287
    },
    {
      "contrastive_loss": 0.4714,
      "epoch": 1.837471783295711,
      "grad_norm": 15.072975158691406,
      "learning_rate": 7.229455837808741e-06,
      "lm_loss": 5.7478,
      "loss": 1.4768,
      "step": 814,
      "text_contrastive_loss": 0.8612,
      "train_positive_log_prob": -85.4531,
      "train_positive_token_accuracy": 0.0744,
      "train_positive_token_prob": 0.0286
    },
    {
      "contrastive_loss": 0.5182,
      "epoch": 1.8397291196388261,
      "grad_norm": 15.159923553466797,
      "learning_rate": 7.222959268113452e-06,
      "lm_loss": 5.5198,
      "loss": 1.5056,
      "step": 815,
      "text_contrastive_loss": 0.871,
      "train_positive_log_prob": -81.5024,
      "train_positive_token_accuracy": 0.0823,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.476,
      "epoch": 1.8419864559819414,
      "grad_norm": 13.744406700134277,
      "learning_rate": 7.216458017668951e-06,
      "lm_loss": 5.6182,
      "loss": 1.4193,
      "step": 816,
      "text_contrastive_loss": 0.763,
      "train_positive_log_prob": -83.7656,
      "train_positive_token_accuracy": 0.0765,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.4769,
      "epoch": 1.8442437923250563,
      "grad_norm": 16.792842864990234,
      "learning_rate": 7.2099521001645225e-06,
      "lm_loss": 5.6844,
      "loss": 1.4495,
      "step": 817,
      "text_contrastive_loss": 0.8084,
      "train_positive_log_prob": -84.5348,
      "train_positive_token_accuracy": 0.0767,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.5317,
      "epoch": 1.8465011286681716,
      "grad_norm": 15.967273712158203,
      "learning_rate": 7.20344152929928e-06,
      "lm_loss": 5.5861,
      "loss": 1.5331,
      "step": 818,
      "text_contrastive_loss": 0.8855,
      "train_positive_log_prob": -82.8577,
      "train_positive_token_accuracy": 0.0809,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.3967,
      "epoch": 1.8487584650112867,
      "grad_norm": 13.10386848449707,
      "learning_rate": 7.19692631878213e-06,
      "lm_loss": 5.7143,
      "loss": 1.3304,
      "step": 819,
      "text_contrastive_loss": 0.7244,
      "train_positive_log_prob": -87.948,
      "train_positive_token_accuracy": 0.0784,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.4624,
      "epoch": 1.8510158013544018,
      "grad_norm": 14.461969375610352,
      "learning_rate": 7.190406482331757e-06,
      "lm_loss": 5.5348,
      "loss": 1.4178,
      "step": 820,
      "text_contrastive_loss": 0.8039,
      "train_positive_log_prob": -80.4716,
      "train_positive_token_accuracy": 0.0731,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.5617,
      "epoch": 1.853273137697517,
      "grad_norm": 18.000465393066406,
      "learning_rate": 7.183882033676579e-06,
      "lm_loss": 5.5423,
      "loss": 1.5223,
      "step": 821,
      "text_contrastive_loss": 0.8127,
      "train_positive_log_prob": -83.162,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.5078,
      "epoch": 1.855530474040632,
      "grad_norm": 15.518335342407227,
      "learning_rate": 7.177352986554729e-06,
      "lm_loss": 5.5988,
      "loss": 1.5518,
      "step": 822,
      "text_contrastive_loss": 0.968,
      "train_positive_log_prob": -84.3963,
      "train_positive_token_accuracy": 0.0806,
      "train_positive_token_prob": 0.0288
    },
    {
      "contrastive_loss": 0.4374,
      "epoch": 1.8577878103837473,
      "grad_norm": 15.986565589904785,
      "learning_rate": 7.1708193547140205e-06,
      "lm_loss": 5.7263,
      "loss": 1.364,
      "step": 823,
      "text_contrastive_loss": 0.708,
      "train_positive_log_prob": -85.1994,
      "train_positive_token_accuracy": 0.0769,
      "train_positive_token_prob": 0.0289
    },
    {
      "contrastive_loss": 0.5156,
      "epoch": 1.8600451467268622,
      "grad_norm": 17.1726131439209,
      "learning_rate": 7.164281151911923e-06,
      "lm_loss": 5.58,
      "loss": 1.4341,
      "step": 824,
      "text_contrastive_loss": 0.721,
      "train_positive_log_prob": -84.4152,
      "train_positive_token_accuracy": 0.0726,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.5864,
      "epoch": 1.8623024830699775,
      "grad_norm": 15.21026611328125,
      "learning_rate": 7.157738391915531e-06,
      "lm_loss": 5.5732,
      "loss": 1.631,
      "step": 825,
      "text_contrastive_loss": 0.9744,
      "train_positive_log_prob": -81.9442,
      "train_positive_token_accuracy": 0.0887,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3965,
      "epoch": 1.8645598194130926,
      "grad_norm": 13.081595420837402,
      "learning_rate": 7.151191088501531e-06,
      "lm_loss": 5.6041,
      "loss": 1.3544,
      "step": 826,
      "text_contrastive_loss": 0.795,
      "train_positive_log_prob": -84.1719,
      "train_positive_token_accuracy": 0.0759,
      "train_positive_token_prob": 0.0288
    },
    {
      "contrastive_loss": 0.5501,
      "epoch": 1.8668171557562077,
      "grad_norm": 15.819600105285645,
      "learning_rate": 7.14463925545618e-06,
      "lm_loss": 5.6947,
      "loss": 1.5524,
      "step": 827,
      "text_contrastive_loss": 0.8656,
      "train_positive_log_prob": -84.9321,
      "train_positive_token_accuracy": 0.076,
      "train_positive_token_prob": 0.028
    },
    {
      "contrastive_loss": 0.4349,
      "epoch": 1.8690744920993227,
      "grad_norm": 15.383347511291504,
      "learning_rate": 7.138082906575271e-06,
      "lm_loss": 5.5701,
      "loss": 1.4005,
      "step": 828,
      "text_contrastive_loss": 0.8173,
      "train_positive_log_prob": -84.0264,
      "train_positive_token_accuracy": 0.0766,
      "train_positive_token_prob": 0.0289
    },
    {
      "contrastive_loss": 0.4265,
      "epoch": 1.8713318284424378,
      "grad_norm": 14.575815200805664,
      "learning_rate": 7.131522055664109e-06,
      "lm_loss": 5.6305,
      "loss": 1.4322,
      "step": 829,
      "text_contrastive_loss": 0.8852,
      "train_positive_log_prob": -82.6329,
      "train_positive_token_accuracy": 0.0759,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.3829,
      "epoch": 1.8735891647855532,
      "grad_norm": 13.55797004699707,
      "learning_rate": 7.124956716537471e-06,
      "lm_loss": 5.5966,
      "loss": 1.3409,
      "step": 830,
      "text_contrastive_loss": 0.7966,
      "train_positive_log_prob": -83.8393,
      "train_positive_token_accuracy": 0.0866,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.4603,
      "epoch": 1.875846501128668,
      "grad_norm": 13.942564964294434,
      "learning_rate": 7.118386903019594e-06,
      "lm_loss": 5.6932,
      "loss": 1.4178,
      "step": 831,
      "text_contrastive_loss": 0.7764,
      "train_positive_log_prob": -81.8914,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.4562,
      "epoch": 1.8781038374717833,
      "grad_norm": 13.51505184173584,
      "learning_rate": 7.111812628944132e-06,
      "lm_loss": 5.6058,
      "loss": 1.4571,
      "step": 832,
      "text_contrastive_loss": 0.8806,
      "train_positive_log_prob": -82.7086,
      "train_positive_token_accuracy": 0.08,
      "train_positive_token_prob": 0.0282
    },
    {
      "contrastive_loss": 0.5499,
      "epoch": 1.8803611738148984,
      "grad_norm": 16.279476165771484,
      "learning_rate": 7.105233908154128e-06,
      "lm_loss": 5.5627,
      "loss": 1.5165,
      "step": 833,
      "text_contrastive_loss": 0.8206,
      "train_positive_log_prob": -82.4866,
      "train_positive_token_accuracy": 0.0785,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.5027,
      "epoch": 1.8826185101580135,
      "grad_norm": 16.14896583557129,
      "learning_rate": 7.098650754501994e-06,
      "lm_loss": 5.637,
      "loss": 1.5065,
      "step": 834,
      "text_contrastive_loss": 0.8802,
      "train_positive_log_prob": -83.8451,
      "train_positive_token_accuracy": 0.0814,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.4518,
      "epoch": 1.8848758465011288,
      "grad_norm": 15.16618824005127,
      "learning_rate": 7.0920631818494745e-06,
      "lm_loss": 5.6372,
      "loss": 1.5074,
      "step": 835,
      "text_contrastive_loss": 0.9837,
      "train_positive_log_prob": -83.4141,
      "train_positive_token_accuracy": 0.0808,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.4711,
      "epoch": 1.8871331828442437,
      "grad_norm": 15.736971855163574,
      "learning_rate": 7.085471204067616e-06,
      "lm_loss": 5.544,
      "loss": 1.4553,
      "step": 836,
      "text_contrastive_loss": 0.8596,
      "train_positive_log_prob": -82.9027,
      "train_positive_token_accuracy": 0.0765,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.3976,
      "epoch": 1.889390519187359,
      "grad_norm": 14.282797813415527,
      "learning_rate": 7.078874835036742e-06,
      "lm_loss": 5.605,
      "loss": 1.3383,
      "step": 837,
      "text_contrastive_loss": 0.7603,
      "train_positive_log_prob": -83.8771,
      "train_positive_token_accuracy": 0.0786,
      "train_positive_token_prob": 0.0288
    },
    {
      "contrastive_loss": 0.5501,
      "epoch": 1.8916478555304739,
      "grad_norm": 16.621746063232422,
      "learning_rate": 7.072274088646425e-06,
      "lm_loss": 5.6455,
      "loss": 1.4939,
      "step": 838,
      "text_contrastive_loss": 0.7583,
      "train_positive_log_prob": -83.645,
      "train_positive_token_accuracy": 0.0794,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.488,
      "epoch": 1.8939051918735892,
      "grad_norm": 16.08866310119629,
      "learning_rate": 7.065668978795449e-06,
      "lm_loss": 5.6173,
      "loss": 1.5136,
      "step": 839,
      "text_contrastive_loss": 0.9277,
      "train_positive_log_prob": -82.9,
      "train_positive_token_accuracy": 0.0791,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.3812,
      "epoch": 1.8961625282167043,
      "grad_norm": 13.742374420166016,
      "learning_rate": 7.059059519391794e-06,
      "lm_loss": 5.5417,
      "loss": 1.2894,
      "step": 840,
      "text_contrastive_loss": 0.708,
      "train_positive_log_prob": -84.0539,
      "train_positive_token_accuracy": 0.0865,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.4737,
      "epoch": 1.8984198645598194,
      "grad_norm": 16.210573196411133,
      "learning_rate": 7.05244572435259e-06,
      "lm_loss": 5.5965,
      "loss": 1.4377,
      "step": 841,
      "text_contrastive_loss": 0.8088,
      "train_positive_log_prob": -82.3539,
      "train_positive_token_accuracy": 0.0719,
      "train_positive_token_prob": 0.0283
    },
    {
      "contrastive_loss": 0.4684,
      "epoch": 1.9006772009029347,
      "grad_norm": 15.303285598754883,
      "learning_rate": 7.045827607604103e-06,
      "lm_loss": 5.6008,
      "loss": 1.3846,
      "step": 842,
      "text_contrastive_loss": 0.7122,
      "train_positive_log_prob": -82.5049,
      "train_positive_token_accuracy": 0.0825,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.4653,
      "epoch": 1.9029345372460496,
      "grad_norm": 14.521797180175781,
      "learning_rate": 7.039205183081694e-06,
      "lm_loss": 5.607,
      "loss": 1.4238,
      "step": 843,
      "text_contrastive_loss": 0.7956,
      "train_positive_log_prob": -83.471,
      "train_positive_token_accuracy": 0.0821,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.4137,
      "epoch": 1.9051918735891649,
      "grad_norm": 13.851727485656738,
      "learning_rate": 7.0325784647298e-06,
      "lm_loss": 5.7506,
      "loss": 1.4192,
      "step": 844,
      "text_contrastive_loss": 0.861,
      "train_positive_log_prob": -86.1233,
      "train_positive_token_accuracy": 0.0822,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.5484,
      "epoch": 1.90744920993228,
      "grad_norm": 14.06857967376709,
      "learning_rate": 7.0259474665018915e-06,
      "lm_loss": 5.5755,
      "loss": 1.4758,
      "step": 845,
      "text_contrastive_loss": 0.7397,
      "train_positive_log_prob": -84.3952,
      "train_positive_token_accuracy": 0.0911,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.5151,
      "epoch": 1.909706546275395,
      "grad_norm": 14.030274391174316,
      "learning_rate": 7.019312202360457e-06,
      "lm_loss": 5.4148,
      "loss": 1.4335,
      "step": 846,
      "text_contrastive_loss": 0.7538,
      "train_positive_log_prob": -79.4206,
      "train_positive_token_accuracy": 0.0913,
      "train_positive_token_prob": 0.0328
    },
    {
      "contrastive_loss": 0.4719,
      "epoch": 1.9119638826185101,
      "grad_norm": 17.120309829711914,
      "learning_rate": 7.012672686276969e-06,
      "lm_loss": 5.519,
      "loss": 1.5222,
      "step": 847,
      "text_contrastive_loss": 0.9969,
      "train_positive_log_prob": -80.3922,
      "train_positive_token_accuracy": 0.0823,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.4682,
      "epoch": 1.9142212189616252,
      "grad_norm": 14.32099437713623,
      "learning_rate": 7.006028932231847e-06,
      "lm_loss": 5.5321,
      "loss": 1.4433,
      "step": 848,
      "text_contrastive_loss": 0.8439,
      "train_positive_log_prob": -81.9454,
      "train_positive_token_accuracy": 0.0857,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.5206,
      "epoch": 1.9164785553047405,
      "grad_norm": 16.19694709777832,
      "learning_rate": 6.999380954214438e-06,
      "lm_loss": 5.4697,
      "loss": 1.4496,
      "step": 849,
      "text_contrastive_loss": 0.7641,
      "train_positive_log_prob": -78.7331,
      "train_positive_token_accuracy": 0.0806,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.4355,
      "epoch": 1.9187358916478554,
      "grad_norm": 15.280251502990723,
      "learning_rate": 6.992728766222982e-06,
      "lm_loss": 5.5791,
      "loss": 1.3745,
      "step": 850,
      "text_contrastive_loss": 0.7621,
      "train_positive_log_prob": -83.1853,
      "train_positive_token_accuracy": 0.0825,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.4448,
      "epoch": 1.9209932279909707,
      "grad_norm": 15.748678207397461,
      "learning_rate": 6.9860723822645825e-06,
      "lm_loss": 5.6034,
      "loss": 1.404,
      "step": 851,
      "text_contrastive_loss": 0.7978,
      "train_positive_log_prob": -82.9328,
      "train_positive_token_accuracy": 0.0765,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.4533,
      "epoch": 1.9232505643340858,
      "grad_norm": 13.725008010864258,
      "learning_rate": 6.979411816355183e-06,
      "lm_loss": 5.5198,
      "loss": 1.4633,
      "step": 852,
      "text_contrastive_loss": 0.9161,
      "train_positive_log_prob": -81.696,
      "train_positive_token_accuracy": 0.0823,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.5777,
      "epoch": 1.925507900677201,
      "grad_norm": 16.754261016845703,
      "learning_rate": 6.972747082519526e-06,
      "lm_loss": 5.6656,
      "loss": 1.6488,
      "step": 853,
      "text_contrastive_loss": 1.009,
      "train_positive_log_prob": -85.5092,
      "train_positive_token_accuracy": 0.0787,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.4732,
      "epoch": 1.927765237020316,
      "grad_norm": 13.837919235229492,
      "learning_rate": 6.966078194791133e-06,
      "lm_loss": 5.4962,
      "loss": 1.3674,
      "step": 854,
      "text_contrastive_loss": 0.6891,
      "train_positive_log_prob": -80.7456,
      "train_positive_token_accuracy": 0.0782,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.4928,
      "epoch": 1.930022573363431,
      "grad_norm": 14.32505989074707,
      "learning_rate": 6.959405167212278e-06,
      "lm_loss": 5.6566,
      "loss": 1.5324,
      "step": 855,
      "text_contrastive_loss": 0.9478,
      "train_positive_log_prob": -81.7352,
      "train_positive_token_accuracy": 0.0781,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.4668,
      "epoch": 1.9322799097065464,
      "grad_norm": 14.705467224121094,
      "learning_rate": 6.952728013833941e-06,
      "lm_loss": 5.6002,
      "loss": 1.4817,
      "step": 856,
      "text_contrastive_loss": 0.9097,
      "train_positive_log_prob": -84.2861,
      "train_positive_token_accuracy": 0.0789,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.5499,
      "epoch": 1.9345372460496613,
      "grad_norm": 15.175885200500488,
      "learning_rate": 6.946046748715796e-06,
      "lm_loss": 5.5416,
      "loss": 1.5401,
      "step": 857,
      "text_contrastive_loss": 0.8722,
      "train_positive_log_prob": -81.0438,
      "train_positive_token_accuracy": 0.0807,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.5026,
      "epoch": 1.9367945823927766,
      "grad_norm": 16.210250854492188,
      "learning_rate": 6.9393613859261755e-06,
      "lm_loss": 5.6146,
      "loss": 1.4351,
      "step": 858,
      "text_contrastive_loss": 0.7421,
      "train_positive_log_prob": -82.7325,
      "train_positive_token_accuracy": 0.0824,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.4726,
      "epoch": 1.9390519187358917,
      "grad_norm": 14.95361328125,
      "learning_rate": 6.932671939542037e-06,
      "lm_loss": 5.5204,
      "loss": 1.4901,
      "step": 859,
      "text_contrastive_loss": 0.931,
      "train_positive_log_prob": -83.196,
      "train_positive_token_accuracy": 0.084,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.4611,
      "epoch": 1.9413092550790068,
      "grad_norm": 13.421514511108398,
      "learning_rate": 6.925978423648941e-06,
      "lm_loss": 5.4818,
      "loss": 1.4622,
      "step": 860,
      "text_contrastive_loss": 0.9058,
      "train_positive_log_prob": -80.0633,
      "train_positive_token_accuracy": 0.0792,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.4955,
      "epoch": 1.9435665914221218,
      "grad_norm": 14.257747650146484,
      "learning_rate": 6.919280852341011e-06,
      "lm_loss": 5.4519,
      "loss": 1.4749,
      "step": 861,
      "text_contrastive_loss": 0.8685,
      "train_positive_log_prob": -79.9278,
      "train_positive_token_accuracy": 0.081,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.4035,
      "epoch": 1.945823927765237,
      "grad_norm": 12.491300582885742,
      "learning_rate": 6.912579239720913e-06,
      "lm_loss": 5.598,
      "loss": 1.3188,
      "step": 862,
      "text_contrastive_loss": 0.711,
      "train_positive_log_prob": -81.0553,
      "train_positive_token_accuracy": 0.0807,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.4694,
      "epoch": 1.9480812641083523,
      "grad_norm": 17.115652084350586,
      "learning_rate": 6.9058735998998224e-06,
      "lm_loss": 5.5599,
      "loss": 1.4206,
      "step": 863,
      "text_contrastive_loss": 0.7905,
      "train_positive_log_prob": -82.4229,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.5671,
      "epoch": 1.9503386004514671,
      "grad_norm": 13.943278312683105,
      "learning_rate": 6.899163946997396e-06,
      "lm_loss": 5.5436,
      "loss": 1.5007,
      "step": 864,
      "text_contrastive_loss": 0.7585,
      "train_positive_log_prob": -81.8299,
      "train_positive_token_accuracy": 0.0794,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.5044,
      "epoch": 1.9525959367945824,
      "grad_norm": 15.068373680114746,
      "learning_rate": 6.892450295141737e-06,
      "lm_loss": 5.5476,
      "loss": 1.4091,
      "step": 865,
      "text_contrastive_loss": 0.6998,
      "train_positive_log_prob": -82.3679,
      "train_positive_token_accuracy": 0.0831,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.4936,
      "epoch": 1.9548532731376975,
      "grad_norm": 13.565226554870605,
      "learning_rate": 6.885732658469374e-06,
      "lm_loss": 5.6471,
      "loss": 1.5469,
      "step": 866,
      "text_contrastive_loss": 0.9771,
      "train_positive_log_prob": -83.9096,
      "train_positive_token_accuracy": 0.0794,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.421,
      "epoch": 1.9571106094808126,
      "grad_norm": 13.99299430847168,
      "learning_rate": 6.8790110511252195e-06,
      "lm_loss": 5.6158,
      "loss": 1.3815,
      "step": 867,
      "text_contrastive_loss": 0.7979,
      "train_positive_log_prob": -83.9282,
      "train_positive_token_accuracy": 0.0768,
      "train_positive_token_prob": 0.0291
    },
    {
      "contrastive_loss": 0.5897,
      "epoch": 1.959367945823928,
      "grad_norm": 15.685070037841797,
      "learning_rate": 6.872285487262555e-06,
      "lm_loss": 5.5884,
      "loss": 1.6263,
      "step": 868,
      "text_contrastive_loss": 0.9556,
      "train_positive_log_prob": -82.8237,
      "train_positive_token_accuracy": 0.0783,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.545,
      "epoch": 1.9616252821670428,
      "grad_norm": 15.854179382324219,
      "learning_rate": 6.865555981042983e-06,
      "lm_loss": 5.6359,
      "loss": 1.5263,
      "step": 869,
      "text_contrastive_loss": 0.8353,
      "train_positive_log_prob": -82.9802,
      "train_positive_token_accuracy": 0.0761,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.4383,
      "epoch": 1.963882618510158,
      "grad_norm": 16.044401168823242,
      "learning_rate": 6.858822546636417e-06,
      "lm_loss": 5.6004,
      "loss": 1.4463,
      "step": 870,
      "text_contrastive_loss": 0.896,
      "train_positive_log_prob": -84.2292,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.4094,
      "epoch": 1.966139954853273,
      "grad_norm": 13.492218971252441,
      "learning_rate": 6.852085198221035e-06,
      "lm_loss": 5.5522,
      "loss": 1.4249,
      "step": 871,
      "text_contrastive_loss": 0.9206,
      "train_positive_log_prob": -81.5495,
      "train_positive_token_accuracy": 0.0745,
      "train_positive_token_prob": 0.0289
    },
    {
      "contrastive_loss": 0.5196,
      "epoch": 1.9683972911963883,
      "grad_norm": 15.734671592712402,
      "learning_rate": 6.845343949983258e-06,
      "lm_loss": 5.5699,
      "loss": 1.5554,
      "step": 872,
      "text_contrastive_loss": 0.9576,
      "train_positive_log_prob": -81.1526,
      "train_positive_token_accuracy": 0.0759,
      "train_positive_token_prob": 0.0281
    },
    {
      "contrastive_loss": 0.4282,
      "epoch": 1.9706546275395034,
      "grad_norm": 14.115796089172363,
      "learning_rate": 6.838598816117725e-06,
      "lm_loss": 5.6147,
      "loss": 1.4456,
      "step": 873,
      "text_contrastive_loss": 0.9118,
      "train_positive_log_prob": -84.8847,
      "train_positive_token_accuracy": 0.0806,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.461,
      "epoch": 1.9729119638826185,
      "grad_norm": 14.111851692199707,
      "learning_rate": 6.831849810827247e-06,
      "lm_loss": 5.4677,
      "loss": 1.4165,
      "step": 874,
      "text_contrastive_loss": 0.8173,
      "train_positive_log_prob": -81.0533,
      "train_positive_token_accuracy": 0.0812,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.5937,
      "epoch": 1.9751693002257338,
      "grad_norm": 16.07346534729004,
      "learning_rate": 6.825096948322791e-06,
      "lm_loss": 5.4724,
      "loss": 1.6095,
      "step": 875,
      "text_contrastive_loss": 0.9372,
      "train_positive_log_prob": -82.13,
      "train_positive_token_accuracy": 0.0778,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.4975,
      "epoch": 1.9774266365688487,
      "grad_norm": 14.203269004821777,
      "learning_rate": 6.818340242823449e-06,
      "lm_loss": 5.5741,
      "loss": 1.433,
      "step": 876,
      "text_contrastive_loss": 0.756,
      "train_positive_log_prob": -82.7429,
      "train_positive_token_accuracy": 0.0803,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.4657,
      "epoch": 1.979683972911964,
      "grad_norm": 13.515305519104004,
      "learning_rate": 6.8115797085564e-06,
      "lm_loss": 5.5507,
      "loss": 1.3591,
      "step": 877,
      "text_contrastive_loss": 0.6766,
      "train_positive_log_prob": -81.7874,
      "train_positive_token_accuracy": 0.0806,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.4917,
      "epoch": 1.981941309255079,
      "grad_norm": 16.76006507873535,
      "learning_rate": 6.804815359756887e-06,
      "lm_loss": 5.6322,
      "loss": 1.472,
      "step": 878,
      "text_contrastive_loss": 0.8342,
      "train_positive_log_prob": -86.0254,
      "train_positive_token_accuracy": 0.0793,
      "train_positive_token_prob": 0.0285
    },
    {
      "contrastive_loss": 0.4057,
      "epoch": 1.9841986455981941,
      "grad_norm": 14.97724437713623,
      "learning_rate": 6.798047210668185e-06,
      "lm_loss": 5.638,
      "loss": 1.3498,
      "step": 879,
      "text_contrastive_loss": 0.7604,
      "train_positive_log_prob": -84.4639,
      "train_positive_token_accuracy": 0.0746,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.4114,
      "epoch": 1.9864559819413092,
      "grad_norm": 14.190703392028809,
      "learning_rate": 6.7912752755415716e-06,
      "lm_loss": 5.6413,
      "loss": 1.452,
      "step": 880,
      "text_contrastive_loss": 0.9531,
      "train_positive_log_prob": -83.3415,
      "train_positive_token_accuracy": 0.082,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.4907,
      "epoch": 1.9887133182844243,
      "grad_norm": 14.56006145477295,
      "learning_rate": 6.7844995686362955e-06,
      "lm_loss": 5.5309,
      "loss": 1.5096,
      "step": 881,
      "text_contrastive_loss": 0.9315,
      "train_positive_log_prob": -82.1201,
      "train_positive_token_accuracy": 0.0868,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.3811,
      "epoch": 1.9909706546275396,
      "grad_norm": 12.781140327453613,
      "learning_rate": 6.777720104219548e-06,
      "lm_loss": 5.4105,
      "loss": 1.3768,
      "step": 882,
      "text_contrastive_loss": 0.9093,
      "train_positive_log_prob": -81.5598,
      "train_positive_token_accuracy": 0.0886,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.55,
      "epoch": 1.9932279909706545,
      "grad_norm": 15.659453392028809,
      "learning_rate": 6.770936896566434e-06,
      "lm_loss": 5.5339,
      "loss": 1.5144,
      "step": 883,
      "text_contrastive_loss": 0.822,
      "train_positive_log_prob": -79.573,
      "train_positive_token_accuracy": 0.0774,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.3921,
      "epoch": 1.9954853273137698,
      "grad_norm": 13.856972694396973,
      "learning_rate": 6.7641499599599355e-06,
      "lm_loss": 5.5311,
      "loss": 1.3099,
      "step": 884,
      "text_contrastive_loss": 0.7294,
      "train_positive_log_prob": -82.2734,
      "train_positive_token_accuracy": 0.0838,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.5687,
      "epoch": 1.997742663656885,
      "grad_norm": 15.394890785217285,
      "learning_rate": 6.757359308690889e-06,
      "lm_loss": 5.6018,
      "loss": 1.6017,
      "step": 885,
      "text_contrastive_loss": 0.9457,
      "train_positive_log_prob": -81.442,
      "train_positive_token_accuracy": 0.079,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.2688,
      "epoch": 2.0,
      "grad_norm": 14.641634941101074,
      "learning_rate": 6.750564957057958e-06,
      "lm_loss": 5.4777,
      "loss": 1.0655,
      "step": 886,
      "text_contrastive_loss": 0.4979,
      "train_positive_log_prob": -81.6036,
      "train_positive_token_accuracy": 0.0776,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.4136,
      "epoch": 2.0022573363431153,
      "grad_norm": 13.20592975616455,
      "learning_rate": 6.743766919367588e-06,
      "lm_loss": 5.6038,
      "loss": 1.3932,
      "step": 887,
      "text_contrastive_loss": 0.8384,
      "train_positive_log_prob": -84.9565,
      "train_positive_token_accuracy": 0.0804,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.4362,
      "epoch": 2.00451467268623,
      "grad_norm": 12.936826705932617,
      "learning_rate": 6.736965209933992e-06,
      "lm_loss": 5.4725,
      "loss": 1.3581,
      "step": 888,
      "text_contrastive_loss": 0.7494,
      "train_positive_log_prob": -80.7032,
      "train_positive_token_accuracy": 0.0774,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.4231,
      "epoch": 2.0067720090293455,
      "grad_norm": 13.164968490600586,
      "learning_rate": 6.730159843079113e-06,
      "lm_loss": 5.5138,
      "loss": 1.4154,
      "step": 889,
      "text_contrastive_loss": 0.8817,
      "train_positive_log_prob": -82.879,
      "train_positive_token_accuracy": 0.0852,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.4867,
      "epoch": 2.0090293453724604,
      "grad_norm": 15.41411018371582,
      "learning_rate": 6.723350833132596e-06,
      "lm_loss": 5.4926,
      "loss": 1.4675,
      "step": 890,
      "text_contrastive_loss": 0.863,
      "train_positive_log_prob": -80.0519,
      "train_positive_token_accuracy": 0.0868,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.365,
      "epoch": 2.0112866817155757,
      "grad_norm": 12.457436561584473,
      "learning_rate": 6.716538194431754e-06,
      "lm_loss": 5.5439,
      "loss": 1.2308,
      "step": 891,
      "text_contrastive_loss": 0.6227,
      "train_positive_log_prob": -83.6553,
      "train_positive_token_accuracy": 0.0896,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.455,
      "epoch": 2.0135440180586905,
      "grad_norm": 14.338443756103516,
      "learning_rate": 6.7097219413215474e-06,
      "lm_loss": 5.6032,
      "loss": 1.4843,
      "step": 892,
      "text_contrastive_loss": 0.9379,
      "train_positive_log_prob": -83.8695,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.458,
      "epoch": 2.015801354401806,
      "grad_norm": 14.148242950439453,
      "learning_rate": 6.702902088154539e-06,
      "lm_loss": 5.6708,
      "loss": 1.4636,
      "step": 893,
      "text_contrastive_loss": 0.8769,
      "train_positive_log_prob": -83.5063,
      "train_positive_token_accuracy": 0.0809,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.4041,
      "epoch": 2.018058690744921,
      "grad_norm": 13.274083137512207,
      "learning_rate": 6.696078649290878e-06,
      "lm_loss": 5.5271,
      "loss": 1.32,
      "step": 894,
      "text_contrastive_loss": 0.7263,
      "train_positive_log_prob": -81.4611,
      "train_positive_token_accuracy": 0.0787,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.3685,
      "epoch": 2.020316027088036,
      "grad_norm": 14.36821460723877,
      "learning_rate": 6.689251639098261e-06,
      "lm_loss": 5.5055,
      "loss": 1.2879,
      "step": 895,
      "text_contrastive_loss": 0.7377,
      "train_positive_log_prob": -80.4559,
      "train_positive_token_accuracy": 0.0901,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.5273,
      "epoch": 2.0225733634311513,
      "grad_norm": 15.859416961669922,
      "learning_rate": 6.682421071951907e-06,
      "lm_loss": 5.6741,
      "loss": 1.5354,
      "step": 896,
      "text_contrastive_loss": 0.8815,
      "train_positive_log_prob": -85.2214,
      "train_positive_token_accuracy": 0.0759,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.3596,
      "epoch": 2.024830699774266,
      "grad_norm": 10.912479400634766,
      "learning_rate": 6.67558696223452e-06,
      "lm_loss": 5.5394,
      "loss": 1.2973,
      "step": 897,
      "text_contrastive_loss": 0.7675,
      "train_positive_log_prob": -81.6643,
      "train_positive_token_accuracy": 0.071,
      "train_positive_token_prob": 0.0287
    },
    {
      "contrastive_loss": 0.4024,
      "epoch": 2.0270880361173815,
      "grad_norm": 14.293588638305664,
      "learning_rate": 6.668749324336268e-06,
      "lm_loss": 5.5751,
      "loss": 1.3196,
      "step": 898,
      "text_contrastive_loss": 0.7193,
      "train_positive_log_prob": -84.0258,
      "train_positive_token_accuracy": 0.0809,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.4079,
      "epoch": 2.0293453724604964,
      "grad_norm": 14.883182525634766,
      "learning_rate": 6.661908172654746e-06,
      "lm_loss": 5.4958,
      "loss": 1.3871,
      "step": 899,
      "text_contrastive_loss": 0.8592,
      "train_positive_log_prob": -81.7061,
      "train_positive_token_accuracy": 0.0772,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.4327,
      "epoch": 2.0316027088036117,
      "grad_norm": 13.25998592376709,
      "learning_rate": 6.65506352159495e-06,
      "lm_loss": 5.5033,
      "loss": 1.4092,
      "step": 900,
      "text_contrastive_loss": 0.8522,
      "train_positive_log_prob": -81.9003,
      "train_positive_token_accuracy": 0.0766,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.4036,
      "epoch": 2.033860045146727,
      "grad_norm": 12.483712196350098,
      "learning_rate": 6.6482153855692395e-06,
      "lm_loss": 5.587,
      "loss": 1.3186,
      "step": 901,
      "text_contrastive_loss": 0.7127,
      "train_positive_log_prob": -82.5823,
      "train_positive_token_accuracy": 0.0758,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.4604,
      "epoch": 2.036117381489842,
      "grad_norm": 13.481544494628906,
      "learning_rate": 6.64136377899732e-06,
      "lm_loss": 5.4519,
      "loss": 1.3855,
      "step": 902,
      "text_contrastive_loss": 0.7599,
      "train_positive_log_prob": -80.2875,
      "train_positive_token_accuracy": 0.091,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.5341,
      "epoch": 2.038374717832957,
      "grad_norm": 15.048299789428711,
      "learning_rate": 6.6345087163061935e-06,
      "lm_loss": 5.5481,
      "loss": 1.5891,
      "step": 903,
      "text_contrastive_loss": 1.0005,
      "train_positive_log_prob": -81.3128,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.4192,
      "epoch": 2.040632054176072,
      "grad_norm": 12.896744728088379,
      "learning_rate": 6.627650211930152e-06,
      "lm_loss": 5.6139,
      "loss": 1.4559,
      "step": 904,
      "text_contrastive_loss": 0.9506,
      "train_positive_log_prob": -83.2017,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.4721,
      "epoch": 2.0428893905191874,
      "grad_norm": 13.316905975341797,
      "learning_rate": 6.620788280310722e-06,
      "lm_loss": 5.5356,
      "loss": 1.4551,
      "step": 905,
      "text_contrastive_loss": 0.859,
      "train_positive_log_prob": -82.946,
      "train_positive_token_accuracy": 0.0743,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3342,
      "epoch": 2.0451467268623027,
      "grad_norm": 12.637482643127441,
      "learning_rate": 6.613922935896659e-06,
      "lm_loss": 5.5852,
      "loss": 1.2455,
      "step": 906,
      "text_contrastive_loss": 0.7056,
      "train_positive_log_prob": -82.7525,
      "train_positive_token_accuracy": 0.0775,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.4832,
      "epoch": 2.0474040632054176,
      "grad_norm": 15.180583000183105,
      "learning_rate": 6.607054193143894e-06,
      "lm_loss": 5.5181,
      "loss": 1.493,
      "step": 907,
      "text_contrastive_loss": 0.916,
      "train_positive_log_prob": -81.5231,
      "train_positive_token_accuracy": 0.0851,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.3542,
      "epoch": 2.049661399548533,
      "grad_norm": 12.676249504089355,
      "learning_rate": 6.600182066515519e-06,
      "lm_loss": 5.5761,
      "loss": 1.2865,
      "step": 908,
      "text_contrastive_loss": 0.7494,
      "train_positive_log_prob": -81.6597,
      "train_positive_token_accuracy": 0.0814,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3492,
      "epoch": 2.0519187358916477,
      "grad_norm": 11.73995590209961,
      "learning_rate": 6.593306570481751e-06,
      "lm_loss": 5.5869,
      "loss": 1.2174,
      "step": 909,
      "text_contrastive_loss": 0.619,
      "train_positive_log_prob": -83.1531,
      "train_positive_token_accuracy": 0.0829,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.4948,
      "epoch": 2.054176072234763,
      "grad_norm": 14.678378105163574,
      "learning_rate": 6.586427719519901e-06,
      "lm_loss": 5.4833,
      "loss": 1.5127,
      "step": 910,
      "text_contrastive_loss": 0.9392,
      "train_positive_log_prob": -80.2189,
      "train_positive_token_accuracy": 0.0732,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.4893,
      "epoch": 2.056433408577878,
      "grad_norm": 14.416768074035645,
      "learning_rate": 6.579545528114344e-06,
      "lm_loss": 5.5788,
      "loss": 1.4894,
      "step": 911,
      "text_contrastive_loss": 0.8845,
      "train_positive_log_prob": -81.7893,
      "train_positive_token_accuracy": 0.0738,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.4469,
      "epoch": 2.0586907449209932,
      "grad_norm": 14.04563045501709,
      "learning_rate": 6.572660010756489e-06,
      "lm_loss": 5.4231,
      "loss": 1.3099,
      "step": 912,
      "text_contrastive_loss": 0.6414,
      "train_positive_log_prob": -79.2252,
      "train_positive_token_accuracy": 0.0872,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.3964,
      "epoch": 2.0609480812641086,
      "grad_norm": 12.90860366821289,
      "learning_rate": 6.565771181944747e-06,
      "lm_loss": 5.3843,
      "loss": 1.2803,
      "step": 913,
      "text_contrastive_loss": 0.691,
      "train_positive_log_prob": -77.2742,
      "train_positive_token_accuracy": 0.0852,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.4296,
      "epoch": 2.0632054176072234,
      "grad_norm": 14.181915283203125,
      "learning_rate": 6.558879056184505e-06,
      "lm_loss": 5.6611,
      "loss": 1.3225,
      "step": 914,
      "text_contrastive_loss": 0.6536,
      "train_positive_log_prob": -84.2789,
      "train_positive_token_accuracy": 0.0722,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.4067,
      "epoch": 2.0654627539503387,
      "grad_norm": 13.280404090881348,
      "learning_rate": 6.551983647988089e-06,
      "lm_loss": 5.5431,
      "loss": 1.3324,
      "step": 915,
      "text_contrastive_loss": 0.7427,
      "train_positive_log_prob": -80.6588,
      "train_positive_token_accuracy": 0.0846,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.5256,
      "epoch": 2.0677200902934536,
      "grad_norm": 15.21223258972168,
      "learning_rate": 6.545084971874738e-06,
      "lm_loss": 5.5956,
      "loss": 1.6109,
      "step": 916,
      "text_contrastive_loss": 1.0513,
      "train_positive_log_prob": -81.8115,
      "train_positive_token_accuracy": 0.0847,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3383,
      "epoch": 2.069977426636569,
      "grad_norm": 12.661205291748047,
      "learning_rate": 6.5381830423705714e-06,
      "lm_loss": 5.5814,
      "loss": 1.2323,
      "step": 917,
      "text_contrastive_loss": 0.6718,
      "train_positive_log_prob": -82.2398,
      "train_positive_token_accuracy": 0.0779,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.4661,
      "epoch": 2.072234762979684,
      "grad_norm": 13.613990783691406,
      "learning_rate": 6.531277874008562e-06,
      "lm_loss": 5.4936,
      "loss": 1.5213,
      "step": 918,
      "text_contrastive_loss": 1.0117,
      "train_positive_log_prob": -81.3131,
      "train_positive_token_accuracy": 0.0866,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.3477,
      "epoch": 2.074492099322799,
      "grad_norm": 14.345525741577148,
      "learning_rate": 6.5243694813284975e-06,
      "lm_loss": 5.7273,
      "loss": 1.2323,
      "step": 919,
      "text_contrastive_loss": 0.6239,
      "train_positive_log_prob": -85.0802,
      "train_positive_token_accuracy": 0.0749,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.3837,
      "epoch": 2.0767494356659144,
      "grad_norm": 12.754055976867676,
      "learning_rate": 6.517457878876958e-06,
      "lm_loss": 5.619,
      "loss": 1.3939,
      "step": 920,
      "text_contrastive_loss": 0.8966,
      "train_positive_log_prob": -82.7717,
      "train_positive_token_accuracy": 0.0721,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.3951,
      "epoch": 2.0790067720090293,
      "grad_norm": 13.097175598144531,
      "learning_rate": 6.510543081207281e-06,
      "lm_loss": 5.5249,
      "loss": 1.3679,
      "step": 921,
      "text_contrastive_loss": 0.8406,
      "train_positive_log_prob": -82.0078,
      "train_positive_token_accuracy": 0.0766,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.5029,
      "epoch": 2.0812641083521446,
      "grad_norm": 16.092721939086914,
      "learning_rate": 6.503625102879534e-06,
      "lm_loss": 5.5724,
      "loss": 1.417,
      "step": 922,
      "text_contrastive_loss": 0.7137,
      "train_positive_log_prob": -83.2183,
      "train_positive_token_accuracy": 0.0795,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3339,
      "epoch": 2.0835214446952595,
      "grad_norm": 13.593026161193848,
      "learning_rate": 6.496703958460479e-06,
      "lm_loss": 5.502,
      "loss": 1.2641,
      "step": 923,
      "text_contrastive_loss": 0.76,
      "train_positive_log_prob": -81.2405,
      "train_positive_token_accuracy": 0.078,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.4719,
      "epoch": 2.0857787810383748,
      "grad_norm": 14.608567237854004,
      "learning_rate": 6.489779662523545e-06,
      "lm_loss": 5.6129,
      "loss": 1.4514,
      "step": 924,
      "text_contrastive_loss": 0.8363,
      "train_positive_log_prob": -84.1724,
      "train_positive_token_accuracy": 0.0758,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.4405,
      "epoch": 2.0880361173814896,
      "grad_norm": 13.26819133758545,
      "learning_rate": 6.4828522296488014e-06,
      "lm_loss": 5.5443,
      "loss": 1.5133,
      "step": 925,
      "text_contrastive_loss": 1.0368,
      "train_positive_log_prob": -82.1472,
      "train_positive_token_accuracy": 0.0759,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.4492,
      "epoch": 2.090293453724605,
      "grad_norm": 14.2427339553833,
      "learning_rate": 6.475921674422917e-06,
      "lm_loss": 5.473,
      "loss": 1.4559,
      "step": 926,
      "text_contrastive_loss": 0.9186,
      "train_positive_log_prob": -80.4308,
      "train_positive_token_accuracy": 0.0819,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.2978,
      "epoch": 2.0925507900677203,
      "grad_norm": 10.867948532104492,
      "learning_rate": 6.4689880114391375e-06,
      "lm_loss": 5.5041,
      "loss": 1.1965,
      "step": 927,
      "text_contrastive_loss": 0.6966,
      "train_positive_log_prob": -81.1004,
      "train_positive_token_accuracy": 0.0756,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.4376,
      "epoch": 2.094808126410835,
      "grad_norm": 15.287556648254395,
      "learning_rate": 6.462051255297255e-06,
      "lm_loss": 5.6055,
      "loss": 1.4279,
      "step": 928,
      "text_contrastive_loss": 0.8596,
      "train_positive_log_prob": -84.4968,
      "train_positive_token_accuracy": 0.0785,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.3481,
      "epoch": 2.0970654627539504,
      "grad_norm": 11.80434799194336,
      "learning_rate": 6.455111420603568e-06,
      "lm_loss": 5.5582,
      "loss": 1.3113,
      "step": 929,
      "text_contrastive_loss": 0.8148,
      "train_positive_log_prob": -81.5301,
      "train_positive_token_accuracy": 0.0807,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.3704,
      "epoch": 2.0993227990970653,
      "grad_norm": 14.718912124633789,
      "learning_rate": 6.448168521970865e-06,
      "lm_loss": 5.5771,
      "loss": 1.2943,
      "step": 930,
      "text_contrastive_loss": 0.7324,
      "train_positive_log_prob": -81.2363,
      "train_positive_token_accuracy": 0.074,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.4146,
      "epoch": 2.1015801354401806,
      "grad_norm": 13.320003509521484,
      "learning_rate": 6.441222574018378e-06,
      "lm_loss": 5.5359,
      "loss": 1.3511,
      "step": 931,
      "text_contrastive_loss": 0.7659,
      "train_positive_log_prob": -81.8475,
      "train_positive_token_accuracy": 0.0729,
      "train_positive_token_prob": 0.0286
    },
    {
      "contrastive_loss": 0.4271,
      "epoch": 2.1038374717832955,
      "grad_norm": 15.628487586975098,
      "learning_rate": 6.434273591371771e-06,
      "lm_loss": 5.6005,
      "loss": 1.4878,
      "step": 932,
      "text_contrastive_loss": 1.0014,
      "train_positive_log_prob": -85.2814,
      "train_positive_token_accuracy": 0.0757,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3987,
      "epoch": 2.106094808126411,
      "grad_norm": 13.400457382202148,
      "learning_rate": 6.427321588663085e-06,
      "lm_loss": 5.5665,
      "loss": 1.339,
      "step": 933,
      "text_contrastive_loss": 0.7674,
      "train_positive_log_prob": -83.3696,
      "train_positive_token_accuracy": 0.0806,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.4431,
      "epoch": 2.108352144469526,
      "grad_norm": 13.66365909576416,
      "learning_rate": 6.42036658053073e-06,
      "lm_loss": 5.6296,
      "loss": 1.3941,
      "step": 934,
      "text_contrastive_loss": 0.7761,
      "train_positive_log_prob": -84.5974,
      "train_positive_token_accuracy": 0.0756,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.4123,
      "epoch": 2.110609480812641,
      "grad_norm": 13.463959693908691,
      "learning_rate": 6.41340858161944e-06,
      "lm_loss": 5.4605,
      "loss": 1.3361,
      "step": 935,
      "text_contrastive_loss": 0.7553,
      "train_positive_log_prob": -80.5011,
      "train_positive_token_accuracy": 0.0808,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.4612,
      "epoch": 2.1128668171557563,
      "grad_norm": 14.150492668151855,
      "learning_rate": 6.406447606580248e-06,
      "lm_loss": 5.58,
      "loss": 1.4585,
      "step": 936,
      "text_contrastive_loss": 0.8786,
      "train_positive_log_prob": -83.7241,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.3639,
      "epoch": 2.115124153498871,
      "grad_norm": 13.390018463134766,
      "learning_rate": 6.399483670070451e-06,
      "lm_loss": 5.5468,
      "loss": 1.2795,
      "step": 937,
      "text_contrastive_loss": 0.7219,
      "train_positive_log_prob": -83.3512,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.3514,
      "epoch": 2.1173814898419865,
      "grad_norm": 13.621367454528809,
      "learning_rate": 6.392516786753586e-06,
      "lm_loss": 5.6154,
      "loss": 1.2941,
      "step": 938,
      "text_contrastive_loss": 0.7623,
      "train_positive_log_prob": -83.423,
      "train_positive_token_accuracy": 0.0678,
      "train_positive_token_prob": 0.0279
    },
    {
      "contrastive_loss": 0.4295,
      "epoch": 2.119638826185102,
      "grad_norm": 14.347768783569336,
      "learning_rate": 6.385546971299389e-06,
      "lm_loss": 5.491,
      "loss": 1.3672,
      "step": 939,
      "text_contrastive_loss": 0.7772,
      "train_positive_log_prob": -79.6387,
      "train_positive_token_accuracy": 0.0757,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3662,
      "epoch": 2.1218961625282167,
      "grad_norm": 11.981817245483398,
      "learning_rate": 6.378574238383776e-06,
      "lm_loss": 5.3599,
      "loss": 1.2613,
      "step": 940,
      "text_contrastive_loss": 0.7182,
      "train_positive_log_prob": -79.4741,
      "train_positive_token_accuracy": 0.0827,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.3346,
      "epoch": 2.124153498871332,
      "grad_norm": 12.558072090148926,
      "learning_rate": 6.3715986026888046e-06,
      "lm_loss": 5.6151,
      "loss": 1.2974,
      "step": 941,
      "text_contrastive_loss": 0.8026,
      "train_positive_log_prob": -83.5464,
      "train_positive_token_accuracy": 0.0835,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.4367,
      "epoch": 2.126410835214447,
      "grad_norm": 13.51561450958252,
      "learning_rate": 6.3646200789026426e-06,
      "lm_loss": 5.5248,
      "loss": 1.422,
      "step": 942,
      "text_contrastive_loss": 0.8657,
      "train_positive_log_prob": -81.407,
      "train_positive_token_accuracy": 0.0764,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.4196,
      "epoch": 2.128668171557562,
      "grad_norm": 12.983024597167969,
      "learning_rate": 6.35763868171954e-06,
      "lm_loss": 5.5869,
      "loss": 1.401,
      "step": 943,
      "text_contrastive_loss": 0.8454,
      "train_positive_log_prob": -79.9807,
      "train_positive_token_accuracy": 0.0775,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3939,
      "epoch": 2.130925507900677,
      "grad_norm": 12.527701377868652,
      "learning_rate": 6.350654425839799e-06,
      "lm_loss": 5.5153,
      "loss": 1.398,
      "step": 944,
      "text_contrastive_loss": 0.9052,
      "train_positive_log_prob": -81.3503,
      "train_positive_token_accuracy": 0.089,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.4015,
      "epoch": 2.1331828442437923,
      "grad_norm": 14.444681167602539,
      "learning_rate": 6.343667325969736e-06,
      "lm_loss": 5.5525,
      "loss": 1.416,
      "step": 945,
      "text_contrastive_loss": 0.9185,
      "train_positive_log_prob": -82.01,
      "train_positive_token_accuracy": 0.0727,
      "train_positive_token_prob": 0.0289
    },
    {
      "contrastive_loss": 0.439,
      "epoch": 2.1354401805869077,
      "grad_norm": 12.963115692138672,
      "learning_rate": 6.336677396821663e-06,
      "lm_loss": 5.4959,
      "loss": 1.4531,
      "step": 946,
      "text_contrastive_loss": 0.9291,
      "train_positive_log_prob": -80.2713,
      "train_positive_token_accuracy": 0.0768,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3516,
      "epoch": 2.1376975169300225,
      "grad_norm": 12.958985328674316,
      "learning_rate": 6.3296846531138445e-06,
      "lm_loss": 5.6031,
      "loss": 1.3727,
      "step": 947,
      "text_contrastive_loss": 0.9215,
      "train_positive_log_prob": -82.3616,
      "train_positive_token_accuracy": 0.0795,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.4498,
      "epoch": 2.139954853273138,
      "grad_norm": 12.821478843688965,
      "learning_rate": 6.322689109570472e-06,
      "lm_loss": 5.6741,
      "loss": 1.4163,
      "step": 948,
      "text_contrastive_loss": 0.7982,
      "train_positive_log_prob": -85.5767,
      "train_positive_token_accuracy": 0.0743,
      "train_positive_token_prob": 0.0287
    },
    {
      "contrastive_loss": 0.473,
      "epoch": 2.1422121896162527,
      "grad_norm": 13.26208782196045,
      "learning_rate": 6.315690780921634e-06,
      "lm_loss": 5.5058,
      "loss": 1.4371,
      "step": 949,
      "text_contrastive_loss": 0.8271,
      "train_positive_log_prob": -81.7886,
      "train_positive_token_accuracy": 0.0755,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.3098,
      "epoch": 2.144469525959368,
      "grad_norm": 12.116291046142578,
      "learning_rate": 6.3086896819032814e-06,
      "lm_loss": 5.6079,
      "loss": 1.2509,
      "step": 950,
      "text_contrastive_loss": 0.7606,
      "train_positive_log_prob": -84.4703,
      "train_positive_token_accuracy": 0.0799,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.4076,
      "epoch": 2.146726862302483,
      "grad_norm": 14.053050994873047,
      "learning_rate": 6.301685827257202e-06,
      "lm_loss": 5.5962,
      "loss": 1.4516,
      "step": 951,
      "text_contrastive_loss": 0.9688,
      "train_positive_log_prob": -82.8739,
      "train_positive_token_accuracy": 0.0805,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.4269,
      "epoch": 2.148984198645598,
      "grad_norm": 13.233821868896484,
      "learning_rate": 6.294679231730983e-06,
      "lm_loss": 5.5696,
      "loss": 1.4272,
      "step": 952,
      "text_contrastive_loss": 0.8866,
      "train_positive_log_prob": -83.3044,
      "train_positive_token_accuracy": 0.0799,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.4512,
      "epoch": 2.1512415349887135,
      "grad_norm": 15.380033493041992,
      "learning_rate": 6.2876699100779815e-06,
      "lm_loss": 5.5953,
      "loss": 1.4521,
      "step": 953,
      "text_contrastive_loss": 0.8827,
      "train_positive_log_prob": -83.3358,
      "train_positive_token_accuracy": 0.0816,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.3405,
      "epoch": 2.1534988713318284,
      "grad_norm": 11.722415924072266,
      "learning_rate": 6.2806578770573e-06,
      "lm_loss": 5.4673,
      "loss": 1.2318,
      "step": 954,
      "text_contrastive_loss": 0.6891,
      "train_positive_log_prob": -81.4551,
      "train_positive_token_accuracy": 0.0742,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.3513,
      "epoch": 2.1557562076749437,
      "grad_norm": 12.42605972290039,
      "learning_rate": 6.273643147433743e-06,
      "lm_loss": 5.4634,
      "loss": 1.2589,
      "step": 955,
      "text_contrastive_loss": 0.7226,
      "train_positive_log_prob": -78.9663,
      "train_positive_token_accuracy": 0.0854,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3209,
      "epoch": 2.1580135440180586,
      "grad_norm": 11.624284744262695,
      "learning_rate": 6.266625735977802e-06,
      "lm_loss": 5.5567,
      "loss": 1.1859,
      "step": 956,
      "text_contrastive_loss": 0.6185,
      "train_positive_log_prob": -83.1381,
      "train_positive_token_accuracy": 0.0757,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.5073,
      "epoch": 2.160270880361174,
      "grad_norm": 15.735133171081543,
      "learning_rate": 6.259605657465607e-06,
      "lm_loss": 5.5571,
      "loss": 1.491,
      "step": 957,
      "text_contrastive_loss": 0.856,
      "train_positive_log_prob": -81.992,
      "train_positive_token_accuracy": 0.0786,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.3826,
      "epoch": 2.1625282167042887,
      "grad_norm": 12.599930763244629,
      "learning_rate": 6.252582926678908e-06,
      "lm_loss": 5.5687,
      "loss": 1.3872,
      "step": 958,
      "text_contrastive_loss": 0.8956,
      "train_positive_log_prob": -84.5738,
      "train_positive_token_accuracy": 0.0766,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.3622,
      "epoch": 2.164785553047404,
      "grad_norm": 13.990739822387695,
      "learning_rate": 6.24555755840504e-06,
      "lm_loss": 5.6309,
      "loss": 1.2896,
      "step": 959,
      "text_contrastive_loss": 0.7287,
      "train_positive_log_prob": -83.6268,
      "train_positive_token_accuracy": 0.0755,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.3748,
      "epoch": 2.1670428893905194,
      "grad_norm": 12.508328437805176,
      "learning_rate": 6.238529567436892e-06,
      "lm_loss": 5.5369,
      "loss": 1.2588,
      "step": 960,
      "text_contrastive_loss": 0.6607,
      "train_positive_log_prob": -82.728,
      "train_positive_token_accuracy": 0.079,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.4102,
      "epoch": 2.1693002257336342,
      "grad_norm": 14.44834041595459,
      "learning_rate": 6.231498968572872e-06,
      "lm_loss": 5.5447,
      "loss": 1.3961,
      "step": 961,
      "text_contrastive_loss": 0.8627,
      "train_positive_log_prob": -84.1846,
      "train_positive_token_accuracy": 0.0813,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.4039,
      "epoch": 2.1715575620767495,
      "grad_norm": 12.841310501098633,
      "learning_rate": 6.224465776616883e-06,
      "lm_loss": 5.5862,
      "loss": 1.312,
      "step": 962,
      "text_contrastive_loss": 0.699,
      "train_positive_log_prob": -83.576,
      "train_positive_token_accuracy": 0.0747,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.413,
      "epoch": 2.1738148984198644,
      "grad_norm": 12.299847602844238,
      "learning_rate": 6.217430006378285e-06,
      "lm_loss": 5.6728,
      "loss": 1.3981,
      "step": 963,
      "text_contrastive_loss": 0.8357,
      "train_positive_log_prob": -83.8058,
      "train_positive_token_accuracy": 0.0793,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.418,
      "epoch": 2.1760722347629797,
      "grad_norm": 14.192880630493164,
      "learning_rate": 6.210391672671869e-06,
      "lm_loss": 5.5752,
      "loss": 1.2848,
      "step": 964,
      "text_contrastive_loss": 0.6185,
      "train_positive_log_prob": -80.6874,
      "train_positive_token_accuracy": 0.0769,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.3756,
      "epoch": 2.1783295711060946,
      "grad_norm": 13.454263687133789,
      "learning_rate": 6.203350790317825e-06,
      "lm_loss": 5.5448,
      "loss": 1.361,
      "step": 965,
      "text_contrastive_loss": 0.8617,
      "train_positive_log_prob": -80.205,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.4535,
      "epoch": 2.18058690744921,
      "grad_norm": 13.363504409790039,
      "learning_rate": 6.196307374141707e-06,
      "lm_loss": 5.5796,
      "loss": 1.4404,
      "step": 966,
      "text_contrastive_loss": 0.858,
      "train_positive_log_prob": -84.37,
      "train_positive_token_accuracy": 0.0741,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.4532,
      "epoch": 2.1828442437923252,
      "grad_norm": 15.592851638793945,
      "learning_rate": 6.189261438974403e-06,
      "lm_loss": 5.6168,
      "loss": 1.4715,
      "step": 967,
      "text_contrastive_loss": 0.9132,
      "train_positive_log_prob": -82.6105,
      "train_positive_token_accuracy": 0.077,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.4228,
      "epoch": 2.18510158013544,
      "grad_norm": 13.822402000427246,
      "learning_rate": 6.1822129996521105e-06,
      "lm_loss": 5.679,
      "loss": 1.4811,
      "step": 968,
      "text_contrastive_loss": 0.9808,
      "train_positive_log_prob": -82.7222,
      "train_positive_token_accuracy": 0.0732,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.3349,
      "epoch": 2.1873589164785554,
      "grad_norm": 12.573756217956543,
      "learning_rate": 6.175162071016295e-06,
      "lm_loss": 5.5118,
      "loss": 1.3064,
      "step": 969,
      "text_contrastive_loss": 0.8408,
      "train_positive_log_prob": -80.7816,
      "train_positive_token_accuracy": 0.0717,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.4184,
      "epoch": 2.1896162528216703,
      "grad_norm": 14.005714416503906,
      "learning_rate": 6.168108667913666e-06,
      "lm_loss": 5.5475,
      "loss": 1.4001,
      "step": 970,
      "text_contrastive_loss": 0.8538,
      "train_positive_log_prob": -82.1283,
      "train_positive_token_accuracy": 0.0736,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3898,
      "epoch": 2.1918735891647856,
      "grad_norm": 12.450246810913086,
      "learning_rate": 6.161052805196141e-06,
      "lm_loss": 5.5428,
      "loss": 1.3176,
      "step": 971,
      "text_contrastive_loss": 0.7471,
      "train_positive_log_prob": -82.1434,
      "train_positive_token_accuracy": 0.0775,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.3987,
      "epoch": 2.194130925507901,
      "grad_norm": 13.179352760314941,
      "learning_rate": 6.15399449772082e-06,
      "lm_loss": 5.5757,
      "loss": 1.283,
      "step": 972,
      "text_contrastive_loss": 0.6534,
      "train_positive_log_prob": -83.6822,
      "train_positive_token_accuracy": 0.0763,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.2804,
      "epoch": 2.1963882618510158,
      "grad_norm": 11.19333267211914,
      "learning_rate": 6.146933760349947e-06,
      "lm_loss": 5.6084,
      "loss": 1.1683,
      "step": 973,
      "text_contrastive_loss": 0.654,
      "train_positive_log_prob": -84.9345,
      "train_positive_token_accuracy": 0.0745,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.4641,
      "epoch": 2.198645598194131,
      "grad_norm": 13.623973846435547,
      "learning_rate": 6.139870607950885e-06,
      "lm_loss": 5.5269,
      "loss": 1.4512,
      "step": 974,
      "text_contrastive_loss": 0.8689,
      "train_positive_log_prob": -81.4148,
      "train_positive_token_accuracy": 0.0711,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3802,
      "epoch": 2.200902934537246,
      "grad_norm": 11.477293014526367,
      "learning_rate": 6.1328050553960804e-06,
      "lm_loss": 5.5935,
      "loss": 1.2467,
      "step": 975,
      "text_contrastive_loss": 0.6142,
      "train_positive_log_prob": -82.6155,
      "train_positive_token_accuracy": 0.0775,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.4762,
      "epoch": 2.2031602708803613,
      "grad_norm": 14.226114273071289,
      "learning_rate": 6.1257371175630375e-06,
      "lm_loss": 5.4978,
      "loss": 1.4596,
      "step": 976,
      "text_contrastive_loss": 0.8673,
      "train_positive_log_prob": -81.0309,
      "train_positive_token_accuracy": 0.0803,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.4986,
      "epoch": 2.205417607223476,
      "grad_norm": 14.475671768188477,
      "learning_rate": 6.118666809334277e-06,
      "lm_loss": 5.582,
      "loss": 1.5774,
      "step": 977,
      "text_contrastive_loss": 1.0414,
      "train_positive_log_prob": -82.0069,
      "train_positive_token_accuracy": 0.0744,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.4309,
      "epoch": 2.2076749435665914,
      "grad_norm": 13.799911499023438,
      "learning_rate": 6.111594145597319e-06,
      "lm_loss": 5.4693,
      "loss": 1.3871,
      "step": 978,
      "text_contrastive_loss": 0.8186,
      "train_positive_log_prob": -80.9287,
      "train_positive_token_accuracy": 0.0808,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.4412,
      "epoch": 2.2099322799097068,
      "grad_norm": 13.247078895568848,
      "learning_rate": 6.104519141244631e-06,
      "lm_loss": 5.4917,
      "loss": 1.395,
      "step": 979,
      "text_contrastive_loss": 0.8093,
      "train_positive_log_prob": -80.4919,
      "train_positive_token_accuracy": 0.0784,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.283,
      "epoch": 2.2121896162528216,
      "grad_norm": 11.904372215270996,
      "learning_rate": 6.0974418111736235e-06,
      "lm_loss": 5.6329,
      "loss": 1.1767,
      "step": 980,
      "text_contrastive_loss": 0.6608,
      "train_positive_log_prob": -84.5426,
      "train_positive_token_accuracy": 0.0799,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.4525,
      "epoch": 2.214446952595937,
      "grad_norm": 14.136039733886719,
      "learning_rate": 6.090362170286591e-06,
      "lm_loss": 5.505,
      "loss": 1.3167,
      "step": 981,
      "text_contrastive_loss": 0.6275,
      "train_positive_log_prob": -81.738,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.4498,
      "epoch": 2.216704288939052,
      "grad_norm": 14.322342872619629,
      "learning_rate": 6.0832802334907044e-06,
      "lm_loss": 5.5289,
      "loss": 1.3687,
      "step": 982,
      "text_contrastive_loss": 0.7321,
      "train_positive_log_prob": -83.043,
      "train_positive_token_accuracy": 0.0861,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3964,
      "epoch": 2.218961625282167,
      "grad_norm": 13.998641967773438,
      "learning_rate": 6.076196015697963e-06,
      "lm_loss": 5.5985,
      "loss": 1.3455,
      "step": 983,
      "text_contrastive_loss": 0.7786,
      "train_positive_log_prob": -85.139,
      "train_positive_token_accuracy": 0.0822,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3999,
      "epoch": 2.221218961625282,
      "grad_norm": 16.890783309936523,
      "learning_rate": 6.069109531825169e-06,
      "lm_loss": 5.5384,
      "loss": 1.3413,
      "step": 984,
      "text_contrastive_loss": 0.7753,
      "train_positive_log_prob": -80.1608,
      "train_positive_token_accuracy": 0.0726,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.4184,
      "epoch": 2.2234762979683973,
      "grad_norm": 14.210457801818848,
      "learning_rate": 6.0620207967939e-06,
      "lm_loss": 5.6143,
      "loss": 1.4467,
      "step": 985,
      "text_contrastive_loss": 0.9337,
      "train_positive_log_prob": -82.2629,
      "train_positive_token_accuracy": 0.0698,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.4921,
      "epoch": 2.2257336343115126,
      "grad_norm": 14.350348472595215,
      "learning_rate": 6.054929825530469e-06,
      "lm_loss": 5.5366,
      "loss": 1.4817,
      "step": 986,
      "text_contrastive_loss": 0.8719,
      "train_positive_log_prob": -82.1436,
      "train_positive_token_accuracy": 0.0761,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.2638,
      "epoch": 2.2279909706546275,
      "grad_norm": 11.467473030090332,
      "learning_rate": 6.047836632965901e-06,
      "lm_loss": 5.6007,
      "loss": 1.1329,
      "step": 987,
      "text_contrastive_loss": 0.6181,
      "train_positive_log_prob": -82.5658,
      "train_positive_token_accuracy": 0.0787,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.4258,
      "epoch": 2.230248306997743,
      "grad_norm": 12.957710266113281,
      "learning_rate": 6.040741234035898e-06,
      "lm_loss": 5.5427,
      "loss": 1.3607,
      "step": 988,
      "text_contrastive_loss": 0.7613,
      "train_positive_log_prob": -81.516,
      "train_positive_token_accuracy": 0.0752,
      "train_positive_token_prob": 0.0287
    },
    {
      "contrastive_loss": 0.4547,
      "epoch": 2.2325056433408577,
      "grad_norm": 14.008922576904297,
      "learning_rate": 6.0336436436808054e-06,
      "lm_loss": 5.5997,
      "loss": 1.5015,
      "step": 989,
      "text_contrastive_loss": 0.9736,
      "train_positive_log_prob": -82.5375,
      "train_positive_token_accuracy": 0.079,
      "train_positive_token_prob": 0.0289
    },
    {
      "contrastive_loss": 0.3921,
      "epoch": 2.234762979683973,
      "grad_norm": 13.679863929748535,
      "learning_rate": 6.026543876845586e-06,
      "lm_loss": 5.5863,
      "loss": 1.3665,
      "step": 990,
      "text_contrastive_loss": 0.8316,
      "train_positive_log_prob": -83.5533,
      "train_positive_token_accuracy": 0.0867,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.3997,
      "epoch": 2.237020316027088,
      "grad_norm": 14.333312034606934,
      "learning_rate": 6.019441948479784e-06,
      "lm_loss": 5.4539,
      "loss": 1.3825,
      "step": 991,
      "text_contrastive_loss": 0.8748,
      "train_positive_log_prob": -80.6994,
      "train_positive_token_accuracy": 0.0763,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.4194,
      "epoch": 2.239277652370203,
      "grad_norm": 13.665488243103027,
      "learning_rate": 6.012337873537494e-06,
      "lm_loss": 5.5423,
      "loss": 1.3489,
      "step": 992,
      "text_contrastive_loss": 0.7507,
      "train_positive_log_prob": -83.0468,
      "train_positive_token_accuracy": 0.0795,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.3387,
      "epoch": 2.2415349887133185,
      "grad_norm": 12.015118598937988,
      "learning_rate": 6.005231666977331e-06,
      "lm_loss": 5.4918,
      "loss": 1.2651,
      "step": 993,
      "text_contrastive_loss": 0.7545,
      "train_positive_log_prob": -81.261,
      "train_positive_token_accuracy": 0.0837,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3605,
      "epoch": 2.2437923250564333,
      "grad_norm": 12.484308242797852,
      "learning_rate": 5.998123343762403e-06,
      "lm_loss": 5.5407,
      "loss": 1.3577,
      "step": 994,
      "text_contrastive_loss": 0.8864,
      "train_positive_log_prob": -81.2947,
      "train_positive_token_accuracy": 0.0809,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.328,
      "epoch": 2.2460496613995486,
      "grad_norm": 12.022150039672852,
      "learning_rate": 5.9910129188602665e-06,
      "lm_loss": 5.6345,
      "loss": 1.321,
      "step": 995,
      "text_contrastive_loss": 0.8593,
      "train_positive_log_prob": -83.0857,
      "train_positive_token_accuracy": 0.0731,
      "train_positive_token_prob": 0.0284
    },
    {
      "contrastive_loss": 0.3676,
      "epoch": 2.2483069977426635,
      "grad_norm": 13.612593650817871,
      "learning_rate": 5.983900407242911e-06,
      "lm_loss": 5.5751,
      "loss": 1.332,
      "step": 996,
      "text_contrastive_loss": 0.8138,
      "train_positive_log_prob": -82.7977,
      "train_positive_token_accuracy": 0.0786,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3833,
      "epoch": 2.250564334085779,
      "grad_norm": 13.808063507080078,
      "learning_rate": 5.976785823886713e-06,
      "lm_loss": 5.6367,
      "loss": 1.3317,
      "step": 997,
      "text_contrastive_loss": 0.7696,
      "train_positive_log_prob": -83.0521,
      "train_positive_token_accuracy": 0.071,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.3783,
      "epoch": 2.2528216704288937,
      "grad_norm": 12.80401611328125,
      "learning_rate": 5.96966918377242e-06,
      "lm_loss": 5.46,
      "loss": 1.3027,
      "step": 998,
      "text_contrastive_loss": 0.7569,
      "train_positive_log_prob": -82.0075,
      "train_positive_token_accuracy": 0.0848,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.3602,
      "epoch": 2.255079006772009,
      "grad_norm": 13.811467170715332,
      "learning_rate": 5.9625505018851e-06,
      "lm_loss": 5.4395,
      "loss": 1.3155,
      "step": 999,
      "text_contrastive_loss": 0.8227,
      "train_positive_log_prob": -82.1437,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.3746,
      "epoch": 2.2573363431151243,
      "grad_norm": 13.968853950500488,
      "learning_rate": 5.955429793214129e-06,
      "lm_loss": 5.6201,
      "loss": 1.2878,
      "step": 1000,
      "text_contrastive_loss": 0.7023,
      "train_positive_log_prob": -82.3269,
      "train_positive_token_accuracy": 0.0813,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.4229,
      "epoch": 2.259593679458239,
      "grad_norm": 14.745325088500977,
      "learning_rate": 5.948307072753146e-06,
      "lm_loss": 5.5213,
      "loss": 1.3778,
      "step": 1001,
      "text_contrastive_loss": 0.8054,
      "train_positive_log_prob": -81.3077,
      "train_positive_token_accuracy": 0.0724,
      "train_positive_token_prob": 0.0288
    },
    {
      "contrastive_loss": 0.3283,
      "epoch": 2.2618510158013545,
      "grad_norm": 14.804535865783691,
      "learning_rate": 5.941182355500028e-06,
      "lm_loss": 5.4092,
      "loss": 1.2485,
      "step": 1002,
      "text_contrastive_loss": 0.7585,
      "train_positive_log_prob": -80.8454,
      "train_positive_token_accuracy": 0.0847,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3974,
      "epoch": 2.2641083521444694,
      "grad_norm": 13.344841957092285,
      "learning_rate": 5.934055656456855e-06,
      "lm_loss": 5.3794,
      "loss": 1.2822,
      "step": 1003,
      "text_contrastive_loss": 0.6938,
      "train_positive_log_prob": -79.8066,
      "train_positive_token_accuracy": 0.0887,
      "train_positive_token_prob": 0.0336
    },
    {
      "contrastive_loss": 0.4088,
      "epoch": 2.2663656884875847,
      "grad_norm": 12.040769577026367,
      "learning_rate": 5.926926990629883e-06,
      "lm_loss": 5.442,
      "loss": 1.3089,
      "step": 1004,
      "text_contrastive_loss": 0.7117,
      "train_positive_log_prob": -79.3547,
      "train_positive_token_accuracy": 0.0838,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.4572,
      "epoch": 2.2686230248307,
      "grad_norm": 13.579639434814453,
      "learning_rate": 5.919796373029504e-06,
      "lm_loss": 5.5397,
      "loss": 1.3674,
      "step": 1005,
      "text_contrastive_loss": 0.7125,
      "train_positive_log_prob": -82.7463,
      "train_positive_token_accuracy": 0.0764,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3127,
      "epoch": 2.270880361173815,
      "grad_norm": 12.374353408813477,
      "learning_rate": 5.912663818670224e-06,
      "lm_loss": 5.5049,
      "loss": 1.2269,
      "step": 1006,
      "text_contrastive_loss": 0.7274,
      "train_positive_log_prob": -81.8819,
      "train_positive_token_accuracy": 0.0825,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3436,
      "epoch": 2.27313769751693,
      "grad_norm": 12.528451919555664,
      "learning_rate": 5.905529342570627e-06,
      "lm_loss": 5.5235,
      "loss": 1.2804,
      "step": 1007,
      "text_contrastive_loss": 0.7689,
      "train_positive_log_prob": -80.7811,
      "train_positive_token_accuracy": 0.0769,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.4227,
      "epoch": 2.275395033860045,
      "grad_norm": 16.113021850585938,
      "learning_rate": 5.898392959753343e-06,
      "lm_loss": 5.4232,
      "loss": 1.3767,
      "step": 1008,
      "text_contrastive_loss": 0.8232,
      "train_positive_log_prob": -80.6083,
      "train_positive_token_accuracy": 0.0746,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.405,
      "epoch": 2.2776523702031604,
      "grad_norm": 13.573700904846191,
      "learning_rate": 5.8912546852450116e-06,
      "lm_loss": 5.5703,
      "loss": 1.3367,
      "step": 1009,
      "text_contrastive_loss": 0.7493,
      "train_positive_log_prob": -82.5105,
      "train_positive_token_accuracy": 0.0694,
      "train_positive_token_prob": 0.0282
    },
    {
      "contrastive_loss": 0.3962,
      "epoch": 2.2799097065462752,
      "grad_norm": 13.823863983154297,
      "learning_rate": 5.8841145340762665e-06,
      "lm_loss": 5.496,
      "loss": 1.3336,
      "step": 1010,
      "text_contrastive_loss": 0.7756,
      "train_positive_log_prob": -81.0228,
      "train_positive_token_accuracy": 0.0829,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.3785,
      "epoch": 2.2821670428893905,
      "grad_norm": 13.912785530090332,
      "learning_rate": 5.876972521281683e-06,
      "lm_loss": 5.5152,
      "loss": 1.2976,
      "step": 1011,
      "text_contrastive_loss": 0.7351,
      "train_positive_log_prob": -81.6721,
      "train_positive_token_accuracy": 0.0783,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3995,
      "epoch": 2.2844243792325054,
      "grad_norm": 14.400925636291504,
      "learning_rate": 5.869828661899761e-06,
      "lm_loss": 5.5173,
      "loss": 1.4048,
      "step": 1012,
      "text_contrastive_loss": 0.9072,
      "train_positive_log_prob": -79.5032,
      "train_positive_token_accuracy": 0.0781,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.4157,
      "epoch": 2.2866817155756207,
      "grad_norm": 14.685125350952148,
      "learning_rate": 5.862682970972888e-06,
      "lm_loss": 5.4966,
      "loss": 1.3343,
      "step": 1013,
      "text_contrastive_loss": 0.7379,
      "train_positive_log_prob": -80.7977,
      "train_positive_token_accuracy": 0.0874,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.4512,
      "epoch": 2.288939051918736,
      "grad_norm": 14.808658599853516,
      "learning_rate": 5.855535463547309e-06,
      "lm_loss": 5.4907,
      "loss": 1.3872,
      "step": 1014,
      "text_contrastive_loss": 0.7739,
      "train_positive_log_prob": -80.4921,
      "train_positive_token_accuracy": 0.0791,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.385,
      "epoch": 2.291196388261851,
      "grad_norm": 13.9261474609375,
      "learning_rate": 5.8483861546730915e-06,
      "lm_loss": 5.6069,
      "loss": 1.2924,
      "step": 1015,
      "text_contrastive_loss": 0.6934,
      "train_positive_log_prob": -82.5283,
      "train_positive_token_accuracy": 0.0719,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.4113,
      "epoch": 2.293453724604966,
      "grad_norm": 14.202028274536133,
      "learning_rate": 5.841235059404097e-06,
      "lm_loss": 5.418,
      "loss": 1.4645,
      "step": 1016,
      "text_contrastive_loss": 1.0228,
      "train_positive_log_prob": -80.362,
      "train_positive_token_accuracy": 0.0877,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.4061,
      "epoch": 2.295711060948081,
      "grad_norm": 13.882689476013184,
      "learning_rate": 5.834082192797948e-06,
      "lm_loss": 5.4877,
      "loss": 1.342,
      "step": 1017,
      "text_contrastive_loss": 0.7742,
      "train_positive_log_prob": -79.5026,
      "train_positive_token_accuracy": 0.0776,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.3712,
      "epoch": 2.2979683972911964,
      "grad_norm": 13.90723991394043,
      "learning_rate": 5.826927569915999e-06,
      "lm_loss": 5.7271,
      "loss": 1.3197,
      "step": 1018,
      "text_contrastive_loss": 0.7516,
      "train_positive_log_prob": -83.911,
      "train_positive_token_accuracy": 0.0725,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.4399,
      "epoch": 2.3002257336343117,
      "grad_norm": 14.244397163391113,
      "learning_rate": 5.819771205823303e-06,
      "lm_loss": 5.6209,
      "loss": 1.3774,
      "step": 1019,
      "text_contrastive_loss": 0.7507,
      "train_positive_log_prob": -83.5902,
      "train_positive_token_accuracy": 0.0791,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.4219,
      "epoch": 2.3024830699774266,
      "grad_norm": 14.064270973205566,
      "learning_rate": 5.812613115588575e-06,
      "lm_loss": 5.4932,
      "loss": 1.3646,
      "step": 1020,
      "text_contrastive_loss": 0.7868,
      "train_positive_log_prob": -79.76,
      "train_positive_token_accuracy": 0.0729,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.3629,
      "epoch": 2.304740406320542,
      "grad_norm": 13.734200477600098,
      "learning_rate": 5.805453314284168e-06,
      "lm_loss": 5.5866,
      "loss": 1.2901,
      "step": 1021,
      "text_contrastive_loss": 0.7371,
      "train_positive_log_prob": -83.4928,
      "train_positive_token_accuracy": 0.0808,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3651,
      "epoch": 2.3069977426636568,
      "grad_norm": 14.739347457885742,
      "learning_rate": 5.7982918169860395e-06,
      "lm_loss": 5.507,
      "loss": 1.3218,
      "step": 1022,
      "text_contrastive_loss": 0.8121,
      "train_positive_log_prob": -81.3155,
      "train_positive_token_accuracy": 0.077,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.3772,
      "epoch": 2.309255079006772,
      "grad_norm": 14.899076461791992,
      "learning_rate": 5.791128638773711e-06,
      "lm_loss": 5.5482,
      "loss": 1.3346,
      "step": 1023,
      "text_contrastive_loss": 0.8051,
      "train_positive_log_prob": -81.4921,
      "train_positive_token_accuracy": 0.0774,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.3936,
      "epoch": 2.311512415349887,
      "grad_norm": 12.669140815734863,
      "learning_rate": 5.783963794730254e-06,
      "lm_loss": 5.5638,
      "loss": 1.3333,
      "step": 1024,
      "text_contrastive_loss": 0.7665,
      "train_positive_log_prob": -82.1736,
      "train_positive_token_accuracy": 0.0772,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.3984,
      "epoch": 2.3137697516930023,
      "grad_norm": 13.517724990844727,
      "learning_rate": 5.776797299942236e-06,
      "lm_loss": 5.6272,
      "loss": 1.3307,
      "step": 1025,
      "text_contrastive_loss": 0.7392,
      "train_positive_log_prob": -83.9292,
      "train_positive_token_accuracy": 0.0794,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.2764,
      "epoch": 2.3160270880361176,
      "grad_norm": 11.722050666809082,
      "learning_rate": 5.7696291694997105e-06,
      "lm_loss": 5.6105,
      "loss": 1.2179,
      "step": 1026,
      "text_contrastive_loss": 0.761,
      "train_positive_log_prob": -84.001,
      "train_positive_token_accuracy": 0.0762,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.4774,
      "epoch": 2.3182844243792324,
      "grad_norm": 14.694503784179688,
      "learning_rate": 5.762459418496169e-06,
      "lm_loss": 5.4532,
      "loss": 1.4874,
      "step": 1027,
      "text_contrastive_loss": 0.9293,
      "train_positive_log_prob": -79.5419,
      "train_positive_token_accuracy": 0.0846,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.4467,
      "epoch": 2.3205417607223477,
      "grad_norm": 13.764412879943848,
      "learning_rate": 5.755288062028519e-06,
      "lm_loss": 5.4687,
      "loss": 1.398,
      "step": 1028,
      "text_contrastive_loss": 0.8088,
      "train_positive_log_prob": -80.6833,
      "train_positive_token_accuracy": 0.0736,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3892,
      "epoch": 2.3227990970654626,
      "grad_norm": 13.268473625183105,
      "learning_rate": 5.748115115197045e-06,
      "lm_loss": 5.5849,
      "loss": 1.335,
      "step": 1029,
      "text_contrastive_loss": 0.7746,
      "train_positive_log_prob": -81.4176,
      "train_positive_token_accuracy": 0.074,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.3528,
      "epoch": 2.325056433408578,
      "grad_norm": 13.001954078674316,
      "learning_rate": 5.740940593105383e-06,
      "lm_loss": 5.5094,
      "loss": 1.2074,
      "step": 1030,
      "text_contrastive_loss": 0.6074,
      "train_positive_log_prob": -81.6473,
      "train_positive_token_accuracy": 0.0778,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.4495,
      "epoch": 2.327313769751693,
      "grad_norm": 14.6458158493042,
      "learning_rate": 5.733764510860482e-06,
      "lm_loss": 5.4524,
      "loss": 1.3742,
      "step": 1031,
      "text_contrastive_loss": 0.759,
      "train_positive_log_prob": -79.1298,
      "train_positive_token_accuracy": 0.0786,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3712,
      "epoch": 2.329571106094808,
      "grad_norm": 13.251885414123535,
      "learning_rate": 5.726586883572584e-06,
      "lm_loss": 5.4817,
      "loss": 1.3233,
      "step": 1032,
      "text_contrastive_loss": 0.808,
      "train_positive_log_prob": -82.1205,
      "train_positive_token_accuracy": 0.073,
      "train_positive_token_prob": 0.0291
    },
    {
      "contrastive_loss": 0.4568,
      "epoch": 2.3318284424379234,
      "grad_norm": 14.308194160461426,
      "learning_rate": 5.719407726355174e-06,
      "lm_loss": 5.5153,
      "loss": 1.424,
      "step": 1033,
      "text_contrastive_loss": 0.8315,
      "train_positive_log_prob": -82.2517,
      "train_positive_token_accuracy": 0.0752,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.4328,
      "epoch": 2.3340857787810383,
      "grad_norm": 14.968981742858887,
      "learning_rate": 5.712227054324968e-06,
      "lm_loss": 5.6012,
      "loss": 1.4212,
      "step": 1034,
      "text_contrastive_loss": 0.8564,
      "train_positive_log_prob": -83.9986,
      "train_positive_token_accuracy": 0.0763,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.4197,
      "epoch": 2.3363431151241536,
      "grad_norm": 14.584522247314453,
      "learning_rate": 5.705044882601862e-06,
      "lm_loss": 5.5542,
      "loss": 1.4009,
      "step": 1035,
      "text_contrastive_loss": 0.8517,
      "train_positive_log_prob": -83.5942,
      "train_positive_token_accuracy": 0.0742,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.4648,
      "epoch": 2.3386004514672685,
      "grad_norm": 14.75509262084961,
      "learning_rate": 5.697861226308923e-06,
      "lm_loss": 5.5775,
      "loss": 1.4913,
      "step": 1036,
      "text_contrastive_loss": 0.9375,
      "train_positive_log_prob": -84.9494,
      "train_positive_token_accuracy": 0.0746,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.4331,
      "epoch": 2.340857787810384,
      "grad_norm": 14.654216766357422,
      "learning_rate": 5.69067610057233e-06,
      "lm_loss": 5.4593,
      "loss": 1.3833,
      "step": 1037,
      "text_contrastive_loss": 0.8085,
      "train_positive_log_prob": -83.1031,
      "train_positive_token_accuracy": 0.0749,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3848,
      "epoch": 2.343115124153499,
      "grad_norm": 12.451683044433594,
      "learning_rate": 5.683489520521365e-06,
      "lm_loss": 5.5258,
      "loss": 1.3609,
      "step": 1038,
      "text_contrastive_loss": 0.8471,
      "train_positive_log_prob": -81.2904,
      "train_positive_token_accuracy": 0.0794,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.4369,
      "epoch": 2.345372460496614,
      "grad_norm": 16.772539138793945,
      "learning_rate": 5.6763015012883686e-06,
      "lm_loss": 5.4578,
      "loss": 1.3977,
      "step": 1039,
      "text_contrastive_loss": 0.8301,
      "train_positive_log_prob": -81.2897,
      "train_positive_token_accuracy": 0.0719,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3251,
      "epoch": 2.3476297968397293,
      "grad_norm": 11.789501190185547,
      "learning_rate": 5.6691120580087126e-06,
      "lm_loss": 5.5235,
      "loss": 1.206,
      "step": 1040,
      "text_contrastive_loss": 0.6571,
      "train_positive_log_prob": -79.0103,
      "train_positive_token_accuracy": 0.0841,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3716,
      "epoch": 2.349887133182844,
      "grad_norm": 12.103446960449219,
      "learning_rate": 5.661921205820767e-06,
      "lm_loss": 5.4876,
      "loss": 1.279,
      "step": 1041,
      "text_contrastive_loss": 0.7173,
      "train_positive_log_prob": -80.6443,
      "train_positive_token_accuracy": 0.0829,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.5042,
      "epoch": 2.3521444695259595,
      "grad_norm": 15.723865509033203,
      "learning_rate": 5.654728959865872e-06,
      "lm_loss": 5.5846,
      "loss": 1.4548,
      "step": 1042,
      "text_contrastive_loss": 0.7843,
      "train_positive_log_prob": -84.0779,
      "train_positive_token_accuracy": 0.0823,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3544,
      "epoch": 2.3544018058690743,
      "grad_norm": 12.008702278137207,
      "learning_rate": 5.647535335288296e-06,
      "lm_loss": 5.5769,
      "loss": 1.2636,
      "step": 1043,
      "text_contrastive_loss": 0.7031,
      "train_positive_log_prob": -84.395,
      "train_positive_token_accuracy": 0.0759,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.3317,
      "epoch": 2.3566591422121896,
      "grad_norm": 11.979096412658691,
      "learning_rate": 5.640340347235215e-06,
      "lm_loss": 5.484,
      "loss": 1.2973,
      "step": 1044,
      "text_contrastive_loss": 0.8344,
      "train_positive_log_prob": -80.8715,
      "train_positive_token_accuracy": 0.0808,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.5111,
      "epoch": 2.3589164785553045,
      "grad_norm": 13.818731307983398,
      "learning_rate": 5.6331440108566735e-06,
      "lm_loss": 5.6814,
      "loss": 1.5727,
      "step": 1045,
      "text_contrastive_loss": 0.987,
      "train_positive_log_prob": -84.6209,
      "train_positive_token_accuracy": 0.0732,
      "train_positive_token_prob": 0.0285
    },
    {
      "contrastive_loss": 0.421,
      "epoch": 2.36117381489842,
      "grad_norm": 14.53001880645752,
      "learning_rate": 5.6259463413055604e-06,
      "lm_loss": 5.5695,
      "loss": 1.4331,
      "step": 1046,
      "text_contrastive_loss": 0.9103,
      "train_positive_log_prob": -81.9357,
      "train_positive_token_accuracy": 0.0791,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.4077,
      "epoch": 2.363431151241535,
      "grad_norm": 14.39387321472168,
      "learning_rate": 5.6187473537375635e-06,
      "lm_loss": 5.6563,
      "loss": 1.3375,
      "step": 1047,
      "text_contrastive_loss": 0.7283,
      "train_positive_log_prob": -81.6435,
      "train_positive_token_accuracy": 0.0725,
      "train_positive_token_prob": 0.0287
    },
    {
      "contrastive_loss": 0.3039,
      "epoch": 2.36568848758465,
      "grad_norm": 11.790059089660645,
      "learning_rate": 5.611547063311152e-06,
      "lm_loss": 5.4627,
      "loss": 1.2354,
      "step": 1048,
      "text_contrastive_loss": 0.7705,
      "train_positive_log_prob": -80.297,
      "train_positive_token_accuracy": 0.0792,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.463,
      "epoch": 2.3679458239277653,
      "grad_norm": 15.357479095458984,
      "learning_rate": 5.604345485187535e-06,
      "lm_loss": 5.5069,
      "loss": 1.4355,
      "step": 1049,
      "text_contrastive_loss": 0.8436,
      "train_positive_log_prob": -84.0033,
      "train_positive_token_accuracy": 0.0835,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.4173,
      "epoch": 2.37020316027088,
      "grad_norm": 12.703691482543945,
      "learning_rate": 5.597142634530639e-06,
      "lm_loss": 5.4058,
      "loss": 1.2865,
      "step": 1050,
      "text_contrastive_loss": 0.6572,
      "train_positive_log_prob": -80.3059,
      "train_positive_token_accuracy": 0.0884,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.4097,
      "epoch": 2.3724604966139955,
      "grad_norm": 14.017982482910156,
      "learning_rate": 5.589938526507059e-06,
      "lm_loss": 5.5373,
      "loss": 1.3326,
      "step": 1051,
      "text_contrastive_loss": 0.7383,
      "train_positive_log_prob": -81.131,
      "train_positive_token_accuracy": 0.069,
      "train_positive_token_prob": 0.0291
    },
    {
      "contrastive_loss": 0.4417,
      "epoch": 2.374717832957111,
      "grad_norm": 17.754405975341797,
      "learning_rate": 5.582733176286048e-06,
      "lm_loss": 5.469,
      "loss": 1.463,
      "step": 1052,
      "text_contrastive_loss": 0.9487,
      "train_positive_log_prob": -79.7064,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.3392,
      "epoch": 2.3769751693002257,
      "grad_norm": 13.985703468322754,
      "learning_rate": 5.575526599039472e-06,
      "lm_loss": 5.3489,
      "loss": 1.2603,
      "step": 1053,
      "text_contrastive_loss": 0.7725,
      "train_positive_log_prob": -79.1122,
      "train_positive_token_accuracy": 0.0816,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.2968,
      "epoch": 2.379232505643341,
      "grad_norm": 12.920539855957031,
      "learning_rate": 5.568318809941777e-06,
      "lm_loss": 5.4325,
      "loss": 1.2386,
      "step": 1054,
      "text_contrastive_loss": 0.7971,
      "train_positive_log_prob": -79.6203,
      "train_positive_token_accuracy": 0.0821,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3871,
      "epoch": 2.381489841986456,
      "grad_norm": 13.54922866821289,
      "learning_rate": 5.561109824169962e-06,
      "lm_loss": 5.4568,
      "loss": 1.2994,
      "step": 1055,
      "text_contrastive_loss": 0.7332,
      "train_positive_log_prob": -83.6655,
      "train_positive_token_accuracy": 0.0818,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3444,
      "epoch": 2.383747178329571,
      "grad_norm": 12.597684860229492,
      "learning_rate": 5.553899656903552e-06,
      "lm_loss": 5.5629,
      "loss": 1.2849,
      "step": 1056,
      "text_contrastive_loss": 0.7684,
      "train_positive_log_prob": -84.6168,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.4698,
      "epoch": 2.386004514672686,
      "grad_norm": 15.04989242553711,
      "learning_rate": 5.546688323324548e-06,
      "lm_loss": 5.5264,
      "loss": 1.4354,
      "step": 1057,
      "text_contrastive_loss": 0.826,
      "train_positive_log_prob": -82.9815,
      "train_positive_token_accuracy": 0.0812,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3774,
      "epoch": 2.3882618510158014,
      "grad_norm": 13.134078025817871,
      "learning_rate": 5.53947583861742e-06,
      "lm_loss": 5.4747,
      "loss": 1.3198,
      "step": 1058,
      "text_contrastive_loss": 0.7899,
      "train_positive_log_prob": -80.108,
      "train_positive_token_accuracy": 0.0854,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.4142,
      "epoch": 2.3905191873589167,
      "grad_norm": 13.905577659606934,
      "learning_rate": 5.5322622179690514e-06,
      "lm_loss": 5.4826,
      "loss": 1.3297,
      "step": 1059,
      "text_contrastive_loss": 0.7345,
      "train_positive_log_prob": -79.4124,
      "train_positive_token_accuracy": 0.0828,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3922,
      "epoch": 2.3927765237020315,
      "grad_norm": 13.758112907409668,
      "learning_rate": 5.525047476568722e-06,
      "lm_loss": 5.5356,
      "loss": 1.3236,
      "step": 1060,
      "text_contrastive_loss": 0.7556,
      "train_positive_log_prob": -82.5415,
      "train_positive_token_accuracy": 0.0812,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.4356,
      "epoch": 2.395033860045147,
      "grad_norm": 13.6261625289917,
      "learning_rate": 5.51783162960807e-06,
      "lm_loss": 5.5089,
      "loss": 1.3608,
      "step": 1061,
      "text_contrastive_loss": 0.7486,
      "train_positive_log_prob": -83.1293,
      "train_positive_token_accuracy": 0.0777,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.4276,
      "epoch": 2.3972911963882617,
      "grad_norm": 15.979998588562012,
      "learning_rate": 5.5106146922810664e-06,
      "lm_loss": 5.5639,
      "loss": 1.4571,
      "step": 1062,
      "text_contrastive_loss": 0.9461,
      "train_positive_log_prob": -83.0504,
      "train_positive_token_accuracy": 0.0748,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.465,
      "epoch": 2.399548532731377,
      "grad_norm": 15.927392959594727,
      "learning_rate": 5.50339667978397e-06,
      "lm_loss": 5.5618,
      "loss": 1.498,
      "step": 1063,
      "text_contrastive_loss": 0.9536,
      "train_positive_log_prob": -82.4492,
      "train_positive_token_accuracy": 0.0748,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.4131,
      "epoch": 2.401805869074492,
      "grad_norm": 14.187854766845703,
      "learning_rate": 5.496177607315312e-06,
      "lm_loss": 5.4653,
      "loss": 1.4196,
      "step": 1064,
      "text_contrastive_loss": 0.92,
      "train_positive_log_prob": -80.7589,
      "train_positive_token_accuracy": 0.0833,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.3869,
      "epoch": 2.404063205417607,
      "grad_norm": 13.354890823364258,
      "learning_rate": 5.488957490075846e-06,
      "lm_loss": 5.4859,
      "loss": 1.3326,
      "step": 1065,
      "text_contrastive_loss": 0.7943,
      "train_positive_log_prob": -80.728,
      "train_positive_token_accuracy": 0.0751,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.4043,
      "epoch": 2.4063205417607225,
      "grad_norm": 13.615126609802246,
      "learning_rate": 5.4817363432685355e-06,
      "lm_loss": 5.6338,
      "loss": 1.3987,
      "step": 1066,
      "text_contrastive_loss": 0.8621,
      "train_positive_log_prob": -84.2201,
      "train_positive_token_accuracy": 0.0787,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.4026,
      "epoch": 2.4085778781038374,
      "grad_norm": 12.85096263885498,
      "learning_rate": 5.474514182098504e-06,
      "lm_loss": 5.4415,
      "loss": 1.3841,
      "step": 1067,
      "text_contrastive_loss": 0.8747,
      "train_positive_log_prob": -80.7073,
      "train_positive_token_accuracy": 0.078,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.445,
      "epoch": 2.4108352144469527,
      "grad_norm": 13.284417152404785,
      "learning_rate": 5.4672910217730155e-06,
      "lm_loss": 5.4465,
      "loss": 1.3754,
      "step": 1068,
      "text_contrastive_loss": 0.7714,
      "train_positive_log_prob": -80.9296,
      "train_positive_token_accuracy": 0.0862,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.4147,
      "epoch": 2.4130925507900676,
      "grad_norm": 13.03646469116211,
      "learning_rate": 5.4600668775014355e-06,
      "lm_loss": 5.5732,
      "loss": 1.3354,
      "step": 1069,
      "text_contrastive_loss": 0.7267,
      "train_positive_log_prob": -82.8376,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.5255,
      "epoch": 2.415349887133183,
      "grad_norm": 13.955810546875,
      "learning_rate": 5.452841764495203e-06,
      "lm_loss": 5.442,
      "loss": 1.4446,
      "step": 1070,
      "text_contrastive_loss": 0.7496,
      "train_positive_log_prob": -78.8667,
      "train_positive_token_accuracy": 0.0801,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3943,
      "epoch": 2.417607223476298,
      "grad_norm": 14.359458923339844,
      "learning_rate": 5.445615697967797e-06,
      "lm_loss": 5.4259,
      "loss": 1.3027,
      "step": 1071,
      "text_contrastive_loss": 0.7316,
      "train_positive_log_prob": -81.8307,
      "train_positive_token_accuracy": 0.0862,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.531,
      "epoch": 2.419864559819413,
      "grad_norm": 14.613249778747559,
      "learning_rate": 5.438388693134702e-06,
      "lm_loss": 5.4336,
      "loss": 1.5319,
      "step": 1072,
      "text_contrastive_loss": 0.9151,
      "train_positive_log_prob": -78.8259,
      "train_positive_token_accuracy": 0.081,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.4024,
      "epoch": 2.4221218961625284,
      "grad_norm": 14.3463773727417,
      "learning_rate": 5.431160765213379e-06,
      "lm_loss": 5.4399,
      "loss": 1.3835,
      "step": 1073,
      "text_contrastive_loss": 0.8743,
      "train_positive_log_prob": -79.1338,
      "train_positive_token_accuracy": 0.0781,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3103,
      "epoch": 2.4243792325056432,
      "grad_norm": 12.166443824768066,
      "learning_rate": 5.423931929423235e-06,
      "lm_loss": 5.4745,
      "loss": 1.1926,
      "step": 1074,
      "text_contrastive_loss": 0.6697,
      "train_positive_log_prob": -80.5519,
      "train_positive_token_accuracy": 0.0791,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3886,
      "epoch": 2.4266365688487586,
      "grad_norm": 12.064868927001953,
      "learning_rate": 5.416702200985585e-06,
      "lm_loss": 5.5676,
      "loss": 1.3866,
      "step": 1075,
      "text_contrastive_loss": 0.8824,
      "train_positive_log_prob": -83.6239,
      "train_positive_token_accuracy": 0.0848,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.409,
      "epoch": 2.4288939051918734,
      "grad_norm": 12.926525115966797,
      "learning_rate": 5.409471595123628e-06,
      "lm_loss": 5.4047,
      "loss": 1.3996,
      "step": 1076,
      "text_contrastive_loss": 0.9003,
      "train_positive_log_prob": -78.577,
      "train_positive_token_accuracy": 0.0763,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.38,
      "epoch": 2.4311512415349887,
      "grad_norm": 13.372188568115234,
      "learning_rate": 5.4022401270624036e-06,
      "lm_loss": 5.494,
      "loss": 1.2331,
      "step": 1077,
      "text_contrastive_loss": 0.6076,
      "train_positive_log_prob": -82.2837,
      "train_positive_token_accuracy": 0.0866,
      "train_positive_token_prob": 0.033
    },
    {
      "contrastive_loss": 0.4095,
      "epoch": 2.4334085778781036,
      "grad_norm": 12.369166374206543,
      "learning_rate": 5.395007812028775e-06,
      "lm_loss": 5.5177,
      "loss": 1.3271,
      "step": 1078,
      "text_contrastive_loss": 0.7316,
      "train_positive_log_prob": -80.5846,
      "train_positive_token_accuracy": 0.0768,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.4923,
      "epoch": 2.435665914221219,
      "grad_norm": 14.51758098602295,
      "learning_rate": 5.387774665251385e-06,
      "lm_loss": 5.5253,
      "loss": 1.4506,
      "step": 1079,
      "text_contrastive_loss": 0.8114,
      "train_positive_log_prob": -83.6361,
      "train_positive_token_accuracy": 0.0795,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3501,
      "epoch": 2.4379232505643342,
      "grad_norm": 12.295190811157227,
      "learning_rate": 5.380540701960627e-06,
      "lm_loss": 5.4864,
      "loss": 1.2486,
      "step": 1080,
      "text_contrastive_loss": 0.6998,
      "train_positive_log_prob": -81.5226,
      "train_positive_token_accuracy": 0.0781,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.4023,
      "epoch": 2.440180586907449,
      "grad_norm": 13.490459442138672,
      "learning_rate": 5.373305937388613e-06,
      "lm_loss": 5.6089,
      "loss": 1.3018,
      "step": 1081,
      "text_contrastive_loss": 0.6773,
      "train_positive_log_prob": -84.2885,
      "train_positive_token_accuracy": 0.0798,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.4213,
      "epoch": 2.4424379232505644,
      "grad_norm": 14.755785942077637,
      "learning_rate": 5.3660703867691475e-06,
      "lm_loss": 5.4886,
      "loss": 1.4683,
      "step": 1082,
      "text_contrastive_loss": 0.9964,
      "train_positive_log_prob": -80.3554,
      "train_positive_token_accuracy": 0.0848,
      "train_positive_token_prob": 0.0325
    },
    {
      "contrastive_loss": 0.3914,
      "epoch": 2.4446952595936793,
      "grad_norm": 13.41222095489502,
      "learning_rate": 5.358834065337684e-06,
      "lm_loss": 5.6045,
      "loss": 1.4521,
      "step": 1083,
      "text_contrastive_loss": 1.0005,
      "train_positive_log_prob": -83.172,
      "train_positive_token_accuracy": 0.0837,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3292,
      "epoch": 2.4469525959367946,
      "grad_norm": 13.041488647460938,
      "learning_rate": 5.3515969883313e-06,
      "lm_loss": 5.5795,
      "loss": 1.293,
      "step": 1084,
      "text_contrastive_loss": 0.8117,
      "train_positive_log_prob": -83.1063,
      "train_positive_token_accuracy": 0.0779,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.3403,
      "epoch": 2.44920993227991,
      "grad_norm": 12.110668182373047,
      "learning_rate": 5.344359170988668e-06,
      "lm_loss": 5.5531,
      "loss": 1.2482,
      "step": 1085,
      "text_contrastive_loss": 0.705,
      "train_positive_log_prob": -81.3447,
      "train_positive_token_accuracy": 0.0792,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.446,
      "epoch": 2.4514672686230248,
      "grad_norm": 15.7059965133667,
      "learning_rate": 5.337120628550016e-06,
      "lm_loss": 5.4581,
      "loss": 1.4751,
      "step": 1086,
      "text_contrastive_loss": 0.9666,
      "train_positive_log_prob": -82.3006,
      "train_positive_token_accuracy": 0.0829,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.4065,
      "epoch": 2.45372460496614,
      "grad_norm": 13.175836563110352,
      "learning_rate": 5.329881376257098e-06,
      "lm_loss": 5.4921,
      "loss": 1.3573,
      "step": 1087,
      "text_contrastive_loss": 0.8032,
      "train_positive_log_prob": -81.1144,
      "train_positive_token_accuracy": 0.0828,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.4353,
      "epoch": 2.455981941309255,
      "grad_norm": 14.727249145507812,
      "learning_rate": 5.322641429353167e-06,
      "lm_loss": 5.4534,
      "loss": 1.4546,
      "step": 1088,
      "text_contrastive_loss": 0.948,
      "train_positive_log_prob": -82.1823,
      "train_positive_token_accuracy": 0.084,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.4149,
      "epoch": 2.4582392776523703,
      "grad_norm": 13.355571746826172,
      "learning_rate": 5.315400803082934e-06,
      "lm_loss": 5.4401,
      "loss": 1.346,
      "step": 1089,
      "text_contrastive_loss": 0.7743,
      "train_positive_log_prob": -80.2823,
      "train_positive_token_accuracy": 0.0765,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.4249,
      "epoch": 2.460496613995485,
      "grad_norm": 15.00484561920166,
      "learning_rate": 5.308159512692544e-06,
      "lm_loss": 5.5169,
      "loss": 1.4481,
      "step": 1090,
      "text_contrastive_loss": 0.943,
      "train_positive_log_prob": -81.1595,
      "train_positive_token_accuracy": 0.0754,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.4935,
      "epoch": 2.4627539503386005,
      "grad_norm": 14.640666007995605,
      "learning_rate": 5.300917573429536e-06,
      "lm_loss": 5.5135,
      "loss": 1.561,
      "step": 1091,
      "text_contrastive_loss": 1.0322,
      "train_positive_log_prob": -78.5885,
      "train_positive_token_accuracy": 0.0776,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.4456,
      "epoch": 2.4650112866817158,
      "grad_norm": 15.940723419189453,
      "learning_rate": 5.293675000542822e-06,
      "lm_loss": 5.6074,
      "loss": 1.4078,
      "step": 1092,
      "text_contrastive_loss": 0.8029,
      "train_positive_log_prob": -83.9385,
      "train_positive_token_accuracy": 0.0804,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.5212,
      "epoch": 2.4672686230248306,
      "grad_norm": 15.651947975158691,
      "learning_rate": 5.286431809282639e-06,
      "lm_loss": 5.4077,
      "loss": 1.5292,
      "step": 1093,
      "text_contrastive_loss": 0.9345,
      "train_positive_log_prob": -80.0202,
      "train_positive_token_accuracy": 0.0826,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.4062,
      "epoch": 2.469525959367946,
      "grad_norm": 13.576653480529785,
      "learning_rate": 5.279188014900537e-06,
      "lm_loss": 5.5198,
      "loss": 1.3024,
      "step": 1094,
      "text_contrastive_loss": 0.6884,
      "train_positive_log_prob": -80.0606,
      "train_positive_token_accuracy": 0.0751,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.4295,
      "epoch": 2.471783295711061,
      "grad_norm": 14.4829683303833,
      "learning_rate": 5.2719436326493255e-06,
      "lm_loss": 5.5134,
      "loss": 1.3906,
      "step": 1095,
      "text_contrastive_loss": 0.8195,
      "train_positive_log_prob": -82.8098,
      "train_positive_token_accuracy": 0.0759,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.4137,
      "epoch": 2.474040632054176,
      "grad_norm": 13.427950859069824,
      "learning_rate": 5.26469867778306e-06,
      "lm_loss": 5.5291,
      "loss": 1.3379,
      "step": 1096,
      "text_contrastive_loss": 0.7426,
      "train_positive_log_prob": -81.4216,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.3779,
      "epoch": 2.476297968397291,
      "grad_norm": 11.763487815856934,
      "learning_rate": 5.257453165556996e-06,
      "lm_loss": 5.4668,
      "loss": 1.2999,
      "step": 1097,
      "text_contrastive_loss": 0.7506,
      "train_positive_log_prob": -81.7218,
      "train_positive_token_accuracy": 0.0886,
      "train_positive_token_prob": 0.0327
    },
    {
      "contrastive_loss": 0.3719,
      "epoch": 2.4785553047404063,
      "grad_norm": 12.892736434936523,
      "learning_rate": 5.2502071112275675e-06,
      "lm_loss": 5.5286,
      "loss": 1.3189,
      "step": 1098,
      "text_contrastive_loss": 0.7883,
      "train_positive_log_prob": -81.9765,
      "train_positive_token_accuracy": 0.0769,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3745,
      "epoch": 2.4808126410835216,
      "grad_norm": 13.535112380981445,
      "learning_rate": 5.242960530052344e-06,
      "lm_loss": 5.4514,
      "loss": 1.3905,
      "step": 1099,
      "text_contrastive_loss": 0.9418,
      "train_positive_log_prob": -79.1948,
      "train_positive_token_accuracy": 0.0828,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3667,
      "epoch": 2.4830699774266365,
      "grad_norm": 12.673724174499512,
      "learning_rate": 5.235713437290012e-06,
      "lm_loss": 5.4863,
      "loss": 1.4022,
      "step": 1100,
      "text_contrastive_loss": 0.9738,
      "train_positive_log_prob": -81.9747,
      "train_positive_token_accuracy": 0.0774,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.3996,
      "epoch": 2.485327313769752,
      "grad_norm": 13.809643745422363,
      "learning_rate": 5.228465848200327e-06,
      "lm_loss": 5.6012,
      "loss": 1.3115,
      "step": 1101,
      "text_contrastive_loss": 0.7034,
      "train_positive_log_prob": -79.9467,
      "train_positive_token_accuracy": 0.0761,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.507,
      "epoch": 2.4875846501128667,
      "grad_norm": 15.798452377319336,
      "learning_rate": 5.221217778044096e-06,
      "lm_loss": 5.4808,
      "loss": 1.4458,
      "step": 1102,
      "text_contrastive_loss": 0.7814,
      "train_positive_log_prob": -80.5246,
      "train_positive_token_accuracy": 0.0763,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3997,
      "epoch": 2.489841986455982,
      "grad_norm": 14.478659629821777,
      "learning_rate": 5.2139692420831325e-06,
      "lm_loss": 5.5925,
      "loss": 1.3412,
      "step": 1103,
      "text_contrastive_loss": 0.7645,
      "train_positive_log_prob": -83.6685,
      "train_positive_token_accuracy": 0.0854,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.4095,
      "epoch": 2.4920993227990973,
      "grad_norm": 12.84615421295166,
      "learning_rate": 5.206720255580241e-06,
      "lm_loss": 5.4692,
      "loss": 1.3362,
      "step": 1104,
      "text_contrastive_loss": 0.7595,
      "train_positive_log_prob": -80.3504,
      "train_positive_token_accuracy": 0.08,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.4717,
      "epoch": 2.494356659142212,
      "grad_norm": 16.172317504882812,
      "learning_rate": 5.199470833799164e-06,
      "lm_loss": 5.573,
      "loss": 1.4538,
      "step": 1105,
      "text_contrastive_loss": 0.8496,
      "train_positive_log_prob": -84.2076,
      "train_positive_token_accuracy": 0.082,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.3017,
      "epoch": 2.4966139954853275,
      "grad_norm": 11.299418449401855,
      "learning_rate": 5.192220992004569e-06,
      "lm_loss": 5.6036,
      "loss": 1.2086,
      "step": 1106,
      "text_contrastive_loss": 0.6931,
      "train_positive_log_prob": -83.639,
      "train_positive_token_accuracy": 0.0803,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.5347,
      "epoch": 2.4988713318284423,
      "grad_norm": 16.39532470703125,
      "learning_rate": 5.184970745461998e-06,
      "lm_loss": 5.6291,
      "loss": 1.507,
      "step": 1107,
      "text_contrastive_loss": 0.8189,
      "train_positive_log_prob": -84.0616,
      "train_positive_token_accuracy": 0.0769,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.3921,
      "epoch": 2.5011286681715577,
      "grad_norm": 12.706552505493164,
      "learning_rate": 5.177720109437857e-06,
      "lm_loss": 5.5586,
      "loss": 1.327,
      "step": 1108,
      "text_contrastive_loss": 0.758,
      "train_positive_log_prob": -83.0968,
      "train_positive_token_accuracy": 0.0784,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.3224,
      "epoch": 2.5033860045146725,
      "grad_norm": 12.656816482543945,
      "learning_rate": 5.170469099199363e-06,
      "lm_loss": 5.4715,
      "loss": 1.2552,
      "step": 1109,
      "text_contrastive_loss": 0.7714,
      "train_positive_log_prob": -78.5682,
      "train_positive_token_accuracy": 0.0881,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.3911,
      "epoch": 2.505643340857788,
      "grad_norm": 14.10784912109375,
      "learning_rate": 5.1632177300145255e-06,
      "lm_loss": 5.4638,
      "loss": 1.3482,
      "step": 1110,
      "text_contrastive_loss": 0.8215,
      "train_positive_log_prob": -81.2746,
      "train_positive_token_accuracy": 0.0827,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.5143,
      "epoch": 2.5079006772009027,
      "grad_norm": 14.490861892700195,
      "learning_rate": 5.155966017152108e-06,
      "lm_loss": 5.4204,
      "loss": 1.4946,
      "step": 1111,
      "text_contrastive_loss": 0.8765,
      "train_positive_log_prob": -81.7895,
      "train_positive_token_accuracy": 0.0834,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.4723,
      "epoch": 2.510158013544018,
      "grad_norm": 15.501252174377441,
      "learning_rate": 5.148713975881598e-06,
      "lm_loss": 5.4654,
      "loss": 1.4282,
      "step": 1112,
      "text_contrastive_loss": 0.8187,
      "train_positive_log_prob": -80.7631,
      "train_positive_token_accuracy": 0.0735,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.406,
      "epoch": 2.5124153498871333,
      "grad_norm": 13.753080368041992,
      "learning_rate": 5.141461621473175e-06,
      "lm_loss": 5.6012,
      "loss": 1.3412,
      "step": 1113,
      "text_contrastive_loss": 0.7502,
      "train_positive_log_prob": -83.4042,
      "train_positive_token_accuracy": 0.0792,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.4504,
      "epoch": 2.514672686230248,
      "grad_norm": 15.42505168914795,
      "learning_rate": 5.1342089691976794e-06,
      "lm_loss": 5.5882,
      "loss": 1.45,
      "step": 1114,
      "text_contrastive_loss": 0.8816,
      "train_positive_log_prob": -81.5714,
      "train_positive_token_accuracy": 0.0789,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3244,
      "epoch": 2.5169300225733635,
      "grad_norm": 11.779836654663086,
      "learning_rate": 5.126956034326573e-06,
      "lm_loss": 5.4373,
      "loss": 1.2967,
      "step": 1115,
      "text_contrastive_loss": 0.8572,
      "train_positive_log_prob": -79.8218,
      "train_positive_token_accuracy": 0.0838,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.3145,
      "epoch": 2.5191873589164784,
      "grad_norm": 12.900495529174805,
      "learning_rate": 5.119702832131922e-06,
      "lm_loss": 5.5437,
      "loss": 1.3138,
      "step": 1116,
      "text_contrastive_loss": 0.8898,
      "train_positive_log_prob": -82.4608,
      "train_positive_token_accuracy": 0.079,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3986,
      "epoch": 2.5214446952595937,
      "grad_norm": 12.25336742401123,
      "learning_rate": 5.112449377886345e-06,
      "lm_loss": 5.4775,
      "loss": 1.372,
      "step": 1117,
      "text_contrastive_loss": 0.8514,
      "train_positive_log_prob": -81.8183,
      "train_positive_token_accuracy": 0.0813,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3882,
      "epoch": 2.523702031602709,
      "grad_norm": 13.7731294631958,
      "learning_rate": 5.105195686863e-06,
      "lm_loss": 5.6046,
      "loss": 1.3206,
      "step": 1118,
      "text_contrastive_loss": 0.744,
      "train_positive_log_prob": -82.9545,
      "train_positive_token_accuracy": 0.0768,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3847,
      "epoch": 2.525959367945824,
      "grad_norm": 13.900425910949707,
      "learning_rate": 5.097941774335538e-06,
      "lm_loss": 5.4642,
      "loss": 1.2978,
      "step": 1119,
      "text_contrastive_loss": 0.7334,
      "train_positive_log_prob": -80.8777,
      "train_positive_token_accuracy": 0.08,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.452,
      "epoch": 2.528216704288939,
      "grad_norm": 13.458977699279785,
      "learning_rate": 5.090687655578078e-06,
      "lm_loss": 5.5858,
      "loss": 1.4618,
      "step": 1120,
      "text_contrastive_loss": 0.9024,
      "train_positive_log_prob": -84.0085,
      "train_positive_token_accuracy": 0.0813,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.4537,
      "epoch": 2.530474040632054,
      "grad_norm": 14.56409740447998,
      "learning_rate": 5.083433345865175e-06,
      "lm_loss": 5.4753,
      "loss": 1.4121,
      "step": 1121,
      "text_contrastive_loss": 0.8219,
      "train_positive_log_prob": -79.843,
      "train_positive_token_accuracy": 0.0879,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.4135,
      "epoch": 2.5327313769751694,
      "grad_norm": 12.935258865356445,
      "learning_rate": 5.076178860471787e-06,
      "lm_loss": 5.491,
      "loss": 1.3246,
      "step": 1122,
      "text_contrastive_loss": 0.724,
      "train_positive_log_prob": -81.0987,
      "train_positive_token_accuracy": 0.0838,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.4647,
      "epoch": 2.5349887133182847,
      "grad_norm": 13.051006317138672,
      "learning_rate": 5.068924214673234e-06,
      "lm_loss": 5.4212,
      "loss": 1.4127,
      "step": 1123,
      "text_contrastive_loss": 0.8117,
      "train_positive_log_prob": -78.798,
      "train_positive_token_accuracy": 0.0806,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.4065,
      "epoch": 2.5372460496613995,
      "grad_norm": 13.802006721496582,
      "learning_rate": 5.061669423745185e-06,
      "lm_loss": 5.4055,
      "loss": 1.3731,
      "step": 1124,
      "text_contrastive_loss": 0.8523,
      "train_positive_log_prob": -79.1395,
      "train_positive_token_accuracy": 0.0934,
      "train_positive_token_prob": 0.0328
    },
    {
      "contrastive_loss": 0.3912,
      "epoch": 2.5395033860045144,
      "grad_norm": 17.284080505371094,
      "learning_rate": 5.054414502963605e-06,
      "lm_loss": 5.5214,
      "loss": 1.3468,
      "step": 1125,
      "text_contrastive_loss": 0.8068,
      "train_positive_log_prob": -82.5795,
      "train_positive_token_accuracy": 0.0838,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3803,
      "epoch": 2.5417607223476297,
      "grad_norm": 12.933502197265625,
      "learning_rate": 5.0471594676047385e-06,
      "lm_loss": 5.5175,
      "loss": 1.3346,
      "step": 1126,
      "text_contrastive_loss": 0.805,
      "train_positive_log_prob": -81.4263,
      "train_positive_token_accuracy": 0.0799,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.3538,
      "epoch": 2.544018058690745,
      "grad_norm": 13.167275428771973,
      "learning_rate": 5.039904332945069e-06,
      "lm_loss": 5.4207,
      "loss": 1.2401,
      "step": 1127,
      "text_contrastive_loss": 0.6885,
      "train_positive_log_prob": -78.5817,
      "train_positive_token_accuracy": 0.0819,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.4452,
      "epoch": 2.54627539503386,
      "grad_norm": 13.284455299377441,
      "learning_rate": 5.03264911426129e-06,
      "lm_loss": 5.3674,
      "loss": 1.3615,
      "step": 1128,
      "text_contrastive_loss": 0.7592,
      "train_positive_log_prob": -78.8357,
      "train_positive_token_accuracy": 0.0819,
      "train_positive_token_prob": 0.0328
    },
    {
      "contrastive_loss": 0.4752,
      "epoch": 2.5485327313769752,
      "grad_norm": 13.207563400268555,
      "learning_rate": 5.025393826830267e-06,
      "lm_loss": 5.5985,
      "loss": 1.4632,
      "step": 1129,
      "text_contrastive_loss": 0.8563,
      "train_positive_log_prob": -82.5769,
      "train_positive_token_accuracy": 0.0726,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.4554,
      "epoch": 2.55079006772009,
      "grad_norm": 12.570433616638184,
      "learning_rate": 5.0181384859290215e-06,
      "lm_loss": 5.5167,
      "loss": 1.406,
      "step": 1130,
      "text_contrastive_loss": 0.7979,
      "train_positive_log_prob": -81.6048,
      "train_positive_token_accuracy": 0.0842,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.4251,
      "epoch": 2.5530474040632054,
      "grad_norm": 14.778961181640625,
      "learning_rate": 5.010883106834676e-06,
      "lm_loss": 5.431,
      "loss": 1.4212,
      "step": 1131,
      "text_contrastive_loss": 0.906,
      "train_positive_log_prob": -81.4475,
      "train_positive_token_accuracy": 0.0792,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.4831,
      "epoch": 2.5553047404063207,
      "grad_norm": 16.45122528076172,
      "learning_rate": 5.003627704824438e-06,
      "lm_loss": 5.486,
      "loss": 1.4543,
      "step": 1132,
      "text_contrastive_loss": 0.8453,
      "train_positive_log_prob": -79.9769,
      "train_positive_token_accuracy": 0.0879,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.4322,
      "epoch": 2.5575620767494356,
      "grad_norm": 13.629638671875,
      "learning_rate": 4.996372295175563e-06,
      "lm_loss": 5.5377,
      "loss": 1.3377,
      "step": 1133,
      "text_contrastive_loss": 0.7035,
      "train_positive_log_prob": -82.1578,
      "train_positive_token_accuracy": 0.0753,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.3606,
      "epoch": 2.559819413092551,
      "grad_norm": 12.509625434875488,
      "learning_rate": 4.989116893165325e-06,
      "lm_loss": 5.4185,
      "loss": 1.2522,
      "step": 1134,
      "text_contrastive_loss": 0.6995,
      "train_positive_log_prob": -80.2624,
      "train_positive_token_accuracy": 0.0789,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.5114,
      "epoch": 2.5620767494356658,
      "grad_norm": 14.97032642364502,
      "learning_rate": 4.981861514070979e-06,
      "lm_loss": 5.4655,
      "loss": 1.5477,
      "step": 1135,
      "text_contrastive_loss": 0.9796,
      "train_positive_log_prob": -80.784,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.4696,
      "epoch": 2.564334085778781,
      "grad_norm": 14.087468147277832,
      "learning_rate": 4.974606173169733e-06,
      "lm_loss": 5.4949,
      "loss": 1.4966,
      "step": 1136,
      "text_contrastive_loss": 0.9551,
      "train_positive_log_prob": -81.6525,
      "train_positive_token_accuracy": 0.0846,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.4408,
      "epoch": 2.5665914221218964,
      "grad_norm": 14.633216857910156,
      "learning_rate": 4.9673508857387115e-06,
      "lm_loss": 5.4085,
      "loss": 1.367,
      "step": 1137,
      "text_contrastive_loss": 0.7706,
      "train_positive_log_prob": -80.0688,
      "train_positive_token_accuracy": 0.0813,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.4022,
      "epoch": 2.5688487584650113,
      "grad_norm": 13.321537017822266,
      "learning_rate": 4.9600956670549324e-06,
      "lm_loss": 5.5587,
      "loss": 1.357,
      "step": 1138,
      "text_contrastive_loss": 0.7979,
      "train_positive_log_prob": -83.3167,
      "train_positive_token_accuracy": 0.0792,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.4094,
      "epoch": 2.5711060948081266,
      "grad_norm": 12.423505783081055,
      "learning_rate": 4.952840532395262e-06,
      "lm_loss": 5.5307,
      "loss": 1.3634,
      "step": 1139,
      "text_contrastive_loss": 0.802,
      "train_positive_log_prob": -84.0355,
      "train_positive_token_accuracy": 0.0825,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.4286,
      "epoch": 2.5733634311512414,
      "grad_norm": 11.84165096282959,
      "learning_rate": 4.945585497036396e-06,
      "lm_loss": 5.498,
      "loss": 1.3595,
      "step": 1140,
      "text_contrastive_loss": 0.7622,
      "train_positive_log_prob": -79.1176,
      "train_positive_token_accuracy": 0.0784,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.4206,
      "epoch": 2.5756207674943568,
      "grad_norm": 15.220135688781738,
      "learning_rate": 4.938330576254817e-06,
      "lm_loss": 5.4074,
      "loss": 1.3147,
      "step": 1141,
      "text_contrastive_loss": 0.7069,
      "train_positive_log_prob": -81.8945,
      "train_positive_token_accuracy": 0.0838,
      "train_positive_token_prob": 0.0325
    },
    {
      "contrastive_loss": 0.4548,
      "epoch": 2.5778781038374716,
      "grad_norm": 14.214799880981445,
      "learning_rate": 4.931075785326767e-06,
      "lm_loss": 5.4122,
      "loss": 1.397,
      "step": 1142,
      "text_contrastive_loss": 0.8019,
      "train_positive_log_prob": -81.3377,
      "train_positive_token_accuracy": 0.0806,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.4385,
      "epoch": 2.580135440180587,
      "grad_norm": 14.037745475769043,
      "learning_rate": 4.9238211395282156e-06,
      "lm_loss": 5.3858,
      "loss": 1.3714,
      "step": 1143,
      "text_contrastive_loss": 0.7886,
      "train_positive_log_prob": -77.5022,
      "train_positive_token_accuracy": 0.0806,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3381,
      "epoch": 2.582392776523702,
      "grad_norm": 12.56302547454834,
      "learning_rate": 4.9165666541348265e-06,
      "lm_loss": 5.5134,
      "loss": 1.2769,
      "step": 1144,
      "text_contrastive_loss": 0.775,
      "train_positive_log_prob": -82.2902,
      "train_positive_token_accuracy": 0.0762,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.5162,
      "epoch": 2.584650112866817,
      "grad_norm": 14.678620338439941,
      "learning_rate": 4.909312344421923e-06,
      "lm_loss": 5.5333,
      "loss": 1.5027,
      "step": 1145,
      "text_contrastive_loss": 0.8663,
      "train_positive_log_prob": -82.6164,
      "train_positive_token_accuracy": 0.0798,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.4099,
      "epoch": 2.5869074492099324,
      "grad_norm": 12.928022384643555,
      "learning_rate": 4.902058225664465e-06,
      "lm_loss": 5.4,
      "loss": 1.3616,
      "step": 1146,
      "text_contrastive_loss": 0.8232,
      "train_positive_log_prob": -77.6009,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.384,
      "epoch": 2.5891647855530473,
      "grad_norm": 15.01064682006836,
      "learning_rate": 4.8948043131370025e-06,
      "lm_loss": 5.579,
      "loss": 1.3733,
      "step": 1147,
      "text_contrastive_loss": 0.8629,
      "train_positive_log_prob": -81.9003,
      "train_positive_token_accuracy": 0.0718,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.3578,
      "epoch": 2.5914221218961626,
      "grad_norm": 12.526495933532715,
      "learning_rate": 4.887550622113657e-06,
      "lm_loss": 5.4814,
      "loss": 1.2652,
      "step": 1148,
      "text_contrastive_loss": 0.7185,
      "train_positive_log_prob": -79.1582,
      "train_positive_token_accuracy": 0.0855,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3802,
      "epoch": 2.5936794582392775,
      "grad_norm": 13.175474166870117,
      "learning_rate": 4.88029716786808e-06,
      "lm_loss": 5.5614,
      "loss": 1.306,
      "step": 1149,
      "text_contrastive_loss": 0.7393,
      "train_positive_log_prob": -83.9876,
      "train_positive_token_accuracy": 0.0846,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.4673,
      "epoch": 2.595936794582393,
      "grad_norm": 14.35908317565918,
      "learning_rate": 4.873043965673427e-06,
      "lm_loss": 5.4945,
      "loss": 1.4968,
      "step": 1150,
      "text_contrastive_loss": 0.9602,
      "train_positive_log_prob": -82.3151,
      "train_positive_token_accuracy": 0.0834,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.5568,
      "epoch": 2.598194130925508,
      "grad_norm": 16.738842010498047,
      "learning_rate": 4.8657910308023205e-06,
      "lm_loss": 5.543,
      "loss": 1.5993,
      "step": 1151,
      "text_contrastive_loss": 0.9765,
      "train_positive_log_prob": -81.545,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.2944,
      "epoch": 2.600451467268623,
      "grad_norm": 11.679557800292969,
      "learning_rate": 4.858538378526825e-06,
      "lm_loss": 5.3878,
      "loss": 1.1371,
      "step": 1152,
      "text_contrastive_loss": 0.608,
      "train_positive_log_prob": -78.2674,
      "train_positive_token_accuracy": 0.0828,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.4211,
      "epoch": 2.6027088036117383,
      "grad_norm": 13.320328712463379,
      "learning_rate": 4.851286024118402e-06,
      "lm_loss": 5.4369,
      "loss": 1.3524,
      "step": 1153,
      "text_contrastive_loss": 0.7754,
      "train_positive_log_prob": -79.2371,
      "train_positive_token_accuracy": 0.0799,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.4434,
      "epoch": 2.604966139954853,
      "grad_norm": 13.534639358520508,
      "learning_rate": 4.844033982847893e-06,
      "lm_loss": 5.5048,
      "loss": 1.4477,
      "step": 1154,
      "text_contrastive_loss": 0.9077,
      "train_positive_log_prob": -79.8649,
      "train_positive_token_accuracy": 0.0872,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3938,
      "epoch": 2.6072234762979685,
      "grad_norm": 13.858593940734863,
      "learning_rate": 4.836782269985475e-06,
      "lm_loss": 5.4346,
      "loss": 1.3776,
      "step": 1155,
      "text_contrastive_loss": 0.8806,
      "train_positive_log_prob": -77.2752,
      "train_positive_token_accuracy": 0.0835,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3625,
      "epoch": 2.609480812641084,
      "grad_norm": 13.214495658874512,
      "learning_rate": 4.829530900800638e-06,
      "lm_loss": 5.5532,
      "loss": 1.3706,
      "step": 1156,
      "text_contrastive_loss": 0.9055,
      "train_positive_log_prob": -81.7372,
      "train_positive_token_accuracy": 0.0791,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.413,
      "epoch": 2.6117381489841986,
      "grad_norm": 12.766947746276855,
      "learning_rate": 4.8222798905621445e-06,
      "lm_loss": 5.5154,
      "loss": 1.3809,
      "step": 1157,
      "text_contrastive_loss": 0.8327,
      "train_positive_log_prob": -80.6022,
      "train_positive_token_accuracy": 0.0909,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3759,
      "epoch": 2.6139954853273135,
      "grad_norm": 12.303617477416992,
      "learning_rate": 4.815029254538003e-06,
      "lm_loss": 5.4039,
      "loss": 1.2761,
      "step": 1158,
      "text_contrastive_loss": 0.7196,
      "train_positive_log_prob": -79.8339,
      "train_positive_token_accuracy": 0.0806,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.5641,
      "epoch": 2.616252821670429,
      "grad_norm": 19.04671287536621,
      "learning_rate": 4.807779007995434e-06,
      "lm_loss": 5.4291,
      "loss": 1.5322,
      "step": 1159,
      "text_contrastive_loss": 0.8503,
      "train_positive_log_prob": -78.6826,
      "train_positive_token_accuracy": 0.0771,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.281,
      "epoch": 2.618510158013544,
      "grad_norm": 12.744864463806152,
      "learning_rate": 4.800529166200837e-06,
      "lm_loss": 5.4667,
      "loss": 1.2207,
      "step": 1160,
      "text_contrastive_loss": 0.786,
      "train_positive_log_prob": -80.6122,
      "train_positive_token_accuracy": 0.0885,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.3549,
      "epoch": 2.620767494356659,
      "grad_norm": 12.05768871307373,
      "learning_rate": 4.7932797444197604e-06,
      "lm_loss": 5.5462,
      "loss": 1.3312,
      "step": 1161,
      "text_contrastive_loss": 0.8434,
      "train_positive_log_prob": -86.4028,
      "train_positive_token_accuracy": 0.0762,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.4921,
      "epoch": 2.6230248306997743,
      "grad_norm": 14.103909492492676,
      "learning_rate": 4.786030757916868e-06,
      "lm_loss": 5.6144,
      "loss": 1.5285,
      "step": 1162,
      "text_contrastive_loss": 0.95,
      "train_positive_log_prob": -83.7298,
      "train_positive_token_accuracy": 0.0833,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.4334,
      "epoch": 2.625282167042889,
      "grad_norm": 14.57669448852539,
      "learning_rate": 4.778782221955907e-06,
      "lm_loss": 5.4605,
      "loss": 1.3621,
      "step": 1163,
      "text_contrastive_loss": 0.7652,
      "train_positive_log_prob": -79.8643,
      "train_positive_token_accuracy": 0.0871,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.3722,
      "epoch": 2.6275395033860045,
      "grad_norm": 12.507501602172852,
      "learning_rate": 4.771534151799676e-06,
      "lm_loss": 5.4844,
      "loss": 1.3388,
      "step": 1164,
      "text_contrastive_loss": 0.8363,
      "train_positive_log_prob": -82.9281,
      "train_positive_token_accuracy": 0.0814,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.4389,
      "epoch": 2.62979683972912,
      "grad_norm": 12.930022239685059,
      "learning_rate": 4.76428656270999e-06,
      "lm_loss": 5.6183,
      "loss": 1.3833,
      "step": 1165,
      "text_contrastive_loss": 0.7651,
      "train_positive_log_prob": -83.1572,
      "train_positive_token_accuracy": 0.0876,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3787,
      "epoch": 2.6320541760722347,
      "grad_norm": 13.370924949645996,
      "learning_rate": 4.757039469947658e-06,
      "lm_loss": 5.4669,
      "loss": 1.3421,
      "step": 1166,
      "text_contrastive_loss": 0.8335,
      "train_positive_log_prob": -82.8615,
      "train_positive_token_accuracy": 0.0822,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3548,
      "epoch": 2.63431151241535,
      "grad_norm": 12.020811080932617,
      "learning_rate": 4.7497928887724325e-06,
      "lm_loss": 5.5947,
      "loss": 1.3023,
      "step": 1167,
      "text_contrastive_loss": 0.7759,
      "train_positive_log_prob": -82.8732,
      "train_positive_token_accuracy": 0.0744,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3843,
      "epoch": 2.636568848758465,
      "grad_norm": 13.562383651733398,
      "learning_rate": 4.7425468344430035e-06,
      "lm_loss": 5.414,
      "loss": 1.308,
      "step": 1168,
      "text_contrastive_loss": 0.7646,
      "train_positive_log_prob": -78.5706,
      "train_positive_token_accuracy": 0.0869,
      "train_positive_token_prob": 0.0328
    },
    {
      "contrastive_loss": 0.3854,
      "epoch": 2.63882618510158,
      "grad_norm": 13.353324890136719,
      "learning_rate": 4.73530132221694e-06,
      "lm_loss": 5.4934,
      "loss": 1.2806,
      "step": 1169,
      "text_contrastive_loss": 0.6917,
      "train_positive_log_prob": -82.8743,
      "train_positive_token_accuracy": 0.073,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.4047,
      "epoch": 2.6410835214446955,
      "grad_norm": 12.325431823730469,
      "learning_rate": 4.7280563673506745e-06,
      "lm_loss": 5.5337,
      "loss": 1.3067,
      "step": 1170,
      "text_contrastive_loss": 0.6973,
      "train_positive_log_prob": -83.8513,
      "train_positive_token_accuracy": 0.0835,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3929,
      "epoch": 2.6433408577878104,
      "grad_norm": 13.446803092956543,
      "learning_rate": 4.720811985099464e-06,
      "lm_loss": 5.5289,
      "loss": 1.3735,
      "step": 1171,
      "text_contrastive_loss": 0.8554,
      "train_positive_log_prob": -81.7638,
      "train_positive_token_accuracy": 0.0763,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.4169,
      "epoch": 2.6455981941309257,
      "grad_norm": 14.658951759338379,
      "learning_rate": 4.713568190717362e-06,
      "lm_loss": 5.5548,
      "loss": 1.3923,
      "step": 1172,
      "text_contrastive_loss": 0.8398,
      "train_positive_log_prob": -82.1077,
      "train_positive_token_accuracy": 0.074,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3331,
      "epoch": 2.6478555304740405,
      "grad_norm": 12.803595542907715,
      "learning_rate": 4.70632499945718e-06,
      "lm_loss": 5.4746,
      "loss": 1.2921,
      "step": 1173,
      "text_contrastive_loss": 0.823,
      "train_positive_log_prob": -81.1941,
      "train_positive_token_accuracy": 0.0822,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.3947,
      "epoch": 2.650112866817156,
      "grad_norm": 12.50619888305664,
      "learning_rate": 4.699082426570465e-06,
      "lm_loss": 5.4903,
      "loss": 1.4064,
      "step": 1174,
      "text_contrastive_loss": 0.9254,
      "train_positive_log_prob": -80.32,
      "train_positive_token_accuracy": 0.0764,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.44,
      "epoch": 2.6523702031602707,
      "grad_norm": 14.369673728942871,
      "learning_rate": 4.6918404873074574e-06,
      "lm_loss": 5.4898,
      "loss": 1.4245,
      "step": 1175,
      "text_contrastive_loss": 0.8712,
      "train_positive_log_prob": -82.52,
      "train_positive_token_accuracy": 0.0777,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3754,
      "epoch": 2.654627539503386,
      "grad_norm": 13.80208683013916,
      "learning_rate": 4.684599196917067e-06,
      "lm_loss": 5.5597,
      "loss": 1.2702,
      "step": 1176,
      "text_contrastive_loss": 0.6776,
      "train_positive_log_prob": -82.3752,
      "train_positive_token_accuracy": 0.0767,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.3706,
      "epoch": 2.656884875846501,
      "grad_norm": 12.911955833435059,
      "learning_rate": 4.677358570646834e-06,
      "lm_loss": 5.5187,
      "loss": 1.3116,
      "step": 1177,
      "text_contrastive_loss": 0.7783,
      "train_positive_log_prob": -81.4929,
      "train_positive_token_accuracy": 0.0772,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.4332,
      "epoch": 2.659142212189616,
      "grad_norm": 15.9859619140625,
      "learning_rate": 4.670118623742904e-06,
      "lm_loss": 5.5002,
      "loss": 1.3827,
      "step": 1178,
      "text_contrastive_loss": 0.7989,
      "train_positive_log_prob": -81.5136,
      "train_positive_token_accuracy": 0.0751,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.4035,
      "epoch": 2.6613995485327315,
      "grad_norm": 13.92669677734375,
      "learning_rate": 4.662879371449987e-06,
      "lm_loss": 5.4052,
      "loss": 1.3365,
      "step": 1179,
      "text_contrastive_loss": 0.785,
      "train_positive_log_prob": -80.3391,
      "train_positive_token_accuracy": 0.0894,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3522,
      "epoch": 2.6636568848758464,
      "grad_norm": 12.085784912109375,
      "learning_rate": 4.655640829011335e-06,
      "lm_loss": 5.4177,
      "loss": 1.2824,
      "step": 1180,
      "text_contrastive_loss": 0.7769,
      "train_positive_log_prob": -79.3168,
      "train_positive_token_accuracy": 0.0827,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3313,
      "epoch": 2.6659142212189617,
      "grad_norm": 14.719451904296875,
      "learning_rate": 4.6484030116687014e-06,
      "lm_loss": 5.4578,
      "loss": 1.2345,
      "step": 1181,
      "text_contrastive_loss": 0.7148,
      "train_positive_log_prob": -82.1677,
      "train_positive_token_accuracy": 0.0885,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.412,
      "epoch": 2.6681715575620766,
      "grad_norm": 13.43020248413086,
      "learning_rate": 4.64116593466232e-06,
      "lm_loss": 5.4645,
      "loss": 1.323,
      "step": 1182,
      "text_contrastive_loss": 0.7291,
      "train_positive_log_prob": -80.4075,
      "train_positive_token_accuracy": 0.0828,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.2984,
      "epoch": 2.670428893905192,
      "grad_norm": 10.837862968444824,
      "learning_rate": 4.633929613230855e-06,
      "lm_loss": 5.4854,
      "loss": 1.2469,
      "step": 1183,
      "text_contrastive_loss": 0.7997,
      "train_positive_log_prob": -81.377,
      "train_positive_token_accuracy": 0.0835,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3425,
      "epoch": 2.672686230248307,
      "grad_norm": 13.186177253723145,
      "learning_rate": 4.626694062611387e-06,
      "lm_loss": 5.4388,
      "loss": 1.2543,
      "step": 1184,
      "text_contrastive_loss": 0.7358,
      "train_positive_log_prob": -82.0034,
      "train_positive_token_accuracy": 0.0785,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3589,
      "epoch": 2.674943566591422,
      "grad_norm": 14.815667152404785,
      "learning_rate": 4.619459298039373e-06,
      "lm_loss": 5.4746,
      "loss": 1.2642,
      "step": 1185,
      "text_contrastive_loss": 0.7156,
      "train_positive_log_prob": -81.8436,
      "train_positive_token_accuracy": 0.0768,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.3478,
      "epoch": 2.6772009029345374,
      "grad_norm": 12.600274085998535,
      "learning_rate": 4.612225334748616e-06,
      "lm_loss": 5.4998,
      "loss": 1.221,
      "step": 1186,
      "text_contrastive_loss": 0.6465,
      "train_positive_log_prob": -84.3436,
      "train_positive_token_accuracy": 0.0759,
      "train_positive_token_prob": 0.0289
    },
    {
      "contrastive_loss": 0.3724,
      "epoch": 2.6794582392776523,
      "grad_norm": 15.114175796508789,
      "learning_rate": 4.6049921879712254e-06,
      "lm_loss": 5.4519,
      "loss": 1.2851,
      "step": 1187,
      "text_contrastive_loss": 0.735,
      "train_positive_log_prob": -81.2476,
      "train_positive_token_accuracy": 0.0747,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.4161,
      "epoch": 2.6817155756207676,
      "grad_norm": 13.305334091186523,
      "learning_rate": 4.597759872937597e-06,
      "lm_loss": 5.4548,
      "loss": 1.3299,
      "step": 1188,
      "text_contrastive_loss": 0.7367,
      "train_positive_log_prob": -79.6501,
      "train_positive_token_accuracy": 0.0787,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.4597,
      "epoch": 2.683972911963883,
      "grad_norm": 14.347017288208008,
      "learning_rate": 4.590528404876374e-06,
      "lm_loss": 5.4162,
      "loss": 1.434,
      "step": 1189,
      "text_contrastive_loss": 0.8654,
      "train_positive_log_prob": -80.7159,
      "train_positive_token_accuracy": 0.0805,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.4394,
      "epoch": 2.6862302483069977,
      "grad_norm": 13.696542739868164,
      "learning_rate": 4.5832977990144165e-06,
      "lm_loss": 5.5338,
      "loss": 1.4007,
      "step": 1190,
      "text_contrastive_loss": 0.8157,
      "train_positive_log_prob": -80.281,
      "train_positive_token_accuracy": 0.0794,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3796,
      "epoch": 2.6884875846501126,
      "grad_norm": 13.836771965026855,
      "learning_rate": 4.5760680705767665e-06,
      "lm_loss": 5.4323,
      "loss": 1.2444,
      "step": 1191,
      "text_contrastive_loss": 0.643,
      "train_positive_log_prob": -80.3787,
      "train_positive_token_accuracy": 0.0763,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.3582,
      "epoch": 2.690744920993228,
      "grad_norm": 12.787728309631348,
      "learning_rate": 4.5688392347866226e-06,
      "lm_loss": 5.4675,
      "loss": 1.2694,
      "step": 1192,
      "text_contrastive_loss": 0.7288,
      "train_positive_log_prob": -81.4767,
      "train_positive_token_accuracy": 0.0823,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.3826,
      "epoch": 2.6930022573363432,
      "grad_norm": 15.042651176452637,
      "learning_rate": 4.561611306865299e-06,
      "lm_loss": 5.52,
      "loss": 1.3041,
      "step": 1193,
      "text_contrastive_loss": 0.739,
      "train_positive_log_prob": -82.8433,
      "train_positive_token_accuracy": 0.0836,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3628,
      "epoch": 2.695259593679458,
      "grad_norm": 14.1420259475708,
      "learning_rate": 4.554384302032204e-06,
      "lm_loss": 5.4638,
      "loss": 1.2849,
      "step": 1194,
      "text_contrastive_loss": 0.7515,
      "train_positive_log_prob": -80.3323,
      "train_positive_token_accuracy": 0.0784,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.4446,
      "epoch": 2.6975169300225734,
      "grad_norm": 13.776505470275879,
      "learning_rate": 4.547158235504797e-06,
      "lm_loss": 5.5475,
      "loss": 1.4387,
      "step": 1195,
      "text_contrastive_loss": 0.8788,
      "train_positive_log_prob": -82.4971,
      "train_positive_token_accuracy": 0.0817,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.4056,
      "epoch": 2.6997742663656883,
      "grad_norm": 12.939220428466797,
      "learning_rate": 4.539933122498566e-06,
      "lm_loss": 5.4481,
      "loss": 1.4169,
      "step": 1196,
      "text_contrastive_loss": 0.9331,
      "train_positive_log_prob": -81.5348,
      "train_positive_token_accuracy": 0.0872,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.2591,
      "epoch": 2.7020316027088036,
      "grad_norm": 13.445302963256836,
      "learning_rate": 4.532708978226987e-06,
      "lm_loss": 5.4251,
      "loss": 1.1482,
      "step": 1197,
      "text_contrastive_loss": 0.6932,
      "train_positive_log_prob": -79.4235,
      "train_positive_token_accuracy": 0.0787,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.3294,
      "epoch": 2.704288939051919,
      "grad_norm": 13.859259605407715,
      "learning_rate": 4.525485817901499e-06,
      "lm_loss": 5.5595,
      "loss": 1.322,
      "step": 1198,
      "text_contrastive_loss": 0.8732,
      "train_positive_log_prob": -84.2796,
      "train_positive_token_accuracy": 0.083,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.4215,
      "epoch": 2.706546275395034,
      "grad_norm": 13.112929344177246,
      "learning_rate": 4.518263656731468e-06,
      "lm_loss": 5.4648,
      "loss": 1.4274,
      "step": 1199,
      "text_contrastive_loss": 0.9189,
      "train_positive_log_prob": -81.3507,
      "train_positive_token_accuracy": 0.0819,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.4598,
      "epoch": 2.708803611738149,
      "grad_norm": 14.781098365783691,
      "learning_rate": 4.511042509924157e-06,
      "lm_loss": 5.4864,
      "loss": 1.4536,
      "step": 1200,
      "text_contrastive_loss": 0.8903,
      "train_positive_log_prob": -81.4819,
      "train_positive_token_accuracy": 0.0826,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.466,
      "epoch": 2.711060948081264,
      "grad_norm": 13.943631172180176,
      "learning_rate": 4.5038223926846905e-06,
      "lm_loss": 5.5542,
      "loss": 1.4778,
      "step": 1201,
      "text_contrastive_loss": 0.9129,
      "train_positive_log_prob": -82.2697,
      "train_positive_token_accuracy": 0.077,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.4928,
      "epoch": 2.7133182844243793,
      "grad_norm": 16.02657699584961,
      "learning_rate": 4.49660332021603e-06,
      "lm_loss": 5.4441,
      "loss": 1.5473,
      "step": 1202,
      "text_contrastive_loss": 1.0202,
      "train_positive_log_prob": -78.9585,
      "train_positive_token_accuracy": 0.077,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.4403,
      "epoch": 2.7155756207674946,
      "grad_norm": 13.975430488586426,
      "learning_rate": 4.489385307718934e-06,
      "lm_loss": 5.5006,
      "loss": 1.3872,
      "step": 1203,
      "text_contrastive_loss": 0.7938,
      "train_positive_log_prob": -79.5866,
      "train_positive_token_accuracy": 0.0753,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3518,
      "epoch": 2.7178329571106095,
      "grad_norm": 12.17011833190918,
      "learning_rate": 4.482168370391931e-06,
      "lm_loss": 5.5338,
      "loss": 1.2789,
      "step": 1204,
      "text_contrastive_loss": 0.7475,
      "train_positive_log_prob": -82.0157,
      "train_positive_token_accuracy": 0.0767,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.3903,
      "epoch": 2.7200902934537243,
      "grad_norm": 12.984262466430664,
      "learning_rate": 4.47495252343128e-06,
      "lm_loss": 5.4867,
      "loss": 1.3849,
      "step": 1205,
      "text_contrastive_loss": 0.8918,
      "train_positive_log_prob": -83.6725,
      "train_positive_token_accuracy": 0.074,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.3598,
      "epoch": 2.7223476297968396,
      "grad_norm": 12.920084953308105,
      "learning_rate": 4.467737782030951e-06,
      "lm_loss": 5.5757,
      "loss": 1.2959,
      "step": 1206,
      "text_contrastive_loss": 0.757,
      "train_positive_log_prob": -84.0959,
      "train_positive_token_accuracy": 0.0795,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3264,
      "epoch": 2.724604966139955,
      "grad_norm": 11.541576385498047,
      "learning_rate": 4.460524161382582e-06,
      "lm_loss": 5.4051,
      "loss": 1.2119,
      "step": 1207,
      "text_contrastive_loss": 0.69,
      "train_positive_log_prob": -80.4468,
      "train_positive_token_accuracy": 0.0816,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.3539,
      "epoch": 2.72686230248307,
      "grad_norm": 11.823294639587402,
      "learning_rate": 4.453311676675453e-06,
      "lm_loss": 5.4722,
      "loss": 1.2801,
      "step": 1208,
      "text_contrastive_loss": 0.7579,
      "train_positive_log_prob": -82.7221,
      "train_positive_token_accuracy": 0.077,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.4779,
      "epoch": 2.729119638826185,
      "grad_norm": 14.281606674194336,
      "learning_rate": 4.44610034309645e-06,
      "lm_loss": 5.4242,
      "loss": 1.4065,
      "step": 1209,
      "text_contrastive_loss": 0.7723,
      "train_positive_log_prob": -81.3001,
      "train_positive_token_accuracy": 0.0769,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.302,
      "epoch": 2.7313769751693,
      "grad_norm": 12.948177337646484,
      "learning_rate": 4.438890175830039e-06,
      "lm_loss": 5.6239,
      "loss": 1.2313,
      "step": 1210,
      "text_contrastive_loss": 0.7339,
      "train_positive_log_prob": -83.7198,
      "train_positive_token_accuracy": 0.0745,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.4556,
      "epoch": 2.7336343115124153,
      "grad_norm": 14.67457389831543,
      "learning_rate": 4.431681190058224e-06,
      "lm_loss": 5.4522,
      "loss": 1.4931,
      "step": 1211,
      "text_contrastive_loss": 0.9847,
      "train_positive_log_prob": -80.8,
      "train_positive_token_accuracy": 0.0799,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3598,
      "epoch": 2.7358916478555306,
      "grad_norm": 12.027202606201172,
      "learning_rate": 4.42447340096053e-06,
      "lm_loss": 5.4147,
      "loss": 1.2419,
      "step": 1212,
      "text_contrastive_loss": 0.6813,
      "train_positive_log_prob": -81.1546,
      "train_positive_token_accuracy": 0.0779,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.4033,
      "epoch": 2.7381489841986455,
      "grad_norm": 14.120058059692383,
      "learning_rate": 4.417266823713953e-06,
      "lm_loss": 5.4045,
      "loss": 1.2572,
      "step": 1213,
      "text_contrastive_loss": 0.6269,
      "train_positive_log_prob": -79.2515,
      "train_positive_token_accuracy": 0.0794,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.2904,
      "epoch": 2.740406320541761,
      "grad_norm": 12.804457664489746,
      "learning_rate": 4.410061473492943e-06,
      "lm_loss": 5.4582,
      "loss": 1.1259,
      "step": 1214,
      "text_contrastive_loss": 0.5793,
      "train_positive_log_prob": -81.4885,
      "train_positive_token_accuracy": 0.0816,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3212,
      "epoch": 2.7426636568848757,
      "grad_norm": 13.253199577331543,
      "learning_rate": 4.402857365469364e-06,
      "lm_loss": 5.4853,
      "loss": 1.222,
      "step": 1215,
      "text_contrastive_loss": 0.7047,
      "train_positive_log_prob": -81.0128,
      "train_positive_token_accuracy": 0.0823,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.323,
      "epoch": 2.744920993227991,
      "grad_norm": 11.764541625976562,
      "learning_rate": 4.3956545148124665e-06,
      "lm_loss": 5.4371,
      "loss": 1.2995,
      "step": 1216,
      "text_contrastive_loss": 0.8655,
      "train_positive_log_prob": -78.6893,
      "train_positive_token_accuracy": 0.0745,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.4459,
      "epoch": 2.7471783295711063,
      "grad_norm": 15.989474296569824,
      "learning_rate": 4.38845293668885e-06,
      "lm_loss": 5.5859,
      "loss": 1.4857,
      "step": 1217,
      "text_contrastive_loss": 0.9625,
      "train_positive_log_prob": -82.6797,
      "train_positive_token_accuracy": 0.076,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.4463,
      "epoch": 2.749435665914221,
      "grad_norm": 13.160560607910156,
      "learning_rate": 4.381252646262437e-06,
      "lm_loss": 5.495,
      "loss": 1.4399,
      "step": 1218,
      "text_contrastive_loss": 0.8884,
      "train_positive_log_prob": -81.8069,
      "train_positive_token_accuracy": 0.0777,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.458,
      "epoch": 2.7516930022573365,
      "grad_norm": 13.962233543395996,
      "learning_rate": 4.37405365869444e-06,
      "lm_loss": 5.536,
      "loss": 1.4393,
      "step": 1219,
      "text_contrastive_loss": 0.8555,
      "train_positive_log_prob": -82.6745,
      "train_positive_token_accuracy": 0.0779,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3958,
      "epoch": 2.7539503386004514,
      "grad_norm": 12.87502384185791,
      "learning_rate": 4.366855989143326e-06,
      "lm_loss": 5.4709,
      "loss": 1.3712,
      "step": 1220,
      "text_contrastive_loss": 0.8567,
      "train_positive_log_prob": -79.346,
      "train_positive_token_accuracy": 0.0784,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3836,
      "epoch": 2.7562076749435667,
      "grad_norm": 13.692797660827637,
      "learning_rate": 4.359659652764786e-06,
      "lm_loss": 5.503,
      "loss": 1.2807,
      "step": 1221,
      "text_contrastive_loss": 0.6935,
      "train_positive_log_prob": -83.8341,
      "train_positive_token_accuracy": 0.0826,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.4108,
      "epoch": 2.758465011286682,
      "grad_norm": 13.382584571838379,
      "learning_rate": 4.352464664711706e-06,
      "lm_loss": 5.4594,
      "loss": 1.3804,
      "step": 1222,
      "text_contrastive_loss": 0.8473,
      "train_positive_log_prob": -80.6109,
      "train_positive_token_accuracy": 0.0839,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.4057,
      "epoch": 2.760722347629797,
      "grad_norm": 13.29274845123291,
      "learning_rate": 4.345271040134129e-06,
      "lm_loss": 5.4844,
      "loss": 1.3814,
      "step": 1223,
      "text_contrastive_loss": 0.8546,
      "train_positive_log_prob": -80.6597,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.4529,
      "epoch": 2.7629796839729117,
      "grad_norm": 15.336840629577637,
      "learning_rate": 4.338078794179234e-06,
      "lm_loss": 5.5096,
      "loss": 1.3576,
      "step": 1224,
      "text_contrastive_loss": 0.7076,
      "train_positive_log_prob": -81.0948,
      "train_positive_token_accuracy": 0.0809,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.5451,
      "epoch": 2.765237020316027,
      "grad_norm": 16.94712257385254,
      "learning_rate": 4.330887941991288e-06,
      "lm_loss": 5.5441,
      "loss": 1.481,
      "step": 1225,
      "text_contrastive_loss": 0.7628,
      "train_positive_log_prob": -83.5034,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.417,
      "epoch": 2.7674943566591423,
      "grad_norm": 12.747289657592773,
      "learning_rate": 4.323698498711634e-06,
      "lm_loss": 5.5802,
      "loss": 1.3896,
      "step": 1226,
      "text_contrastive_loss": 0.8292,
      "train_positive_log_prob": -78.9744,
      "train_positive_token_accuracy": 0.0687,
      "train_positive_token_prob": 0.028
    },
    {
      "contrastive_loss": 0.3553,
      "epoch": 2.769751693002257,
      "grad_norm": 12.221817016601562,
      "learning_rate": 4.316510479478636e-06,
      "lm_loss": 5.3802,
      "loss": 1.241,
      "step": 1227,
      "text_contrastive_loss": 0.6953,
      "train_positive_log_prob": -78.3718,
      "train_positive_token_accuracy": 0.077,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.43,
      "epoch": 2.7720090293453725,
      "grad_norm": 14.645344734191895,
      "learning_rate": 4.309323899427671e-06,
      "lm_loss": 5.373,
      "loss": 1.3379,
      "step": 1228,
      "text_contrastive_loss": 0.7412,
      "train_positive_log_prob": -78.6221,
      "train_positive_token_accuracy": 0.0841,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.3726,
      "epoch": 2.7742663656884874,
      "grad_norm": 13.062455177307129,
      "learning_rate": 4.302138773691079e-06,
      "lm_loss": 5.4994,
      "loss": 1.2907,
      "step": 1229,
      "text_contrastive_loss": 0.7365,
      "train_positive_log_prob": -80.9441,
      "train_positive_token_accuracy": 0.0865,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.387,
      "epoch": 2.7765237020316027,
      "grad_norm": 14.095149040222168,
      "learning_rate": 4.294955117398139e-06,
      "lm_loss": 5.4908,
      "loss": 1.3432,
      "step": 1230,
      "text_contrastive_loss": 0.8143,
      "train_positive_log_prob": -78.8683,
      "train_positive_token_accuracy": 0.0713,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.3265,
      "epoch": 2.778781038374718,
      "grad_norm": 11.440180778503418,
      "learning_rate": 4.287772945675035e-06,
      "lm_loss": 5.5812,
      "loss": 1.255,
      "step": 1231,
      "text_contrastive_loss": 0.7408,
      "train_positive_log_prob": -83.3722,
      "train_positive_token_accuracy": 0.073,
      "train_positive_token_prob": 0.0285
    },
    {
      "contrastive_loss": 0.3287,
      "epoch": 2.781038374717833,
      "grad_norm": 14.378003120422363,
      "learning_rate": 4.280592273644829e-06,
      "lm_loss": 5.4748,
      "loss": 1.2209,
      "step": 1232,
      "text_contrastive_loss": 0.6894,
      "train_positive_log_prob": -81.1622,
      "train_positive_token_accuracy": 0.0854,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.3838,
      "epoch": 2.783295711060948,
      "grad_norm": 12.829390525817871,
      "learning_rate": 4.273413116427419e-06,
      "lm_loss": 5.5321,
      "loss": 1.2876,
      "step": 1233,
      "text_contrastive_loss": 0.7011,
      "train_positive_log_prob": -81.3172,
      "train_positive_token_accuracy": 0.0755,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.3836,
      "epoch": 2.785553047404063,
      "grad_norm": 14.463422775268555,
      "learning_rate": 4.26623548913952e-06,
      "lm_loss": 5.4773,
      "loss": 1.3214,
      "step": 1234,
      "text_contrastive_loss": 0.7801,
      "train_positive_log_prob": -80.2541,
      "train_positive_token_accuracy": 0.0792,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.4316,
      "epoch": 2.7878103837471784,
      "grad_norm": 13.590999603271484,
      "learning_rate": 4.259059406894619e-06,
      "lm_loss": 5.469,
      "loss": 1.394,
      "step": 1235,
      "text_contrastive_loss": 0.8311,
      "train_positive_log_prob": -82.0576,
      "train_positive_token_accuracy": 0.0752,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.471,
      "epoch": 2.7900677200902937,
      "grad_norm": 16.407123565673828,
      "learning_rate": 4.251884884802956e-06,
      "lm_loss": 5.3534,
      "loss": 1.3968,
      "step": 1236,
      "text_contrastive_loss": 0.781,
      "train_positive_log_prob": -77.7943,
      "train_positive_token_accuracy": 0.0836,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.3975,
      "epoch": 2.7923250564334086,
      "grad_norm": 15.670740127563477,
      "learning_rate": 4.2447119379714805e-06,
      "lm_loss": 5.5612,
      "loss": 1.3721,
      "step": 1237,
      "text_contrastive_loss": 0.837,
      "train_positive_log_prob": -84.1368,
      "train_positive_token_accuracy": 0.0855,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.3886,
      "epoch": 2.7945823927765234,
      "grad_norm": 13.319536209106445,
      "learning_rate": 4.237540581503831e-06,
      "lm_loss": 5.4816,
      "loss": 1.3145,
      "step": 1238,
      "text_contrastive_loss": 0.7554,
      "train_positive_log_prob": -81.7738,
      "train_positive_token_accuracy": 0.0762,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3296,
      "epoch": 2.7968397291196387,
      "grad_norm": 12.3250093460083,
      "learning_rate": 4.23037083050029e-06,
      "lm_loss": 5.5549,
      "loss": 1.2445,
      "step": 1239,
      "text_contrastive_loss": 0.719,
      "train_positive_log_prob": -80.176,
      "train_positive_token_accuracy": 0.0733,
      "train_positive_token_prob": 0.0286
    },
    {
      "contrastive_loss": 0.3645,
      "epoch": 2.799097065462754,
      "grad_norm": 14.026390075683594,
      "learning_rate": 4.223202700057765e-06,
      "lm_loss": 5.495,
      "loss": 1.3381,
      "step": 1240,
      "text_contrastive_loss": 0.8483,
      "train_positive_log_prob": -82.6181,
      "train_positive_token_accuracy": 0.0725,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.3603,
      "epoch": 2.801354401805869,
      "grad_norm": 13.461612701416016,
      "learning_rate": 4.216036205269748e-06,
      "lm_loss": 5.4894,
      "loss": 1.3684,
      "step": 1241,
      "text_contrastive_loss": 0.9183,
      "train_positive_log_prob": -80.9055,
      "train_positive_token_accuracy": 0.0808,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3864,
      "epoch": 2.8036117381489842,
      "grad_norm": 14.413870811462402,
      "learning_rate": 4.20887136122629e-06,
      "lm_loss": 5.4594,
      "loss": 1.3118,
      "step": 1242,
      "text_contrastive_loss": 0.7589,
      "train_positive_log_prob": -82.2732,
      "train_positive_token_accuracy": 0.0811,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.528,
      "epoch": 2.805869074492099,
      "grad_norm": 14.763357162475586,
      "learning_rate": 4.201708183013963e-06,
      "lm_loss": 5.4852,
      "loss": 1.5696,
      "step": 1243,
      "text_contrastive_loss": 0.9862,
      "train_positive_log_prob": -80.1133,
      "train_positive_token_accuracy": 0.069,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.3172,
      "epoch": 2.8081264108352144,
      "grad_norm": 12.26828384399414,
      "learning_rate": 4.1945466857158336e-06,
      "lm_loss": 5.4729,
      "loss": 1.2154,
      "step": 1244,
      "text_contrastive_loss": 0.7019,
      "train_positive_log_prob": -81.5168,
      "train_positive_token_accuracy": 0.0831,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.4179,
      "epoch": 2.8103837471783297,
      "grad_norm": 14.41314697265625,
      "learning_rate": 4.187386884411426e-06,
      "lm_loss": 5.5451,
      "loss": 1.3693,
      "step": 1245,
      "text_contrastive_loss": 0.7936,
      "train_positive_log_prob": -82.1774,
      "train_positive_token_accuracy": 0.0754,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3806,
      "epoch": 2.8126410835214446,
      "grad_norm": 13.032148361206055,
      "learning_rate": 4.1802287941767e-06,
      "lm_loss": 5.4604,
      "loss": 1.393,
      "step": 1246,
      "text_contrastive_loss": 0.9328,
      "train_positive_log_prob": -83.8024,
      "train_positive_token_accuracy": 0.0687,
      "train_positive_token_prob": 0.0289
    },
    {
      "contrastive_loss": 0.3946,
      "epoch": 2.81489841986456,
      "grad_norm": 12.606998443603516,
      "learning_rate": 4.173072430084002e-06,
      "lm_loss": 5.5478,
      "loss": 1.4201,
      "step": 1247,
      "text_contrastive_loss": 0.9413,
      "train_positive_log_prob": -81.8941,
      "train_positive_token_accuracy": 0.0765,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.3443,
      "epoch": 2.8171557562076748,
      "grad_norm": 13.290593147277832,
      "learning_rate": 4.165917807202055e-06,
      "lm_loss": 5.4816,
      "loss": 1.1863,
      "step": 1248,
      "text_contrastive_loss": 0.5876,
      "train_positive_log_prob": -82.2438,
      "train_positive_token_accuracy": 0.082,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3559,
      "epoch": 2.81941309255079,
      "grad_norm": 15.067920684814453,
      "learning_rate": 4.1587649405959065e-06,
      "lm_loss": 5.4577,
      "loss": 1.2862,
      "step": 1249,
      "text_contrastive_loss": 0.769,
      "train_positive_log_prob": -81.1565,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3956,
      "epoch": 2.8216704288939054,
      "grad_norm": 13.614821434020996,
      "learning_rate": 4.151613845326912e-06,
      "lm_loss": 5.5196,
      "loss": 1.3751,
      "step": 1250,
      "text_contrastive_loss": 0.855,
      "train_positive_log_prob": -81.5007,
      "train_positive_token_accuracy": 0.0775,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.4083,
      "epoch": 2.8239277652370203,
      "grad_norm": 13.48894214630127,
      "learning_rate": 4.144464536452693e-06,
      "lm_loss": 5.5392,
      "loss": 1.3301,
      "step": 1251,
      "text_contrastive_loss": 0.7358,
      "train_positive_log_prob": -81.4915,
      "train_positive_token_accuracy": 0.0842,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3992,
      "epoch": 2.8261851015801356,
      "grad_norm": 14.535311698913574,
      "learning_rate": 4.137317029027111e-06,
      "lm_loss": 5.3489,
      "loss": 1.3523,
      "step": 1252,
      "text_contrastive_loss": 0.8362,
      "train_positive_log_prob": -78.2249,
      "train_positive_token_accuracy": 0.0871,
      "train_positive_token_prob": 0.0336
    },
    {
      "contrastive_loss": 0.4001,
      "epoch": 2.8284424379232505,
      "grad_norm": 14.125943183898926,
      "learning_rate": 4.1301713381002394e-06,
      "lm_loss": 5.3939,
      "loss": 1.3438,
      "step": 1253,
      "text_contrastive_loss": 0.8087,
      "train_positive_log_prob": -78.3838,
      "train_positive_token_accuracy": 0.0801,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.4122,
      "epoch": 2.8306997742663658,
      "grad_norm": 13.226386070251465,
      "learning_rate": 4.123027478718318e-06,
      "lm_loss": 5.4704,
      "loss": 1.2586,
      "step": 1254,
      "text_contrastive_loss": 0.5988,
      "train_positive_log_prob": -81.179,
      "train_positive_token_accuracy": 0.0827,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.4602,
      "epoch": 2.832957110609481,
      "grad_norm": 15.592979431152344,
      "learning_rate": 4.115885465923734e-06,
      "lm_loss": 5.4381,
      "loss": 1.4722,
      "step": 1255,
      "text_contrastive_loss": 0.9363,
      "train_positive_log_prob": -79.7466,
      "train_positive_token_accuracy": 0.0783,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.4278,
      "epoch": 2.835214446952596,
      "grad_norm": 12.926896095275879,
      "learning_rate": 4.108745314754989e-06,
      "lm_loss": 5.4251,
      "loss": 1.3541,
      "step": 1256,
      "text_contrastive_loss": 0.7677,
      "train_positive_log_prob": -79.3661,
      "train_positive_token_accuracy": 0.0836,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.3709,
      "epoch": 2.837471783295711,
      "grad_norm": 13.7692289352417,
      "learning_rate": 4.101607040246659e-06,
      "lm_loss": 5.4437,
      "loss": 1.3455,
      "step": 1257,
      "text_contrastive_loss": 0.8605,
      "train_positive_log_prob": -82.8454,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.3342,
      "epoch": 2.839729119638826,
      "grad_norm": 12.793830871582031,
      "learning_rate": 4.094470657429374e-06,
      "lm_loss": 5.534,
      "loss": 1.3307,
      "step": 1258,
      "text_contrastive_loss": 0.8862,
      "train_positive_log_prob": -81.7586,
      "train_positive_token_accuracy": 0.0738,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.3849,
      "epoch": 2.8419864559819414,
      "grad_norm": 13.881046295166016,
      "learning_rate": 4.087336181329777e-06,
      "lm_loss": 5.522,
      "loss": 1.361,
      "step": 1259,
      "text_contrastive_loss": 0.8478,
      "train_positive_log_prob": -81.2762,
      "train_positive_token_accuracy": 0.0754,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.3566,
      "epoch": 2.8442437923250563,
      "grad_norm": 14.135019302368164,
      "learning_rate": 4.080203626970498e-06,
      "lm_loss": 5.496,
      "loss": 1.3519,
      "step": 1260,
      "text_contrastive_loss": 0.8914,
      "train_positive_log_prob": -79.7113,
      "train_positive_token_accuracy": 0.0811,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3855,
      "epoch": 2.8465011286681716,
      "grad_norm": 12.44174575805664,
      "learning_rate": 4.0730730093701185e-06,
      "lm_loss": 5.4494,
      "loss": 1.3268,
      "step": 1261,
      "text_contrastive_loss": 0.7927,
      "train_positive_log_prob": -81.718,
      "train_positive_token_accuracy": 0.0836,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.385,
      "epoch": 2.8487584650112865,
      "grad_norm": 13.774590492248535,
      "learning_rate": 4.065944343543146e-06,
      "lm_loss": 5.5729,
      "loss": 1.324,
      "step": 1262,
      "text_contrastive_loss": 0.7635,
      "train_positive_log_prob": -84.2902,
      "train_positive_token_accuracy": 0.0728,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.4697,
      "epoch": 2.851015801354402,
      "grad_norm": 14.798707962036133,
      "learning_rate": 4.058817644499973e-06,
      "lm_loss": 5.5272,
      "loss": 1.4702,
      "step": 1263,
      "text_contrastive_loss": 0.8955,
      "train_positive_log_prob": -81.6971,
      "train_positive_token_accuracy": 0.0801,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.3595,
      "epoch": 2.853273137697517,
      "grad_norm": 13.562416076660156,
      "learning_rate": 4.051692927246857e-06,
      "lm_loss": 5.4418,
      "loss": 1.2281,
      "step": 1264,
      "text_contrastive_loss": 0.6488,
      "train_positive_log_prob": -80.7348,
      "train_positive_token_accuracy": 0.0732,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3096,
      "epoch": 2.855530474040632,
      "grad_norm": 17.661754608154297,
      "learning_rate": 4.044570206785874e-06,
      "lm_loss": 5.4619,
      "loss": 1.2246,
      "step": 1265,
      "text_contrastive_loss": 0.7378,
      "train_positive_log_prob": -81.3721,
      "train_positive_token_accuracy": 0.0847,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3508,
      "epoch": 2.8577878103837473,
      "grad_norm": 11.998366355895996,
      "learning_rate": 4.037449498114903e-06,
      "lm_loss": 5.5167,
      "loss": 1.2914,
      "step": 1266,
      "text_contrastive_loss": 0.7778,
      "train_positive_log_prob": -80.4452,
      "train_positive_token_accuracy": 0.0841,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.4118,
      "epoch": 2.860045146726862,
      "grad_norm": 13.16608715057373,
      "learning_rate": 4.0303308162275835e-06,
      "lm_loss": 5.5101,
      "loss": 1.2936,
      "step": 1267,
      "text_contrastive_loss": 0.6617,
      "train_positive_log_prob": -83.7504,
      "train_positive_token_accuracy": 0.0835,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3707,
      "epoch": 2.8623024830699775,
      "grad_norm": 13.012649536132812,
      "learning_rate": 4.0232141761132894e-06,
      "lm_loss": 5.3819,
      "loss": 1.3016,
      "step": 1268,
      "text_contrastive_loss": 0.7854,
      "train_positive_log_prob": -78.8567,
      "train_positive_token_accuracy": 0.0794,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.448,
      "epoch": 2.864559819413093,
      "grad_norm": 13.595703125,
      "learning_rate": 4.016099592757091e-06,
      "lm_loss": 5.4091,
      "loss": 1.356,
      "step": 1269,
      "text_contrastive_loss": 0.7342,
      "train_positive_log_prob": -80.2895,
      "train_positive_token_accuracy": 0.0861,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.403,
      "epoch": 2.8668171557562077,
      "grad_norm": 14.504570960998535,
      "learning_rate": 4.008987081139734e-06,
      "lm_loss": 5.5506,
      "loss": 1.2738,
      "step": 1270,
      "text_contrastive_loss": 0.6315,
      "train_positive_log_prob": -84.0404,
      "train_positive_token_accuracy": 0.0751,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.4179,
      "epoch": 2.8690744920993225,
      "grad_norm": 13.538684844970703,
      "learning_rate": 4.0018766562375984e-06,
      "lm_loss": 5.4326,
      "loss": 1.3779,
      "step": 1271,
      "text_contrastive_loss": 0.8334,
      "train_positive_log_prob": -79.6862,
      "train_positive_token_accuracy": 0.0732,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3203,
      "epoch": 2.871331828442438,
      "grad_norm": 12.093225479125977,
      "learning_rate": 3.994768333022669e-06,
      "lm_loss": 5.4881,
      "loss": 1.2175,
      "step": 1272,
      "text_contrastive_loss": 0.6968,
      "train_positive_log_prob": -82.7274,
      "train_positive_token_accuracy": 0.0816,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.4798,
      "epoch": 2.873589164785553,
      "grad_norm": 13.538874626159668,
      "learning_rate": 3.987662126462507e-06,
      "lm_loss": 5.5,
      "loss": 1.4636,
      "step": 1273,
      "text_contrastive_loss": 0.8675,
      "train_positive_log_prob": -82.9512,
      "train_positive_token_accuracy": 0.0899,
      "train_positive_token_prob": 0.0329
    },
    {
      "contrastive_loss": 0.3468,
      "epoch": 2.875846501128668,
      "grad_norm": 13.044873237609863,
      "learning_rate": 3.980558051520218e-06,
      "lm_loss": 5.4294,
      "loss": 1.2759,
      "step": 1274,
      "text_contrastive_loss": 0.7723,
      "train_positive_log_prob": -80.0195,
      "train_positive_token_accuracy": 0.0819,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.4188,
      "epoch": 2.8781038374717833,
      "grad_norm": 13.54685115814209,
      "learning_rate": 3.973456123154415e-06,
      "lm_loss": 5.5064,
      "loss": 1.3055,
      "step": 1275,
      "text_contrastive_loss": 0.6722,
      "train_positive_log_prob": -80.8951,
      "train_positive_token_accuracy": 0.0739,
      "train_positive_token_prob": 0.0291
    },
    {
      "contrastive_loss": 0.3674,
      "epoch": 2.880361173814898,
      "grad_norm": 12.428534507751465,
      "learning_rate": 3.966356356319196e-06,
      "lm_loss": 5.4792,
      "loss": 1.3216,
      "step": 1276,
      "text_contrastive_loss": 0.8125,
      "train_positive_log_prob": -80.7279,
      "train_positive_token_accuracy": 0.0901,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.3046,
      "epoch": 2.8826185101580135,
      "grad_norm": 12.955340385437012,
      "learning_rate": 3.959258765964104e-06,
      "lm_loss": 5.424,
      "loss": 1.2631,
      "step": 1277,
      "text_contrastive_loss": 0.8323,
      "train_positive_log_prob": -80.3458,
      "train_positive_token_accuracy": 0.0767,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.3902,
      "epoch": 2.884875846501129,
      "grad_norm": 13.668217658996582,
      "learning_rate": 3.9521633670341005e-06,
      "lm_loss": 5.3886,
      "loss": 1.3276,
      "step": 1278,
      "text_contrastive_loss": 0.7971,
      "train_positive_log_prob": -78.66,
      "train_positive_token_accuracy": 0.0749,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.425,
      "epoch": 2.8871331828442437,
      "grad_norm": 13.321612358093262,
      "learning_rate": 3.9450701744695325e-06,
      "lm_loss": 5.4796,
      "loss": 1.3674,
      "step": 1279,
      "text_contrastive_loss": 0.7889,
      "train_positive_log_prob": -80.8224,
      "train_positive_token_accuracy": 0.0694,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.3522,
      "epoch": 2.889390519187359,
      "grad_norm": 13.01663589477539,
      "learning_rate": 3.937979203206103e-06,
      "lm_loss": 5.4883,
      "loss": 1.282,
      "step": 1280,
      "text_contrastive_loss": 0.7619,
      "train_positive_log_prob": -83.9133,
      "train_positive_token_accuracy": 0.0818,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3696,
      "epoch": 2.891647855530474,
      "grad_norm": 13.588577270507812,
      "learning_rate": 3.930890468174833e-06,
      "lm_loss": 5.4038,
      "loss": 1.3386,
      "step": 1281,
      "text_contrastive_loss": 0.8573,
      "train_positive_log_prob": -79.368,
      "train_positive_token_accuracy": 0.0795,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3557,
      "epoch": 2.893905191873589,
      "grad_norm": 13.21587085723877,
      "learning_rate": 3.92380398430204e-06,
      "lm_loss": 5.4722,
      "loss": 1.2746,
      "step": 1282,
      "text_contrastive_loss": 0.7435,
      "train_positive_log_prob": -82.4255,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.537,
      "epoch": 2.8961625282167045,
      "grad_norm": 19.4005069732666,
      "learning_rate": 3.916719766509297e-06,
      "lm_loss": 5.4177,
      "loss": 1.5095,
      "step": 1283,
      "text_contrastive_loss": 0.8615,
      "train_positive_log_prob": -81.1866,
      "train_positive_token_accuracy": 0.0834,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.4378,
      "epoch": 2.8984198645598194,
      "grad_norm": 13.29963207244873,
      "learning_rate": 3.9096378297134115e-06,
      "lm_loss": 5.4622,
      "loss": 1.3843,
      "step": 1284,
      "text_contrastive_loss": 0.8007,
      "train_positive_log_prob": -82.2274,
      "train_positive_token_accuracy": 0.0763,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.4134,
      "epoch": 2.9006772009029347,
      "grad_norm": 14.635613441467285,
      "learning_rate": 3.90255818882638e-06,
      "lm_loss": 5.5322,
      "loss": 1.3805,
      "step": 1285,
      "text_contrastive_loss": 0.8276,
      "train_positive_log_prob": -82.7852,
      "train_positive_token_accuracy": 0.077,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.3661,
      "epoch": 2.9029345372460496,
      "grad_norm": 13.061617851257324,
      "learning_rate": 3.89548085875537e-06,
      "lm_loss": 5.5022,
      "loss": 1.323,
      "step": 1286,
      "text_contrastive_loss": 0.8134,
      "train_positive_log_prob": -81.9035,
      "train_positive_token_accuracy": 0.0766,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3357,
      "epoch": 2.905191873589165,
      "grad_norm": 12.666919708251953,
      "learning_rate": 3.888405854402684e-06,
      "lm_loss": 5.4253,
      "loss": 1.3267,
      "step": 1287,
      "text_contrastive_loss": 0.8969,
      "train_positive_log_prob": -80.2976,
      "train_positive_token_accuracy": 0.0771,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3475,
      "epoch": 2.90744920993228,
      "grad_norm": 13.228978157043457,
      "learning_rate": 3.881333190665723e-06,
      "lm_loss": 5.3635,
      "loss": 1.2762,
      "step": 1288,
      "text_contrastive_loss": 0.7846,
      "train_positive_log_prob": -78.2085,
      "train_positive_token_accuracy": 0.0771,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.4708,
      "epoch": 2.909706546275395,
      "grad_norm": 14.680399894714355,
      "learning_rate": 3.8742628824369624e-06,
      "lm_loss": 5.4786,
      "loss": 1.4769,
      "step": 1289,
      "text_contrastive_loss": 0.9165,
      "train_positive_log_prob": -80.2097,
      "train_positive_token_accuracy": 0.0784,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.3534,
      "epoch": 2.91196388261851,
      "grad_norm": 13.228862762451172,
      "learning_rate": 3.86719494460392e-06,
      "lm_loss": 5.5715,
      "loss": 1.3555,
      "step": 1290,
      "text_contrastive_loss": 0.89,
      "train_positive_log_prob": -85.337,
      "train_positive_token_accuracy": 0.0803,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.4133,
      "epoch": 2.9142212189616252,
      "grad_norm": 13.871740341186523,
      "learning_rate": 3.8601293920491165e-06,
      "lm_loss": 5.5267,
      "loss": 1.4243,
      "step": 1291,
      "text_contrastive_loss": 0.9166,
      "train_positive_log_prob": -81.6705,
      "train_positive_token_accuracy": 0.0745,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.4165,
      "epoch": 2.9164785553047405,
      "grad_norm": 15.968809127807617,
      "learning_rate": 3.853066239650055e-06,
      "lm_loss": 5.4784,
      "loss": 1.3839,
      "step": 1292,
      "text_contrastive_loss": 0.839,
      "train_positive_log_prob": -79.4615,
      "train_positive_token_accuracy": 0.0811,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.4197,
      "epoch": 2.9187358916478554,
      "grad_norm": 12.649006843566895,
      "learning_rate": 3.846005502279182e-06,
      "lm_loss": 5.4442,
      "loss": 1.3339,
      "step": 1293,
      "text_contrastive_loss": 0.7395,
      "train_positive_log_prob": -79.8937,
      "train_positive_token_accuracy": 0.0692,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.411,
      "epoch": 2.9209932279909707,
      "grad_norm": 14.790594100952148,
      "learning_rate": 3.83894719480386e-06,
      "lm_loss": 5.406,
      "loss": 1.4348,
      "step": 1294,
      "text_contrastive_loss": 0.9665,
      "train_positive_log_prob": -79.2114,
      "train_positive_token_accuracy": 0.0845,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.3828,
      "epoch": 2.9232505643340856,
      "grad_norm": 14.072667121887207,
      "learning_rate": 3.8318913320863355e-06,
      "lm_loss": 5.4848,
      "loss": 1.3355,
      "step": 1295,
      "text_contrastive_loss": 0.8084,
      "train_positive_log_prob": -80.8594,
      "train_positive_token_accuracy": 0.0843,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.4175,
      "epoch": 2.925507900677201,
      "grad_norm": 13.898487091064453,
      "learning_rate": 3.8248379289837065e-06,
      "lm_loss": 5.3468,
      "loss": 1.3991,
      "step": 1296,
      "text_contrastive_loss": 0.8939,
      "train_positive_log_prob": -79.5724,
      "train_positive_token_accuracy": 0.0848,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.4609,
      "epoch": 2.927765237020316,
      "grad_norm": 13.965980529785156,
      "learning_rate": 3.81778700034789e-06,
      "lm_loss": 5.4237,
      "loss": 1.4498,
      "step": 1297,
      "text_contrastive_loss": 0.8931,
      "train_positive_log_prob": -81.0562,
      "train_positive_token_accuracy": 0.0821,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3697,
      "epoch": 2.930022573363431,
      "grad_norm": 12.408913612365723,
      "learning_rate": 3.810738561025599e-06,
      "lm_loss": 5.4696,
      "loss": 1.2408,
      "step": 1298,
      "text_contrastive_loss": 0.6483,
      "train_positive_log_prob": -81.6389,
      "train_positive_token_accuracy": 0.0817,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.4593,
      "epoch": 2.9322799097065464,
      "grad_norm": 15.56299877166748,
      "learning_rate": 3.803692625858295e-06,
      "lm_loss": 5.4225,
      "loss": 1.4455,
      "step": 1299,
      "text_contrastive_loss": 0.8881,
      "train_positive_log_prob": -79.0296,
      "train_positive_token_accuracy": 0.0729,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3879,
      "epoch": 2.9345372460496613,
      "grad_norm": 13.634164810180664,
      "learning_rate": 3.7966492096821773e-06,
      "lm_loss": 5.3584,
      "loss": 1.3445,
      "step": 1300,
      "text_contrastive_loss": 0.8417,
      "train_positive_log_prob": -79.0793,
      "train_positive_token_accuracy": 0.0866,
      "train_positive_token_prob": 0.0328
    },
    {
      "contrastive_loss": 0.3294,
      "epoch": 2.9367945823927766,
      "grad_norm": 12.517492294311523,
      "learning_rate": 3.7896083273281324e-06,
      "lm_loss": 5.5431,
      "loss": 1.2618,
      "step": 1301,
      "text_contrastive_loss": 0.7561,
      "train_positive_log_prob": -83.3422,
      "train_positive_token_accuracy": 0.0751,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3643,
      "epoch": 2.939051918735892,
      "grad_norm": 12.977726936340332,
      "learning_rate": 3.7825699936217183e-06,
      "lm_loss": 5.3936,
      "loss": 1.2703,
      "step": 1302,
      "text_contrastive_loss": 0.7331,
      "train_positive_log_prob": -81.6006,
      "train_positive_token_accuracy": 0.0824,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.472,
      "epoch": 2.9413092550790068,
      "grad_norm": 13.751511573791504,
      "learning_rate": 3.7755342233831188e-06,
      "lm_loss": 5.3611,
      "loss": 1.4255,
      "step": 1303,
      "text_contrastive_loss": 0.8348,
      "train_positive_log_prob": -76.8571,
      "train_positive_token_accuracy": 0.0817,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.4073,
      "epoch": 2.9435665914221216,
      "grad_norm": 12.415215492248535,
      "learning_rate": 3.7685010314271287e-06,
      "lm_loss": 5.4899,
      "loss": 1.4332,
      "step": 1304,
      "text_contrastive_loss": 0.9538,
      "train_positive_log_prob": -80.5901,
      "train_positive_token_accuracy": 0.0804,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3922,
      "epoch": 2.945823927765237,
      "grad_norm": 13.121039390563965,
      "learning_rate": 3.761470432563109e-06,
      "lm_loss": 5.5034,
      "loss": 1.3051,
      "step": 1305,
      "text_contrastive_loss": 0.7251,
      "train_positive_log_prob": -79.223,
      "train_positive_token_accuracy": 0.0784,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.4288,
      "epoch": 2.9480812641083523,
      "grad_norm": 15.263547897338867,
      "learning_rate": 3.75444244159496e-06,
      "lm_loss": 5.5268,
      "loss": 1.3455,
      "step": 1306,
      "text_contrastive_loss": 0.7279,
      "train_positive_log_prob": -82.5366,
      "train_positive_token_accuracy": 0.0735,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.4629,
      "epoch": 2.950338600451467,
      "grad_norm": 14.952875137329102,
      "learning_rate": 3.747417073321092e-06,
      "lm_loss": 5.4919,
      "loss": 1.3997,
      "step": 1307,
      "text_contrastive_loss": 0.7752,
      "train_positive_log_prob": -82.9756,
      "train_positive_token_accuracy": 0.0871,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.3381,
      "epoch": 2.9525959367945824,
      "grad_norm": 11.371570587158203,
      "learning_rate": 3.740394342534394e-06,
      "lm_loss": 5.4794,
      "loss": 1.2245,
      "step": 1308,
      "text_contrastive_loss": 0.6769,
      "train_positive_log_prob": -80.9922,
      "train_positive_token_accuracy": 0.0859,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.4304,
      "epoch": 2.9548532731376973,
      "grad_norm": 13.57474422454834,
      "learning_rate": 3.7333742640221994e-06,
      "lm_loss": 5.4681,
      "loss": 1.3594,
      "step": 1309,
      "text_contrastive_loss": 0.7643,
      "train_positive_log_prob": -79.3054,
      "train_positive_token_accuracy": 0.0813,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.4388,
      "epoch": 2.9571106094808126,
      "grad_norm": 13.896613121032715,
      "learning_rate": 3.7263568525662574e-06,
      "lm_loss": 5.4731,
      "loss": 1.4319,
      "step": 1310,
      "text_contrastive_loss": 0.8916,
      "train_positive_log_prob": -80.9411,
      "train_positive_token_accuracy": 0.0766,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3984,
      "epoch": 2.959367945823928,
      "grad_norm": 13.689347267150879,
      "learning_rate": 3.7193421229427017e-06,
      "lm_loss": 5.4342,
      "loss": 1.3539,
      "step": 1311,
      "text_contrastive_loss": 0.8243,
      "train_positive_log_prob": -80.1957,
      "train_positive_token_accuracy": 0.0749,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.4104,
      "epoch": 2.961625282167043,
      "grad_norm": 12.509286880493164,
      "learning_rate": 3.7123300899220193e-06,
      "lm_loss": 5.4652,
      "loss": 1.4173,
      "step": 1312,
      "text_contrastive_loss": 0.9208,
      "train_positive_log_prob": -80.463,
      "train_positive_token_accuracy": 0.0777,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3278,
      "epoch": 2.963882618510158,
      "grad_norm": 13.086407661437988,
      "learning_rate": 3.7053207682690184e-06,
      "lm_loss": 5.4826,
      "loss": 1.248,
      "step": 1313,
      "text_contrastive_loss": 0.7438,
      "train_positive_log_prob": -80.9728,
      "train_positive_token_accuracy": 0.0733,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.3843,
      "epoch": 2.966139954853273,
      "grad_norm": 13.62429428100586,
      "learning_rate": 3.698314172742799e-06,
      "lm_loss": 5.5315,
      "loss": 1.3238,
      "step": 1314,
      "text_contrastive_loss": 0.7727,
      "train_positive_log_prob": -83.8359,
      "train_positive_token_accuracy": 0.0745,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3883,
      "epoch": 2.9683972911963883,
      "grad_norm": 13.54436206817627,
      "learning_rate": 3.691310318096719e-06,
      "lm_loss": 5.4706,
      "loss": 1.3076,
      "step": 1315,
      "text_contrastive_loss": 0.7445,
      "train_positive_log_prob": -81.0807,
      "train_positive_token_accuracy": 0.0793,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.4113,
      "epoch": 2.9706546275395036,
      "grad_norm": 14.356799125671387,
      "learning_rate": 3.684309219078368e-06,
      "lm_loss": 5.4237,
      "loss": 1.3562,
      "step": 1316,
      "text_contrastive_loss": 0.805,
      "train_positive_log_prob": -79.9827,
      "train_positive_token_accuracy": 0.0834,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3692,
      "epoch": 2.9729119638826185,
      "grad_norm": 12.790327072143555,
      "learning_rate": 3.6773108904295294e-06,
      "lm_loss": 5.5083,
      "loss": 1.3468,
      "step": 1317,
      "text_contrastive_loss": 0.8535,
      "train_positive_log_prob": -81.4411,
      "train_positive_token_accuracy": 0.0753,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.4919,
      "epoch": 2.975169300225734,
      "grad_norm": 15.638786315917969,
      "learning_rate": 3.6703153468861585e-06,
      "lm_loss": 5.4922,
      "loss": 1.4634,
      "step": 1318,
      "text_contrastive_loss": 0.8445,
      "train_positive_log_prob": -80.9253,
      "train_positive_token_accuracy": 0.0764,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3833,
      "epoch": 2.9774266365688487,
      "grad_norm": 13.907543182373047,
      "learning_rate": 3.663322603178339e-06,
      "lm_loss": 5.4327,
      "loss": 1.3383,
      "step": 1319,
      "text_contrastive_loss": 0.8236,
      "train_positive_log_prob": -79.5552,
      "train_positive_token_accuracy": 0.0816,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.4456,
      "epoch": 2.979683972911964,
      "grad_norm": 15.450085639953613,
      "learning_rate": 3.6563326740302664e-06,
      "lm_loss": 5.3582,
      "loss": 1.376,
      "step": 1320,
      "text_contrastive_loss": 0.789,
      "train_positive_log_prob": -78.1743,
      "train_positive_token_accuracy": 0.0822,
      "train_positive_token_prob": 0.0325
    },
    {
      "contrastive_loss": 0.5325,
      "epoch": 2.9819413092550793,
      "grad_norm": 15.420104026794434,
      "learning_rate": 3.6493455741602035e-06,
      "lm_loss": 5.3336,
      "loss": 1.4703,
      "step": 1321,
      "text_contrastive_loss": 0.8088,
      "train_positive_log_prob": -76.8342,
      "train_positive_token_accuracy": 0.07,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3127,
      "epoch": 2.984198645598194,
      "grad_norm": 12.451042175292969,
      "learning_rate": 3.642361318280461e-06,
      "lm_loss": 5.3186,
      "loss": 1.1322,
      "step": 1322,
      "text_contrastive_loss": 0.5753,
      "train_positive_log_prob": -77.7446,
      "train_positive_token_accuracy": 0.0826,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.3505,
      "epoch": 2.986455981941309,
      "grad_norm": 14.17666244506836,
      "learning_rate": 3.635379921097359e-06,
      "lm_loss": 5.5185,
      "loss": 1.3039,
      "step": 1323,
      "text_contrastive_loss": 0.803,
      "train_positive_log_prob": -82.8006,
      "train_positive_token_accuracy": 0.0804,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.429,
      "epoch": 2.9887133182844243,
      "grad_norm": 14.575055122375488,
      "learning_rate": 3.6284013973111962e-06,
      "lm_loss": 5.5456,
      "loss": 1.4523,
      "step": 1324,
      "text_contrastive_loss": 0.9376,
      "train_positive_log_prob": -84.5287,
      "train_positive_token_accuracy": 0.0774,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3925,
      "epoch": 2.9909706546275396,
      "grad_norm": 13.344061851501465,
      "learning_rate": 3.621425761616224e-06,
      "lm_loss": 5.4794,
      "loss": 1.4149,
      "step": 1325,
      "text_contrastive_loss": 0.949,
      "train_positive_log_prob": -79.8587,
      "train_positive_token_accuracy": 0.0793,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.3149,
      "epoch": 2.9932279909706545,
      "grad_norm": 12.432286262512207,
      "learning_rate": 3.614453028700613e-06,
      "lm_loss": 5.3891,
      "loss": 1.2165,
      "step": 1326,
      "text_contrastive_loss": 0.7255,
      "train_positive_log_prob": -81.1121,
      "train_positive_token_accuracy": 0.0833,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.4244,
      "epoch": 2.99548532731377,
      "grad_norm": 15.432085037231445,
      "learning_rate": 3.6074832132464165e-06,
      "lm_loss": 5.51,
      "loss": 1.403,
      "step": 1327,
      "text_contrastive_loss": 0.8552,
      "train_positive_log_prob": -80.9323,
      "train_positive_token_accuracy": 0.0758,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.4511,
      "epoch": 2.9977426636568847,
      "grad_norm": 13.709569931030273,
      "learning_rate": 3.600516329929551e-06,
      "lm_loss": 5.5615,
      "loss": 1.5017,
      "step": 1328,
      "text_contrastive_loss": 0.9889,
      "train_positive_log_prob": -82.1145,
      "train_positive_token_accuracy": 0.0833,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.2199,
      "epoch": 3.0,
      "grad_norm": 15.74201488494873,
      "learning_rate": 3.5935523934197537e-06,
      "lm_loss": 5.72,
      "loss": 1.1429,
      "step": 1329,
      "text_contrastive_loss": 0.7019,
      "train_positive_log_prob": -85.5839,
      "train_positive_token_accuracy": 0.0757,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.3658,
      "epoch": 3.0022573363431153,
      "grad_norm": 11.930546760559082,
      "learning_rate": 3.5865914183805606e-06,
      "lm_loss": 5.4776,
      "loss": 1.28,
      "step": 1330,
      "text_contrastive_loss": 0.7328,
      "train_positive_log_prob": -80.9713,
      "train_positive_token_accuracy": 0.0832,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.3738,
      "epoch": 3.00451467268623,
      "grad_norm": 13.340158462524414,
      "learning_rate": 3.5796334194692704e-06,
      "lm_loss": 5.5184,
      "loss": 1.3609,
      "step": 1331,
      "text_contrastive_loss": 0.8704,
      "train_positive_log_prob": -80.6185,
      "train_positive_token_accuracy": 0.0824,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.3602,
      "epoch": 3.0067720090293455,
      "grad_norm": 13.094389915466309,
      "learning_rate": 3.572678411336916e-06,
      "lm_loss": 5.5166,
      "loss": 1.3185,
      "step": 1332,
      "text_contrastive_loss": 0.8133,
      "train_positive_log_prob": -83.9795,
      "train_positive_token_accuracy": 0.0811,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3618,
      "epoch": 3.0090293453724604,
      "grad_norm": 11.701364517211914,
      "learning_rate": 3.5657264086282317e-06,
      "lm_loss": 5.5034,
      "loss": 1.3167,
      "step": 1333,
      "text_contrastive_loss": 0.8092,
      "train_positive_log_prob": -80.2494,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3653,
      "epoch": 3.0112866817155757,
      "grad_norm": 12.613174438476562,
      "learning_rate": 3.5587774259816234e-06,
      "lm_loss": 5.5188,
      "loss": 1.2239,
      "step": 1334,
      "text_contrastive_loss": 0.6134,
      "train_positive_log_prob": -83.8572,
      "train_positive_token_accuracy": 0.0784,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.4612,
      "epoch": 3.0135440180586905,
      "grad_norm": 13.1723051071167,
      "learning_rate": 3.5518314780291384e-06,
      "lm_loss": 5.3698,
      "loss": 1.4429,
      "step": 1335,
      "text_contrastive_loss": 0.8895,
      "train_positive_log_prob": -77.969,
      "train_positive_token_accuracy": 0.0853,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.3863,
      "epoch": 3.015801354401806,
      "grad_norm": 14.11274528503418,
      "learning_rate": 3.544888579396435e-06,
      "lm_loss": 5.5404,
      "loss": 1.4009,
      "step": 1336,
      "text_contrastive_loss": 0.9211,
      "train_positive_log_prob": -82.5646,
      "train_positive_token_accuracy": 0.0818,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.4541,
      "epoch": 3.018058690744921,
      "grad_norm": 13.539448738098145,
      "learning_rate": 3.5379487447027483e-06,
      "lm_loss": 5.4008,
      "loss": 1.3906,
      "step": 1337,
      "text_contrastive_loss": 0.7927,
      "train_positive_log_prob": -79.4109,
      "train_positive_token_accuracy": 0.084,
      "train_positive_token_prob": 0.0328
    },
    {
      "contrastive_loss": 0.3204,
      "epoch": 3.020316027088036,
      "grad_norm": 13.184925079345703,
      "learning_rate": 3.5310119885608625e-06,
      "lm_loss": 5.4394,
      "loss": 1.201,
      "step": 1338,
      "text_contrastive_loss": 0.6732,
      "train_positive_log_prob": -81.7632,
      "train_positive_token_accuracy": 0.0845,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3226,
      "epoch": 3.0225733634311513,
      "grad_norm": 12.389911651611328,
      "learning_rate": 3.524078325577084e-06,
      "lm_loss": 5.4582,
      "loss": 1.2632,
      "step": 1339,
      "text_contrastive_loss": 0.7895,
      "train_positive_log_prob": -79.844,
      "train_positive_token_accuracy": 0.0745,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.4028,
      "epoch": 3.024830699774266,
      "grad_norm": 12.883729934692383,
      "learning_rate": 3.517147770351199e-06,
      "lm_loss": 5.5024,
      "loss": 1.2784,
      "step": 1340,
      "text_contrastive_loss": 0.6506,
      "train_positive_log_prob": -83.6307,
      "train_positive_token_accuracy": 0.0771,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3951,
      "epoch": 3.0270880361173815,
      "grad_norm": 13.757779121398926,
      "learning_rate": 3.5102203374764555e-06,
      "lm_loss": 5.4794,
      "loss": 1.3501,
      "step": 1341,
      "text_contrastive_loss": 0.8143,
      "train_positive_log_prob": -81.5234,
      "train_positive_token_accuracy": 0.0759,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.3902,
      "epoch": 3.0293453724604964,
      "grad_norm": 13.152948379516602,
      "learning_rate": 3.503296041539522e-06,
      "lm_loss": 5.463,
      "loss": 1.3479,
      "step": 1342,
      "text_contrastive_loss": 0.8229,
      "train_positive_log_prob": -79.1949,
      "train_positive_token_accuracy": 0.0825,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3784,
      "epoch": 3.0316027088036117,
      "grad_norm": 14.146117210388184,
      "learning_rate": 3.496374897120467e-06,
      "lm_loss": 5.3706,
      "loss": 1.3808,
      "step": 1343,
      "text_contrastive_loss": 0.9308,
      "train_positive_log_prob": -79.1308,
      "train_positive_token_accuracy": 0.0769,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.339,
      "epoch": 3.033860045146727,
      "grad_norm": 13.40596866607666,
      "learning_rate": 3.4894569187927204e-06,
      "lm_loss": 5.4452,
      "loss": 1.2462,
      "step": 1344,
      "text_contrastive_loss": 0.7253,
      "train_positive_log_prob": -81.1677,
      "train_positive_token_accuracy": 0.0747,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3532,
      "epoch": 3.036117381489842,
      "grad_norm": 11.866280555725098,
      "learning_rate": 3.4825421211230437e-06,
      "lm_loss": 5.592,
      "loss": 1.2654,
      "step": 1345,
      "text_contrastive_loss": 0.706,
      "train_positive_log_prob": -81.9996,
      "train_positive_token_accuracy": 0.0794,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.3521,
      "epoch": 3.038374717832957,
      "grad_norm": 13.236581802368164,
      "learning_rate": 3.4756305186715046e-06,
      "lm_loss": 5.6645,
      "loss": 1.3594,
      "step": 1346,
      "text_contrastive_loss": 0.8817,
      "train_positive_log_prob": -82.9983,
      "train_positive_token_accuracy": 0.0762,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.3935,
      "epoch": 3.040632054176072,
      "grad_norm": 13.178022384643555,
      "learning_rate": 3.4687221259914394e-06,
      "lm_loss": 5.3523,
      "loss": 1.2888,
      "step": 1347,
      "text_contrastive_loss": 0.72,
      "train_positive_log_prob": -79.9937,
      "train_positive_token_accuracy": 0.087,
      "train_positive_token_prob": 0.0327
    },
    {
      "contrastive_loss": 0.351,
      "epoch": 3.0428893905191874,
      "grad_norm": 12.475351333618164,
      "learning_rate": 3.461816957629429e-06,
      "lm_loss": 5.4678,
      "loss": 1.2952,
      "step": 1348,
      "text_contrastive_loss": 0.7947,
      "train_positive_log_prob": -80.0998,
      "train_positive_token_accuracy": 0.0781,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3867,
      "epoch": 3.0451467268623027,
      "grad_norm": 13.979798316955566,
      "learning_rate": 3.4549150281252635e-06,
      "lm_loss": 5.4229,
      "loss": 1.2811,
      "step": 1349,
      "text_contrastive_loss": 0.7044,
      "train_positive_log_prob": -82.2965,
      "train_positive_token_accuracy": 0.0806,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.4382,
      "epoch": 3.0474040632054176,
      "grad_norm": 14.120234489440918,
      "learning_rate": 3.448016352011914e-06,
      "lm_loss": 5.4526,
      "loss": 1.3429,
      "step": 1350,
      "text_contrastive_loss": 0.7189,
      "train_positive_log_prob": -82.156,
      "train_positive_token_accuracy": 0.0812,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3654,
      "epoch": 3.049661399548533,
      "grad_norm": 13.496257781982422,
      "learning_rate": 3.441120943815497e-06,
      "lm_loss": 5.519,
      "loss": 1.2871,
      "step": 1351,
      "text_contrastive_loss": 0.7395,
      "train_positive_log_prob": -81.361,
      "train_positive_token_accuracy": 0.0782,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3403,
      "epoch": 3.0519187358916477,
      "grad_norm": 12.333660125732422,
      "learning_rate": 3.4342288180552556e-06,
      "lm_loss": 5.473,
      "loss": 1.2689,
      "step": 1352,
      "text_contrastive_loss": 0.7626,
      "train_positive_log_prob": -82.3552,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.3127,
      "epoch": 3.054176072234763,
      "grad_norm": 13.125408172607422,
      "learning_rate": 3.427339989243514e-06,
      "lm_loss": 5.5455,
      "loss": 1.2053,
      "step": 1353,
      "text_contrastive_loss": 0.6761,
      "train_positive_log_prob": -82.2894,
      "train_positive_token_accuracy": 0.0803,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3042,
      "epoch": 3.056433408577878,
      "grad_norm": 12.157618522644043,
      "learning_rate": 3.420454471885659e-06,
      "lm_loss": 5.4026,
      "loss": 1.1991,
      "step": 1354,
      "text_contrastive_loss": 0.7093,
      "train_positive_log_prob": -80.3646,
      "train_positive_token_accuracy": 0.0798,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.3313,
      "epoch": 3.0586907449209932,
      "grad_norm": 12.843050956726074,
      "learning_rate": 3.4135722804801004e-06,
      "lm_loss": 5.5239,
      "loss": 1.3181,
      "step": 1355,
      "text_contrastive_loss": 0.8687,
      "train_positive_log_prob": -83.3554,
      "train_positive_token_accuracy": 0.0795,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.4004,
      "epoch": 3.0609480812641086,
      "grad_norm": 13.376944541931152,
      "learning_rate": 3.4066934295182496e-06,
      "lm_loss": 5.486,
      "loss": 1.3579,
      "step": 1356,
      "text_contrastive_loss": 0.8178,
      "train_positive_log_prob": -80.9617,
      "train_positive_token_accuracy": 0.0757,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.2915,
      "epoch": 3.0632054176072234,
      "grad_norm": 11.282021522521973,
      "learning_rate": 3.3998179334844823e-06,
      "lm_loss": 5.434,
      "loss": 1.1963,
      "step": 1357,
      "text_contrastive_loss": 0.7228,
      "train_positive_log_prob": -79.7677,
      "train_positive_token_accuracy": 0.0809,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.3228,
      "epoch": 3.0654627539503387,
      "grad_norm": 12.650505065917969,
      "learning_rate": 3.3929458068561073e-06,
      "lm_loss": 5.4454,
      "loss": 1.2475,
      "step": 1358,
      "text_contrastive_loss": 0.7602,
      "train_positive_log_prob": -80.2805,
      "train_positive_token_accuracy": 0.0763,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.2636,
      "epoch": 3.0677200902934536,
      "grad_norm": 9.988433837890625,
      "learning_rate": 3.3860770641033417e-06,
      "lm_loss": 5.4381,
      "loss": 1.1262,
      "step": 1359,
      "text_contrastive_loss": 0.6376,
      "train_positive_log_prob": -80.5692,
      "train_positive_token_accuracy": 0.0838,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.3212,
      "epoch": 3.069977426636569,
      "grad_norm": 11.848210334777832,
      "learning_rate": 3.379211719689278e-06,
      "lm_loss": 5.594,
      "loss": 1.2387,
      "step": 1360,
      "text_contrastive_loss": 0.7162,
      "train_positive_log_prob": -82.9027,
      "train_positive_token_accuracy": 0.0783,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3626,
      "epoch": 3.072234762979684,
      "grad_norm": 10.95758056640625,
      "learning_rate": 3.37234978806985e-06,
      "lm_loss": 5.3868,
      "loss": 1.2913,
      "step": 1361,
      "text_contrastive_loss": 0.7801,
      "train_positive_log_prob": -79.472,
      "train_positive_token_accuracy": 0.0786,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3916,
      "epoch": 3.074492099322799,
      "grad_norm": 13.328849792480469,
      "learning_rate": 3.365491283693807e-06,
      "lm_loss": 5.3945,
      "loss": 1.3312,
      "step": 1362,
      "text_contrastive_loss": 0.8004,
      "train_positive_log_prob": -79.738,
      "train_positive_token_accuracy": 0.07,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.3169,
      "epoch": 3.0767494356659144,
      "grad_norm": 13.587141036987305,
      "learning_rate": 3.358636221002682e-06,
      "lm_loss": 5.6589,
      "loss": 1.2384,
      "step": 1363,
      "text_contrastive_loss": 0.7112,
      "train_positive_log_prob": -82.4115,
      "train_positive_token_accuracy": 0.0716,
      "train_positive_token_prob": 0.0282
    },
    {
      "contrastive_loss": 0.3051,
      "epoch": 3.0790067720090293,
      "grad_norm": 12.500545501708984,
      "learning_rate": 3.351784614430761e-06,
      "lm_loss": 5.5049,
      "loss": 1.1483,
      "step": 1364,
      "text_contrastive_loss": 0.5855,
      "train_positive_log_prob": -81.1327,
      "train_positive_token_accuracy": 0.0792,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3559,
      "epoch": 3.0812641083521446,
      "grad_norm": 12.103687286376953,
      "learning_rate": 3.3449364784050515e-06,
      "lm_loss": 5.3643,
      "loss": 1.3086,
      "step": 1365,
      "text_contrastive_loss": 0.8326,
      "train_positive_log_prob": -77.4453,
      "train_positive_token_accuracy": 0.085,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3158,
      "epoch": 3.0835214446952595,
      "grad_norm": 11.751809120178223,
      "learning_rate": 3.3380918273452557e-06,
      "lm_loss": 5.5469,
      "loss": 1.2403,
      "step": 1366,
      "text_contrastive_loss": 0.7396,
      "train_positive_log_prob": -80.1322,
      "train_positive_token_accuracy": 0.08,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3473,
      "epoch": 3.0857787810383748,
      "grad_norm": 12.667675018310547,
      "learning_rate": 3.3312506756637343e-06,
      "lm_loss": 5.4027,
      "loss": 1.2416,
      "step": 1367,
      "text_contrastive_loss": 0.708,
      "train_positive_log_prob": -79.3535,
      "train_positive_token_accuracy": 0.0856,
      "train_positive_token_prob": 0.0336
    },
    {
      "contrastive_loss": 0.3494,
      "epoch": 3.0880361173814896,
      "grad_norm": 12.610095977783203,
      "learning_rate": 3.324413037765483e-06,
      "lm_loss": 5.4826,
      "loss": 1.2818,
      "step": 1368,
      "text_contrastive_loss": 0.7682,
      "train_positive_log_prob": -81.7262,
      "train_positive_token_accuracy": 0.0731,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.3474,
      "epoch": 3.090293453724605,
      "grad_norm": 12.490227699279785,
      "learning_rate": 3.317578928048096e-06,
      "lm_loss": 5.4635,
      "loss": 1.2691,
      "step": 1369,
      "text_contrastive_loss": 0.7506,
      "train_positive_log_prob": -79.4187,
      "train_positive_token_accuracy": 0.0801,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3186,
      "epoch": 3.0925507900677203,
      "grad_norm": 11.91494083404541,
      "learning_rate": 3.310748360901741e-06,
      "lm_loss": 5.4459,
      "loss": 1.2483,
      "step": 1370,
      "text_contrastive_loss": 0.7703,
      "train_positive_log_prob": -82.114,
      "train_positive_token_accuracy": 0.0746,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.2689,
      "epoch": 3.094808126410835,
      "grad_norm": 11.223448753356934,
      "learning_rate": 3.303921350709124e-06,
      "lm_loss": 5.4636,
      "loss": 1.1267,
      "step": 1371,
      "text_contrastive_loss": 0.623,
      "train_positive_log_prob": -81.017,
      "train_positive_token_accuracy": 0.0822,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3487,
      "epoch": 3.0970654627539504,
      "grad_norm": 13.478902816772461,
      "learning_rate": 3.2970979118454616e-06,
      "lm_loss": 5.3685,
      "loss": 1.2357,
      "step": 1372,
      "text_contrastive_loss": 0.7001,
      "train_positive_log_prob": -77.8859,
      "train_positive_token_accuracy": 0.073,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3186,
      "epoch": 3.0993227990970653,
      "grad_norm": 12.863323211669922,
      "learning_rate": 3.2902780586784542e-06,
      "lm_loss": 5.5203,
      "loss": 1.2499,
      "step": 1373,
      "text_contrastive_loss": 0.7586,
      "train_positive_log_prob": -83.4758,
      "train_positive_token_accuracy": 0.0821,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.4542,
      "epoch": 3.1015801354401806,
      "grad_norm": 13.092292785644531,
      "learning_rate": 3.283461805568246e-06,
      "lm_loss": 5.558,
      "loss": 1.4925,
      "step": 1374,
      "text_contrastive_loss": 0.965,
      "train_positive_log_prob": -80.5689,
      "train_positive_token_accuracy": 0.0746,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.3399,
      "epoch": 3.1038374717832955,
      "grad_norm": 13.868317604064941,
      "learning_rate": 3.276649166867406e-06,
      "lm_loss": 5.4346,
      "loss": 1.2537,
      "step": 1375,
      "text_contrastive_loss": 0.7407,
      "train_positive_log_prob": -81.5861,
      "train_positive_token_accuracy": 0.0807,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.277,
      "epoch": 3.106094808126411,
      "grad_norm": 13.915228843688965,
      "learning_rate": 3.2698401569208883e-06,
      "lm_loss": 5.3979,
      "loss": 1.1634,
      "step": 1376,
      "text_contrastive_loss": 0.6931,
      "train_positive_log_prob": -78.2482,
      "train_positive_token_accuracy": 0.0822,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.3218,
      "epoch": 3.108352144469526,
      "grad_norm": 12.190874099731445,
      "learning_rate": 3.2630347900660094e-06,
      "lm_loss": 5.4945,
      "loss": 1.2482,
      "step": 1377,
      "text_contrastive_loss": 0.7537,
      "train_positive_log_prob": -84.8713,
      "train_positive_token_accuracy": 0.0736,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.393,
      "epoch": 3.110609480812641,
      "grad_norm": 16.03776741027832,
      "learning_rate": 3.256233080632414e-06,
      "lm_loss": 5.4982,
      "loss": 1.2598,
      "step": 1378,
      "text_contrastive_loss": 0.6339,
      "train_positive_log_prob": -80.8671,
      "train_positive_token_accuracy": 0.0752,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3224,
      "epoch": 3.1128668171557563,
      "grad_norm": 12.586856842041016,
      "learning_rate": 3.249435042942043e-06,
      "lm_loss": 5.5305,
      "loss": 1.2621,
      "step": 1379,
      "text_contrastive_loss": 0.7733,
      "train_positive_log_prob": -82.0357,
      "train_positive_token_accuracy": 0.0777,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3985,
      "epoch": 3.115124153498871,
      "grad_norm": 14.891951560974121,
      "learning_rate": 3.242640691309111e-06,
      "lm_loss": 5.5164,
      "loss": 1.2823,
      "step": 1380,
      "text_contrastive_loss": 0.6644,
      "train_positive_log_prob": -81.1738,
      "train_positive_token_accuracy": 0.0847,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.4055,
      "epoch": 3.1173814898419865,
      "grad_norm": 12.801090240478516,
      "learning_rate": 3.235850040040066e-06,
      "lm_loss": 5.5175,
      "loss": 1.3029,
      "step": 1381,
      "text_contrastive_loss": 0.6913,
      "train_positive_log_prob": -80.9535,
      "train_positive_token_accuracy": 0.0692,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.3894,
      "epoch": 3.119638826185102,
      "grad_norm": 12.925514221191406,
      "learning_rate": 3.2290631034335684e-06,
      "lm_loss": 5.5601,
      "loss": 1.3589,
      "step": 1382,
      "text_contrastive_loss": 0.827,
      "train_positive_log_prob": -82.1636,
      "train_positive_token_accuracy": 0.0809,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.3831,
      "epoch": 3.1218961625282167,
      "grad_norm": 14.680846214294434,
      "learning_rate": 3.2222798957804524e-06,
      "lm_loss": 5.4835,
      "loss": 1.3201,
      "step": 1383,
      "text_contrastive_loss": 0.7772,
      "train_positive_log_prob": -80.6224,
      "train_positive_token_accuracy": 0.081,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.4091,
      "epoch": 3.124153498871332,
      "grad_norm": 13.382808685302734,
      "learning_rate": 3.215500431363706e-06,
      "lm_loss": 5.3571,
      "loss": 1.3014,
      "step": 1384,
      "text_contrastive_loss": 0.7131,
      "train_positive_log_prob": -81.1287,
      "train_positive_token_accuracy": 0.0851,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.3998,
      "epoch": 3.126410835214447,
      "grad_norm": 14.403979301452637,
      "learning_rate": 3.20872472445843e-06,
      "lm_loss": 5.4311,
      "loss": 1.3408,
      "step": 1385,
      "text_contrastive_loss": 0.7959,
      "train_positive_log_prob": -81.9782,
      "train_positive_token_accuracy": 0.0795,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.4187,
      "epoch": 3.128668171557562,
      "grad_norm": 13.64614200592041,
      "learning_rate": 3.2019527893318177e-06,
      "lm_loss": 5.4454,
      "loss": 1.3283,
      "step": 1386,
      "text_contrastive_loss": 0.73,
      "train_positive_log_prob": -80.7853,
      "train_positive_token_accuracy": 0.08,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3547,
      "epoch": 3.130925507900677,
      "grad_norm": 13.358538627624512,
      "learning_rate": 3.195184640243115e-06,
      "lm_loss": 5.389,
      "loss": 1.2771,
      "step": 1387,
      "text_contrastive_loss": 0.7669,
      "train_positive_log_prob": -79.9754,
      "train_positive_token_accuracy": 0.0811,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.4172,
      "epoch": 3.1331828442437923,
      "grad_norm": 13.655101776123047,
      "learning_rate": 3.1884202914436024e-06,
      "lm_loss": 5.4035,
      "loss": 1.3725,
      "step": 1388,
      "text_contrastive_loss": 0.83,
      "train_positive_log_prob": -78.4975,
      "train_positive_token_accuracy": 0.0777,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3278,
      "epoch": 3.1354401805869077,
      "grad_norm": 13.055240631103516,
      "learning_rate": 3.1816597571765517e-06,
      "lm_loss": 5.4432,
      "loss": 1.26,
      "step": 1389,
      "text_contrastive_loss": 0.7758,
      "train_positive_log_prob": -78.3704,
      "train_positive_token_accuracy": 0.0705,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.3732,
      "epoch": 3.1376975169300225,
      "grad_norm": 12.419684410095215,
      "learning_rate": 3.1749030516772084e-06,
      "lm_loss": 5.4468,
      "loss": 1.2938,
      "step": 1390,
      "text_contrastive_loss": 0.7517,
      "train_positive_log_prob": -78.2568,
      "train_positive_token_accuracy": 0.0791,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.315,
      "epoch": 3.139954853273138,
      "grad_norm": 12.234460830688477,
      "learning_rate": 3.168150189172754e-06,
      "lm_loss": 5.5778,
      "loss": 1.3066,
      "step": 1391,
      "text_contrastive_loss": 0.8676,
      "train_positive_log_prob": -82.2528,
      "train_positive_token_accuracy": 0.0758,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.42,
      "epoch": 3.1422121896162527,
      "grad_norm": 15.418965339660645,
      "learning_rate": 3.1614011838822755e-06,
      "lm_loss": 5.5274,
      "loss": 1.3568,
      "step": 1392,
      "text_contrastive_loss": 0.7681,
      "train_positive_log_prob": -85.1195,
      "train_positive_token_accuracy": 0.0852,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.35,
      "epoch": 3.144469525959368,
      "grad_norm": 12.588700294494629,
      "learning_rate": 3.154656050016742e-06,
      "lm_loss": 5.4606,
      "loss": 1.3457,
      "step": 1393,
      "text_contrastive_loss": 0.8991,
      "train_positive_log_prob": -77.8964,
      "train_positive_token_accuracy": 0.072,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.4081,
      "epoch": 3.146726862302483,
      "grad_norm": 12.298941612243652,
      "learning_rate": 3.1479148017789673e-06,
      "lm_loss": 5.4516,
      "loss": 1.3663,
      "step": 1394,
      "text_contrastive_loss": 0.8262,
      "train_positive_log_prob": -80.6229,
      "train_positive_token_accuracy": 0.0827,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3261,
      "epoch": 3.148984198645598,
      "grad_norm": 12.022645950317383,
      "learning_rate": 3.1411774533635854e-06,
      "lm_loss": 5.4987,
      "loss": 1.2973,
      "step": 1395,
      "text_contrastive_loss": 0.8427,
      "train_positive_log_prob": -80.6333,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3343,
      "epoch": 3.1512415349887135,
      "grad_norm": 13.233353614807129,
      "learning_rate": 3.134444018957019e-06,
      "lm_loss": 5.4817,
      "loss": 1.2546,
      "step": 1396,
      "text_contrastive_loss": 0.7443,
      "train_positive_log_prob": -80.0181,
      "train_positive_token_accuracy": 0.0757,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3314,
      "epoch": 3.1534988713318284,
      "grad_norm": 12.329866409301758,
      "learning_rate": 3.1277145127374475e-06,
      "lm_loss": 5.5327,
      "loss": 1.3321,
      "step": 1397,
      "text_contrastive_loss": 0.8949,
      "train_positive_log_prob": -81.3933,
      "train_positive_token_accuracy": 0.0786,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3466,
      "epoch": 3.1557562076749437,
      "grad_norm": 12.84872055053711,
      "learning_rate": 3.1209889488747813e-06,
      "lm_loss": 5.5437,
      "loss": 1.2957,
      "step": 1398,
      "text_contrastive_loss": 0.7895,
      "train_positive_log_prob": -80.0808,
      "train_positive_token_accuracy": 0.0749,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.4117,
      "epoch": 3.1580135440180586,
      "grad_norm": 12.506523132324219,
      "learning_rate": 3.114267341530627e-06,
      "lm_loss": 5.3671,
      "loss": 1.3359,
      "step": 1399,
      "text_contrastive_loss": 0.7751,
      "train_positive_log_prob": -78.6646,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.0327
    },
    {
      "contrastive_loss": 0.3116,
      "epoch": 3.160270880361174,
      "grad_norm": 10.065068244934082,
      "learning_rate": 3.1075497048582635e-06,
      "lm_loss": 5.3999,
      "loss": 1.1763,
      "step": 1400,
      "text_contrastive_loss": 0.6494,
      "train_positive_log_prob": -79.4544,
      "train_positive_token_accuracy": 0.0751,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.4121,
      "epoch": 3.1625282167042887,
      "grad_norm": 14.337026596069336,
      "learning_rate": 3.1008360530026053e-06,
      "lm_loss": 5.4837,
      "loss": 1.4533,
      "step": 1401,
      "text_contrastive_loss": 0.9857,
      "train_positive_log_prob": -80.9288,
      "train_positive_token_accuracy": 0.0762,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3098,
      "epoch": 3.164785553047404,
      "grad_norm": 10.524683952331543,
      "learning_rate": 3.0941264001001796e-06,
      "lm_loss": 5.3857,
      "loss": 1.2367,
      "step": 1402,
      "text_contrastive_loss": 0.7766,
      "train_positive_log_prob": -77.7547,
      "train_positive_token_accuracy": 0.0809,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3242,
      "epoch": 3.1670428893905194,
      "grad_norm": 11.856847763061523,
      "learning_rate": 3.0874207602790895e-06,
      "lm_loss": 5.3839,
      "loss": 1.2118,
      "step": 1403,
      "text_contrastive_loss": 0.6985,
      "train_positive_log_prob": -80.0418,
      "train_positive_token_accuracy": 0.0875,
      "train_positive_token_prob": 0.0327
    },
    {
      "contrastive_loss": 0.3542,
      "epoch": 3.1693002257336342,
      "grad_norm": 12.487991333007812,
      "learning_rate": 3.0807191476589926e-06,
      "lm_loss": 5.5494,
      "loss": 1.2514,
      "step": 1404,
      "text_contrastive_loss": 0.6846,
      "train_positive_log_prob": -79.9678,
      "train_positive_token_accuracy": 0.0718,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.33,
      "epoch": 3.1715575620767495,
      "grad_norm": 11.98621940612793,
      "learning_rate": 3.0740215763510617e-06,
      "lm_loss": 5.4704,
      "loss": 1.1744,
      "step": 1405,
      "text_contrastive_loss": 0.5947,
      "train_positive_log_prob": -82.208,
      "train_positive_token_accuracy": 0.0847,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3767,
      "epoch": 3.1738148984198644,
      "grad_norm": 12.796917915344238,
      "learning_rate": 3.0673280604579623e-06,
      "lm_loss": 5.3512,
      "loss": 1.2458,
      "step": 1406,
      "text_contrastive_loss": 0.6681,
      "train_positive_log_prob": -80.8947,
      "train_positive_token_accuracy": 0.0778,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3847,
      "epoch": 3.1760722347629797,
      "grad_norm": 13.427523612976074,
      "learning_rate": 3.0606386140738253e-06,
      "lm_loss": 5.4488,
      "loss": 1.3555,
      "step": 1407,
      "text_contrastive_loss": 0.8518,
      "train_positive_log_prob": -79.1584,
      "train_positive_token_accuracy": 0.0836,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.3398,
      "epoch": 3.1783295711060946,
      "grad_norm": 13.557880401611328,
      "learning_rate": 3.053953251284205e-06,
      "lm_loss": 5.5605,
      "loss": 1.2408,
      "step": 1408,
      "text_contrastive_loss": 0.6898,
      "train_positive_log_prob": -82.2353,
      "train_positive_token_accuracy": 0.0816,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3556,
      "epoch": 3.18058690744921,
      "grad_norm": 14.373702049255371,
      "learning_rate": 3.047271986166061e-06,
      "lm_loss": 5.4678,
      "loss": 1.2604,
      "step": 1409,
      "text_contrastive_loss": 0.716,
      "train_positive_log_prob": -81.8427,
      "train_positive_token_accuracy": 0.0718,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.345,
      "epoch": 3.1828442437923252,
      "grad_norm": 11.91468334197998,
      "learning_rate": 3.0405948327877233e-06,
      "lm_loss": 5.3607,
      "loss": 1.2759,
      "step": 1410,
      "text_contrastive_loss": 0.7897,
      "train_positive_log_prob": -80.8505,
      "train_positive_token_accuracy": 0.0847,
      "train_positive_token_prob": 0.033
    },
    {
      "contrastive_loss": 0.4006,
      "epoch": 3.18510158013544,
      "grad_norm": 13.399833679199219,
      "learning_rate": 3.033921805208867e-06,
      "lm_loss": 5.3878,
      "loss": 1.381,
      "step": 1411,
      "text_contrastive_loss": 0.8832,
      "train_positive_log_prob": -79.133,
      "train_positive_token_accuracy": 0.0947,
      "train_positive_token_prob": 0.0336
    },
    {
      "contrastive_loss": 0.3332,
      "epoch": 3.1873589164785554,
      "grad_norm": 12.890466690063477,
      "learning_rate": 3.027252917480476e-06,
      "lm_loss": 5.4102,
      "loss": 1.3251,
      "step": 1412,
      "text_contrastive_loss": 0.9019,
      "train_positive_log_prob": -81.1528,
      "train_positive_token_accuracy": 0.0794,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.385,
      "epoch": 3.1896162528216703,
      "grad_norm": 12.853949546813965,
      "learning_rate": 3.0205881836448186e-06,
      "lm_loss": 5.5684,
      "loss": 1.3291,
      "step": 1413,
      "text_contrastive_loss": 0.7746,
      "train_positive_log_prob": -81.4836,
      "train_positive_token_accuracy": 0.0766,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.3598,
      "epoch": 3.1918735891647856,
      "grad_norm": 12.92371654510498,
      "learning_rate": 3.0139276177354188e-06,
      "lm_loss": 5.5097,
      "loss": 1.2993,
      "step": 1414,
      "text_contrastive_loss": 0.777,
      "train_positive_log_prob": -83.2085,
      "train_positive_token_accuracy": 0.0763,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.4402,
      "epoch": 3.194130925507901,
      "grad_norm": 13.475192070007324,
      "learning_rate": 3.00727123377702e-06,
      "lm_loss": 5.5492,
      "loss": 1.4162,
      "step": 1415,
      "text_contrastive_loss": 0.842,
      "train_positive_log_prob": -81.8248,
      "train_positive_token_accuracy": 0.0803,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3644,
      "epoch": 3.1963882618510158,
      "grad_norm": 12.008026123046875,
      "learning_rate": 3.0006190457855643e-06,
      "lm_loss": 5.601,
      "loss": 1.2532,
      "step": 1416,
      "text_contrastive_loss": 0.6573,
      "train_positive_log_prob": -83.509,
      "train_positive_token_accuracy": 0.0783,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3773,
      "epoch": 3.198645598194131,
      "grad_norm": 14.657004356384277,
      "learning_rate": 2.9939710677681545e-06,
      "lm_loss": 5.381,
      "loss": 1.366,
      "step": 1417,
      "text_contrastive_loss": 0.9011,
      "train_positive_log_prob": -78.0458,
      "train_positive_token_accuracy": 0.0883,
      "train_positive_token_prob": 0.0329
    },
    {
      "contrastive_loss": 0.3871,
      "epoch": 3.200902934537246,
      "grad_norm": 13.67013931274414,
      "learning_rate": 2.987327313723033e-06,
      "lm_loss": 5.4455,
      "loss": 1.3054,
      "step": 1418,
      "text_contrastive_loss": 0.7476,
      "train_positive_log_prob": -80.408,
      "train_positive_token_accuracy": 0.0836,
      "train_positive_token_prob": 0.0326
    },
    {
      "contrastive_loss": 0.2903,
      "epoch": 3.2031602708803613,
      "grad_norm": 11.551298141479492,
      "learning_rate": 2.980687797639543e-06,
      "lm_loss": 5.5571,
      "loss": 1.213,
      "step": 1419,
      "text_contrastive_loss": 0.734,
      "train_positive_log_prob": -82.0749,
      "train_positive_token_accuracy": 0.0733,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.2487,
      "epoch": 3.205417607223476,
      "grad_norm": 12.48673152923584,
      "learning_rate": 2.9740525334981105e-06,
      "lm_loss": 5.4923,
      "loss": 1.0882,
      "step": 1420,
      "text_contrastive_loss": 0.5806,
      "train_positive_log_prob": -82.8596,
      "train_positive_token_accuracy": 0.0795,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.3514,
      "epoch": 3.2076749435665914,
      "grad_norm": 12.990463256835938,
      "learning_rate": 2.967421535270203e-06,
      "lm_loss": 5.4214,
      "loss": 1.2665,
      "step": 1421,
      "text_contrastive_loss": 0.7459,
      "train_positive_log_prob": -80.2828,
      "train_positive_token_accuracy": 0.0778,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.4896,
      "epoch": 3.2099322799097068,
      "grad_norm": 14.413372993469238,
      "learning_rate": 2.9607948169183077e-06,
      "lm_loss": 5.4621,
      "loss": 1.508,
      "step": 1422,
      "text_contrastive_loss": 0.9445,
      "train_positive_log_prob": -81.3139,
      "train_positive_token_accuracy": 0.0769,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3271,
      "epoch": 3.2121896162528216,
      "grad_norm": 11.85389232635498,
      "learning_rate": 2.9541723923958975e-06,
      "lm_loss": 5.5029,
      "loss": 1.2189,
      "step": 1423,
      "text_contrastive_loss": 0.6829,
      "train_positive_log_prob": -80.7847,
      "train_positive_token_accuracy": 0.0772,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.3824,
      "epoch": 3.214446952595937,
      "grad_norm": 14.24267864227295,
      "learning_rate": 2.94755427564741e-06,
      "lm_loss": 5.4803,
      "loss": 1.2923,
      "step": 1424,
      "text_contrastive_loss": 0.7238,
      "train_positive_log_prob": -79.1848,
      "train_positive_token_accuracy": 0.0714,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3116,
      "epoch": 3.216704288939052,
      "grad_norm": 14.309062004089355,
      "learning_rate": 2.9409404806082077e-06,
      "lm_loss": 5.4512,
      "loss": 1.2347,
      "step": 1425,
      "text_contrastive_loss": 0.756,
      "train_positive_log_prob": -80.8562,
      "train_positive_token_accuracy": 0.0801,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.3572,
      "epoch": 3.218961625282167,
      "grad_norm": 12.542322158813477,
      "learning_rate": 2.934331021204551e-06,
      "lm_loss": 5.2986,
      "loss": 1.2568,
      "step": 1426,
      "text_contrastive_loss": 0.7394,
      "train_positive_log_prob": -76.5596,
      "train_positive_token_accuracy": 0.0816,
      "train_positive_token_prob": 0.0326
    },
    {
      "contrastive_loss": 0.2844,
      "epoch": 3.221218961625282,
      "grad_norm": 12.496670722961426,
      "learning_rate": 2.9277259113535774e-06,
      "lm_loss": 5.3349,
      "loss": 1.1623,
      "step": 1427,
      "text_contrastive_loss": 0.6887,
      "train_positive_log_prob": -78.5639,
      "train_positive_token_accuracy": 0.0867,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.3231,
      "epoch": 3.2234762979683973,
      "grad_norm": 11.824939727783203,
      "learning_rate": 2.9211251649632587e-06,
      "lm_loss": 5.3371,
      "loss": 1.2538,
      "step": 1428,
      "text_contrastive_loss": 0.7941,
      "train_positive_log_prob": -77.8502,
      "train_positive_token_accuracy": 0.0803,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3712,
      "epoch": 3.2257336343115126,
      "grad_norm": 11.644871711730957,
      "learning_rate": 2.9145287959323852e-06,
      "lm_loss": 5.3979,
      "loss": 1.2941,
      "step": 1429,
      "text_contrastive_loss": 0.7664,
      "train_positive_log_prob": -79.6593,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3841,
      "epoch": 3.2279909706546275,
      "grad_norm": 13.726497650146484,
      "learning_rate": 2.9079368181505263e-06,
      "lm_loss": 5.4403,
      "loss": 1.297,
      "step": 1430,
      "text_contrastive_loss": 0.7378,
      "train_positive_log_prob": -81.8983,
      "train_positive_token_accuracy": 0.0742,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3483,
      "epoch": 3.230248306997743,
      "grad_norm": 12.534523963928223,
      "learning_rate": 2.9013492454980074e-06,
      "lm_loss": 5.462,
      "loss": 1.2156,
      "step": 1431,
      "text_contrastive_loss": 0.6423,
      "train_positive_log_prob": -81.3106,
      "train_positive_token_accuracy": 0.0782,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3205,
      "epoch": 3.2325056433408577,
      "grad_norm": 13.735143661499023,
      "learning_rate": 2.894766091845873e-06,
      "lm_loss": 5.5468,
      "loss": 1.3025,
      "step": 1432,
      "text_contrastive_loss": 0.8546,
      "train_positive_log_prob": -81.8362,
      "train_positive_token_accuracy": 0.081,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.4447,
      "epoch": 3.234762979683973,
      "grad_norm": 13.709455490112305,
      "learning_rate": 2.88818737105587e-06,
      "lm_loss": 5.4476,
      "loss": 1.4373,
      "step": 1433,
      "text_contrastive_loss": 0.8956,
      "train_positive_log_prob": -79.8429,
      "train_positive_token_accuracy": 0.0776,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3643,
      "epoch": 3.237020316027088,
      "grad_norm": 12.758544921875,
      "learning_rate": 2.881613096980407e-06,
      "lm_loss": 5.3446,
      "loss": 1.2587,
      "step": 1434,
      "text_contrastive_loss": 0.7199,
      "train_positive_log_prob": -78.6063,
      "train_positive_token_accuracy": 0.0749,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.3323,
      "epoch": 3.239277652370203,
      "grad_norm": 11.02239990234375,
      "learning_rate": 2.8750432834625312e-06,
      "lm_loss": 5.4404,
      "loss": 1.2886,
      "step": 1435,
      "text_contrastive_loss": 0.8245,
      "train_positive_log_prob": -79.8645,
      "train_positive_token_accuracy": 0.082,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3958,
      "epoch": 3.2415349887133185,
      "grad_norm": 12.147967338562012,
      "learning_rate": 2.8684779443358945e-06,
      "lm_loss": 5.5308,
      "loss": 1.4073,
      "step": 1436,
      "text_contrastive_loss": 0.9167,
      "train_positive_log_prob": -82.5297,
      "train_positive_token_accuracy": 0.0738,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.4072,
      "epoch": 3.2437923250564333,
      "grad_norm": 14.212855339050293,
      "learning_rate": 2.861917093424731e-06,
      "lm_loss": 5.4445,
      "loss": 1.3616,
      "step": 1437,
      "text_contrastive_loss": 0.82,
      "train_positive_log_prob": -80.8615,
      "train_positive_token_accuracy": 0.0915,
      "train_positive_token_prob": 0.033
    },
    {
      "contrastive_loss": 0.2872,
      "epoch": 3.2460496613995486,
      "grad_norm": 12.381433486938477,
      "learning_rate": 2.855360744543822e-06,
      "lm_loss": 5.3889,
      "loss": 1.2503,
      "step": 1438,
      "text_contrastive_loss": 0.8485,
      "train_positive_log_prob": -79.6539,
      "train_positive_token_accuracy": 0.0758,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3199,
      "epoch": 3.2483069977426635,
      "grad_norm": 11.74273681640625,
      "learning_rate": 2.8488089114984725e-06,
      "lm_loss": 5.4003,
      "loss": 1.232,
      "step": 1439,
      "text_contrastive_loss": 0.7442,
      "train_positive_log_prob": -79.4633,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.35,
      "epoch": 3.250564334085779,
      "grad_norm": 15.128246307373047,
      "learning_rate": 2.84226160808447e-06,
      "lm_loss": 5.5677,
      "loss": 1.2912,
      "step": 1440,
      "text_contrastive_loss": 0.7687,
      "train_positive_log_prob": -81.7769,
      "train_positive_token_accuracy": 0.0771,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.4455,
      "epoch": 3.2528216704288937,
      "grad_norm": 13.406543731689453,
      "learning_rate": 2.835718848088076e-06,
      "lm_loss": 5.3907,
      "loss": 1.4058,
      "step": 1441,
      "text_contrastive_loss": 0.8426,
      "train_positive_log_prob": -79.1141,
      "train_positive_token_accuracy": 0.0743,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3688,
      "epoch": 3.255079006772009,
      "grad_norm": 13.79780101776123,
      "learning_rate": 2.8291806452859803e-06,
      "lm_loss": 5.4736,
      "loss": 1.2779,
      "step": 1442,
      "text_contrastive_loss": 0.7234,
      "train_positive_log_prob": -81.1199,
      "train_positive_token_accuracy": 0.0754,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.4313,
      "epoch": 3.2573363431151243,
      "grad_norm": 13.48995590209961,
      "learning_rate": 2.822647013445272e-06,
      "lm_loss": 5.4517,
      "loss": 1.4321,
      "step": 1443,
      "text_contrastive_loss": 0.9113,
      "train_positive_log_prob": -80.9007,
      "train_positive_token_accuracy": 0.0848,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.4178,
      "epoch": 3.259593679458239,
      "grad_norm": 12.608296394348145,
      "learning_rate": 2.8161179663234215e-06,
      "lm_loss": 5.4421,
      "loss": 1.4412,
      "step": 1444,
      "text_contrastive_loss": 0.9585,
      "train_positive_log_prob": -78.8458,
      "train_positive_token_accuracy": 0.0828,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3429,
      "epoch": 3.2618510158013545,
      "grad_norm": 13.537189483642578,
      "learning_rate": 2.809593517668243e-06,
      "lm_loss": 5.4962,
      "loss": 1.2356,
      "step": 1445,
      "text_contrastive_loss": 0.6861,
      "train_positive_log_prob": -81.7324,
      "train_positive_token_accuracy": 0.0735,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.4539,
      "epoch": 3.2641083521444694,
      "grad_norm": 13.725131034851074,
      "learning_rate": 2.8030736812178717e-06,
      "lm_loss": 5.3237,
      "loss": 1.4391,
      "step": 1446,
      "text_contrastive_loss": 0.9055,
      "train_positive_log_prob": -77.3248,
      "train_positive_token_accuracy": 0.0775,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3744,
      "epoch": 3.2663656884875847,
      "grad_norm": 11.78693962097168,
      "learning_rate": 2.796558470700723e-06,
      "lm_loss": 5.4174,
      "loss": 1.3114,
      "step": 1447,
      "text_contrastive_loss": 0.7906,
      "train_positive_log_prob": -80.3055,
      "train_positive_token_accuracy": 0.0775,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.4067,
      "epoch": 3.2686230248307,
      "grad_norm": 13.28524398803711,
      "learning_rate": 2.790047899835479e-06,
      "lm_loss": 5.3779,
      "loss": 1.3395,
      "step": 1448,
      "text_contrastive_loss": 0.79,
      "train_positive_log_prob": -79.4962,
      "train_positive_token_accuracy": 0.0848,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.2783,
      "epoch": 3.270880361173815,
      "grad_norm": 12.303711891174316,
      "learning_rate": 2.7835419823310507e-06,
      "lm_loss": 5.4055,
      "loss": 1.2131,
      "step": 1449,
      "text_contrastive_loss": 0.7884,
      "train_positive_log_prob": -80.2617,
      "train_positive_token_accuracy": 0.0805,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3437,
      "epoch": 3.27313769751693,
      "grad_norm": 13.363794326782227,
      "learning_rate": 2.777040731886549e-06,
      "lm_loss": 5.4144,
      "loss": 1.2726,
      "step": 1450,
      "text_contrastive_loss": 0.7748,
      "train_positive_log_prob": -78.5691,
      "train_positive_token_accuracy": 0.0848,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.2811,
      "epoch": 3.275395033860045,
      "grad_norm": 11.258405685424805,
      "learning_rate": 2.770544162191261e-06,
      "lm_loss": 5.482,
      "loss": 1.1472,
      "step": 1451,
      "text_contrastive_loss": 0.6357,
      "train_positive_log_prob": -82.4903,
      "train_positive_token_accuracy": 0.0833,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.4406,
      "epoch": 3.2776523702031604,
      "grad_norm": 14.803228378295898,
      "learning_rate": 2.7640522869246134e-06,
      "lm_loss": 5.3397,
      "loss": 1.279,
      "step": 1452,
      "text_contrastive_loss": 0.6088,
      "train_positive_log_prob": -79.1339,
      "train_positive_token_accuracy": 0.0778,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.3859,
      "epoch": 3.2799097065462752,
      "grad_norm": 13.501431465148926,
      "learning_rate": 2.7575651197561504e-06,
      "lm_loss": 5.3894,
      "loss": 1.3452,
      "step": 1453,
      "text_contrastive_loss": 0.8405,
      "train_positive_log_prob": -79.9568,
      "train_positive_token_accuracy": 0.0835,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.3808,
      "epoch": 3.2821670428893905,
      "grad_norm": 14.436296463012695,
      "learning_rate": 2.7510826743455037e-06,
      "lm_loss": 5.5601,
      "loss": 1.3028,
      "step": 1454,
      "text_contrastive_loss": 0.732,
      "train_positive_log_prob": -84.3522,
      "train_positive_token_accuracy": 0.0687,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.409,
      "epoch": 3.2844243792325054,
      "grad_norm": 12.698884963989258,
      "learning_rate": 2.744604964342364e-06,
      "lm_loss": 5.4928,
      "loss": 1.3901,
      "step": 1455,
      "text_contrastive_loss": 0.8636,
      "train_positive_log_prob": -80.6962,
      "train_positive_token_accuracy": 0.083,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3434,
      "epoch": 3.2866817155756207,
      "grad_norm": 11.475349426269531,
      "learning_rate": 2.7381320033864434e-06,
      "lm_loss": 5.362,
      "loss": 1.2695,
      "step": 1456,
      "text_contrastive_loss": 0.7796,
      "train_positive_log_prob": -79.7615,
      "train_positive_token_accuracy": 0.0827,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3341,
      "epoch": 3.288939051918736,
      "grad_norm": 14.666303634643555,
      "learning_rate": 2.7316638051074605e-06,
      "lm_loss": 5.4915,
      "loss": 1.2745,
      "step": 1457,
      "text_contrastive_loss": 0.7826,
      "train_positive_log_prob": -80.5508,
      "train_positive_token_accuracy": 0.0778,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3401,
      "epoch": 3.291196388261851,
      "grad_norm": 12.208320617675781,
      "learning_rate": 2.72520038312511e-06,
      "lm_loss": 5.4205,
      "loss": 1.2532,
      "step": 1458,
      "text_contrastive_loss": 0.742,
      "train_positive_log_prob": -81.0321,
      "train_positive_token_accuracy": 0.0813,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.4117,
      "epoch": 3.293453724604966,
      "grad_norm": 12.495759010314941,
      "learning_rate": 2.7187417510490176e-06,
      "lm_loss": 5.4519,
      "loss": 1.291,
      "step": 1459,
      "text_contrastive_loss": 0.6683,
      "train_positive_log_prob": -79.7556,
      "train_positive_token_accuracy": 0.0764,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3489,
      "epoch": 3.295711060948081,
      "grad_norm": 13.113781929016113,
      "learning_rate": 2.7122879224787315e-06,
      "lm_loss": 5.5168,
      "loss": 1.3154,
      "step": 1460,
      "text_contrastive_loss": 0.8298,
      "train_positive_log_prob": -81.1143,
      "train_positive_token_accuracy": 0.0765,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.4159,
      "epoch": 3.2979683972911964,
      "grad_norm": 14.516807556152344,
      "learning_rate": 2.7058389110036835e-06,
      "lm_loss": 5.4569,
      "loss": 1.3379,
      "step": 1461,
      "text_contrastive_loss": 0.7525,
      "train_positive_log_prob": -80.6998,
      "train_positive_token_accuracy": 0.0769,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3721,
      "epoch": 3.3002257336343117,
      "grad_norm": 13.572257995605469,
      "learning_rate": 2.6993947302031643e-06,
      "lm_loss": 5.3711,
      "loss": 1.3431,
      "step": 1462,
      "text_contrastive_loss": 0.8677,
      "train_positive_log_prob": -78.4474,
      "train_positive_token_accuracy": 0.0746,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.2717,
      "epoch": 3.3024830699774266,
      "grad_norm": 13.101594924926758,
      "learning_rate": 2.692955393646286e-06,
      "lm_loss": 5.4834,
      "loss": 1.1467,
      "step": 1463,
      "text_contrastive_loss": 0.6534,
      "train_positive_log_prob": -81.4774,
      "train_positive_token_accuracy": 0.0794,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.2453,
      "epoch": 3.304740406320542,
      "grad_norm": 10.775618553161621,
      "learning_rate": 2.686520914891968e-06,
      "lm_loss": 5.4244,
      "loss": 1.0384,
      "step": 1464,
      "text_contrastive_loss": 0.5013,
      "train_positive_log_prob": -79.544,
      "train_positive_token_accuracy": 0.0855,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3128,
      "epoch": 3.3069977426636568,
      "grad_norm": 12.02344799041748,
      "learning_rate": 2.6800913074888984e-06,
      "lm_loss": 5.5315,
      "loss": 1.2232,
      "step": 1465,
      "text_contrastive_loss": 0.7147,
      "train_positive_log_prob": -83.6844,
      "train_positive_token_accuracy": 0.086,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.2648,
      "epoch": 3.309255079006772,
      "grad_norm": 12.345636367797852,
      "learning_rate": 2.6736665849755073e-06,
      "lm_loss": 5.3845,
      "loss": 1.1419,
      "step": 1466,
      "text_contrastive_loss": 0.6773,
      "train_positive_log_prob": -78.5004,
      "train_positive_token_accuracy": 0.0804,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.4061,
      "epoch": 3.311512415349887,
      "grad_norm": 13.457576751708984,
      "learning_rate": 2.6672467608799413e-06,
      "lm_loss": 5.5098,
      "loss": 1.4108,
      "step": 1467,
      "text_contrastive_loss": 0.9074,
      "train_positive_log_prob": -81.1176,
      "train_positive_token_accuracy": 0.0616,
      "train_positive_token_prob": 0.0283
    },
    {
      "contrastive_loss": 0.3699,
      "epoch": 3.3137697516930023,
      "grad_norm": 11.337600708007812,
      "learning_rate": 2.660831848720028e-06,
      "lm_loss": 5.4729,
      "loss": 1.2979,
      "step": 1468,
      "text_contrastive_loss": 0.7615,
      "train_positive_log_prob": -80.855,
      "train_positive_token_accuracy": 0.0775,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3656,
      "epoch": 3.3160270880361176,
      "grad_norm": 15.963889122009277,
      "learning_rate": 2.654421862003256e-06,
      "lm_loss": 5.5073,
      "loss": 1.2696,
      "step": 1469,
      "text_contrastive_loss": 0.7065,
      "train_positive_log_prob": -80.6729,
      "train_positive_token_accuracy": 0.0781,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3328,
      "epoch": 3.3182844243792324,
      "grad_norm": 14.723601341247559,
      "learning_rate": 2.648016814226742e-06,
      "lm_loss": 5.6061,
      "loss": 1.2822,
      "step": 1470,
      "text_contrastive_loss": 0.7774,
      "train_positive_log_prob": -84.2654,
      "train_positive_token_accuracy": 0.0772,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.3971,
      "epoch": 3.3205417607223477,
      "grad_norm": 13.494935035705566,
      "learning_rate": 2.6416167188772052e-06,
      "lm_loss": 5.5084,
      "loss": 1.3143,
      "step": 1471,
      "text_contrastive_loss": 0.7327,
      "train_positive_log_prob": -82.9996,
      "train_positive_token_accuracy": 0.0749,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.4222,
      "epoch": 3.3227990970654626,
      "grad_norm": 14.805681228637695,
      "learning_rate": 2.6352215894309306e-06,
      "lm_loss": 5.3873,
      "loss": 1.3416,
      "step": 1472,
      "text_contrastive_loss": 0.7614,
      "train_positive_log_prob": -80.7656,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.0329
    },
    {
      "contrastive_loss": 0.3509,
      "epoch": 3.325056433408578,
      "grad_norm": 11.837203025817871,
      "learning_rate": 2.6288314393537522e-06,
      "lm_loss": 5.5112,
      "loss": 1.2858,
      "step": 1473,
      "text_contrastive_loss": 0.7677,
      "train_positive_log_prob": -81.2044,
      "train_positive_token_accuracy": 0.0774,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.37,
      "epoch": 3.327313769751693,
      "grad_norm": 10.98058032989502,
      "learning_rate": 2.6224462821010185e-06,
      "lm_loss": 5.5068,
      "loss": 1.3272,
      "step": 1474,
      "text_contrastive_loss": 0.813,
      "train_positive_log_prob": -81.1172,
      "train_positive_token_accuracy": 0.079,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.3709,
      "epoch": 3.329571106094808,
      "grad_norm": 13.405777931213379,
      "learning_rate": 2.616066131117563e-06,
      "lm_loss": 5.4351,
      "loss": 1.2832,
      "step": 1475,
      "text_contrastive_loss": 0.7377,
      "train_positive_log_prob": -79.0198,
      "train_positive_token_accuracy": 0.0768,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3743,
      "epoch": 3.3318284424379234,
      "grad_norm": 12.909689903259277,
      "learning_rate": 2.6096909998376794e-06,
      "lm_loss": 5.4092,
      "loss": 1.3607,
      "step": 1476,
      "text_contrastive_loss": 0.8911,
      "train_positive_log_prob": -81.3746,
      "train_positive_token_accuracy": 0.0826,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3567,
      "epoch": 3.3340857787810383,
      "grad_norm": 13.252840995788574,
      "learning_rate": 2.6033209016850926e-06,
      "lm_loss": 5.5363,
      "loss": 1.2997,
      "step": 1477,
      "text_contrastive_loss": 0.7788,
      "train_positive_log_prob": -83.0844,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3543,
      "epoch": 3.3363431151241536,
      "grad_norm": 13.664084434509277,
      "learning_rate": 2.596955850072928e-06,
      "lm_loss": 5.4244,
      "loss": 1.281,
      "step": 1478,
      "text_contrastive_loss": 0.7686,
      "train_positive_log_prob": -80.1999,
      "train_positive_token_accuracy": 0.084,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.4341,
      "epoch": 3.3386004514672685,
      "grad_norm": 13.399177551269531,
      "learning_rate": 2.5905958584036826e-06,
      "lm_loss": 5.6161,
      "loss": 1.4276,
      "step": 1479,
      "text_contrastive_loss": 0.8638,
      "train_positive_log_prob": -82.4656,
      "train_positive_token_accuracy": 0.075,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.2895,
      "epoch": 3.340857787810384,
      "grad_norm": 11.764191627502441,
      "learning_rate": 2.5842409400692026e-06,
      "lm_loss": 5.5003,
      "loss": 1.1359,
      "step": 1480,
      "text_contrastive_loss": 0.5926,
      "train_positive_log_prob": -80.0552,
      "train_positive_token_accuracy": 0.0805,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3489,
      "epoch": 3.343115124153499,
      "grad_norm": 12.049650192260742,
      "learning_rate": 2.577891108450651e-06,
      "lm_loss": 5.5732,
      "loss": 1.29,
      "step": 1481,
      "text_contrastive_loss": 0.7676,
      "train_positive_log_prob": -82.0718,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3308,
      "epoch": 3.345372460496614,
      "grad_norm": 12.622515678405762,
      "learning_rate": 2.571546376918479e-06,
      "lm_loss": 5.4124,
      "loss": 1.2243,
      "step": 1482,
      "text_contrastive_loss": 0.7046,
      "train_positive_log_prob": -80.6769,
      "train_positive_token_accuracy": 0.077,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.394,
      "epoch": 3.3476297968397293,
      "grad_norm": 12.73591136932373,
      "learning_rate": 2.5652067588324015e-06,
      "lm_loss": 5.4206,
      "loss": 1.3097,
      "step": 1483,
      "text_contrastive_loss": 0.7472,
      "train_positive_log_prob": -78.704,
      "train_positive_token_accuracy": 0.0815,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.3021,
      "epoch": 3.349887133182844,
      "grad_norm": 11.945694923400879,
      "learning_rate": 2.55887226754136e-06,
      "lm_loss": 5.3205,
      "loss": 1.2063,
      "step": 1484,
      "text_contrastive_loss": 0.7444,
      "train_positive_log_prob": -76.5879,
      "train_positive_token_accuracy": 0.0822,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.4803,
      "epoch": 3.3521444695259595,
      "grad_norm": 15.011481285095215,
      "learning_rate": 2.552542916383507e-06,
      "lm_loss": 5.4357,
      "loss": 1.4293,
      "step": 1485,
      "text_contrastive_loss": 0.811,
      "train_positive_log_prob": -77.498,
      "train_positive_token_accuracy": 0.0767,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3935,
      "epoch": 3.3544018058690743,
      "grad_norm": 13.88205337524414,
      "learning_rate": 2.5462187186861697e-06,
      "lm_loss": 5.4123,
      "loss": 1.3346,
      "step": 1486,
      "text_contrastive_loss": 0.7997,
      "train_positive_log_prob": -81.2688,
      "train_positive_token_accuracy": 0.0778,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.2888,
      "epoch": 3.3566591422121896,
      "grad_norm": 12.709901809692383,
      "learning_rate": 2.5398996877658256e-06,
      "lm_loss": 5.5052,
      "loss": 1.1977,
      "step": 1487,
      "text_contrastive_loss": 0.7167,
      "train_positive_log_prob": -82.7467,
      "train_positive_token_accuracy": 0.0769,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.3343,
      "epoch": 3.3589164785553045,
      "grad_norm": 13.5997314453125,
      "learning_rate": 2.5335858369280674e-06,
      "lm_loss": 5.4754,
      "loss": 1.2899,
      "step": 1488,
      "text_contrastive_loss": 0.8161,
      "train_positive_log_prob": -83.2009,
      "train_positive_token_accuracy": 0.0761,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3642,
      "epoch": 3.36117381489842,
      "grad_norm": 14.860273361206055,
      "learning_rate": 2.5272771794675866e-06,
      "lm_loss": 5.6381,
      "loss": 1.3343,
      "step": 1489,
      "text_contrastive_loss": 0.8127,
      "train_positive_log_prob": -85.8706,
      "train_positive_token_accuracy": 0.0823,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.4325,
      "epoch": 3.363431151241535,
      "grad_norm": 13.943649291992188,
      "learning_rate": 2.5209737286681367e-06,
      "lm_loss": 5.4665,
      "loss": 1.4305,
      "step": 1490,
      "text_contrastive_loss": 0.9026,
      "train_positive_log_prob": -81.1655,
      "train_positive_token_accuracy": 0.0768,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3042,
      "epoch": 3.36568848758465,
      "grad_norm": 13.22721004486084,
      "learning_rate": 2.514675497802508e-06,
      "lm_loss": 5.5726,
      "loss": 1.2556,
      "step": 1491,
      "text_contrastive_loss": 0.7882,
      "train_positive_log_prob": -81.8022,
      "train_positive_token_accuracy": 0.0716,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.2725,
      "epoch": 3.3679458239277653,
      "grad_norm": 11.849224090576172,
      "learning_rate": 2.508382500132499e-06,
      "lm_loss": 5.4561,
      "loss": 1.1832,
      "step": 1492,
      "text_contrastive_loss": 0.7303,
      "train_positive_log_prob": -80.438,
      "train_positive_token_accuracy": 0.0819,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.2978,
      "epoch": 3.37020316027088,
      "grad_norm": 11.576635360717773,
      "learning_rate": 2.50209474890889e-06,
      "lm_loss": 5.3959,
      "loss": 1.2757,
      "step": 1493,
      "text_contrastive_loss": 0.8767,
      "train_positive_log_prob": -79.1441,
      "train_positive_token_accuracy": 0.0743,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3249,
      "epoch": 3.3724604966139955,
      "grad_norm": 11.91965103149414,
      "learning_rate": 2.495812257371416e-06,
      "lm_loss": 5.4112,
      "loss": 1.2655,
      "step": 1494,
      "text_contrastive_loss": 0.799,
      "train_positive_log_prob": -79.7504,
      "train_positive_token_accuracy": 0.071,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.3196,
      "epoch": 3.374717832957111,
      "grad_norm": 13.613897323608398,
      "learning_rate": 2.4895350387487304e-06,
      "lm_loss": 5.6697,
      "loss": 1.268,
      "step": 1495,
      "text_contrastive_loss": 0.7627,
      "train_positive_log_prob": -83.8037,
      "train_positive_token_accuracy": 0.0756,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.2848,
      "epoch": 3.3769751693002257,
      "grad_norm": 11.66408920288086,
      "learning_rate": 2.4832631062583906e-06,
      "lm_loss": 5.3674,
      "loss": 1.2153,
      "step": 1496,
      "text_contrastive_loss": 0.7875,
      "train_positive_log_prob": -78.7697,
      "train_positive_token_accuracy": 0.0844,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.4446,
      "epoch": 3.379232505643341,
      "grad_norm": 14.82761001586914,
      "learning_rate": 2.47699647310682e-06,
      "lm_loss": 5.5406,
      "loss": 1.4435,
      "step": 1497,
      "text_contrastive_loss": 0.8895,
      "train_positive_log_prob": -85.1436,
      "train_positive_token_accuracy": 0.0752,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.2982,
      "epoch": 3.381489841986456,
      "grad_norm": 12.15412425994873,
      "learning_rate": 2.470735152489287e-06,
      "lm_loss": 5.4322,
      "loss": 1.132,
      "step": 1498,
      "text_contrastive_loss": 0.5812,
      "train_positive_log_prob": -81.164,
      "train_positive_token_accuracy": 0.0821,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.4166,
      "epoch": 3.383747178329571,
      "grad_norm": 13.879342079162598,
      "learning_rate": 2.4644791575898665e-06,
      "lm_loss": 5.5772,
      "loss": 1.408,
      "step": 1499,
      "text_contrastive_loss": 0.8673,
      "train_positive_log_prob": -83.6823,
      "train_positive_token_accuracy": 0.0778,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3544,
      "epoch": 3.386004514672686,
      "grad_norm": 12.835597038269043,
      "learning_rate": 2.4582285015814263e-06,
      "lm_loss": 5.4296,
      "loss": 1.2063,
      "step": 1500,
      "text_contrastive_loss": 0.6179,
      "train_positive_log_prob": -81.8884,
      "train_positive_token_accuracy": 0.0852,
      "train_positive_token_prob": 0.0325
    },
    {
      "contrastive_loss": 0.3748,
      "epoch": 3.3882618510158014,
      "grad_norm": 13.693696022033691,
      "learning_rate": 2.4519831976255892e-06,
      "lm_loss": 5.4888,
      "loss": 1.3371,
      "step": 1501,
      "text_contrastive_loss": 0.8268,
      "train_positive_log_prob": -80.0147,
      "train_positive_token_accuracy": 0.0852,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.2995,
      "epoch": 3.3905191873589167,
      "grad_norm": 12.810537338256836,
      "learning_rate": 2.445743258872711e-06,
      "lm_loss": 5.4853,
      "loss": 1.1739,
      "step": 1502,
      "text_contrastive_loss": 0.6517,
      "train_positive_log_prob": -81.5071,
      "train_positive_token_accuracy": 0.0823,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.4199,
      "epoch": 3.3927765237020315,
      "grad_norm": 14.674356460571289,
      "learning_rate": 2.4395086984618486e-06,
      "lm_loss": 5.5201,
      "loss": 1.3155,
      "step": 1503,
      "text_contrastive_loss": 0.6871,
      "train_positive_log_prob": -80.0595,
      "train_positive_token_accuracy": 0.0843,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3283,
      "epoch": 3.395033860045147,
      "grad_norm": 12.382877349853516,
      "learning_rate": 2.433279529520732e-06,
      "lm_loss": 5.4382,
      "loss": 1.2897,
      "step": 1504,
      "text_contrastive_loss": 0.8351,
      "train_positive_log_prob": -80.4912,
      "train_positive_token_accuracy": 0.0847,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.3853,
      "epoch": 3.3972911963882617,
      "grad_norm": 12.306852340698242,
      "learning_rate": 2.427055765165741e-06,
      "lm_loss": 5.4388,
      "loss": 1.349,
      "step": 1505,
      "text_contrastive_loss": 0.8397,
      "train_positive_log_prob": -80.6895,
      "train_positive_token_accuracy": 0.0814,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3212,
      "epoch": 3.399548532731377,
      "grad_norm": 12.166094779968262,
      "learning_rate": 2.420837418501876e-06,
      "lm_loss": 5.4144,
      "loss": 1.2337,
      "step": 1506,
      "text_contrastive_loss": 0.7421,
      "train_positive_log_prob": -78.0012,
      "train_positive_token_accuracy": 0.0759,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3518,
      "epoch": 3.401805869074492,
      "grad_norm": 13.620892524719238,
      "learning_rate": 2.414624502622731e-06,
      "lm_loss": 5.4152,
      "loss": 1.2913,
      "step": 1507,
      "text_contrastive_loss": 0.7961,
      "train_positive_log_prob": -78.2482,
      "train_positive_token_accuracy": 0.0806,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3299,
      "epoch": 3.404063205417607,
      "grad_norm": 13.580711364746094,
      "learning_rate": 2.408417030610457e-06,
      "lm_loss": 5.4487,
      "loss": 1.2475,
      "step": 1508,
      "text_contrastive_loss": 0.7454,
      "train_positive_log_prob": -79.8124,
      "train_positive_token_accuracy": 0.08,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3441,
      "epoch": 3.4063205417607225,
      "grad_norm": 14.103965759277344,
      "learning_rate": 2.4022150155357526e-06,
      "lm_loss": 5.5329,
      "loss": 1.2285,
      "step": 1509,
      "text_contrastive_loss": 0.6623,
      "train_positive_log_prob": -84.2947,
      "train_positive_token_accuracy": 0.0851,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3348,
      "epoch": 3.4085778781038374,
      "grad_norm": 11.755352020263672,
      "learning_rate": 2.396018470457821e-06,
      "lm_loss": 5.3933,
      "loss": 1.2474,
      "step": 1510,
      "text_contrastive_loss": 0.7464,
      "train_positive_log_prob": -79.17,
      "train_positive_token_accuracy": 0.0732,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.2613,
      "epoch": 3.4108352144469527,
      "grad_norm": 11.900193214416504,
      "learning_rate": 2.389827408424345e-06,
      "lm_loss": 5.5358,
      "loss": 1.1067,
      "step": 1511,
      "text_contrastive_loss": 0.5836,
      "train_positive_log_prob": -83.0834,
      "train_positive_token_accuracy": 0.0703,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.4277,
      "epoch": 3.4130925507900676,
      "grad_norm": 13.55148983001709,
      "learning_rate": 2.3836418424714665e-06,
      "lm_loss": 5.3704,
      "loss": 1.4147,
      "step": 1512,
      "text_contrastive_loss": 0.8998,
      "train_positive_log_prob": -79.9694,
      "train_positive_token_accuracy": 0.0871,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.334,
      "epoch": 3.415349887133183,
      "grad_norm": 13.18596363067627,
      "learning_rate": 2.377461785623752e-06,
      "lm_loss": 5.4027,
      "loss": 1.2542,
      "step": 1513,
      "text_contrastive_loss": 0.7598,
      "train_positive_log_prob": -78.9727,
      "train_positive_token_accuracy": 0.0746,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.3319,
      "epoch": 3.417607223476298,
      "grad_norm": 11.630409240722656,
      "learning_rate": 2.3712872508941714e-06,
      "lm_loss": 5.4037,
      "loss": 1.201,
      "step": 1514,
      "text_contrastive_loss": 0.6573,
      "train_positive_log_prob": -82.0735,
      "train_positive_token_accuracy": 0.0782,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3014,
      "epoch": 3.419864559819413,
      "grad_norm": 12.5684175491333,
      "learning_rate": 2.3651182512840604e-06,
      "lm_loss": 5.4383,
      "loss": 1.2324,
      "step": 1515,
      "text_contrastive_loss": 0.7744,
      "train_positive_log_prob": -80.0666,
      "train_positive_token_accuracy": 0.0748,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.3529,
      "epoch": 3.4221218961625284,
      "grad_norm": 12.914410591125488,
      "learning_rate": 2.358954799783106e-06,
      "lm_loss": 5.3883,
      "loss": 1.3431,
      "step": 1516,
      "text_contrastive_loss": 0.9027,
      "train_positive_log_prob": -79.8367,
      "train_positive_token_accuracy": 0.0838,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3331,
      "epoch": 3.4243792325056432,
      "grad_norm": 12.251567840576172,
      "learning_rate": 2.3527969093693105e-06,
      "lm_loss": 5.4971,
      "loss": 1.2708,
      "step": 1517,
      "text_contrastive_loss": 0.776,
      "train_positive_log_prob": -81.6597,
      "train_positive_token_accuracy": 0.0823,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3004,
      "epoch": 3.4266365688487586,
      "grad_norm": 10.214859008789062,
      "learning_rate": 2.346644593008966e-06,
      "lm_loss": 5.4587,
      "loss": 1.1951,
      "step": 1518,
      "text_contrastive_loss": 0.6976,
      "train_positive_log_prob": -82.3247,
      "train_positive_token_accuracy": 0.08,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.2544,
      "epoch": 3.4288939051918734,
      "grad_norm": 10.06332015991211,
      "learning_rate": 2.3404978636566312e-06,
      "lm_loss": 5.4006,
      "loss": 1.1817,
      "step": 1519,
      "text_contrastive_loss": 0.7745,
      "train_positive_log_prob": -81.0751,
      "train_positive_token_accuracy": 0.0869,
      "train_positive_token_prob": 0.0329
    },
    {
      "contrastive_loss": 0.3118,
      "epoch": 3.4311512415349887,
      "grad_norm": 13.071005821228027,
      "learning_rate": 2.3343567342550933e-06,
      "lm_loss": 5.4059,
      "loss": 1.2044,
      "step": 1520,
      "text_contrastive_loss": 0.7041,
      "train_positive_log_prob": -78.6801,
      "train_positive_token_accuracy": 0.08,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3277,
      "epoch": 3.4334085778781036,
      "grad_norm": 12.789834022521973,
      "learning_rate": 2.328221217735355e-06,
      "lm_loss": 5.375,
      "loss": 1.1977,
      "step": 1521,
      "text_contrastive_loss": 0.6651,
      "train_positive_log_prob": -78.9508,
      "train_positive_token_accuracy": 0.0809,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3906,
      "epoch": 3.435665914221219,
      "grad_norm": 13.222907066345215,
      "learning_rate": 2.322091327016597e-06,
      "lm_loss": 5.3968,
      "loss": 1.3879,
      "step": 1522,
      "text_contrastive_loss": 0.9151,
      "train_positive_log_prob": -79.7508,
      "train_positive_token_accuracy": 0.0708,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3174,
      "epoch": 3.4379232505643342,
      "grad_norm": 12.321296691894531,
      "learning_rate": 2.3159670750061563e-06,
      "lm_loss": 5.4893,
      "loss": 1.1912,
      "step": 1523,
      "text_contrastive_loss": 0.6497,
      "train_positive_log_prob": -81.9154,
      "train_positive_token_accuracy": 0.0804,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3686,
      "epoch": 3.440180586907449,
      "grad_norm": 13.29212474822998,
      "learning_rate": 2.3098484745994933e-06,
      "lm_loss": 5.4338,
      "loss": 1.3172,
      "step": 1524,
      "text_contrastive_loss": 0.8103,
      "train_positive_log_prob": -83.5652,
      "train_positive_token_accuracy": 0.0813,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.333,
      "epoch": 3.4424379232505644,
      "grad_norm": 12.273012161254883,
      "learning_rate": 2.3037355386801683e-06,
      "lm_loss": 5.4078,
      "loss": 1.1939,
      "step": 1525,
      "text_contrastive_loss": 0.6401,
      "train_positive_log_prob": -79.3,
      "train_positive_token_accuracy": 0.086,
      "train_positive_token_prob": 0.0327
    },
    {
      "contrastive_loss": 0.3193,
      "epoch": 3.4446952595936793,
      "grad_norm": 12.238722801208496,
      "learning_rate": 2.2976282801198237e-06,
      "lm_loss": 5.4413,
      "loss": 1.1801,
      "step": 1526,
      "text_contrastive_loss": 0.6333,
      "train_positive_log_prob": -80.6191,
      "train_positive_token_accuracy": 0.0786,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.4134,
      "epoch": 3.4469525959367946,
      "grad_norm": 14.603962898254395,
      "learning_rate": 2.2915267117781328e-06,
      "lm_loss": 5.4377,
      "loss": 1.411,
      "step": 1527,
      "text_contrastive_loss": 0.9075,
      "train_positive_log_prob": -80.6354,
      "train_positive_token_accuracy": 0.0748,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3273,
      "epoch": 3.44920993227991,
      "grad_norm": 11.297776222229004,
      "learning_rate": 2.2854308465027963e-06,
      "lm_loss": 5.4826,
      "loss": 1.2509,
      "step": 1528,
      "text_contrastive_loss": 0.7507,
      "train_positive_log_prob": -82.5831,
      "train_positive_token_accuracy": 0.0807,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3296,
      "epoch": 3.4514672686230248,
      "grad_norm": 11.965171813964844,
      "learning_rate": 2.279340697129505e-06,
      "lm_loss": 5.4535,
      "loss": 1.2298,
      "step": 1529,
      "text_contrastive_loss": 0.7097,
      "train_positive_log_prob": -81.1854,
      "train_positive_token_accuracy": 0.0798,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.3828,
      "epoch": 3.45372460496614,
      "grad_norm": 12.046841621398926,
      "learning_rate": 2.2732562764819157e-06,
      "lm_loss": 5.464,
      "loss": 1.297,
      "step": 1530,
      "text_contrastive_loss": 0.7356,
      "train_positive_log_prob": -81.8761,
      "train_positive_token_accuracy": 0.0777,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.28,
      "epoch": 3.455981941309255,
      "grad_norm": 11.561809539794922,
      "learning_rate": 2.267177597371616e-06,
      "lm_loss": 5.3392,
      "loss": 1.1943,
      "step": 1531,
      "text_contrastive_loss": 0.7609,
      "train_positive_log_prob": -81.2562,
      "train_positive_token_accuracy": 0.0809,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.3527,
      "epoch": 3.4582392776523703,
      "grad_norm": 12.117919921875,
      "learning_rate": 2.26110467259811e-06,
      "lm_loss": 5.3512,
      "loss": 1.28,
      "step": 1532,
      "text_contrastive_loss": 0.7843,
      "train_positive_log_prob": -78.7829,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.3226,
      "epoch": 3.460496613995485,
      "grad_norm": 12.698823928833008,
      "learning_rate": 2.255037514948785e-06,
      "lm_loss": 5.3925,
      "loss": 1.2494,
      "step": 1533,
      "text_contrastive_loss": 0.7751,
      "train_positive_log_prob": -79.8815,
      "train_positive_token_accuracy": 0.0807,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.4023,
      "epoch": 3.4627539503386005,
      "grad_norm": 14.095758438110352,
      "learning_rate": 2.2489761371988826e-06,
      "lm_loss": 5.3957,
      "loss": 1.3687,
      "step": 1534,
      "text_contrastive_loss": 0.8536,
      "train_positive_log_prob": -81.2942,
      "train_positive_token_accuracy": 0.0822,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3481,
      "epoch": 3.4650112866817158,
      "grad_norm": 15.021332740783691,
      "learning_rate": 2.242920552111473e-06,
      "lm_loss": 5.4151,
      "loss": 1.2612,
      "step": 1535,
      "text_contrastive_loss": 0.7432,
      "train_positive_log_prob": -80.8017,
      "train_positive_token_accuracy": 0.084,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.396,
      "epoch": 3.4672686230248306,
      "grad_norm": 13.502098083496094,
      "learning_rate": 2.236870772437433e-06,
      "lm_loss": 5.5101,
      "loss": 1.3803,
      "step": 1536,
      "text_contrastive_loss": 0.8665,
      "train_positive_log_prob": -81.231,
      "train_positive_token_accuracy": 0.0787,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.2971,
      "epoch": 3.469525959367946,
      "grad_norm": 12.005428314208984,
      "learning_rate": 2.2308268109154126e-06,
      "lm_loss": 5.5129,
      "loss": 1.2055,
      "step": 1537,
      "text_contrastive_loss": 0.7142,
      "train_positive_log_prob": -82.2874,
      "train_positive_token_accuracy": 0.0866,
      "train_positive_token_prob": 0.0326
    },
    {
      "contrastive_loss": 0.3754,
      "epoch": 3.471783295711061,
      "grad_norm": 13.41446590423584,
      "learning_rate": 2.224788680271811e-06,
      "lm_loss": 5.4155,
      "loss": 1.2911,
      "step": 1538,
      "text_contrastive_loss": 0.7482,
      "train_positive_log_prob": -81.6538,
      "train_positive_token_accuracy": 0.0736,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3638,
      "epoch": 3.474040632054176,
      "grad_norm": 13.16190242767334,
      "learning_rate": 2.218756393220753e-06,
      "lm_loss": 5.5326,
      "loss": 1.3163,
      "step": 1539,
      "text_contrastive_loss": 0.7986,
      "train_positive_log_prob": -81.5605,
      "train_positive_token_accuracy": 0.0702,
      "train_positive_token_prob": 0.0289
    },
    {
      "contrastive_loss": 0.2733,
      "epoch": 3.476297968397291,
      "grad_norm": 10.823956489562988,
      "learning_rate": 2.212729962464051e-06,
      "lm_loss": 5.47,
      "loss": 1.1562,
      "step": 1540,
      "text_contrastive_loss": 0.6717,
      "train_positive_log_prob": -80.3626,
      "train_positive_token_accuracy": 0.0861,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.405,
      "epoch": 3.4785553047404063,
      "grad_norm": 14.038640975952148,
      "learning_rate": 2.2067094006911943e-06,
      "lm_loss": 5.3801,
      "loss": 1.4272,
      "step": 1541,
      "text_contrastive_loss": 0.9683,
      "train_positive_log_prob": -79.5456,
      "train_positive_token_accuracy": 0.0785,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.3287,
      "epoch": 3.4808126410835216,
      "grad_norm": 13.235450744628906,
      "learning_rate": 2.2006947205793107e-06,
      "lm_loss": 5.3748,
      "loss": 1.2709,
      "step": 1542,
      "text_contrastive_loss": 0.8094,
      "train_positive_log_prob": -78.2735,
      "train_positive_token_accuracy": 0.0738,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3925,
      "epoch": 3.4830699774266365,
      "grad_norm": 14.204618453979492,
      "learning_rate": 2.1946859347931442e-06,
      "lm_loss": 5.3882,
      "loss": 1.3161,
      "step": 1543,
      "text_contrastive_loss": 0.7695,
      "train_positive_log_prob": -78.5115,
      "train_positive_token_accuracy": 0.081,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.3835,
      "epoch": 3.485327313769752,
      "grad_norm": 13.043344497680664,
      "learning_rate": 2.1886830559850264e-06,
      "lm_loss": 5.4147,
      "loss": 1.2988,
      "step": 1544,
      "text_contrastive_loss": 0.7475,
      "train_positive_log_prob": -78.4912,
      "train_positive_token_accuracy": 0.0811,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3383,
      "epoch": 3.4875846501128667,
      "grad_norm": 12.718570709228516,
      "learning_rate": 2.182686096794852e-06,
      "lm_loss": 5.4734,
      "loss": 1.2949,
      "step": 1545,
      "text_contrastive_loss": 0.8186,
      "train_positive_log_prob": -81.6073,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3272,
      "epoch": 3.489841986455982,
      "grad_norm": 13.003602027893066,
      "learning_rate": 2.176695069850053e-06,
      "lm_loss": 5.4558,
      "loss": 1.2379,
      "step": 1546,
      "text_contrastive_loss": 0.7304,
      "train_positive_log_prob": -79.3602,
      "train_positive_token_accuracy": 0.0755,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.381,
      "epoch": 3.4920993227990973,
      "grad_norm": 12.9622802734375,
      "learning_rate": 2.1707099877655634e-06,
      "lm_loss": 5.3894,
      "loss": 1.313,
      "step": 1547,
      "text_contrastive_loss": 0.7862,
      "train_positive_log_prob": -80.6528,
      "train_positive_token_accuracy": 0.0856,
      "train_positive_token_prob": 0.0328
    },
    {
      "contrastive_loss": 0.2719,
      "epoch": 3.494356659142212,
      "grad_norm": 11.515785217285156,
      "learning_rate": 2.1647308631438068e-06,
      "lm_loss": 5.355,
      "loss": 1.1773,
      "step": 1548,
      "text_contrastive_loss": 0.7398,
      "train_positive_log_prob": -79.4634,
      "train_positive_token_accuracy": 0.0771,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3136,
      "epoch": 3.4966139954853275,
      "grad_norm": 12.572412490844727,
      "learning_rate": 2.1587577085746596e-06,
      "lm_loss": 5.4211,
      "loss": 1.207,
      "step": 1549,
      "text_contrastive_loss": 0.7026,
      "train_positive_log_prob": -78.3742,
      "train_positive_token_accuracy": 0.0857,
      "train_positive_token_prob": 0.0326
    },
    {
      "contrastive_loss": 0.4071,
      "epoch": 3.4988713318284423,
      "grad_norm": 14.194604873657227,
      "learning_rate": 2.1527905366354292e-06,
      "lm_loss": 5.3583,
      "loss": 1.2807,
      "step": 1550,
      "text_contrastive_loss": 0.6754,
      "train_positive_log_prob": -79.7542,
      "train_positive_token_accuracy": 0.0798,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.4247,
      "epoch": 3.5011286681715577,
      "grad_norm": 13.476304054260254,
      "learning_rate": 2.14682935989082e-06,
      "lm_loss": 5.4014,
      "loss": 1.3515,
      "step": 1551,
      "text_contrastive_loss": 0.7734,
      "train_positive_log_prob": -78.1968,
      "train_positive_token_accuracy": 0.0747,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3518,
      "epoch": 3.5033860045146725,
      "grad_norm": 13.156872749328613,
      "learning_rate": 2.14087419089292e-06,
      "lm_loss": 5.4082,
      "loss": 1.2955,
      "step": 1552,
      "text_contrastive_loss": 0.8059,
      "train_positive_log_prob": -80.3814,
      "train_positive_token_accuracy": 0.0739,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.2908,
      "epoch": 3.505643340857788,
      "grad_norm": 10.860188484191895,
      "learning_rate": 2.1349250421811622e-06,
      "lm_loss": 5.363,
      "loss": 1.1888,
      "step": 1553,
      "text_contrastive_loss": 0.7233,
      "train_positive_log_prob": -80.6763,
      "train_positive_token_accuracy": 0.0875,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.2545,
      "epoch": 3.5079006772009027,
      "grad_norm": 12.058480262756348,
      "learning_rate": 2.1289819262823065e-06,
      "lm_loss": 5.4619,
      "loss": 1.1261,
      "step": 1554,
      "text_contrastive_loss": 0.6509,
      "train_positive_log_prob": -79.2798,
      "train_positive_token_accuracy": 0.0732,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3596,
      "epoch": 3.510158013544018,
      "grad_norm": 13.837039947509766,
      "learning_rate": 2.1230448557104087e-06,
      "lm_loss": 5.3523,
      "loss": 1.2333,
      "step": 1555,
      "text_contrastive_loss": 0.677,
      "train_positive_log_prob": -77.6497,
      "train_positive_token_accuracy": 0.0819,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3374,
      "epoch": 3.5124153498871333,
      "grad_norm": 12.251382827758789,
      "learning_rate": 2.117113842966792e-06,
      "lm_loss": 5.5174,
      "loss": 1.2319,
      "step": 1556,
      "text_contrastive_loss": 0.6856,
      "train_positive_log_prob": -81.2086,
      "train_positive_token_accuracy": 0.0717,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.4057,
      "epoch": 3.514672686230248,
      "grad_norm": 14.353056907653809,
      "learning_rate": 2.111188900540028e-06,
      "lm_loss": 5.4156,
      "loss": 1.3047,
      "step": 1557,
      "text_contrastive_loss": 0.7148,
      "train_positive_log_prob": -82.061,
      "train_positive_token_accuracy": 0.0874,
      "train_positive_token_prob": 0.033
    },
    {
      "contrastive_loss": 0.2912,
      "epoch": 3.5169300225733635,
      "grad_norm": 11.068564414978027,
      "learning_rate": 2.1052700409059057e-06,
      "lm_loss": 5.4744,
      "loss": 1.1417,
      "step": 1558,
      "text_contrastive_loss": 0.6062,
      "train_positive_log_prob": -82.2836,
      "train_positive_token_accuracy": 0.0865,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3046,
      "epoch": 3.5191873589164784,
      "grad_norm": 13.130361557006836,
      "learning_rate": 2.0993572765274044e-06,
      "lm_loss": 5.5494,
      "loss": 1.2723,
      "step": 1559,
      "text_contrastive_loss": 0.8256,
      "train_positive_log_prob": -81.2457,
      "train_positive_token_accuracy": 0.0771,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.2906,
      "epoch": 3.5214446952595937,
      "grad_norm": 12.09571647644043,
      "learning_rate": 2.093450619854671e-06,
      "lm_loss": 5.5055,
      "loss": 1.2892,
      "step": 1560,
      "text_contrastive_loss": 0.8961,
      "train_positive_log_prob": -81.8086,
      "train_positive_token_accuracy": 0.082,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3227,
      "epoch": 3.523702031602709,
      "grad_norm": 13.012192726135254,
      "learning_rate": 2.08755008332499e-06,
      "lm_loss": 5.4112,
      "loss": 1.2437,
      "step": 1561,
      "text_contrastive_loss": 0.7597,
      "train_positive_log_prob": -80.7691,
      "train_positive_token_accuracy": 0.0827,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3763,
      "epoch": 3.525959367945824,
      "grad_norm": 12.909019470214844,
      "learning_rate": 2.0816556793627624e-06,
      "lm_loss": 5.4735,
      "loss": 1.3205,
      "step": 1562,
      "text_contrastive_loss": 0.7937,
      "train_positive_log_prob": -82.2685,
      "train_positive_token_accuracy": 0.0767,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.4227,
      "epoch": 3.528216704288939,
      "grad_norm": 13.299750328063965,
      "learning_rate": 2.0757674203794696e-06,
      "lm_loss": 5.5079,
      "loss": 1.3434,
      "step": 1563,
      "text_contrastive_loss": 0.7399,
      "train_positive_log_prob": -83.6727,
      "train_positive_token_accuracy": 0.082,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3216,
      "epoch": 3.530474040632054,
      "grad_norm": 12.418523788452148,
      "learning_rate": 2.06988531877366e-06,
      "lm_loss": 5.2851,
      "loss": 1.2102,
      "step": 1564,
      "text_contrastive_loss": 0.7203,
      "train_positive_log_prob": -77.6855,
      "train_positive_token_accuracy": 0.0813,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.363,
      "epoch": 3.5327313769751694,
      "grad_norm": 13.458390235900879,
      "learning_rate": 2.064009386930915e-06,
      "lm_loss": 5.4871,
      "loss": 1.335,
      "step": 1565,
      "text_contrastive_loss": 0.8465,
      "train_positive_log_prob": -80.0791,
      "train_positive_token_accuracy": 0.0865,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.4234,
      "epoch": 3.5349887133182847,
      "grad_norm": 14.445470809936523,
      "learning_rate": 2.0581396372238254e-06,
      "lm_loss": 5.3556,
      "loss": 1.3654,
      "step": 1566,
      "text_contrastive_loss": 0.813,
      "train_positive_log_prob": -78.2426,
      "train_positive_token_accuracy": 0.0861,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.334,
      "epoch": 3.5372460496613995,
      "grad_norm": 13.816523551940918,
      "learning_rate": 2.0522760820119615e-06,
      "lm_loss": 5.5365,
      "loss": 1.2447,
      "step": 1567,
      "text_contrastive_loss": 0.7139,
      "train_positive_log_prob": -83.1731,
      "train_positive_token_accuracy": 0.0751,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.476,
      "epoch": 3.5395033860045144,
      "grad_norm": 19.811813354492188,
      "learning_rate": 2.046418733641853e-06,
      "lm_loss": 5.5185,
      "loss": 1.467,
      "step": 1568,
      "text_contrastive_loss": 0.8782,
      "train_positive_log_prob": -81.4403,
      "train_positive_token_accuracy": 0.0778,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.3066,
      "epoch": 3.5417607223476297,
      "grad_norm": 13.256061553955078,
      "learning_rate": 2.04056760444696e-06,
      "lm_loss": 5.5971,
      "loss": 1.2254,
      "step": 1569,
      "text_contrastive_loss": 0.7181,
      "train_positive_log_prob": -86.1801,
      "train_positive_token_accuracy": 0.0743,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.3382,
      "epoch": 3.544018058690745,
      "grad_norm": 12.53451919555664,
      "learning_rate": 2.0347227067476478e-06,
      "lm_loss": 5.3916,
      "loss": 1.3546,
      "step": 1570,
      "text_contrastive_loss": 0.9546,
      "train_positive_log_prob": -78.4316,
      "train_positive_token_accuracy": 0.0787,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.2651,
      "epoch": 3.54627539503386,
      "grad_norm": 10.973885536193848,
      "learning_rate": 2.02888405285116e-06,
      "lm_loss": 5.453,
      "loss": 1.1802,
      "step": 1571,
      "text_contrastive_loss": 0.7395,
      "train_positive_log_prob": -78.8148,
      "train_positive_token_accuracy": 0.0776,
      "train_positive_token_prob": 0.0292
    },
    {
      "contrastive_loss": 0.4255,
      "epoch": 3.5485327313769752,
      "grad_norm": 13.16361141204834,
      "learning_rate": 2.02305165505159e-06,
      "lm_loss": 5.4203,
      "loss": 1.3472,
      "step": 1572,
      "text_contrastive_loss": 0.7595,
      "train_positive_log_prob": -80.8616,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.4142,
      "epoch": 3.55079006772009,
      "grad_norm": 12.9069242477417,
      "learning_rate": 2.0172255256298623e-06,
      "lm_loss": 5.4061,
      "loss": 1.3569,
      "step": 1573,
      "text_contrastive_loss": 0.8041,
      "train_positive_log_prob": -80.7105,
      "train_positive_token_accuracy": 0.0824,
      "train_positive_token_prob": 0.0326
    },
    {
      "contrastive_loss": 0.2933,
      "epoch": 3.5530474040632054,
      "grad_norm": 11.150215148925781,
      "learning_rate": 2.0114056768537005e-06,
      "lm_loss": 5.3737,
      "loss": 1.1758,
      "step": 1574,
      "text_contrastive_loss": 0.6903,
      "train_positive_log_prob": -80.7757,
      "train_positive_token_accuracy": 0.0769,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3092,
      "epoch": 3.5553047404063207,
      "grad_norm": 12.42226791381836,
      "learning_rate": 2.005592120977606e-06,
      "lm_loss": 5.4272,
      "loss": 1.2342,
      "step": 1575,
      "text_contrastive_loss": 0.7646,
      "train_positive_log_prob": -78.9682,
      "train_positive_token_accuracy": 0.0762,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.4042,
      "epoch": 3.5575620767494356,
      "grad_norm": 14.214899063110352,
      "learning_rate": 1.9997848702428226e-06,
      "lm_loss": 5.367,
      "loss": 1.4012,
      "step": 1576,
      "text_contrastive_loss": 0.9206,
      "train_positive_log_prob": -79.9787,
      "train_positive_token_accuracy": 0.0813,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.4442,
      "epoch": 3.559819413092551,
      "grad_norm": 14.875078201293945,
      "learning_rate": 1.9939839368773267e-06,
      "lm_loss": 5.6141,
      "loss": 1.4326,
      "step": 1577,
      "text_contrastive_loss": 0.854,
      "train_positive_log_prob": -82.4555,
      "train_positive_token_accuracy": 0.0684,
      "train_positive_token_prob": 0.0285
    },
    {
      "contrastive_loss": 0.4697,
      "epoch": 3.5620767494356658,
      "grad_norm": 16.47623634338379,
      "learning_rate": 1.9881893330957893e-06,
      "lm_loss": 5.4831,
      "loss": 1.5145,
      "step": 1578,
      "text_contrastive_loss": 0.993,
      "train_positive_log_prob": -81.1563,
      "train_positive_token_accuracy": 0.0839,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.2458,
      "epoch": 3.564334085778781,
      "grad_norm": 11.240406036376953,
      "learning_rate": 1.982401071099549e-06,
      "lm_loss": 5.398,
      "loss": 1.2106,
      "step": 1579,
      "text_contrastive_loss": 0.85,
      "train_positive_log_prob": -82.7413,
      "train_positive_token_accuracy": 0.079,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.4033,
      "epoch": 3.5665914221218964,
      "grad_norm": 12.894579887390137,
      "learning_rate": 1.9766191630765964e-06,
      "lm_loss": 5.5431,
      "loss": 1.3649,
      "step": 1580,
      "text_contrastive_loss": 0.8146,
      "train_positive_log_prob": -82.0647,
      "train_positive_token_accuracy": 0.0789,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.247,
      "epoch": 3.5688487584650113,
      "grad_norm": 9.906365394592285,
      "learning_rate": 1.970843621201541e-06,
      "lm_loss": 5.4329,
      "loss": 1.1266,
      "step": 1581,
      "text_contrastive_loss": 0.6727,
      "train_positive_log_prob": -79.608,
      "train_positive_token_accuracy": 0.0792,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3531,
      "epoch": 3.5711060948081266,
      "grad_norm": 12.115705490112305,
      "learning_rate": 1.9650744576355894e-06,
      "lm_loss": 5.4785,
      "loss": 1.2599,
      "step": 1582,
      "text_contrastive_loss": 0.7179,
      "train_positive_log_prob": -81.0406,
      "train_positive_token_accuracy": 0.079,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.4402,
      "epoch": 3.5733634311512414,
      "grad_norm": 13.983465194702148,
      "learning_rate": 1.959311684526513e-06,
      "lm_loss": 5.51,
      "loss": 1.4032,
      "step": 1583,
      "text_contrastive_loss": 0.824,
      "train_positive_log_prob": -84.3919,
      "train_positive_token_accuracy": 0.0824,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.375,
      "epoch": 3.5756207674943568,
      "grad_norm": 14.502907752990723,
      "learning_rate": 1.9535553140086322e-06,
      "lm_loss": 5.3925,
      "loss": 1.332,
      "step": 1584,
      "text_contrastive_loss": 0.8354,
      "train_positive_log_prob": -79.2664,
      "train_positive_token_accuracy": 0.0781,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3706,
      "epoch": 3.5778781038374716,
      "grad_norm": 13.97586441040039,
      "learning_rate": 1.9478053582027826e-06,
      "lm_loss": 5.4038,
      "loss": 1.3105,
      "step": 1585,
      "text_contrastive_loss": 0.7989,
      "train_positive_log_prob": -78.4822,
      "train_positive_token_accuracy": 0.0842,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.4562,
      "epoch": 3.580135440180587,
      "grad_norm": 13.074638366699219,
      "learning_rate": 1.9420618292162974e-06,
      "lm_loss": 5.3254,
      "loss": 1.4316,
      "step": 1586,
      "text_contrastive_loss": 0.8857,
      "train_positive_log_prob": -80.0487,
      "train_positive_token_accuracy": 0.0819,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3422,
      "epoch": 3.582392776523702,
      "grad_norm": 12.996149063110352,
      "learning_rate": 1.9363247391429695e-06,
      "lm_loss": 5.4479,
      "loss": 1.2868,
      "step": 1587,
      "text_contrastive_loss": 0.7995,
      "train_positive_log_prob": -81.6767,
      "train_positive_token_accuracy": 0.081,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3504,
      "epoch": 3.584650112866817,
      "grad_norm": 11.976738929748535,
      "learning_rate": 1.93059410006304e-06,
      "lm_loss": 5.5108,
      "loss": 1.2936,
      "step": 1588,
      "text_contrastive_loss": 0.7843,
      "train_positive_log_prob": -83.397,
      "train_positive_token_accuracy": 0.0772,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.3133,
      "epoch": 3.5869074492099324,
      "grad_norm": 11.291887283325195,
      "learning_rate": 1.924869924043165e-06,
      "lm_loss": 5.395,
      "loss": 1.2656,
      "step": 1589,
      "text_contrastive_loss": 0.8256,
      "train_positive_log_prob": -80.3596,
      "train_positive_token_accuracy": 0.0768,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3633,
      "epoch": 3.5891647855530473,
      "grad_norm": 12.659879684448242,
      "learning_rate": 1.919152223136391e-06,
      "lm_loss": 5.439,
      "loss": 1.3022,
      "step": 1590,
      "text_contrastive_loss": 0.7901,
      "train_positive_log_prob": -80.862,
      "train_positive_token_accuracy": 0.0793,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3369,
      "epoch": 3.5914221218961626,
      "grad_norm": 14.898165702819824,
      "learning_rate": 1.913441009382133e-06,
      "lm_loss": 5.3437,
      "loss": 1.2725,
      "step": 1591,
      "text_contrastive_loss": 0.8025,
      "train_positive_log_prob": -79.0059,
      "train_positive_token_accuracy": 0.0859,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.3797,
      "epoch": 3.5936794582392775,
      "grad_norm": 12.831680297851562,
      "learning_rate": 1.9077362948061404e-06,
      "lm_loss": 5.3652,
      "loss": 1.3208,
      "step": 1592,
      "text_contrastive_loss": 0.8091,
      "train_positive_log_prob": -79.7247,
      "train_positive_token_accuracy": 0.077,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.4952,
      "epoch": 3.595936794582393,
      "grad_norm": 13.396096229553223,
      "learning_rate": 1.902038091420481e-06,
      "lm_loss": 5.4346,
      "loss": 1.4233,
      "step": 1593,
      "text_contrastive_loss": 0.7693,
      "train_positive_log_prob": -79.5267,
      "train_positive_token_accuracy": 0.0769,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3941,
      "epoch": 3.598194130925508,
      "grad_norm": 12.79609203338623,
      "learning_rate": 1.8963464112235185e-06,
      "lm_loss": 5.4697,
      "loss": 1.3875,
      "step": 1594,
      "text_contrastive_loss": 0.8929,
      "train_positive_log_prob": -79.6926,
      "train_positive_token_accuracy": 0.0721,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.4476,
      "epoch": 3.600451467268623,
      "grad_norm": 16.410837173461914,
      "learning_rate": 1.8906612661998698e-06,
      "lm_loss": 5.4431,
      "loss": 1.3496,
      "step": 1595,
      "text_contrastive_loss": 0.7154,
      "train_positive_log_prob": -80.6089,
      "train_positive_token_accuracy": 0.0762,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3113,
      "epoch": 3.6027088036117383,
      "grad_norm": 12.252687454223633,
      "learning_rate": 1.884982668320398e-06,
      "lm_loss": 5.4839,
      "loss": 1.2253,
      "step": 1596,
      "text_contrastive_loss": 0.7312,
      "train_positive_log_prob": -78.9801,
      "train_positive_token_accuracy": 0.0803,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3348,
      "epoch": 3.604966139954853,
      "grad_norm": 12.842511177062988,
      "learning_rate": 1.8793106295421797e-06,
      "lm_loss": 5.3726,
      "loss": 1.2057,
      "step": 1597,
      "text_contrastive_loss": 0.6672,
      "train_positive_log_prob": -77.9063,
      "train_positive_token_accuracy": 0.0787,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.393,
      "epoch": 3.6072234762979685,
      "grad_norm": 14.63471794128418,
      "learning_rate": 1.873645161808481e-06,
      "lm_loss": 5.366,
      "loss": 1.3704,
      "step": 1598,
      "text_contrastive_loss": 0.8817,
      "train_positive_log_prob": -78.3344,
      "train_positive_token_accuracy": 0.0789,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3527,
      "epoch": 3.609480812641084,
      "grad_norm": 12.025533676147461,
      "learning_rate": 1.8679862770487273e-06,
      "lm_loss": 5.3853,
      "loss": 1.2841,
      "step": 1599,
      "text_contrastive_loss": 0.7857,
      "train_positive_log_prob": -78.1483,
      "train_positive_token_accuracy": 0.0851,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3843,
      "epoch": 3.6117381489841986,
      "grad_norm": 12.612031936645508,
      "learning_rate": 1.8623339871784869e-06,
      "lm_loss": 5.4367,
      "loss": 1.3268,
      "step": 1600,
      "text_contrastive_loss": 0.7976,
      "train_positive_log_prob": -82.7199,
      "train_positive_token_accuracy": 0.0778,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.4264,
      "epoch": 3.6139954853273135,
      "grad_norm": 13.698080062866211,
      "learning_rate": 1.8566883040994411e-06,
      "lm_loss": 5.4968,
      "loss": 1.4221,
      "step": 1601,
      "text_contrastive_loss": 0.8921,
      "train_positive_log_prob": -82.1514,
      "train_positive_token_accuracy": 0.081,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3204,
      "epoch": 3.616252821670429,
      "grad_norm": 11.818224906921387,
      "learning_rate": 1.8510492396993595e-06,
      "lm_loss": 5.4532,
      "loss": 1.2383,
      "step": 1602,
      "text_contrastive_loss": 0.7451,
      "train_positive_log_prob": -83.2307,
      "train_positive_token_accuracy": 0.0755,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.4045,
      "epoch": 3.618510158013544,
      "grad_norm": 13.190351486206055,
      "learning_rate": 1.8454168058520732e-06,
      "lm_loss": 5.505,
      "loss": 1.395,
      "step": 1603,
      "text_contrastive_loss": 0.8799,
      "train_positive_log_prob": -80.2099,
      "train_positive_token_accuracy": 0.078,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3406,
      "epoch": 3.620767494356659,
      "grad_norm": 12.244832038879395,
      "learning_rate": 1.8397910144174536e-06,
      "lm_loss": 5.3262,
      "loss": 1.2494,
      "step": 1604,
      "text_contrastive_loss": 0.7522,
      "train_positive_log_prob": -78.8156,
      "train_positive_token_accuracy": 0.0808,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3156,
      "epoch": 3.6230248306997743,
      "grad_norm": 11.526841163635254,
      "learning_rate": 1.8341718772413852e-06,
      "lm_loss": 5.3288,
      "loss": 1.164,
      "step": 1605,
      "text_contrastive_loss": 0.6311,
      "train_positive_log_prob": -77.2912,
      "train_positive_token_accuracy": 0.0762,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.4301,
      "epoch": 3.625282167042889,
      "grad_norm": 12.750216484069824,
      "learning_rate": 1.8285594061557421e-06,
      "lm_loss": 5.3824,
      "loss": 1.3957,
      "step": 1606,
      "text_contrastive_loss": 0.8547,
      "train_positive_log_prob": -80.4866,
      "train_positive_token_accuracy": 0.0836,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.35,
      "epoch": 3.6275395033860045,
      "grad_norm": 13.768596649169922,
      "learning_rate": 1.822953612978362e-06,
      "lm_loss": 5.5162,
      "loss": 1.2777,
      "step": 1607,
      "text_contrastive_loss": 0.7523,
      "train_positive_log_prob": -83.8653,
      "train_positive_token_accuracy": 0.0763,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.405,
      "epoch": 3.62979683972912,
      "grad_norm": 12.813020706176758,
      "learning_rate": 1.817354509513017e-06,
      "lm_loss": 5.3793,
      "loss": 1.3395,
      "step": 1608,
      "text_contrastive_loss": 0.7932,
      "train_positive_log_prob": -80.2028,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3253,
      "epoch": 3.6320541760722347,
      "grad_norm": 12.3479585647583,
      "learning_rate": 1.8117621075493979e-06,
      "lm_loss": 5.4163,
      "loss": 1.2278,
      "step": 1609,
      "text_contrastive_loss": 0.7216,
      "train_positive_log_prob": -81.3096,
      "train_positive_token_accuracy": 0.0835,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.3627,
      "epoch": 3.63431151241535,
      "grad_norm": 11.790746688842773,
      "learning_rate": 1.8061764188630831e-06,
      "lm_loss": 5.3681,
      "loss": 1.3255,
      "step": 1610,
      "text_contrastive_loss": 0.852,
      "train_positive_log_prob": -79.1585,
      "train_positive_token_accuracy": 0.0832,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.2754,
      "epoch": 3.636568848758465,
      "grad_norm": 12.158080101013184,
      "learning_rate": 1.8005974552155158e-06,
      "lm_loss": 5.4961,
      "loss": 1.1475,
      "step": 1611,
      "text_contrastive_loss": 0.645,
      "train_positive_log_prob": -81.9944,
      "train_positive_token_accuracy": 0.078,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3996,
      "epoch": 3.63882618510158,
      "grad_norm": 12.607054710388184,
      "learning_rate": 1.7950252283539776e-06,
      "lm_loss": 5.5635,
      "loss": 1.389,
      "step": 1612,
      "text_contrastive_loss": 0.8661,
      "train_positive_log_prob": -83.1741,
      "train_positive_token_accuracy": 0.0745,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3213,
      "epoch": 3.6410835214446955,
      "grad_norm": 12.518956184387207,
      "learning_rate": 1.7894597500115657e-06,
      "lm_loss": 5.4323,
      "loss": 1.2948,
      "step": 1613,
      "text_contrastive_loss": 0.8607,
      "train_positive_log_prob": -80.6768,
      "train_positive_token_accuracy": 0.0789,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.4059,
      "epoch": 3.6433408577878104,
      "grad_norm": 15.735977172851562,
      "learning_rate": 1.7839010319071687e-06,
      "lm_loss": 5.3952,
      "loss": 1.3833,
      "step": 1614,
      "text_contrastive_loss": 0.8759,
      "train_positive_log_prob": -79.9137,
      "train_positive_token_accuracy": 0.0736,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3427,
      "epoch": 3.6455981941309257,
      "grad_norm": 13.345383644104004,
      "learning_rate": 1.7783490857454354e-06,
      "lm_loss": 5.5722,
      "loss": 1.3157,
      "step": 1615,
      "text_contrastive_loss": 0.8315,
      "train_positive_log_prob": -81.996,
      "train_positive_token_accuracy": 0.0746,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.3228,
      "epoch": 3.6478555304740405,
      "grad_norm": 11.710400581359863,
      "learning_rate": 1.7728039232167603e-06,
      "lm_loss": 5.4267,
      "loss": 1.2269,
      "step": 1616,
      "text_contrastive_loss": 0.7228,
      "train_positive_log_prob": -80.3139,
      "train_positive_token_accuracy": 0.0779,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3458,
      "epoch": 3.650112866817156,
      "grad_norm": 13.129372596740723,
      "learning_rate": 1.7672655559972535e-06,
      "lm_loss": 5.3303,
      "loss": 1.2842,
      "step": 1617,
      "text_contrastive_loss": 0.8108,
      "train_positive_log_prob": -78.4302,
      "train_positive_token_accuracy": 0.0826,
      "train_positive_token_prob": 0.0325
    },
    {
      "contrastive_loss": 0.2252,
      "epoch": 3.6523702031602707,
      "grad_norm": 11.039880752563477,
      "learning_rate": 1.7617339957487167e-06,
      "lm_loss": 5.4686,
      "loss": 1.1095,
      "step": 1618,
      "text_contrastive_loss": 0.675,
      "train_positive_log_prob": -82.1572,
      "train_positive_token_accuracy": 0.0728,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3163,
      "epoch": 3.654627539503386,
      "grad_norm": 12.812634468078613,
      "learning_rate": 1.7562092541186144e-06,
      "lm_loss": 5.4963,
      "loss": 1.2151,
      "step": 1619,
      "text_contrastive_loss": 0.6984,
      "train_positive_log_prob": -82.1396,
      "train_positive_token_accuracy": 0.0737,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.3392,
      "epoch": 3.656884875846501,
      "grad_norm": 12.152909278869629,
      "learning_rate": 1.750691342740058e-06,
      "lm_loss": 5.4037,
      "loss": 1.256,
      "step": 1620,
      "text_contrastive_loss": 0.7528,
      "train_positive_log_prob": -78.6081,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.4045,
      "epoch": 3.659142212189616,
      "grad_norm": 13.417214393615723,
      "learning_rate": 1.7451802732317763e-06,
      "lm_loss": 5.3755,
      "loss": 1.2903,
      "step": 1621,
      "text_contrastive_loss": 0.6964,
      "train_positive_log_prob": -81.431,
      "train_positive_token_accuracy": 0.0768,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.2638,
      "epoch": 3.6613995485327315,
      "grad_norm": 12.613409996032715,
      "learning_rate": 1.7396760571980902e-06,
      "lm_loss": 5.3969,
      "loss": 1.1343,
      "step": 1622,
      "text_contrastive_loss": 0.6617,
      "train_positive_log_prob": -80.4469,
      "train_positive_token_accuracy": 0.0768,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.323,
      "epoch": 3.6636568848758464,
      "grad_norm": 12.255485534667969,
      "learning_rate": 1.7341787062288928e-06,
      "lm_loss": 5.4692,
      "loss": 1.2179,
      "step": 1623,
      "text_contrastive_loss": 0.6959,
      "train_positive_log_prob": -81.6074,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3705,
      "epoch": 3.6659142212189617,
      "grad_norm": 12.568070411682129,
      "learning_rate": 1.7286882318996162e-06,
      "lm_loss": 5.4433,
      "loss": 1.3011,
      "step": 1624,
      "text_contrastive_loss": 0.7726,
      "train_positive_log_prob": -80.1824,
      "train_positive_token_accuracy": 0.0833,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.3608,
      "epoch": 3.6681715575620766,
      "grad_norm": 13.158514976501465,
      "learning_rate": 1.7232046457712164e-06,
      "lm_loss": 5.3833,
      "loss": 1.2496,
      "step": 1625,
      "text_contrastive_loss": 0.701,
      "train_positive_log_prob": -79.0456,
      "train_positive_token_accuracy": 0.0827,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.4922,
      "epoch": 3.670428893905192,
      "grad_norm": 14.625532150268555,
      "learning_rate": 1.7177279593901463e-06,
      "lm_loss": 5.4737,
      "loss": 1.4734,
      "step": 1626,
      "text_contrastive_loss": 0.8676,
      "train_positive_log_prob": -83.0258,
      "train_positive_token_accuracy": 0.085,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.3635,
      "epoch": 3.672686230248307,
      "grad_norm": 12.785755157470703,
      "learning_rate": 1.712258184288328e-06,
      "lm_loss": 5.4296,
      "loss": 1.3403,
      "step": 1627,
      "text_contrastive_loss": 0.8677,
      "train_positive_log_prob": -80.3248,
      "train_positive_token_accuracy": 0.0792,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3955,
      "epoch": 3.674943566591422,
      "grad_norm": 13.993976593017578,
      "learning_rate": 1.7067953319831327e-06,
      "lm_loss": 5.2966,
      "loss": 1.3552,
      "step": 1628,
      "text_contrastive_loss": 0.86,
      "train_positive_log_prob": -77.5487,
      "train_positive_token_accuracy": 0.0846,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.4406,
      "epoch": 3.6772009029345374,
      "grad_norm": 13.542219161987305,
      "learning_rate": 1.7013394139773537e-06,
      "lm_loss": 5.4495,
      "loss": 1.4451,
      "step": 1629,
      "text_contrastive_loss": 0.9192,
      "train_positive_log_prob": -80.5491,
      "train_positive_token_accuracy": 0.0872,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3337,
      "epoch": 3.6794582392776523,
      "grad_norm": 13.1850004196167,
      "learning_rate": 1.6958904417591853e-06,
      "lm_loss": 5.512,
      "loss": 1.304,
      "step": 1630,
      "text_contrastive_loss": 0.8381,
      "train_positive_log_prob": -83.4348,
      "train_positive_token_accuracy": 0.0815,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.4008,
      "epoch": 3.6817155756207676,
      "grad_norm": 15.749434471130371,
      "learning_rate": 1.6904484268021915e-06,
      "lm_loss": 5.4665,
      "loss": 1.37,
      "step": 1631,
      "text_contrastive_loss": 0.8451,
      "train_positive_log_prob": -82.7882,
      "train_positive_token_accuracy": 0.0818,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3308,
      "epoch": 3.683972911963883,
      "grad_norm": 12.142971992492676,
      "learning_rate": 1.6850133805652907e-06,
      "lm_loss": 5.5244,
      "loss": 1.3,
      "step": 1632,
      "text_contrastive_loss": 0.8335,
      "train_positive_log_prob": -81.2679,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.2907,
      "epoch": 3.6862302483069977,
      "grad_norm": 11.403225898742676,
      "learning_rate": 1.6795853144927282e-06,
      "lm_loss": 5.5715,
      "loss": 1.2479,
      "step": 1633,
      "text_contrastive_loss": 0.8001,
      "train_positive_log_prob": -82.9844,
      "train_positive_token_accuracy": 0.0758,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.288,
      "epoch": 3.6884875846501126,
      "grad_norm": 11.208388328552246,
      "learning_rate": 1.6741642400140513e-06,
      "lm_loss": 5.3952,
      "loss": 1.1811,
      "step": 1634,
      "text_contrastive_loss": 0.7071,
      "train_positive_log_prob": -81.3046,
      "train_positive_token_accuracy": 0.0816,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.2854,
      "epoch": 3.690744920993228,
      "grad_norm": 11.621232032775879,
      "learning_rate": 1.668750168544081e-06,
      "lm_loss": 5.4768,
      "loss": 1.2199,
      "step": 1635,
      "text_contrastive_loss": 0.7738,
      "train_positive_log_prob": -80.2905,
      "train_positive_token_accuracy": 0.0781,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.2686,
      "epoch": 3.6930022573363432,
      "grad_norm": 12.664349555969238,
      "learning_rate": 1.663343111482898e-06,
      "lm_loss": 5.3322,
      "loss": 1.1795,
      "step": 1636,
      "text_contrastive_loss": 0.7554,
      "train_positive_log_prob": -77.382,
      "train_positive_token_accuracy": 0.0848,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.3965,
      "epoch": 3.695259593679458,
      "grad_norm": 14.684317588806152,
      "learning_rate": 1.657943080215812e-06,
      "lm_loss": 5.3943,
      "loss": 1.3464,
      "step": 1637,
      "text_contrastive_loss": 0.821,
      "train_positive_log_prob": -80.0894,
      "train_positive_token_accuracy": 0.0805,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3085,
      "epoch": 3.6975169300225734,
      "grad_norm": 11.034719467163086,
      "learning_rate": 1.6525500861133386e-06,
      "lm_loss": 5.572,
      "loss": 1.2273,
      "step": 1638,
      "text_contrastive_loss": 0.7231,
      "train_positive_log_prob": -81.7108,
      "train_positive_token_accuracy": 0.0799,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3509,
      "epoch": 3.6997742663656883,
      "grad_norm": 12.65117359161377,
      "learning_rate": 1.6471641405311727e-06,
      "lm_loss": 5.2876,
      "loss": 1.2707,
      "step": 1639,
      "text_contrastive_loss": 0.782,
      "train_positive_log_prob": -78.7163,
      "train_positive_token_accuracy": 0.0813,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.3723,
      "epoch": 3.7020316027088036,
      "grad_norm": 13.441893577575684,
      "learning_rate": 1.641785254810172e-06,
      "lm_loss": 5.4307,
      "loss": 1.3708,
      "step": 1640,
      "text_contrastive_loss": 0.9108,
      "train_positive_log_prob": -82.3086,
      "train_positive_token_accuracy": 0.0781,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.3514,
      "epoch": 3.704288939051919,
      "grad_norm": 12.292375564575195,
      "learning_rate": 1.636413440276326e-06,
      "lm_loss": 5.3901,
      "loss": 1.3336,
      "step": 1641,
      "text_contrastive_loss": 0.8864,
      "train_positive_log_prob": -82.3257,
      "train_positive_token_accuracy": 0.0826,
      "train_positive_token_prob": 0.033
    },
    {
      "contrastive_loss": 0.3296,
      "epoch": 3.706546275395034,
      "grad_norm": 10.76636791229248,
      "learning_rate": 1.631048708240736e-06,
      "lm_loss": 5.5827,
      "loss": 1.2695,
      "step": 1642,
      "text_contrastive_loss": 0.7634,
      "train_positive_log_prob": -83.3671,
      "train_positive_token_accuracy": 0.0735,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.3658,
      "epoch": 3.708803611738149,
      "grad_norm": 12.98524284362793,
      "learning_rate": 1.6256910699995921e-06,
      "lm_loss": 5.3857,
      "loss": 1.2606,
      "step": 1643,
      "text_contrastive_loss": 0.7125,
      "train_positive_log_prob": -79.5867,
      "train_positive_token_accuracy": 0.0814,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.2739,
      "epoch": 3.711060948081264,
      "grad_norm": 11.67389965057373,
      "learning_rate": 1.620340536834139e-06,
      "lm_loss": 5.4435,
      "loss": 1.1752,
      "step": 1644,
      "text_contrastive_loss": 0.7139,
      "train_positive_log_prob": -80.2704,
      "train_positive_token_accuracy": 0.0879,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.3485,
      "epoch": 3.7133182844243793,
      "grad_norm": 13.924629211425781,
      "learning_rate": 1.6149971200106723e-06,
      "lm_loss": 5.3682,
      "loss": 1.3164,
      "step": 1645,
      "text_contrastive_loss": 0.8623,
      "train_positive_log_prob": -78.9962,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.3264,
      "epoch": 3.7155756207674946,
      "grad_norm": 12.823797225952148,
      "learning_rate": 1.6096608307804973e-06,
      "lm_loss": 5.3535,
      "loss": 1.2174,
      "step": 1646,
      "text_contrastive_loss": 0.7113,
      "train_positive_log_prob": -77.8711,
      "train_positive_token_accuracy": 0.0738,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3062,
      "epoch": 3.7178329571106095,
      "grad_norm": 10.675153732299805,
      "learning_rate": 1.604331680379908e-06,
      "lm_loss": 5.4783,
      "loss": 1.164,
      "step": 1647,
      "text_contrastive_loss": 0.6199,
      "train_positive_log_prob": -82.2805,
      "train_positive_token_accuracy": 0.0793,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.2938,
      "epoch": 3.7200902934537243,
      "grad_norm": 12.304091453552246,
      "learning_rate": 1.599009680030173e-06,
      "lm_loss": 5.5528,
      "loss": 1.287,
      "step": 1648,
      "text_contrastive_loss": 0.8758,
      "train_positive_log_prob": -83.4886,
      "train_positive_token_accuracy": 0.0787,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3439,
      "epoch": 3.7223476297968396,
      "grad_norm": 14.056093215942383,
      "learning_rate": 1.5936948409375007e-06,
      "lm_loss": 5.4472,
      "loss": 1.3349,
      "step": 1649,
      "text_contrastive_loss": 0.8927,
      "train_positive_log_prob": -81.1791,
      "train_positive_token_accuracy": 0.0868,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.4163,
      "epoch": 3.724604966139955,
      "grad_norm": 14.787209510803223,
      "learning_rate": 1.5883871742930257e-06,
      "lm_loss": 5.5042,
      "loss": 1.332,
      "step": 1650,
      "text_contrastive_loss": 0.7305,
      "train_positive_log_prob": -81.7026,
      "train_positive_token_accuracy": 0.0745,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.33,
      "epoch": 3.72686230248307,
      "grad_norm": 11.547781944274902,
      "learning_rate": 1.5830866912727722e-06,
      "lm_loss": 5.4291,
      "loss": 1.2499,
      "step": 1651,
      "text_contrastive_loss": 0.754,
      "train_positive_log_prob": -81.3766,
      "train_positive_token_accuracy": 0.0847,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.3267,
      "epoch": 3.729119638826185,
      "grad_norm": 13.305439949035645,
      "learning_rate": 1.5777934030376445e-06,
      "lm_loss": 5.3766,
      "loss": 1.2505,
      "step": 1652,
      "text_contrastive_loss": 0.7724,
      "train_positive_log_prob": -81.0454,
      "train_positive_token_accuracy": 0.085,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.2939,
      "epoch": 3.7313769751693,
      "grad_norm": 12.66342830657959,
      "learning_rate": 1.5725073207333963e-06,
      "lm_loss": 5.3784,
      "loss": 1.2235,
      "step": 1653,
      "text_contrastive_loss": 0.7835,
      "train_positive_log_prob": -82.6539,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.4104,
      "epoch": 3.7336343115124153,
      "grad_norm": 15.684741020202637,
      "learning_rate": 1.5672284554906087e-06,
      "lm_loss": 5.4047,
      "loss": 1.2856,
      "step": 1654,
      "text_contrastive_loss": 0.6696,
      "train_positive_log_prob": -80.0549,
      "train_positive_token_accuracy": 0.0741,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.4281,
      "epoch": 3.7358916478555306,
      "grad_norm": 13.89198112487793,
      "learning_rate": 1.561956818424661e-06,
      "lm_loss": 5.4007,
      "loss": 1.2978,
      "step": 1655,
      "text_contrastive_loss": 0.6592,
      "train_positive_log_prob": -79.3625,
      "train_positive_token_accuracy": 0.0763,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3196,
      "epoch": 3.7381489841986455,
      "grad_norm": 13.032612800598145,
      "learning_rate": 1.5566924206357187e-06,
      "lm_loss": 5.4677,
      "loss": 1.1883,
      "step": 1656,
      "text_contrastive_loss": 0.6438,
      "train_positive_log_prob": -80.6727,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3218,
      "epoch": 3.740406320541761,
      "grad_norm": 12.352042198181152,
      "learning_rate": 1.5514352732087024e-06,
      "lm_loss": 5.4245,
      "loss": 1.1817,
      "step": 1657,
      "text_contrastive_loss": 0.6349,
      "train_positive_log_prob": -80.8585,
      "train_positive_token_accuracy": 0.076,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.4089,
      "epoch": 3.7426636568848757,
      "grad_norm": 12.809606552124023,
      "learning_rate": 1.5461853872132648e-06,
      "lm_loss": 5.3186,
      "loss": 1.357,
      "step": 1658,
      "text_contrastive_loss": 0.8324,
      "train_positive_log_prob": -78.6184,
      "train_positive_token_accuracy": 0.0801,
      "train_positive_token_prob": 0.0325
    },
    {
      "contrastive_loss": 0.332,
      "epoch": 3.744920993227991,
      "grad_norm": 11.478001594543457,
      "learning_rate": 1.5409427737037713e-06,
      "lm_loss": 5.3247,
      "loss": 1.2684,
      "step": 1659,
      "text_contrastive_loss": 0.8079,
      "train_positive_log_prob": -79.268,
      "train_positive_token_accuracy": 0.0811,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.323,
      "epoch": 3.7471783295711063,
      "grad_norm": 12.333276748657227,
      "learning_rate": 1.5357074437192688e-06,
      "lm_loss": 5.4138,
      "loss": 1.2353,
      "step": 1660,
      "text_contrastive_loss": 0.7418,
      "train_positive_log_prob": -78.9674,
      "train_positive_token_accuracy": 0.0769,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.328,
      "epoch": 3.749435665914221,
      "grad_norm": 12.25515079498291,
      "learning_rate": 1.5304794082834713e-06,
      "lm_loss": 5.4627,
      "loss": 1.2554,
      "step": 1661,
      "text_contrastive_loss": 0.7624,
      "train_positive_log_prob": -81.5765,
      "train_positive_token_accuracy": 0.0829,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3308,
      "epoch": 3.7516930022573365,
      "grad_norm": 12.769706726074219,
      "learning_rate": 1.5252586784047374e-06,
      "lm_loss": 5.51,
      "loss": 1.3018,
      "step": 1662,
      "text_contrastive_loss": 0.84,
      "train_positive_log_prob": -81.1442,
      "train_positive_token_accuracy": 0.0857,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.3141,
      "epoch": 3.7539503386004514,
      "grad_norm": 12.316630363464355,
      "learning_rate": 1.520045265076034e-06,
      "lm_loss": 5.4767,
      "loss": 1.1797,
      "step": 1663,
      "text_contrastive_loss": 0.6358,
      "train_positive_log_prob": -80.6773,
      "train_positive_token_accuracy": 0.0777,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3972,
      "epoch": 3.7562076749435667,
      "grad_norm": 13.816545486450195,
      "learning_rate": 1.5148391792749272e-06,
      "lm_loss": 5.3845,
      "loss": 1.3507,
      "step": 1664,
      "text_contrastive_loss": 0.83,
      "train_positive_log_prob": -78.4986,
      "train_positive_token_accuracy": 0.0825,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.3861,
      "epoch": 3.758465011286682,
      "grad_norm": 13.981588363647461,
      "learning_rate": 1.5096404319635533e-06,
      "lm_loss": 5.2765,
      "loss": 1.3035,
      "step": 1665,
      "text_contrastive_loss": 0.7795,
      "train_positive_log_prob": -76.9203,
      "train_positive_token_accuracy": 0.0898,
      "train_positive_token_prob": 0.033
    },
    {
      "contrastive_loss": 0.2476,
      "epoch": 3.760722347629797,
      "grad_norm": 11.404553413391113,
      "learning_rate": 1.5044490340885987e-06,
      "lm_loss": 5.3712,
      "loss": 1.074,
      "step": 1666,
      "text_contrastive_loss": 0.5786,
      "train_positive_log_prob": -79.924,
      "train_positive_token_accuracy": 0.0835,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.457,
      "epoch": 3.7629796839729117,
      "grad_norm": 15.401835441589355,
      "learning_rate": 1.4992649965812673e-06,
      "lm_loss": 5.3711,
      "loss": 1.4998,
      "step": 1667,
      "text_contrastive_loss": 1.0114,
      "train_positive_log_prob": -76.517,
      "train_positive_token_accuracy": 0.076,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.3473,
      "epoch": 3.765237020316027,
      "grad_norm": 10.748358726501465,
      "learning_rate": 1.4940883303572724e-06,
      "lm_loss": 5.3974,
      "loss": 1.2184,
      "step": 1668,
      "text_contrastive_loss": 0.6627,
      "train_positive_log_prob": -81.0664,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.4122,
      "epoch": 3.7674943566591423,
      "grad_norm": 13.811262130737305,
      "learning_rate": 1.4889190463168019e-06,
      "lm_loss": 5.3403,
      "loss": 1.322,
      "step": 1669,
      "text_contrastive_loss": 0.7516,
      "train_positive_log_prob": -79.8809,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.35,
      "epoch": 3.769751693002257,
      "grad_norm": 12.622626304626465,
      "learning_rate": 1.483757155344503e-06,
      "lm_loss": 5.5095,
      "loss": 1.3177,
      "step": 1670,
      "text_contrastive_loss": 0.8334,
      "train_positive_log_prob": -80.8839,
      "train_positive_token_accuracy": 0.078,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.3512,
      "epoch": 3.7720090293453725,
      "grad_norm": 12.186516761779785,
      "learning_rate": 1.47860266830945e-06,
      "lm_loss": 5.4187,
      "loss": 1.2653,
      "step": 1671,
      "text_contrastive_loss": 0.7445,
      "train_positive_log_prob": -80.6886,
      "train_positive_token_accuracy": 0.0829,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.2086,
      "epoch": 3.7742663656884874,
      "grad_norm": 9.920574188232422,
      "learning_rate": 1.473455596065133e-06,
      "lm_loss": 5.4306,
      "loss": 1.0939,
      "step": 1672,
      "text_contrastive_loss": 0.6845,
      "train_positive_log_prob": -80.9688,
      "train_positive_token_accuracy": 0.0757,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3392,
      "epoch": 3.7765237020316027,
      "grad_norm": 13.45047378540039,
      "learning_rate": 1.4683159494494259e-06,
      "lm_loss": 5.4241,
      "loss": 1.2624,
      "step": 1673,
      "text_contrastive_loss": 0.7617,
      "train_positive_log_prob": -79.3485,
      "train_positive_token_accuracy": 0.0769,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.4582,
      "epoch": 3.778781038374718,
      "grad_norm": 13.828205108642578,
      "learning_rate": 1.4631837392845694e-06,
      "lm_loss": 5.4319,
      "loss": 1.4233,
      "step": 1674,
      "text_contrastive_loss": 0.8439,
      "train_positive_log_prob": -81.7987,
      "train_positive_token_accuracy": 0.0744,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.315,
      "epoch": 3.781038374717833,
      "grad_norm": 11.94735336303711,
      "learning_rate": 1.4580589763771413e-06,
      "lm_loss": 5.4159,
      "loss": 1.2562,
      "step": 1675,
      "text_contrastive_loss": 0.7992,
      "train_positive_log_prob": -80.2065,
      "train_positive_token_accuracy": 0.0807,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3714,
      "epoch": 3.783295711060948,
      "grad_norm": 12.656200408935547,
      "learning_rate": 1.4529416715180434e-06,
      "lm_loss": 5.364,
      "loss": 1.2811,
      "step": 1676,
      "text_contrastive_loss": 0.7466,
      "train_positive_log_prob": -80.126,
      "train_positive_token_accuracy": 0.0803,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.2728,
      "epoch": 3.785553047404063,
      "grad_norm": 11.234992027282715,
      "learning_rate": 1.44783183548247e-06,
      "lm_loss": 5.4339,
      "loss": 1.1794,
      "step": 1677,
      "text_contrastive_loss": 0.7265,
      "train_positive_log_prob": -80.6438,
      "train_positive_token_accuracy": 0.0834,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.25,
      "epoch": 3.7878103837471784,
      "grad_norm": 11.732927322387695,
      "learning_rate": 1.4427294790298902e-06,
      "lm_loss": 5.4708,
      "loss": 1.1064,
      "step": 1678,
      "text_contrastive_loss": 0.6186,
      "train_positive_log_prob": -81.0598,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3426,
      "epoch": 3.7900677200902937,
      "grad_norm": 12.424903869628906,
      "learning_rate": 1.4376346129040243e-06,
      "lm_loss": 5.4416,
      "loss": 1.2513,
      "step": 1679,
      "text_contrastive_loss": 0.7291,
      "train_positive_log_prob": -79.2237,
      "train_positive_token_accuracy": 0.0783,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3115,
      "epoch": 3.7923250564334086,
      "grad_norm": 12.56497859954834,
      "learning_rate": 1.432547247832819e-06,
      "lm_loss": 5.417,
      "loss": 1.2445,
      "step": 1680,
      "text_contrastive_loss": 0.7825,
      "train_positive_log_prob": -79.5602,
      "train_positive_token_accuracy": 0.0742,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3519,
      "epoch": 3.7945823927765234,
      "grad_norm": 14.999197959899902,
      "learning_rate": 1.4274673945284278e-06,
      "lm_loss": 5.364,
      "loss": 1.2122,
      "step": 1681,
      "text_contrastive_loss": 0.6477,
      "train_positive_log_prob": -80.2879,
      "train_positive_token_accuracy": 0.0856,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.3976,
      "epoch": 3.7968397291196387,
      "grad_norm": 13.894376754760742,
      "learning_rate": 1.422395063687188e-06,
      "lm_loss": 5.4293,
      "loss": 1.304,
      "step": 1682,
      "text_contrastive_loss": 0.7269,
      "train_positive_log_prob": -80.506,
      "train_positive_token_accuracy": 0.0819,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.4057,
      "epoch": 3.799097065462754,
      "grad_norm": 16.181882858276367,
      "learning_rate": 1.4173302659895938e-06,
      "lm_loss": 5.462,
      "loss": 1.2965,
      "step": 1683,
      "text_contrastive_loss": 0.6891,
      "train_positive_log_prob": -80.4106,
      "train_positive_token_accuracy": 0.0844,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.3992,
      "epoch": 3.801354401805869,
      "grad_norm": 14.336645126342773,
      "learning_rate": 1.4122730121002808e-06,
      "lm_loss": 5.4631,
      "loss": 1.3213,
      "step": 1684,
      "text_contrastive_loss": 0.7515,
      "train_positive_log_prob": -81.6489,
      "train_positive_token_accuracy": 0.0831,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.2964,
      "epoch": 3.8036117381489842,
      "grad_norm": 12.844280242919922,
      "learning_rate": 1.4072233126679985e-06,
      "lm_loss": 5.5475,
      "loss": 1.1546,
      "step": 1685,
      "text_contrastive_loss": 0.607,
      "train_positive_log_prob": -82.3203,
      "train_positive_token_accuracy": 0.0769,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.3262,
      "epoch": 3.805869074492099,
      "grad_norm": 12.758700370788574,
      "learning_rate": 1.4021811783255912e-06,
      "lm_loss": 5.5363,
      "loss": 1.271,
      "step": 1686,
      "text_contrastive_loss": 0.7824,
      "train_positive_log_prob": -80.8609,
      "train_positive_token_accuracy": 0.0765,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.396,
      "epoch": 3.8081264108352144,
      "grad_norm": 14.34169864654541,
      "learning_rate": 1.3971466196899697e-06,
      "lm_loss": 5.5657,
      "loss": 1.3342,
      "step": 1687,
      "text_contrastive_loss": 0.7632,
      "train_positive_log_prob": -81.226,
      "train_positive_token_accuracy": 0.0739,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.3609,
      "epoch": 3.8103837471783297,
      "grad_norm": 12.53060531616211,
      "learning_rate": 1.3921196473620975e-06,
      "lm_loss": 5.4392,
      "loss": 1.2955,
      "step": 1688,
      "text_contrastive_loss": 0.7814,
      "train_positive_log_prob": -80.1826,
      "train_positive_token_accuracy": 0.0779,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.299,
      "epoch": 3.8126410835214446,
      "grad_norm": 12.580927848815918,
      "learning_rate": 1.3871002719269616e-06,
      "lm_loss": 5.4511,
      "loss": 1.2568,
      "step": 1689,
      "text_contrastive_loss": 0.8254,
      "train_positive_log_prob": -78.8264,
      "train_positive_token_accuracy": 0.0851,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3082,
      "epoch": 3.81489841986456,
      "grad_norm": 12.516457557678223,
      "learning_rate": 1.3820885039535564e-06,
      "lm_loss": 5.3756,
      "loss": 1.1959,
      "step": 1690,
      "text_contrastive_loss": 0.7003,
      "train_positive_log_prob": -79.8363,
      "train_positive_token_accuracy": 0.081,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.3517,
      "epoch": 3.8171557562076748,
      "grad_norm": 14.051484107971191,
      "learning_rate": 1.3770843539948508e-06,
      "lm_loss": 5.5623,
      "loss": 1.2839,
      "step": 1691,
      "text_contrastive_loss": 0.7519,
      "train_positive_log_prob": -81.8257,
      "train_positive_token_accuracy": 0.0751,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3826,
      "epoch": 3.81941309255079,
      "grad_norm": 14.231378555297852,
      "learning_rate": 1.3720878325877785e-06,
      "lm_loss": 5.3975,
      "loss": 1.3171,
      "step": 1692,
      "text_contrastive_loss": 0.7895,
      "train_positive_log_prob": -79.0235,
      "train_positive_token_accuracy": 0.0771,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3225,
      "epoch": 3.8216704288939054,
      "grad_norm": 12.910148620605469,
      "learning_rate": 1.3670989502532089e-06,
      "lm_loss": 5.4088,
      "loss": 1.3314,
      "step": 1693,
      "text_contrastive_loss": 0.936,
      "train_positive_log_prob": -81.1756,
      "train_positive_token_accuracy": 0.0809,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3416,
      "epoch": 3.8239277652370203,
      "grad_norm": 13.2839994430542,
      "learning_rate": 1.362117717495926e-06,
      "lm_loss": 5.3833,
      "loss": 1.3033,
      "step": 1694,
      "text_contrastive_loss": 0.8468,
      "train_positive_log_prob": -78.4732,
      "train_positive_token_accuracy": 0.078,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.3205,
      "epoch": 3.8261851015801356,
      "grad_norm": 12.949750900268555,
      "learning_rate": 1.3571441448046086e-06,
      "lm_loss": 5.4558,
      "loss": 1.2313,
      "step": 1695,
      "text_contrastive_loss": 0.7303,
      "train_positive_log_prob": -81.1304,
      "train_positive_token_accuracy": 0.0835,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.3511,
      "epoch": 3.8284424379232505,
      "grad_norm": 12.76517105102539,
      "learning_rate": 1.3521782426517988e-06,
      "lm_loss": 5.4325,
      "loss": 1.3254,
      "step": 1696,
      "text_contrastive_loss": 0.8621,
      "train_positive_log_prob": -79.7705,
      "train_positive_token_accuracy": 0.0766,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3614,
      "epoch": 3.8306997742663658,
      "grad_norm": 12.243370056152344,
      "learning_rate": 1.3472200214938974e-06,
      "lm_loss": 5.552,
      "loss": 1.2319,
      "step": 1697,
      "text_contrastive_loss": 0.6306,
      "train_positive_log_prob": -84.407,
      "train_positive_token_accuracy": 0.0804,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.4119,
      "epoch": 3.832957110609481,
      "grad_norm": 12.767352104187012,
      "learning_rate": 1.3422694917711276e-06,
      "lm_loss": 5.4179,
      "loss": 1.3749,
      "step": 1698,
      "text_contrastive_loss": 0.8424,
      "train_positive_log_prob": -79.1732,
      "train_positive_token_accuracy": 0.0826,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3748,
      "epoch": 3.835214446952596,
      "grad_norm": 13.090643882751465,
      "learning_rate": 1.3373266639075134e-06,
      "lm_loss": 5.3874,
      "loss": 1.281,
      "step": 1699,
      "text_contrastive_loss": 0.7349,
      "train_positive_log_prob": -81.1469,
      "train_positive_token_accuracy": 0.083,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.3711,
      "epoch": 3.837471783295711,
      "grad_norm": 13.619207382202148,
      "learning_rate": 1.3323915483108662e-06,
      "lm_loss": 5.3796,
      "loss": 1.2763,
      "step": 1700,
      "text_contrastive_loss": 0.7345,
      "train_positive_log_prob": -76.7199,
      "train_positive_token_accuracy": 0.0715,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.4036,
      "epoch": 3.839729119638826,
      "grad_norm": 12.674599647521973,
      "learning_rate": 1.3274641553727568e-06,
      "lm_loss": 5.448,
      "loss": 1.3591,
      "step": 1701,
      "text_contrastive_loss": 0.8213,
      "train_positive_log_prob": -80.635,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.4016,
      "epoch": 3.8419864559819414,
      "grad_norm": 12.56098747253418,
      "learning_rate": 1.3225444954684962e-06,
      "lm_loss": 5.2899,
      "loss": 1.3108,
      "step": 1702,
      "text_contrastive_loss": 0.7605,
      "train_positive_log_prob": -78.6033,
      "train_positive_token_accuracy": 0.0833,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.3472,
      "epoch": 3.8442437923250563,
      "grad_norm": 14.493278503417969,
      "learning_rate": 1.3176325789571075e-06,
      "lm_loss": 5.4614,
      "loss": 1.3333,
      "step": 1703,
      "text_contrastive_loss": 0.88,
      "train_positive_log_prob": -80.5549,
      "train_positive_token_accuracy": 0.0845,
      "train_positive_token_prob": 0.0325
    },
    {
      "contrastive_loss": 0.3124,
      "epoch": 3.8465011286681716,
      "grad_norm": 11.662528991699219,
      "learning_rate": 1.3127284161813153e-06,
      "lm_loss": 5.4786,
      "loss": 1.2094,
      "step": 1704,
      "text_contrastive_loss": 0.6982,
      "train_positive_log_prob": -80.4709,
      "train_positive_token_accuracy": 0.0766,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3871,
      "epoch": 3.8487584650112865,
      "grad_norm": 14.075630187988281,
      "learning_rate": 1.3078320174675141e-06,
      "lm_loss": 5.5057,
      "loss": 1.3205,
      "step": 1705,
      "text_contrastive_loss": 0.7656,
      "train_positive_log_prob": -82.8651,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3457,
      "epoch": 3.851015801354402,
      "grad_norm": 12.885129928588867,
      "learning_rate": 1.3029433931257524e-06,
      "lm_loss": 5.4407,
      "loss": 1.2158,
      "step": 1706,
      "text_contrastive_loss": 0.652,
      "train_positive_log_prob": -80.4417,
      "train_positive_token_accuracy": 0.0819,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.3413,
      "epoch": 3.853273137697517,
      "grad_norm": 11.768065452575684,
      "learning_rate": 1.2980625534497037e-06,
      "lm_loss": 5.3975,
      "loss": 1.3302,
      "step": 1707,
      "text_contrastive_loss": 0.8983,
      "train_positive_log_prob": -79.2785,
      "train_positive_token_accuracy": 0.0753,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3214,
      "epoch": 3.855530474040632,
      "grad_norm": 12.895203590393066,
      "learning_rate": 1.2931895087166551e-06,
      "lm_loss": 5.4598,
      "loss": 1.2505,
      "step": 1708,
      "text_contrastive_loss": 0.7661,
      "train_positive_log_prob": -79.2966,
      "train_positive_token_accuracy": 0.0753,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.302,
      "epoch": 3.8577878103837473,
      "grad_norm": 11.259028434753418,
      "learning_rate": 1.2883242691874792e-06,
      "lm_loss": 5.463,
      "loss": 1.1774,
      "step": 1709,
      "text_contrastive_loss": 0.6582,
      "train_positive_log_prob": -81.8617,
      "train_positive_token_accuracy": 0.0769,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3362,
      "epoch": 3.860045146726862,
      "grad_norm": 11.875563621520996,
      "learning_rate": 1.2834668451066118e-06,
      "lm_loss": 5.3559,
      "loss": 1.2851,
      "step": 1710,
      "text_contrastive_loss": 0.8265,
      "train_positive_log_prob": -82.1208,
      "train_positive_token_accuracy": 0.0752,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3458,
      "epoch": 3.8623024830699775,
      "grad_norm": 12.237154960632324,
      "learning_rate": 1.2786172467020357e-06,
      "lm_loss": 5.3498,
      "loss": 1.2954,
      "step": 1711,
      "text_contrastive_loss": 0.8292,
      "train_positive_log_prob": -79.0932,
      "train_positive_token_accuracy": 0.0766,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.4083,
      "epoch": 3.864559819413093,
      "grad_norm": 12.66635799407959,
      "learning_rate": 1.2737754841852501e-06,
      "lm_loss": 5.4067,
      "loss": 1.3198,
      "step": 1712,
      "text_contrastive_loss": 0.7417,
      "train_positive_log_prob": -80.6356,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.2855,
      "epoch": 3.8668171557562077,
      "grad_norm": 10.827605247497559,
      "learning_rate": 1.2689415677512574e-06,
      "lm_loss": 5.3867,
      "loss": 1.2149,
      "step": 1713,
      "text_contrastive_loss": 0.7815,
      "train_positive_log_prob": -79.4785,
      "train_positive_token_accuracy": 0.0792,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3468,
      "epoch": 3.8690744920993225,
      "grad_norm": 11.63575553894043,
      "learning_rate": 1.2641155075785444e-06,
      "lm_loss": 5.4074,
      "loss": 1.3457,
      "step": 1714,
      "text_contrastive_loss": 0.9163,
      "train_positive_log_prob": -81.1527,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3969,
      "epoch": 3.871331828442438,
      "grad_norm": 15.075748443603516,
      "learning_rate": 1.259297313829046e-06,
      "lm_loss": 5.369,
      "loss": 1.2969,
      "step": 1715,
      "text_contrastive_loss": 0.7261,
      "train_positive_log_prob": -78.6736,
      "train_positive_token_accuracy": 0.0762,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3952,
      "epoch": 3.873589164785553,
      "grad_norm": 13.494071960449219,
      "learning_rate": 1.2544869966481389e-06,
      "lm_loss": 5.5108,
      "loss": 1.2823,
      "step": 1716,
      "text_contrastive_loss": 0.6721,
      "train_positive_log_prob": -82.2344,
      "train_positive_token_accuracy": 0.0759,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.3156,
      "epoch": 3.875846501128668,
      "grad_norm": 11.806557655334473,
      "learning_rate": 1.249684566164614e-06,
      "lm_loss": 5.3949,
      "loss": 1.2527,
      "step": 1717,
      "text_contrastive_loss": 0.7952,
      "train_positive_log_prob": -78.6854,
      "train_positive_token_accuracy": 0.0814,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.4408,
      "epoch": 3.8781038374717833,
      "grad_norm": 13.002033233642578,
      "learning_rate": 1.2448900324906559e-06,
      "lm_loss": 5.3442,
      "loss": 1.3488,
      "step": 1718,
      "text_contrastive_loss": 0.7472,
      "train_positive_log_prob": -78.8077,
      "train_positive_token_accuracy": 0.0879,
      "train_positive_token_prob": 0.0327
    },
    {
      "contrastive_loss": 0.3307,
      "epoch": 3.880361173814898,
      "grad_norm": 13.509655952453613,
      "learning_rate": 1.2401034057218181e-06,
      "lm_loss": 5.316,
      "loss": 1.2387,
      "step": 1719,
      "text_contrastive_loss": 0.7528,
      "train_positive_log_prob": -76.8592,
      "train_positive_token_accuracy": 0.0778,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.363,
      "epoch": 3.8826185101580135,
      "grad_norm": 12.645583152770996,
      "learning_rate": 1.2353246959370086e-06,
      "lm_loss": 5.4205,
      "loss": 1.3323,
      "step": 1720,
      "text_contrastive_loss": 0.8545,
      "train_positive_log_prob": -80.23,
      "train_positive_token_accuracy": 0.0744,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.42,
      "epoch": 3.884875846501129,
      "grad_norm": 14.332308769226074,
      "learning_rate": 1.2305539131984646e-06,
      "lm_loss": 5.3384,
      "loss": 1.4214,
      "step": 1721,
      "text_contrastive_loss": 0.9351,
      "train_positive_log_prob": -78.5759,
      "train_positive_token_accuracy": 0.0767,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.4018,
      "epoch": 3.8871331828442437,
      "grad_norm": 12.9856538772583,
      "learning_rate": 1.2257910675517315e-06,
      "lm_loss": 5.3644,
      "loss": 1.3223,
      "step": 1722,
      "text_contrastive_loss": 0.7682,
      "train_positive_log_prob": -78.6038,
      "train_positive_token_accuracy": 0.0792,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.2709,
      "epoch": 3.889390519187359,
      "grad_norm": 11.077625274658203,
      "learning_rate": 1.22103616902564e-06,
      "lm_loss": 5.3067,
      "loss": 1.2186,
      "step": 1723,
      "text_contrastive_loss": 0.8341,
      "train_positive_log_prob": -78.7509,
      "train_positive_token_accuracy": 0.0795,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.4286,
      "epoch": 3.891647855530474,
      "grad_norm": 15.429622650146484,
      "learning_rate": 1.21628922763229e-06,
      "lm_loss": 5.379,
      "loss": 1.3677,
      "step": 1724,
      "text_contrastive_loss": 0.8025,
      "train_positive_log_prob": -78.4739,
      "train_positive_token_accuracy": 0.0758,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3737,
      "epoch": 3.893905191873589,
      "grad_norm": 13.851577758789062,
      "learning_rate": 1.2115502533670253e-06,
      "lm_loss": 5.6074,
      "loss": 1.3588,
      "step": 1725,
      "text_contrastive_loss": 0.8488,
      "train_positive_log_prob": -84.7875,
      "train_positive_token_accuracy": 0.0783,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.4194,
      "epoch": 3.8961625282167045,
      "grad_norm": 15.04504108428955,
      "learning_rate": 1.2068192562084146e-06,
      "lm_loss": 5.2553,
      "loss": 1.3722,
      "step": 1726,
      "text_contrastive_loss": 0.8546,
      "train_positive_log_prob": -76.0595,
      "train_positive_token_accuracy": 0.0856,
      "train_positive_token_prob": 0.0333
    },
    {
      "contrastive_loss": 0.3581,
      "epoch": 3.8984198645598194,
      "grad_norm": 11.373048782348633,
      "learning_rate": 1.2020962461182268e-06,
      "lm_loss": 5.4031,
      "loss": 1.2518,
      "step": 1727,
      "text_contrastive_loss": 0.7068,
      "train_positive_log_prob": -80.0769,
      "train_positive_token_accuracy": 0.0793,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.338,
      "epoch": 3.9006772009029347,
      "grad_norm": 15.169660568237305,
      "learning_rate": 1.1973812330414159e-06,
      "lm_loss": 5.3122,
      "loss": 1.2035,
      "step": 1728,
      "text_contrastive_loss": 0.6685,
      "train_positive_log_prob": -77.5892,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3461,
      "epoch": 3.9029345372460496,
      "grad_norm": 13.146347045898438,
      "learning_rate": 1.1926742269060965e-06,
      "lm_loss": 5.4069,
      "loss": 1.2937,
      "step": 1729,
      "text_contrastive_loss": 0.8137,
      "train_positive_log_prob": -79.2476,
      "train_positive_token_accuracy": 0.0858,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3596,
      "epoch": 3.905191873589165,
      "grad_norm": 12.391469955444336,
      "learning_rate": 1.1879752376235231e-06,
      "lm_loss": 5.4798,
      "loss": 1.2918,
      "step": 1730,
      "text_contrastive_loss": 0.7685,
      "train_positive_log_prob": -78.5356,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.3001,
      "epoch": 3.90744920993228,
      "grad_norm": 11.998820304870605,
      "learning_rate": 1.1832842750880702e-06,
      "lm_loss": 5.4692,
      "loss": 1.2222,
      "step": 1731,
      "text_contrastive_loss": 0.7504,
      "train_positive_log_prob": -79.6677,
      "train_positive_token_accuracy": 0.081,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.3504,
      "epoch": 3.909706546275395,
      "grad_norm": 12.540609359741211,
      "learning_rate": 1.1786013491772103e-06,
      "lm_loss": 5.3138,
      "loss": 1.277,
      "step": 1732,
      "text_contrastive_loss": 0.7904,
      "train_positive_log_prob": -78.1723,
      "train_positive_token_accuracy": 0.082,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.3438,
      "epoch": 3.91196388261851,
      "grad_norm": 13.03815746307373,
      "learning_rate": 1.173926469751493e-06,
      "lm_loss": 5.4266,
      "loss": 1.2882,
      "step": 1733,
      "text_contrastive_loss": 0.8034,
      "train_positive_log_prob": -79.0074,
      "train_positive_token_accuracy": 0.0816,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3338,
      "epoch": 3.9142212189616252,
      "grad_norm": 10.691126823425293,
      "learning_rate": 1.1692596466545275e-06,
      "lm_loss": 5.4213,
      "loss": 1.2798,
      "step": 1734,
      "text_contrastive_loss": 0.8077,
      "train_positive_log_prob": -81.9562,
      "train_positive_token_accuracy": 0.0879,
      "train_positive_token_prob": 0.0329
    },
    {
      "contrastive_loss": 0.2949,
      "epoch": 3.9164785553047405,
      "grad_norm": 11.609354972839355,
      "learning_rate": 1.1646008897129546e-06,
      "lm_loss": 5.4752,
      "loss": 1.2089,
      "step": 1735,
      "text_contrastive_loss": 0.733,
      "train_positive_log_prob": -80.8048,
      "train_positive_token_accuracy": 0.0776,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3394,
      "epoch": 3.9187358916478554,
      "grad_norm": 13.065603256225586,
      "learning_rate": 1.1599502087364345e-06,
      "lm_loss": 5.5566,
      "loss": 1.227,
      "step": 1736,
      "text_contrastive_loss": 0.6638,
      "train_positive_log_prob": -83.5901,
      "train_positive_token_accuracy": 0.0779,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.4092,
      "epoch": 3.9209932279909707,
      "grad_norm": 12.14127254486084,
      "learning_rate": 1.1553076135176222e-06,
      "lm_loss": 5.5211,
      "loss": 1.3445,
      "step": 1737,
      "text_contrastive_loss": 0.7664,
      "train_positive_log_prob": -80.5517,
      "train_positive_token_accuracy": 0.0779,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.328,
      "epoch": 3.9232505643340856,
      "grad_norm": 12.346694946289062,
      "learning_rate": 1.1506731138321474e-06,
      "lm_loss": 5.4705,
      "loss": 1.2178,
      "step": 1738,
      "text_contrastive_loss": 0.6854,
      "train_positive_log_prob": -78.8081,
      "train_positive_token_accuracy": 0.0784,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.4454,
      "epoch": 3.925507900677201,
      "grad_norm": 14.74303913116455,
      "learning_rate": 1.1460467194385889e-06,
      "lm_loss": 5.3789,
      "loss": 1.4999,
      "step": 1739,
      "text_contrastive_loss": 1.0331,
      "train_positive_log_prob": -79.3599,
      "train_positive_token_accuracy": 0.0855,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.329,
      "epoch": 3.927765237020316,
      "grad_norm": 12.316030502319336,
      "learning_rate": 1.1414284400784643e-06,
      "lm_loss": 5.4031,
      "loss": 1.2632,
      "step": 1740,
      "text_contrastive_loss": 0.7877,
      "train_positive_log_prob": -80.7535,
      "train_positive_token_accuracy": 0.0763,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.2936,
      "epoch": 3.930022573363431,
      "grad_norm": 11.090486526489258,
      "learning_rate": 1.1368182854762005e-06,
      "lm_loss": 5.49,
      "loss": 1.1501,
      "step": 1741,
      "text_contrastive_loss": 0.6149,
      "train_positive_log_prob": -81.276,
      "train_positive_token_accuracy": 0.0799,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.2949,
      "epoch": 3.9322799097065464,
      "grad_norm": 13.24492359161377,
      "learning_rate": 1.13221626533912e-06,
      "lm_loss": 5.5015,
      "loss": 1.1919,
      "step": 1742,
      "text_contrastive_loss": 0.6937,
      "train_positive_log_prob": -81.1034,
      "train_positive_token_accuracy": 0.0827,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.364,
      "epoch": 3.9345372460496613,
      "grad_norm": 13.615939140319824,
      "learning_rate": 1.1276223893574123e-06,
      "lm_loss": 5.4266,
      "loss": 1.3158,
      "step": 1743,
      "text_contrastive_loss": 0.8182,
      "train_positive_log_prob": -80.4011,
      "train_positive_token_accuracy": 0.0801,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3225,
      "epoch": 3.9367945823927766,
      "grad_norm": 11.627803802490234,
      "learning_rate": 1.1230366672041216e-06,
      "lm_loss": 5.4074,
      "loss": 1.2319,
      "step": 1744,
      "text_contrastive_loss": 0.7374,
      "train_positive_log_prob": -80.5205,
      "train_positive_token_accuracy": 0.0816,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3449,
      "epoch": 3.939051918735892,
      "grad_norm": 12.176133155822754,
      "learning_rate": 1.118459108535122e-06,
      "lm_loss": 5.3875,
      "loss": 1.253,
      "step": 1745,
      "text_contrastive_loss": 0.7387,
      "train_positive_log_prob": -78.4769,
      "train_positive_token_accuracy": 0.078,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3757,
      "epoch": 3.9413092550790068,
      "grad_norm": 13.935737609863281,
      "learning_rate": 1.1138897229890995e-06,
      "lm_loss": 5.4409,
      "loss": 1.3611,
      "step": 1746,
      "text_contrastive_loss": 0.8826,
      "train_positive_log_prob": -82.0837,
      "train_positive_token_accuracy": 0.0738,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.4234,
      "epoch": 3.9435665914221216,
      "grad_norm": 13.838811874389648,
      "learning_rate": 1.109328520187528e-06,
      "lm_loss": 5.5712,
      "loss": 1.3746,
      "step": 1747,
      "text_contrastive_loss": 0.7882,
      "train_positive_log_prob": -85.3294,
      "train_positive_token_accuracy": 0.0778,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.2762,
      "epoch": 3.945823927765237,
      "grad_norm": 12.039306640625,
      "learning_rate": 1.1047755097346541e-06,
      "lm_loss": 5.4991,
      "loss": 1.2051,
      "step": 1748,
      "text_contrastive_loss": 0.7581,
      "train_positive_log_prob": -79.0288,
      "train_positive_token_accuracy": 0.0805,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.4056,
      "epoch": 3.9480812641083523,
      "grad_norm": 13.613452911376953,
      "learning_rate": 1.100230701217473e-06,
      "lm_loss": 5.2911,
      "loss": 1.3528,
      "step": 1749,
      "text_contrastive_loss": 0.8362,
      "train_positive_log_prob": -78.0426,
      "train_positive_token_accuracy": 0.0779,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.396,
      "epoch": 3.950338600451467,
      "grad_norm": 13.061095237731934,
      "learning_rate": 1.0956941042057106e-06,
      "lm_loss": 5.4116,
      "loss": 1.3418,
      "step": 1750,
      "text_contrastive_loss": 0.8094,
      "train_positive_log_prob": -80.1604,
      "train_positive_token_accuracy": 0.0789,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3525,
      "epoch": 3.9525959367945824,
      "grad_norm": 12.334271430969238,
      "learning_rate": 1.091165728251799e-06,
      "lm_loss": 5.4548,
      "loss": 1.3168,
      "step": 1751,
      "text_contrastive_loss": 0.8377,
      "train_positive_log_prob": -81.3919,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3543,
      "epoch": 3.9548532731376973,
      "grad_norm": 12.155789375305176,
      "learning_rate": 1.0866455828908634e-06,
      "lm_loss": 5.4532,
      "loss": 1.259,
      "step": 1752,
      "text_contrastive_loss": 0.7188,
      "train_positive_log_prob": -80.2235,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3295,
      "epoch": 3.9571106094808126,
      "grad_norm": 12.81718635559082,
      "learning_rate": 1.082133677640697e-06,
      "lm_loss": 5.4989,
      "loss": 1.2976,
      "step": 1753,
      "text_contrastive_loss": 0.8363,
      "train_positive_log_prob": -80.4246,
      "train_positive_token_accuracy": 0.0839,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.3965,
      "epoch": 3.959367945823928,
      "grad_norm": 12.595709800720215,
      "learning_rate": 1.0776300220017437e-06,
      "lm_loss": 5.3596,
      "loss": 1.3096,
      "step": 1754,
      "text_contrastive_loss": 0.7541,
      "train_positive_log_prob": -78.7819,
      "train_positive_token_accuracy": 0.0749,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.3105,
      "epoch": 3.961625282167043,
      "grad_norm": 12.437752723693848,
      "learning_rate": 1.0731346254570735e-06,
      "lm_loss": 5.309,
      "loss": 1.1856,
      "step": 1755,
      "text_contrastive_loss": 0.6882,
      "train_positive_log_prob": -79.6097,
      "train_positive_token_accuracy": 0.0815,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3407,
      "epoch": 3.963882618510158,
      "grad_norm": 12.470183372497559,
      "learning_rate": 1.068647497472368e-06,
      "lm_loss": 5.4084,
      "loss": 1.2713,
      "step": 1756,
      "text_contrastive_loss": 0.7796,
      "train_positive_log_prob": -81.0631,
      "train_positive_token_accuracy": 0.0771,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3843,
      "epoch": 3.966139954853273,
      "grad_norm": 13.911128044128418,
      "learning_rate": 1.064168647495899e-06,
      "lm_loss": 5.3746,
      "loss": 1.402,
      "step": 1757,
      "text_contrastive_loss": 0.9603,
      "train_positive_log_prob": -78.6689,
      "train_positive_token_accuracy": 0.0833,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.4406,
      "epoch": 3.9683972911963883,
      "grad_norm": 13.523388862609863,
      "learning_rate": 1.0596980849585065e-06,
      "lm_loss": 5.4742,
      "loss": 1.3819,
      "step": 1758,
      "text_contrastive_loss": 0.7877,
      "train_positive_log_prob": -81.9404,
      "train_positive_token_accuracy": 0.0817,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.3728,
      "epoch": 3.9706546275395036,
      "grad_norm": 11.978256225585938,
      "learning_rate": 1.0552358192735784e-06,
      "lm_loss": 5.4214,
      "loss": 1.2713,
      "step": 1759,
      "text_contrastive_loss": 0.7129,
      "train_positive_log_prob": -80.2448,
      "train_positive_token_accuracy": 0.0765,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3509,
      "epoch": 3.9729119638826185,
      "grad_norm": 13.259681701660156,
      "learning_rate": 1.0507818598370355e-06,
      "lm_loss": 5.5014,
      "loss": 1.2838,
      "step": 1760,
      "text_contrastive_loss": 0.7654,
      "train_positive_log_prob": -84.0741,
      "train_positive_token_accuracy": 0.08,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.2749,
      "epoch": 3.975169300225734,
      "grad_norm": 11.363874435424805,
      "learning_rate": 1.0463362160273076e-06,
      "lm_loss": 5.3643,
      "loss": 1.1328,
      "step": 1761,
      "text_contrastive_loss": 0.6431,
      "train_positive_log_prob": -78.9544,
      "train_positive_token_accuracy": 0.0924,
      "train_positive_token_prob": 0.0338
    },
    {
      "contrastive_loss": 0.3748,
      "epoch": 3.9774266365688487,
      "grad_norm": 14.530806541442871,
      "learning_rate": 1.0418988972053162e-06,
      "lm_loss": 5.4712,
      "loss": 1.3201,
      "step": 1762,
      "text_contrastive_loss": 0.7964,
      "train_positive_log_prob": -80.787,
      "train_positive_token_accuracy": 0.0836,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3066,
      "epoch": 3.979683972911964,
      "grad_norm": 11.596506118774414,
      "learning_rate": 1.037469912714449e-06,
      "lm_loss": 5.514,
      "loss": 1.1512,
      "step": 1763,
      "text_contrastive_loss": 0.5864,
      "train_positive_log_prob": -82.52,
      "train_positive_token_accuracy": 0.0744,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.3326,
      "epoch": 3.9819413092550793,
      "grad_norm": 11.653396606445312,
      "learning_rate": 1.0330492718805469e-06,
      "lm_loss": 5.3952,
      "loss": 1.2658,
      "step": 1764,
      "text_contrastive_loss": 0.7874,
      "train_positive_log_prob": -79.6244,
      "train_positive_token_accuracy": 0.0847,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3091,
      "epoch": 3.984198645598194,
      "grad_norm": 11.884001731872559,
      "learning_rate": 1.0286369840118859e-06,
      "lm_loss": 5.4429,
      "loss": 1.2438,
      "step": 1765,
      "text_contrastive_loss": 0.7809,
      "train_positive_log_prob": -80.0117,
      "train_positive_token_accuracy": 0.0837,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3805,
      "epoch": 3.986455981941309,
      "grad_norm": 12.454544067382812,
      "learning_rate": 1.0242330583991507e-06,
      "lm_loss": 5.3399,
      "loss": 1.3185,
      "step": 1766,
      "text_contrastive_loss": 0.8081,
      "train_positive_log_prob": -78.6391,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.3767,
      "epoch": 3.9887133182844243,
      "grad_norm": 12.833479881286621,
      "learning_rate": 1.0198375043154142e-06,
      "lm_loss": 5.4009,
      "loss": 1.3558,
      "step": 1767,
      "text_contrastive_loss": 0.878,
      "train_positive_log_prob": -80.6305,
      "train_positive_token_accuracy": 0.0835,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3691,
      "epoch": 3.9909706546275396,
      "grad_norm": 12.905168533325195,
      "learning_rate": 1.0154503310161269e-06,
      "lm_loss": 5.4242,
      "loss": 1.3104,
      "step": 1768,
      "text_contrastive_loss": 0.7979,
      "train_positive_log_prob": -80.2363,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3059,
      "epoch": 3.9932279909706545,
      "grad_norm": 12.173251152038574,
      "learning_rate": 1.0110715477390915e-06,
      "lm_loss": 5.3933,
      "loss": 1.2139,
      "step": 1769,
      "text_contrastive_loss": 0.7373,
      "train_positive_log_prob": -77.7596,
      "train_positive_token_accuracy": 0.0848,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.4048,
      "epoch": 3.99548532731377,
      "grad_norm": 14.231117248535156,
      "learning_rate": 1.006701163704445e-06,
      "lm_loss": 5.4088,
      "loss": 1.3117,
      "step": 1770,
      "text_contrastive_loss": 0.7321,
      "train_positive_log_prob": -79.8205,
      "train_positive_token_accuracy": 0.0826,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3499,
      "epoch": 3.9977426636568847,
      "grad_norm": 11.990342140197754,
      "learning_rate": 1.0023391881146349e-06,
      "lm_loss": 5.3637,
      "loss": 1.2202,
      "step": 1771,
      "text_contrastive_loss": 0.6679,
      "train_positive_log_prob": -77.0421,
      "train_positive_token_accuracy": 0.0786,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.2676,
      "epoch": 4.0,
      "grad_norm": 16.870798110961914,
      "learning_rate": 9.97985630154407e-07,
      "lm_loss": 5.3585,
      "loss": 1.0692,
      "step": 1772,
      "text_contrastive_loss": 0.5314,
      "train_positive_log_prob": -81.5245,
      "train_positive_token_accuracy": 0.0721,
      "train_positive_token_prob": 0.0329
    },
    {
      "contrastive_loss": 0.2872,
      "epoch": 4.002257336343115,
      "grad_norm": 11.189910888671875,
      "learning_rate": 9.936404989907828e-07,
      "lm_loss": 5.3116,
      "loss": 1.1275,
      "step": 1773,
      "text_contrastive_loss": 0.6182,
      "train_positive_log_prob": -78.2434,
      "train_positive_token_accuracy": 0.0825,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.3803,
      "epoch": 4.004514672686231,
      "grad_norm": 12.189338684082031,
      "learning_rate": 9.89303803773039e-07,
      "lm_loss": 5.3849,
      "loss": 1.3012,
      "step": 1774,
      "text_contrastive_loss": 0.7648,
      "train_positive_log_prob": -78.9199,
      "train_positive_token_accuracy": 0.0784,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3288,
      "epoch": 4.006772009029345,
      "grad_norm": 12.933213233947754,
      "learning_rate": 9.849755536326866e-07,
      "lm_loss": 5.5121,
      "loss": 1.2728,
      "step": 1775,
      "text_contrastive_loss": 0.7855,
      "train_positive_log_prob": -81.8424,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.4056,
      "epoch": 4.00902934537246,
      "grad_norm": 13.452736854553223,
      "learning_rate": 9.806557576834591e-07,
      "lm_loss": 5.4404,
      "loss": 1.4022,
      "step": 1776,
      "text_contrastive_loss": 0.9051,
      "train_positive_log_prob": -83.0911,
      "train_positive_token_accuracy": 0.0772,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.2943,
      "epoch": 4.011286681715576,
      "grad_norm": 12.769314765930176,
      "learning_rate": 9.763444250212855e-07,
      "lm_loss": 5.428,
      "loss": 1.1971,
      "step": 1777,
      "text_contrastive_loss": 0.7199,
      "train_positive_log_prob": -82.0135,
      "train_positive_token_accuracy": 0.0812,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.2997,
      "epoch": 4.013544018058691,
      "grad_norm": 11.931144714355469,
      "learning_rate": 9.72041564724277e-07,
      "lm_loss": 5.364,
      "loss": 1.2264,
      "step": 1778,
      "text_contrastive_loss": 0.7807,
      "train_positive_log_prob": -79.0299,
      "train_positive_token_accuracy": 0.0819,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3139,
      "epoch": 4.015801354401806,
      "grad_norm": 11.053611755371094,
      "learning_rate": 9.677471858526998e-07,
      "lm_loss": 5.3535,
      "loss": 1.1793,
      "step": 1779,
      "text_contrastive_loss": 0.66,
      "train_positive_log_prob": -78.2922,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3045,
      "epoch": 4.018058690744921,
      "grad_norm": 11.501129150390625,
      "learning_rate": 9.63461297448966e-07,
      "lm_loss": 5.3834,
      "loss": 1.1888,
      "step": 1780,
      "text_contrastive_loss": 0.692,
      "train_positive_log_prob": -79.2184,
      "train_positive_token_accuracy": 0.077,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3407,
      "epoch": 4.020316027088036,
      "grad_norm": 11.808714866638184,
      "learning_rate": 9.59183908537607e-07,
      "lm_loss": 5.4025,
      "loss": 1.2999,
      "step": 1781,
      "text_contrastive_loss": 0.8379,
      "train_positive_log_prob": -81.1241,
      "train_positive_token_accuracy": 0.078,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.2588,
      "epoch": 4.022573363431151,
      "grad_norm": 11.92055606842041,
      "learning_rate": 9.549150281252633e-07,
      "lm_loss": 5.5553,
      "loss": 1.2261,
      "step": 1782,
      "text_contrastive_loss": 0.8235,
      "train_positive_log_prob": -82.0505,
      "train_positive_token_accuracy": 0.0857,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.291,
      "epoch": 4.024830699774267,
      "grad_norm": 12.611737251281738,
      "learning_rate": 9.506546652006504e-07,
      "lm_loss": 5.4777,
      "loss": 1.1931,
      "step": 1783,
      "text_contrastive_loss": 0.7086,
      "train_positive_log_prob": -80.498,
      "train_positive_token_accuracy": 0.0835,
      "train_positive_token_prob": 0.0326
    },
    {
      "contrastive_loss": 0.2773,
      "epoch": 4.027088036117381,
      "grad_norm": 12.908148765563965,
      "learning_rate": 9.464028287345551e-07,
      "lm_loss": 5.3637,
      "loss": 1.1642,
      "step": 1784,
      "text_contrastive_loss": 0.701,
      "train_positive_log_prob": -80.2282,
      "train_positive_token_accuracy": 0.0867,
      "train_positive_token_prob": 0.0337
    },
    {
      "contrastive_loss": 0.3173,
      "epoch": 4.029345372460496,
      "grad_norm": 12.55312728881836,
      "learning_rate": 9.421595276798084e-07,
      "lm_loss": 5.4336,
      "loss": 1.2938,
      "step": 1785,
      "text_contrastive_loss": 0.8663,
      "train_positive_log_prob": -79.3157,
      "train_positive_token_accuracy": 0.0822,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3421,
      "epoch": 4.031602708803612,
      "grad_norm": 11.590481758117676,
      "learning_rate": 9.379247709712725e-07,
      "lm_loss": 5.4248,
      "loss": 1.2646,
      "step": 1786,
      "text_contrastive_loss": 0.76,
      "train_positive_log_prob": -79.2937,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3528,
      "epoch": 4.033860045146727,
      "grad_norm": 12.176640510559082,
      "learning_rate": 9.336985675258109e-07,
      "lm_loss": 5.3758,
      "loss": 1.272,
      "step": 1787,
      "text_contrastive_loss": 0.7634,
      "train_positive_log_prob": -78.9601,
      "train_positive_token_accuracy": 0.075,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3157,
      "epoch": 4.036117381489842,
      "grad_norm": 10.483368873596191,
      "learning_rate": 9.294809262422838e-07,
      "lm_loss": 5.5366,
      "loss": 1.2181,
      "step": 1788,
      "text_contrastive_loss": 0.6974,
      "train_positive_log_prob": -83.069,
      "train_positive_token_accuracy": 0.0755,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3807,
      "epoch": 4.038374717832957,
      "grad_norm": 12.850399017333984,
      "learning_rate": 9.2527185600152e-07,
      "lm_loss": 5.4233,
      "loss": 1.2933,
      "step": 1789,
      "text_contrastive_loss": 0.7405,
      "train_positive_log_prob": -80.9718,
      "train_positive_token_accuracy": 0.0808,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.2874,
      "epoch": 4.040632054176072,
      "grad_norm": 11.69446849822998,
      "learning_rate": 9.210713656663023e-07,
      "lm_loss": 5.4678,
      "loss": 1.1651,
      "step": 1790,
      "text_contrastive_loss": 0.6618,
      "train_positive_log_prob": -81.3588,
      "train_positive_token_accuracy": 0.0798,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.2661,
      "epoch": 4.042889390519187,
      "grad_norm": 12.541370391845703,
      "learning_rate": 9.168794640813428e-07,
      "lm_loss": 5.3809,
      "loss": 1.1086,
      "step": 1791,
      "text_contrastive_loss": 0.6087,
      "train_positive_log_prob": -80.3596,
      "train_positive_token_accuracy": 0.073,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3006,
      "epoch": 4.045146726862303,
      "grad_norm": 10.826007843017578,
      "learning_rate": 9.126961600732742e-07,
      "lm_loss": 5.3845,
      "loss": 1.1871,
      "step": 1792,
      "text_contrastive_loss": 0.6963,
      "train_positive_log_prob": -80.2194,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3142,
      "epoch": 4.047404063205418,
      "grad_norm": 12.387374877929688,
      "learning_rate": 9.085214624506228e-07,
      "lm_loss": 5.3772,
      "loss": 1.279,
      "step": 1793,
      "text_contrastive_loss": 0.8542,
      "train_positive_log_prob": -79.5236,
      "train_positive_token_accuracy": 0.0885,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.3609,
      "epoch": 4.049661399548532,
      "grad_norm": 12.288250923156738,
      "learning_rate": 9.043553800037952e-07,
      "lm_loss": 5.333,
      "loss": 1.2493,
      "step": 1794,
      "text_contrastive_loss": 0.7101,
      "train_positive_log_prob": -77.4764,
      "train_positive_token_accuracy": 0.0815,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.2918,
      "epoch": 4.051918735891648,
      "grad_norm": 12.319785118103027,
      "learning_rate": 9.001979215050544e-07,
      "lm_loss": 5.4524,
      "loss": 1.2632,
      "step": 1795,
      "text_contrastive_loss": 0.8522,
      "train_positive_log_prob": -78.662,
      "train_positive_token_accuracy": 0.078,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3115,
      "epoch": 4.054176072234763,
      "grad_norm": 11.470542907714844,
      "learning_rate": 8.960490957085061e-07,
      "lm_loss": 5.3993,
      "loss": 1.1714,
      "step": 1796,
      "text_contrastive_loss": 0.64,
      "train_positive_log_prob": -79.3431,
      "train_positive_token_accuracy": 0.0845,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.4237,
      "epoch": 4.056433408577878,
      "grad_norm": 13.134199142456055,
      "learning_rate": 8.919089113500795e-07,
      "lm_loss": 5.328,
      "loss": 1.3832,
      "step": 1797,
      "text_contrastive_loss": 0.8535,
      "train_positive_log_prob": -78.2578,
      "train_positive_token_accuracy": 0.0952,
      "train_positive_token_prob": 0.0341
    },
    {
      "contrastive_loss": 0.332,
      "epoch": 4.058690744920993,
      "grad_norm": 12.321008682250977,
      "learning_rate": 8.877773771475074e-07,
      "lm_loss": 5.464,
      "loss": 1.219,
      "step": 1798,
      "text_contrastive_loss": 0.6812,
      "train_positive_log_prob": -81.9166,
      "train_positive_token_accuracy": 0.0761,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3589,
      "epoch": 4.060948081264108,
      "grad_norm": 12.761566162109375,
      "learning_rate": 8.836545018003084e-07,
      "lm_loss": 5.5035,
      "loss": 1.3069,
      "step": 1799,
      "text_contrastive_loss": 0.7954,
      "train_positive_log_prob": -81.6488,
      "train_positive_token_accuracy": 0.0737,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.3871,
      "epoch": 4.063205417607223,
      "grad_norm": 14.203374862670898,
      "learning_rate": 8.795402939897679e-07,
      "lm_loss": 5.4265,
      "loss": 1.3561,
      "step": 1800,
      "text_contrastive_loss": 0.8529,
      "train_positive_log_prob": -82.095,
      "train_positive_token_accuracy": 0.0824,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.3719,
      "epoch": 4.065462753950339,
      "grad_norm": 12.417008399963379,
      "learning_rate": 8.754347623789222e-07,
      "lm_loss": 5.3673,
      "loss": 1.3582,
      "step": 1801,
      "text_contrastive_loss": 0.8992,
      "train_positive_log_prob": -79.7549,
      "train_positive_token_accuracy": 0.0828,
      "train_positive_token_prob": 0.0327
    },
    {
      "contrastive_loss": 0.3008,
      "epoch": 4.067720090293454,
      "grad_norm": 12.664691925048828,
      "learning_rate": 8.713379156125385e-07,
      "lm_loss": 5.4173,
      "loss": 1.1594,
      "step": 1802,
      "text_contrastive_loss": 0.6336,
      "train_positive_log_prob": -78.679,
      "train_positive_token_accuracy": 0.0775,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3289,
      "epoch": 4.0699774266365685,
      "grad_norm": 11.8455228805542,
      "learning_rate": 8.672497623170944e-07,
      "lm_loss": 5.4951,
      "loss": 1.2802,
      "step": 1803,
      "text_contrastive_loss": 0.8035,
      "train_positive_log_prob": -81.2894,
      "train_positive_token_accuracy": 0.0731,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.3138,
      "epoch": 4.072234762979684,
      "grad_norm": 11.705512046813965,
      "learning_rate": 8.631703111007645e-07,
      "lm_loss": 5.5792,
      "loss": 1.2216,
      "step": 1804,
      "text_contrastive_loss": 0.6997,
      "train_positive_log_prob": -81.6147,
      "train_positive_token_accuracy": 0.0692,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.339,
      "epoch": 4.074492099322799,
      "grad_norm": 12.615190505981445,
      "learning_rate": 8.590995705533994e-07,
      "lm_loss": 5.4129,
      "loss": 1.2578,
      "step": 1805,
      "text_contrastive_loss": 0.7548,
      "train_positive_log_prob": -81.3609,
      "train_positive_token_accuracy": 0.0834,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.3092,
      "epoch": 4.076749435665914,
      "grad_norm": 13.203263282775879,
      "learning_rate": 8.550375492465102e-07,
      "lm_loss": 5.383,
      "loss": 1.1657,
      "step": 1806,
      "text_contrastive_loss": 0.6365,
      "train_positive_log_prob": -81.7542,
      "train_positive_token_accuracy": 0.0864,
      "train_positive_token_prob": 0.0326
    },
    {
      "contrastive_loss": 0.429,
      "epoch": 4.07900677200903,
      "grad_norm": 13.173894882202148,
      "learning_rate": 8.509842557332437e-07,
      "lm_loss": 5.4636,
      "loss": 1.3769,
      "step": 1807,
      "text_contrastive_loss": 0.8032,
      "train_positive_log_prob": -82.9379,
      "train_positive_token_accuracy": 0.0823,
      "train_positive_token_prob": 0.0325
    },
    {
      "contrastive_loss": 0.3388,
      "epoch": 4.081264108352144,
      "grad_norm": 13.1770601272583,
      "learning_rate": 8.469396985483724e-07,
      "lm_loss": 5.4438,
      "loss": 1.2609,
      "step": 1808,
      "text_contrastive_loss": 0.7553,
      "train_positive_log_prob": -80.1489,
      "train_positive_token_accuracy": 0.0785,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.2894,
      "epoch": 4.0835214446952595,
      "grad_norm": 10.92631721496582,
      "learning_rate": 8.429038862082734e-07,
      "lm_loss": 5.4675,
      "loss": 1.1805,
      "step": 1809,
      "text_contrastive_loss": 0.6887,
      "train_positive_log_prob": -79.0419,
      "train_positive_token_accuracy": 0.0797,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.2761,
      "epoch": 4.085778781038375,
      "grad_norm": 12.565187454223633,
      "learning_rate": 8.388768272109105e-07,
      "lm_loss": 5.5025,
      "loss": 1.2419,
      "step": 1810,
      "text_contrastive_loss": 0.8312,
      "train_positive_log_prob": -81.0989,
      "train_positive_token_accuracy": 0.0808,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.2982,
      "epoch": 4.08803611738149,
      "grad_norm": 13.363114356994629,
      "learning_rate": 8.34858530035813e-07,
      "lm_loss": 5.4419,
      "loss": 1.1888,
      "step": 1811,
      "text_contrastive_loss": 0.693,
      "train_positive_log_prob": -82.2441,
      "train_positive_token_accuracy": 0.077,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.399,
      "epoch": 4.090293453724605,
      "grad_norm": 12.355823516845703,
      "learning_rate": 8.308490031440641e-07,
      "lm_loss": 5.4626,
      "loss": 1.3551,
      "step": 1812,
      "text_contrastive_loss": 0.8195,
      "train_positive_log_prob": -79.486,
      "train_positive_token_accuracy": 0.077,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.3029,
      "epoch": 4.09255079006772,
      "grad_norm": 12.237298965454102,
      "learning_rate": 8.268482549782797e-07,
      "lm_loss": 5.3363,
      "loss": 1.2381,
      "step": 1813,
      "text_contrastive_loss": 0.8032,
      "train_positive_log_prob": -79.0204,
      "train_positive_token_accuracy": 0.0826,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.2492,
      "epoch": 4.094808126410835,
      "grad_norm": 10.571876525878906,
      "learning_rate": 8.228562939625906e-07,
      "lm_loss": 5.3881,
      "loss": 1.1082,
      "step": 1814,
      "text_contrastive_loss": 0.6403,
      "train_positive_log_prob": -80.406,
      "train_positive_token_accuracy": 0.0783,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.2274,
      "epoch": 4.0970654627539504,
      "grad_norm": 10.091174125671387,
      "learning_rate": 8.188731285026219e-07,
      "lm_loss": 5.4757,
      "loss": 1.0978,
      "step": 1815,
      "text_contrastive_loss": 0.6458,
      "train_positive_log_prob": -82.757,
      "train_positive_token_accuracy": 0.0822,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.2739,
      "epoch": 4.099322799097066,
      "grad_norm": 11.475275039672852,
      "learning_rate": 8.148987669854846e-07,
      "lm_loss": 5.4154,
      "loss": 1.1924,
      "step": 1816,
      "text_contrastive_loss": 0.7539,
      "train_positive_log_prob": -80.2566,
      "train_positive_token_accuracy": 0.0754,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3325,
      "epoch": 4.10158013544018,
      "grad_norm": 11.306097984313965,
      "learning_rate": 8.109332177797469e-07,
      "lm_loss": 5.421,
      "loss": 1.2918,
      "step": 1817,
      "text_contrastive_loss": 0.8344,
      "train_positive_log_prob": -81.6312,
      "train_positive_token_accuracy": 0.082,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.3344,
      "epoch": 4.1038374717832955,
      "grad_norm": 14.255751609802246,
      "learning_rate": 8.069764892354237e-07,
      "lm_loss": 5.4051,
      "loss": 1.3253,
      "step": 1818,
      "text_contrastive_loss": 0.9009,
      "train_positive_log_prob": -81.4146,
      "train_positive_token_accuracy": 0.0858,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.356,
      "epoch": 4.106094808126411,
      "grad_norm": 12.633360862731934,
      "learning_rate": 8.030285896839546e-07,
      "lm_loss": 5.4218,
      "loss": 1.2902,
      "step": 1819,
      "text_contrastive_loss": 0.784,
      "train_positive_log_prob": -80.2421,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.4155,
      "epoch": 4.108352144469526,
      "grad_norm": 14.468781471252441,
      "learning_rate": 7.99089527438191e-07,
      "lm_loss": 5.345,
      "loss": 1.3988,
      "step": 1820,
      "text_contrastive_loss": 0.8976,
      "train_positive_log_prob": -78.7982,
      "train_positive_token_accuracy": 0.0798,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.3224,
      "epoch": 4.110609480812641,
      "grad_norm": 11.905516624450684,
      "learning_rate": 7.951593107923744e-07,
      "lm_loss": 5.4817,
      "loss": 1.2949,
      "step": 1821,
      "text_contrastive_loss": 0.8485,
      "train_positive_log_prob": -82.5006,
      "train_positive_token_accuracy": 0.075,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.2713,
      "epoch": 4.112866817155756,
      "grad_norm": 12.390151023864746,
      "learning_rate": 7.912379480221228e-07,
      "lm_loss": 5.2288,
      "loss": 1.1593,
      "step": 1822,
      "text_contrastive_loss": 0.7302,
      "train_positive_log_prob": -76.511,
      "train_positive_token_accuracy": 0.0838,
      "train_positive_token_prob": 0.0331
    },
    {
      "contrastive_loss": 0.3649,
      "epoch": 4.115124153498871,
      "grad_norm": 12.990013122558594,
      "learning_rate": 7.873254473844077e-07,
      "lm_loss": 5.4546,
      "loss": 1.2952,
      "step": 1823,
      "text_contrastive_loss": 0.7698,
      "train_positive_log_prob": -80.4384,
      "train_positive_token_accuracy": 0.0792,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.2801,
      "epoch": 4.1173814898419865,
      "grad_norm": 10.795341491699219,
      "learning_rate": 7.834218171175428e-07,
      "lm_loss": 5.4179,
      "loss": 1.2101,
      "step": 1824,
      "text_contrastive_loss": 0.7764,
      "train_positive_log_prob": -80.5509,
      "train_positive_token_accuracy": 0.0767,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.2809,
      "epoch": 4.119638826185102,
      "grad_norm": 12.400055885314941,
      "learning_rate": 7.795270654411635e-07,
      "lm_loss": 5.3719,
      "loss": 1.149,
      "step": 1825,
      "text_contrastive_loss": 0.6619,
      "train_positive_log_prob": -81.2927,
      "train_positive_token_accuracy": 0.0738,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.336,
      "epoch": 4.121896162528217,
      "grad_norm": 12.246535301208496,
      "learning_rate": 7.756412005562114e-07,
      "lm_loss": 5.4104,
      "loss": 1.2633,
      "step": 1826,
      "text_contrastive_loss": 0.7724,
      "train_positive_log_prob": -79.3204,
      "train_positive_token_accuracy": 0.0741,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3774,
      "epoch": 4.1241534988713315,
      "grad_norm": 11.641721725463867,
      "learning_rate": 7.717642306449113e-07,
      "lm_loss": 5.5162,
      "loss": 1.3255,
      "step": 1827,
      "text_contrastive_loss": 0.7929,
      "train_positive_log_prob": -80.3963,
      "train_positive_token_accuracy": 0.0734,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3358,
      "epoch": 4.126410835214447,
      "grad_norm": 13.242731094360352,
      "learning_rate": 7.678961638707633e-07,
      "lm_loss": 5.4367,
      "loss": 1.3027,
      "step": 1828,
      "text_contrastive_loss": 0.8465,
      "train_positive_log_prob": -80.8479,
      "train_positive_token_accuracy": 0.0806,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.2636,
      "epoch": 4.128668171557562,
      "grad_norm": 11.120746612548828,
      "learning_rate": 7.640370083785175e-07,
      "lm_loss": 5.3813,
      "loss": 1.1778,
      "step": 1829,
      "text_contrastive_loss": 0.7521,
      "train_positive_log_prob": -79.4193,
      "train_positive_token_accuracy": 0.0844,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.3299,
      "epoch": 4.1309255079006775,
      "grad_norm": 12.61805534362793,
      "learning_rate": 7.601867722941642e-07,
      "lm_loss": 5.4367,
      "loss": 1.2724,
      "step": 1830,
      "text_contrastive_loss": 0.7978,
      "train_positive_log_prob": -81.3806,
      "train_positive_token_accuracy": 0.0885,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.3394,
      "epoch": 4.133182844243792,
      "grad_norm": 12.158013343811035,
      "learning_rate": 7.563454637249056e-07,
      "lm_loss": 5.5159,
      "loss": 1.3072,
      "step": 1831,
      "text_contrastive_loss": 0.8323,
      "train_positive_log_prob": -81.5749,
      "train_positive_token_accuracy": 0.0724,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.3357,
      "epoch": 4.135440180586907,
      "grad_norm": 14.1693754196167,
      "learning_rate": 7.52513090759151e-07,
      "lm_loss": 5.4365,
      "loss": 1.2435,
      "step": 1832,
      "text_contrastive_loss": 0.7282,
      "train_positive_log_prob": -79.2957,
      "train_positive_token_accuracy": 0.0811,
      "train_positive_token_prob": 0.0327
    },
    {
      "contrastive_loss": 0.2817,
      "epoch": 4.1376975169300225,
      "grad_norm": 11.860201835632324,
      "learning_rate": 7.486896614664962e-07,
      "lm_loss": 5.3825,
      "loss": 1.1636,
      "step": 1833,
      "text_contrastive_loss": 0.6873,
      "train_positive_log_prob": -80.5702,
      "train_positive_token_accuracy": 0.0723,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3016,
      "epoch": 4.139954853273138,
      "grad_norm": 11.996062278747559,
      "learning_rate": 7.448751838977014e-07,
      "lm_loss": 5.3753,
      "loss": 1.1836,
      "step": 1834,
      "text_contrastive_loss": 0.689,
      "train_positive_log_prob": -79.5193,
      "train_positive_token_accuracy": 0.075,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3436,
      "epoch": 4.142212189616253,
      "grad_norm": 12.734991073608398,
      "learning_rate": 7.410696660846761e-07,
      "lm_loss": 5.459,
      "loss": 1.2774,
      "step": 1835,
      "text_contrastive_loss": 0.7757,
      "train_positive_log_prob": -81.822,
      "train_positive_token_accuracy": 0.0779,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.4216,
      "epoch": 4.144469525959368,
      "grad_norm": 13.545905113220215,
      "learning_rate": 7.372731160404672e-07,
      "lm_loss": 5.3478,
      "loss": 1.4101,
      "step": 1836,
      "text_contrastive_loss": 0.9075,
      "train_positive_log_prob": -78.5069,
      "train_positive_token_accuracy": 0.0761,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3848,
      "epoch": 4.146726862302483,
      "grad_norm": 11.820764541625977,
      "learning_rate": 7.334855417592385e-07,
      "lm_loss": 5.4283,
      "loss": 1.3046,
      "step": 1837,
      "text_contrastive_loss": 0.7538,
      "train_positive_log_prob": -79.6684,
      "train_positive_token_accuracy": 0.0804,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.2993,
      "epoch": 4.148984198645598,
      "grad_norm": 12.980536460876465,
      "learning_rate": 7.297069512162535e-07,
      "lm_loss": 5.3651,
      "loss": 1.1632,
      "step": 1838,
      "text_contrastive_loss": 0.6548,
      "train_positive_log_prob": -80.013,
      "train_positive_token_accuracy": 0.0904,
      "train_positive_token_prob": 0.033
    },
    {
      "contrastive_loss": 0.278,
      "epoch": 4.1512415349887135,
      "grad_norm": 11.505115509033203,
      "learning_rate": 7.25937352367857e-07,
      "lm_loss": 5.4105,
      "loss": 1.2199,
      "step": 1839,
      "text_contrastive_loss": 0.8017,
      "train_positive_log_prob": -79.675,
      "train_positive_token_accuracy": 0.0785,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.282,
      "epoch": 4.153498871331829,
      "grad_norm": 10.915390014648438,
      "learning_rate": 7.22176753151464e-07,
      "lm_loss": 5.5966,
      "loss": 1.1317,
      "step": 1840,
      "text_contrastive_loss": 0.5801,
      "train_positive_log_prob": -83.7717,
      "train_positive_token_accuracy": 0.0759,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.3058,
      "epoch": 4.155756207674943,
      "grad_norm": 13.515417098999023,
      "learning_rate": 7.184251614855369e-07,
      "lm_loss": 5.4955,
      "loss": 1.1339,
      "step": 1841,
      "text_contrastive_loss": 0.5571,
      "train_positive_log_prob": -81.4449,
      "train_positive_token_accuracy": 0.0764,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.3044,
      "epoch": 4.158013544018059,
      "grad_norm": 11.466351509094238,
      "learning_rate": 7.146825852695749e-07,
      "lm_loss": 5.4374,
      "loss": 1.2048,
      "step": 1842,
      "text_contrastive_loss": 0.7134,
      "train_positive_log_prob": -84.0274,
      "train_positive_token_accuracy": 0.0827,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.3425,
      "epoch": 4.160270880361174,
      "grad_norm": 12.742423057556152,
      "learning_rate": 7.109490323840884e-07,
      "lm_loss": 5.4831,
      "loss": 1.3366,
      "step": 1843,
      "text_contrastive_loss": 0.8915,
      "train_positive_log_prob": -79.186,
      "train_positive_token_accuracy": 0.0747,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.359,
      "epoch": 4.162528216704289,
      "grad_norm": 11.728898048400879,
      "learning_rate": 7.072245106905928e-07,
      "lm_loss": 5.4898,
      "loss": 1.2143,
      "step": 1844,
      "text_contrastive_loss": 0.6126,
      "train_positive_log_prob": -81.4005,
      "train_positive_token_accuracy": 0.0732,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.3609,
      "epoch": 4.164785553047404,
      "grad_norm": 13.316404342651367,
      "learning_rate": 7.035090280315854e-07,
      "lm_loss": 5.5048,
      "loss": 1.3146,
      "step": 1845,
      "text_contrastive_loss": 0.8064,
      "train_positive_log_prob": -81.0004,
      "train_positive_token_accuracy": 0.0847,
      "train_positive_token_prob": 0.0325
    },
    {
      "contrastive_loss": 0.3137,
      "epoch": 4.167042889390519,
      "grad_norm": 11.35694694519043,
      "learning_rate": 6.998025922305313e-07,
      "lm_loss": 5.3996,
      "loss": 1.2345,
      "step": 1846,
      "text_contrastive_loss": 0.7617,
      "train_positive_log_prob": -79.4142,
      "train_positive_token_accuracy": 0.0869,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.3434,
      "epoch": 4.169300225733634,
      "grad_norm": 12.942392349243164,
      "learning_rate": 6.961052110918432e-07,
      "lm_loss": 5.3292,
      "loss": 1.2651,
      "step": 1847,
      "text_contrastive_loss": 0.7775,
      "train_positive_log_prob": -78.1046,
      "train_positive_token_accuracy": 0.0869,
      "train_positive_token_prob": 0.0333
    },
    {
      "contrastive_loss": 0.3689,
      "epoch": 4.1715575620767495,
      "grad_norm": 14.120200157165527,
      "learning_rate": 6.924168924008712e-07,
      "lm_loss": 5.3815,
      "loss": 1.1817,
      "step": 1848,
      "text_contrastive_loss": 0.5493,
      "train_positive_log_prob": -79.3788,
      "train_positive_token_accuracy": 0.0805,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3311,
      "epoch": 4.173814898419865,
      "grad_norm": 13.238189697265625,
      "learning_rate": 6.887376439238813e-07,
      "lm_loss": 5.4392,
      "loss": 1.2551,
      "step": 1849,
      "text_contrastive_loss": 0.7603,
      "train_positive_log_prob": -79.7593,
      "train_positive_token_accuracy": 0.0778,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3229,
      "epoch": 4.176072234762979,
      "grad_norm": 12.410944938659668,
      "learning_rate": 6.850674734080454e-07,
      "lm_loss": 5.4465,
      "loss": 1.2135,
      "step": 1850,
      "text_contrastive_loss": 0.692,
      "train_positive_log_prob": -80.2507,
      "train_positive_token_accuracy": 0.075,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.2852,
      "epoch": 4.178329571106095,
      "grad_norm": 12.629752159118652,
      "learning_rate": 6.814063885814127e-07,
      "lm_loss": 5.4311,
      "loss": 1.181,
      "step": 1851,
      "text_contrastive_loss": 0.7054,
      "train_positive_log_prob": -77.4575,
      "train_positive_token_accuracy": 0.0781,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.2841,
      "epoch": 4.18058690744921,
      "grad_norm": 11.972105979919434,
      "learning_rate": 6.77754397152906e-07,
      "lm_loss": 5.4088,
      "loss": 1.2003,
      "step": 1852,
      "text_contrastive_loss": 0.7507,
      "train_positive_log_prob": -79.0844,
      "train_positive_token_accuracy": 0.0793,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3862,
      "epoch": 4.182844243792325,
      "grad_norm": 13.545210838317871,
      "learning_rate": 6.741115068123017e-07,
      "lm_loss": 5.4036,
      "loss": 1.3364,
      "step": 1853,
      "text_contrastive_loss": 0.8197,
      "train_positive_log_prob": -80.1254,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.364,
      "epoch": 4.1851015801354405,
      "grad_norm": 14.371001243591309,
      "learning_rate": 6.704777252302108e-07,
      "lm_loss": 5.4491,
      "loss": 1.3381,
      "step": 1854,
      "text_contrastive_loss": 0.8583,
      "train_positive_log_prob": -80.6463,
      "train_positive_token_accuracy": 0.0805,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.3535,
      "epoch": 4.187358916478555,
      "grad_norm": 13.23997688293457,
      "learning_rate": 6.66853060058063e-07,
      "lm_loss": 5.4168,
      "loss": 1.3207,
      "step": 1855,
      "text_contrastive_loss": 0.851,
      "train_positive_log_prob": -81.5058,
      "train_positive_token_accuracy": 0.082,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.281,
      "epoch": 4.18961625282167,
      "grad_norm": 10.905323028564453,
      "learning_rate": 6.632375189280948e-07,
      "lm_loss": 5.4921,
      "loss": 1.226,
      "step": 1856,
      "text_contrastive_loss": 0.7916,
      "train_positive_log_prob": -81.1158,
      "train_positive_token_accuracy": 0.0733,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.3916,
      "epoch": 4.191873589164786,
      "grad_norm": 13.602428436279297,
      "learning_rate": 6.596311094533292e-07,
      "lm_loss": 5.487,
      "loss": 1.362,
      "step": 1857,
      "text_contrastive_loss": 0.8434,
      "train_positive_log_prob": -81.5548,
      "train_positive_token_accuracy": 0.0762,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.4214,
      "epoch": 4.194130925507901,
      "grad_norm": 17.563297271728516,
      "learning_rate": 6.56033839227564e-07,
      "lm_loss": 5.3989,
      "loss": 1.3424,
      "step": 1858,
      "text_contrastive_loss": 0.7623,
      "train_positive_log_prob": -80.3679,
      "train_positive_token_accuracy": 0.0819,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.3235,
      "epoch": 4.196388261851016,
      "grad_norm": 12.790931701660156,
      "learning_rate": 6.524457158253472e-07,
      "lm_loss": 5.3472,
      "loss": 1.2644,
      "step": 1859,
      "text_contrastive_loss": 0.8123,
      "train_positive_log_prob": -77.931,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3487,
      "epoch": 4.198645598194131,
      "grad_norm": 12.175345420837402,
      "learning_rate": 6.488667468019727e-07,
      "lm_loss": 5.4997,
      "loss": 1.2886,
      "step": 1860,
      "text_contrastive_loss": 0.7799,
      "train_positive_log_prob": -79.8771,
      "train_positive_token_accuracy": 0.0804,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.2934,
      "epoch": 4.200902934537246,
      "grad_norm": 12.599485397338867,
      "learning_rate": 6.452969396934567e-07,
      "lm_loss": 5.5562,
      "loss": 1.1829,
      "step": 1861,
      "text_contrastive_loss": 0.6678,
      "train_positive_log_prob": -81.6972,
      "train_positive_token_accuracy": 0.0768,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.246,
      "epoch": 4.203160270880361,
      "grad_norm": 11.814972877502441,
      "learning_rate": 6.417363020165235e-07,
      "lm_loss": 5.3403,
      "loss": 1.1099,
      "step": 1862,
      "text_contrastive_loss": 0.6597,
      "train_positive_log_prob": -78.7533,
      "train_positive_token_accuracy": 0.0771,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.2798,
      "epoch": 4.205417607223477,
      "grad_norm": 10.557435989379883,
      "learning_rate": 6.381848412685882e-07,
      "lm_loss": 5.4239,
      "loss": 1.1508,
      "step": 1863,
      "text_contrastive_loss": 0.6571,
      "train_positive_log_prob": -79.5598,
      "train_positive_token_accuracy": 0.0807,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3817,
      "epoch": 4.207674943566591,
      "grad_norm": 12.646711349487305,
      "learning_rate": 6.346425649277454e-07,
      "lm_loss": 5.465,
      "loss": 1.3197,
      "step": 1864,
      "text_contrastive_loss": 0.783,
      "train_positive_log_prob": -81.398,
      "train_positive_token_accuracy": 0.0768,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.3897,
      "epoch": 4.209932279909706,
      "grad_norm": 15.26496696472168,
      "learning_rate": 6.31109480452749e-07,
      "lm_loss": 5.4765,
      "loss": 1.3993,
      "step": 1865,
      "text_contrastive_loss": 0.924,
      "train_positive_log_prob": -81.3361,
      "train_positive_token_accuracy": 0.0768,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3847,
      "epoch": 4.212189616252822,
      "grad_norm": 14.464232444763184,
      "learning_rate": 6.275855952829995e-07,
      "lm_loss": 5.4352,
      "loss": 1.3913,
      "step": 1866,
      "text_contrastive_loss": 0.926,
      "train_positive_log_prob": -80.3427,
      "train_positive_token_accuracy": 0.0809,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3403,
      "epoch": 4.214446952595937,
      "grad_norm": 11.755424499511719,
      "learning_rate": 6.240709168385251e-07,
      "lm_loss": 5.4246,
      "loss": 1.2199,
      "step": 1867,
      "text_contrastive_loss": 0.6741,
      "train_positive_log_prob": -82.3869,
      "train_positive_token_accuracy": 0.0792,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.2969,
      "epoch": 4.216704288939052,
      "grad_norm": 11.180468559265137,
      "learning_rate": 6.2056545251997e-07,
      "lm_loss": 5.4097,
      "loss": 1.2182,
      "step": 1868,
      "text_contrastive_loss": 0.7606,
      "train_positive_log_prob": -80.1256,
      "train_positive_token_accuracy": 0.0765,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.2758,
      "epoch": 4.218961625282167,
      "grad_norm": 11.721410751342773,
      "learning_rate": 6.170692097085751e-07,
      "lm_loss": 5.3654,
      "loss": 1.185,
      "step": 1869,
      "text_contrastive_loss": 0.7453,
      "train_positive_log_prob": -80.0163,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.307,
      "epoch": 4.221218961625282,
      "grad_norm": 11.450150489807129,
      "learning_rate": 6.135821957661658e-07,
      "lm_loss": 5.398,
      "loss": 1.1763,
      "step": 1870,
      "text_contrastive_loss": 0.6591,
      "train_positive_log_prob": -80.4404,
      "train_positive_token_accuracy": 0.0912,
      "train_positive_token_prob": 0.0331
    },
    {
      "contrastive_loss": 0.299,
      "epoch": 4.223476297968397,
      "grad_norm": 11.473048210144043,
      "learning_rate": 6.101044180351318e-07,
      "lm_loss": 5.5531,
      "loss": 1.1965,
      "step": 1871,
      "text_contrastive_loss": 0.6842,
      "train_positive_log_prob": -84.6443,
      "train_positive_token_accuracy": 0.08,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.3191,
      "epoch": 4.225733634311513,
      "grad_norm": 12.278962135314941,
      "learning_rate": 6.066358838384184e-07,
      "lm_loss": 5.4404,
      "loss": 1.2195,
      "step": 1872,
      "text_contrastive_loss": 0.7128,
      "train_positive_log_prob": -81.247,
      "train_positive_token_accuracy": 0.0879,
      "train_positive_token_prob": 0.0328
    },
    {
      "contrastive_loss": 0.3228,
      "epoch": 4.227990970654628,
      "grad_norm": 13.453584671020508,
      "learning_rate": 6.031766004795047e-07,
      "lm_loss": 5.4725,
      "loss": 1.2444,
      "step": 1873,
      "text_contrastive_loss": 0.7486,
      "train_positive_log_prob": -81.2786,
      "train_positive_token_accuracy": 0.0771,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.2865,
      "epoch": 4.230248306997742,
      "grad_norm": 11.51198959350586,
      "learning_rate": 5.997265752423936e-07,
      "lm_loss": 5.4334,
      "loss": 1.1658,
      "step": 1874,
      "text_contrastive_loss": 0.6719,
      "train_positive_log_prob": -80.8198,
      "train_positive_token_accuracy": 0.0748,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.2555,
      "epoch": 4.232505643340858,
      "grad_norm": 12.516451835632324,
      "learning_rate": 5.962858153915896e-07,
      "lm_loss": 5.4485,
      "loss": 1.1855,
      "step": 1875,
      "text_contrastive_loss": 0.7705,
      "train_positive_log_prob": -80.4737,
      "train_positive_token_accuracy": 0.0756,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3613,
      "epoch": 4.234762979683973,
      "grad_norm": 13.309110641479492,
      "learning_rate": 5.928543281720917e-07,
      "lm_loss": 5.4458,
      "loss": 1.3317,
      "step": 1876,
      "text_contrastive_loss": 0.8516,
      "train_positive_log_prob": -81.0071,
      "train_positive_token_accuracy": 0.0838,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.4067,
      "epoch": 4.237020316027088,
      "grad_norm": 14.060657501220703,
      "learning_rate": 5.894321208093712e-07,
      "lm_loss": 5.4241,
      "loss": 1.3882,
      "step": 1877,
      "text_contrastive_loss": 0.8781,
      "train_positive_log_prob": -81.2745,
      "train_positive_token_accuracy": 0.0821,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.2982,
      "epoch": 4.239277652370204,
      "grad_norm": 11.619880676269531,
      "learning_rate": 5.860192005093624e-07,
      "lm_loss": 5.4563,
      "loss": 1.253,
      "step": 1878,
      "text_contrastive_loss": 0.8184,
      "train_positive_log_prob": -81.8854,
      "train_positive_token_accuracy": 0.0781,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.3242,
      "epoch": 4.241534988713318,
      "grad_norm": 12.934802055358887,
      "learning_rate": 5.826155744584405e-07,
      "lm_loss": 5.3315,
      "loss": 1.1814,
      "step": 1879,
      "text_contrastive_loss": 0.6482,
      "train_positive_log_prob": -78.1412,
      "train_positive_token_accuracy": 0.0757,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.2808,
      "epoch": 4.243792325056433,
      "grad_norm": 12.162077903747559,
      "learning_rate": 5.792212498234134e-07,
      "lm_loss": 5.4209,
      "loss": 1.2429,
      "step": 1880,
      "text_contrastive_loss": 0.8402,
      "train_positive_log_prob": -80.4847,
      "train_positive_token_accuracy": 0.0781,
      "train_positive_token_prob": 0.0326
    },
    {
      "contrastive_loss": 0.3423,
      "epoch": 4.246049661399549,
      "grad_norm": 13.890145301818848,
      "learning_rate": 5.758362337515028e-07,
      "lm_loss": 5.4438,
      "loss": 1.3069,
      "step": 1881,
      "text_contrastive_loss": 0.8404,
      "train_positive_log_prob": -81.1248,
      "train_positive_token_accuracy": 0.0781,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.2812,
      "epoch": 4.248306997742664,
      "grad_norm": 13.08950138092041,
      "learning_rate": 5.724605333703303e-07,
      "lm_loss": 5.4289,
      "loss": 1.2036,
      "step": 1882,
      "text_contrastive_loss": 0.7592,
      "train_positive_log_prob": -77.8523,
      "train_positive_token_accuracy": 0.0804,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.3114,
      "epoch": 4.250564334085778,
      "grad_norm": 11.903820037841797,
      "learning_rate": 5.690941557878988e-07,
      "lm_loss": 5.4437,
      "loss": 1.2735,
      "step": 1883,
      "text_contrastive_loss": 0.8354,
      "train_positive_log_prob": -80.6324,
      "train_positive_token_accuracy": 0.0799,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.2795,
      "epoch": 4.252821670428894,
      "grad_norm": 11.589510917663574,
      "learning_rate": 5.657371080925866e-07,
      "lm_loss": 5.413,
      "loss": 1.1392,
      "step": 1884,
      "text_contrastive_loss": 0.6367,
      "train_positive_log_prob": -80.4782,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.2403,
      "epoch": 4.255079006772009,
      "grad_norm": 11.013530731201172,
      "learning_rate": 5.623893973531225e-07,
      "lm_loss": 5.4858,
      "loss": 1.0851,
      "step": 1885,
      "text_contrastive_loss": 0.5923,
      "train_positive_log_prob": -81.4935,
      "train_positive_token_accuracy": 0.076,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.3294,
      "epoch": 4.257336343115124,
      "grad_norm": 11.694799423217773,
      "learning_rate": 5.590510306185765e-07,
      "lm_loss": 5.48,
      "loss": 1.2948,
      "step": 1886,
      "text_contrastive_loss": 0.8349,
      "train_positive_log_prob": -81.6183,
      "train_positive_token_accuracy": 0.0771,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.3281,
      "epoch": 4.25959367945824,
      "grad_norm": 12.523609161376953,
      "learning_rate": 5.557220149183412e-07,
      "lm_loss": 5.4963,
      "loss": 1.3172,
      "step": 1887,
      "text_contrastive_loss": 0.879,
      "train_positive_log_prob": -82.4027,
      "train_positive_token_accuracy": 0.0795,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.3005,
      "epoch": 4.261851015801354,
      "grad_norm": 12.24012565612793,
      "learning_rate": 5.524023572621229e-07,
      "lm_loss": 5.4156,
      "loss": 1.2741,
      "step": 1888,
      "text_contrastive_loss": 0.864,
      "train_positive_log_prob": -80.4733,
      "train_positive_token_accuracy": 0.0786,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.2849,
      "epoch": 4.264108352144469,
      "grad_norm": 12.618743896484375,
      "learning_rate": 5.4909206463992e-07,
      "lm_loss": 5.4421,
      "loss": 1.2187,
      "step": 1889,
      "text_contrastive_loss": 0.7792,
      "train_positive_log_prob": -80.4849,
      "train_positive_token_accuracy": 0.0774,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.3576,
      "epoch": 4.266365688487585,
      "grad_norm": 13.902774810791016,
      "learning_rate": 5.457911440220154e-07,
      "lm_loss": 5.4181,
      "loss": 1.3273,
      "step": 1890,
      "text_contrastive_loss": 0.8559,
      "train_positive_log_prob": -78.8596,
      "train_positive_token_accuracy": 0.0843,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3058,
      "epoch": 4.2686230248307,
      "grad_norm": 12.352407455444336,
      "learning_rate": 5.424996023589524e-07,
      "lm_loss": 5.4638,
      "loss": 1.2143,
      "step": 1891,
      "text_contrastive_loss": 0.7242,
      "train_positive_log_prob": -79.8192,
      "train_positive_token_accuracy": 0.073,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.3377,
      "epoch": 4.270880361173815,
      "grad_norm": 12.394871711730957,
      "learning_rate": 5.392174465815308e-07,
      "lm_loss": 5.448,
      "loss": 1.3208,
      "step": 1892,
      "text_contrastive_loss": 0.8766,
      "train_positive_log_prob": -80.6557,
      "train_positive_token_accuracy": 0.0813,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.2068,
      "epoch": 4.27313769751693,
      "grad_norm": 9.813087463378906,
      "learning_rate": 5.359446836007842e-07,
      "lm_loss": 5.3534,
      "loss": 1.0492,
      "step": 1893,
      "text_contrastive_loss": 0.6142,
      "train_positive_log_prob": -78.2899,
      "train_positive_token_accuracy": 0.0751,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.308,
      "epoch": 4.275395033860045,
      "grad_norm": 13.385251998901367,
      "learning_rate": 5.326813203079706e-07,
      "lm_loss": 5.4145,
      "loss": 1.2336,
      "step": 1894,
      "text_contrastive_loss": 0.7683,
      "train_positive_log_prob": -78.9948,
      "train_positive_token_accuracy": 0.0795,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3299,
      "epoch": 4.27765237020316,
      "grad_norm": 13.189772605895996,
      "learning_rate": 5.294273635745517e-07,
      "lm_loss": 5.4433,
      "loss": 1.3282,
      "step": 1895,
      "text_contrastive_loss": 0.9078,
      "train_positive_log_prob": -81.1639,
      "train_positive_token_accuracy": 0.0828,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.5198,
      "epoch": 4.279909706546276,
      "grad_norm": 13.458785057067871,
      "learning_rate": 5.261828202521868e-07,
      "lm_loss": 5.4507,
      "loss": 1.5369,
      "step": 1896,
      "text_contrastive_loss": 0.9441,
      "train_positive_log_prob": -82.8307,
      "train_positive_token_accuracy": 0.0794,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.2221,
      "epoch": 4.282167042889391,
      "grad_norm": 11.057195663452148,
      "learning_rate": 5.229476971727115e-07,
      "lm_loss": 5.3322,
      "loss": 1.0969,
      "step": 1897,
      "text_contrastive_loss": 0.6831,
      "train_positive_log_prob": -78.2909,
      "train_positive_token_accuracy": 0.0835,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.2966,
      "epoch": 4.284424379232505,
      "grad_norm": 11.43716812133789,
      "learning_rate": 5.197220011481274e-07,
      "lm_loss": 5.392,
      "loss": 1.2066,
      "step": 1898,
      "text_contrastive_loss": 0.7415,
      "train_positive_log_prob": -79.3838,
      "train_positive_token_accuracy": 0.0812,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.3147,
      "epoch": 4.286681715575621,
      "grad_norm": 11.857349395751953,
      "learning_rate": 5.165057389705835e-07,
      "lm_loss": 5.4267,
      "loss": 1.2362,
      "step": 1899,
      "text_contrastive_loss": 0.7577,
      "train_positive_log_prob": -79.3457,
      "train_positive_token_accuracy": 0.0793,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3817,
      "epoch": 4.288939051918736,
      "grad_norm": 13.249019622802734,
      "learning_rate": 5.132989174123659e-07,
      "lm_loss": 5.4844,
      "loss": 1.3051,
      "step": 1900,
      "text_contrastive_loss": 0.7499,
      "train_positive_log_prob": -81.2181,
      "train_positive_token_accuracy": 0.0839,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3436,
      "epoch": 4.291196388261851,
      "grad_norm": 12.054238319396973,
      "learning_rate": 5.101015432258843e-07,
      "lm_loss": 5.4637,
      "loss": 1.2689,
      "step": 1901,
      "text_contrastive_loss": 0.7578,
      "train_positive_log_prob": -80.6345,
      "train_positive_token_accuracy": 0.0786,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3196,
      "epoch": 4.293453724604966,
      "grad_norm": 13.134613990783691,
      "learning_rate": 5.069136231436539e-07,
      "lm_loss": 5.4477,
      "loss": 1.2169,
      "step": 1902,
      "text_contrastive_loss": 0.7051,
      "train_positive_log_prob": -79.6727,
      "train_positive_token_accuracy": 0.0751,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3759,
      "epoch": 4.295711060948081,
      "grad_norm": 11.014152526855469,
      "learning_rate": 5.037351638782812e-07,
      "lm_loss": 5.4405,
      "loss": 1.3688,
      "step": 1903,
      "text_contrastive_loss": 0.8977,
      "train_positive_log_prob": -79.6974,
      "train_positive_token_accuracy": 0.08,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.372,
      "epoch": 4.297968397291196,
      "grad_norm": 13.289602279663086,
      "learning_rate": 5.00566172122453e-07,
      "lm_loss": 5.3684,
      "loss": 1.3195,
      "step": 1904,
      "text_contrastive_loss": 0.8213,
      "train_positive_log_prob": -79.671,
      "train_positive_token_accuracy": 0.0769,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.4053,
      "epoch": 4.300225733634312,
      "grad_norm": 12.918943405151367,
      "learning_rate": 4.97406654548922e-07,
      "lm_loss": 5.3847,
      "loss": 1.4672,
      "step": 1905,
      "text_contrastive_loss": 1.0468,
      "train_positive_log_prob": -77.8223,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.4202,
      "epoch": 4.302483069977427,
      "grad_norm": 13.597563743591309,
      "learning_rate": 4.942566178104924e-07,
      "lm_loss": 5.3564,
      "loss": 1.3696,
      "step": 1906,
      "text_contrastive_loss": 0.8275,
      "train_positive_log_prob": -78.4392,
      "train_positive_token_accuracy": 0.0814,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.3352,
      "epoch": 4.3047404063205414,
      "grad_norm": 11.399421691894531,
      "learning_rate": 4.911160685400008e-07,
      "lm_loss": 5.4272,
      "loss": 1.2694,
      "step": 1907,
      "text_contrastive_loss": 0.7829,
      "train_positive_log_prob": -82.2106,
      "train_positive_token_accuracy": 0.0745,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.303,
      "epoch": 4.306997742663657,
      "grad_norm": 11.587457656860352,
      "learning_rate": 4.879850133503106e-07,
      "lm_loss": 5.4226,
      "loss": 1.2057,
      "step": 1908,
      "text_contrastive_loss": 0.7208,
      "train_positive_log_prob": -81.3343,
      "train_positive_token_accuracy": 0.0771,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.292,
      "epoch": 4.309255079006772,
      "grad_norm": 12.802454948425293,
      "learning_rate": 4.848634588342932e-07,
      "lm_loss": 5.4536,
      "loss": 1.213,
      "step": 1909,
      "text_contrastive_loss": 0.7512,
      "train_positive_log_prob": -81.4406,
      "train_positive_token_accuracy": 0.0813,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3073,
      "epoch": 4.311512415349887,
      "grad_norm": 11.837509155273438,
      "learning_rate": 4.817514115648164e-07,
      "lm_loss": 5.395,
      "loss": 1.1225,
      "step": 1910,
      "text_contrastive_loss": 0.5514,
      "train_positive_log_prob": -78.6523,
      "train_positive_token_accuracy": 0.0689,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.292,
      "epoch": 4.313769751693002,
      "grad_norm": 11.047466278076172,
      "learning_rate": 4.786488780947246e-07,
      "lm_loss": 5.3851,
      "loss": 1.172,
      "step": 1911,
      "text_contrastive_loss": 0.683,
      "train_positive_log_prob": -80.1441,
      "train_positive_token_accuracy": 0.0759,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3286,
      "epoch": 4.316027088036117,
      "grad_norm": 12.600515365600586,
      "learning_rate": 4.755558649568337e-07,
      "lm_loss": 5.5078,
      "loss": 1.2175,
      "step": 1912,
      "text_contrastive_loss": 0.6761,
      "train_positive_log_prob": -81.7156,
      "train_positive_token_accuracy": 0.08,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3419,
      "epoch": 4.318284424379232,
      "grad_norm": 12.97014045715332,
      "learning_rate": 4.7247237866391236e-07,
      "lm_loss": 5.3523,
      "loss": 1.2679,
      "step": 1913,
      "text_contrastive_loss": 0.7815,
      "train_positive_log_prob": -78.1133,
      "train_positive_token_accuracy": 0.0766,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.234,
      "epoch": 4.320541760722348,
      "grad_norm": 10.707161903381348,
      "learning_rate": 4.6939842570867034e-07,
      "lm_loss": 5.2514,
      "loss": 1.0834,
      "step": 1914,
      "text_contrastive_loss": 0.6484,
      "train_positive_log_prob": -76.5977,
      "train_positive_token_accuracy": 0.0833,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.2927,
      "epoch": 4.322799097065463,
      "grad_norm": 12.018651008605957,
      "learning_rate": 4.663340125637389e-07,
      "lm_loss": 5.4897,
      "loss": 1.1458,
      "step": 1915,
      "text_contrastive_loss": 0.6083,
      "train_positive_log_prob": -84.1907,
      "train_positive_token_accuracy": 0.0853,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3639,
      "epoch": 4.3250564334085775,
      "grad_norm": 13.130175590515137,
      "learning_rate": 4.6327914568166763e-07,
      "lm_loss": 5.552,
      "loss": 1.3112,
      "step": 1916,
      "text_contrastive_loss": 0.7841,
      "train_positive_log_prob": -83.5838,
      "train_positive_token_accuracy": 0.0763,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.3794,
      "epoch": 4.327313769751693,
      "grad_norm": 11.861356735229492,
      "learning_rate": 4.6023383149490066e-07,
      "lm_loss": 5.353,
      "loss": 1.3373,
      "step": 1917,
      "text_contrastive_loss": 0.8452,
      "train_positive_log_prob": -79.5412,
      "train_positive_token_accuracy": 0.082,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.3183,
      "epoch": 4.329571106094808,
      "grad_norm": 12.326311111450195,
      "learning_rate": 4.571980764157724e-07,
      "lm_loss": 5.4535,
      "loss": 1.2567,
      "step": 1918,
      "text_contrastive_loss": 0.7861,
      "train_positive_log_prob": -80.2063,
      "train_positive_token_accuracy": 0.0786,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3893,
      "epoch": 4.331828442437923,
      "grad_norm": 12.893142700195312,
      "learning_rate": 4.5417188683648417e-07,
      "lm_loss": 5.4157,
      "loss": 1.3553,
      "step": 1919,
      "text_contrastive_loss": 0.8489,
      "train_positive_log_prob": -81.1094,
      "train_positive_token_accuracy": 0.0887,
      "train_positive_token_prob": 0.0326
    },
    {
      "contrastive_loss": 0.3404,
      "epoch": 4.334085778781039,
      "grad_norm": 13.110001564025879,
      "learning_rate": 4.511552691290988e-07,
      "lm_loss": 5.3956,
      "loss": 1.2924,
      "step": 1920,
      "text_contrastive_loss": 0.8249,
      "train_positive_log_prob": -80.4881,
      "train_positive_token_accuracy": 0.0806,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3375,
      "epoch": 4.336343115124153,
      "grad_norm": 12.533656120300293,
      "learning_rate": 4.4814822964552363e-07,
      "lm_loss": 5.3953,
      "loss": 1.2804,
      "step": 1921,
      "text_contrastive_loss": 0.8069,
      "train_positive_log_prob": -81.7771,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.4657,
      "epoch": 4.3386004514672685,
      "grad_norm": 17.309276580810547,
      "learning_rate": 4.4515077471749767e-07,
      "lm_loss": 5.3767,
      "loss": 1.4613,
      "step": 1922,
      "text_contrastive_loss": 0.9159,
      "train_positive_log_prob": -77.839,
      "train_positive_token_accuracy": 0.0799,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.4217,
      "epoch": 4.340857787810384,
      "grad_norm": 14.05301284790039,
      "learning_rate": 4.421629106565778e-07,
      "lm_loss": 5.4113,
      "loss": 1.3483,
      "step": 1923,
      "text_contrastive_loss": 0.7708,
      "train_positive_log_prob": -80.115,
      "train_positive_token_accuracy": 0.081,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3704,
      "epoch": 4.343115124153499,
      "grad_norm": 15.06448745727539,
      "learning_rate": 4.391846437541258e-07,
      "lm_loss": 5.548,
      "loss": 1.3138,
      "step": 1924,
      "text_contrastive_loss": 0.7773,
      "train_positive_log_prob": -82.7872,
      "train_positive_token_accuracy": 0.0694,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.3323,
      "epoch": 4.345372460496614,
      "grad_norm": 12.01858139038086,
      "learning_rate": 4.362159802812971e-07,
      "lm_loss": 5.3856,
      "loss": 1.2592,
      "step": 1925,
      "text_contrastive_loss": 0.7768,
      "train_positive_log_prob": -78.4668,
      "train_positive_token_accuracy": 0.0791,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3896,
      "epoch": 4.347629796839729,
      "grad_norm": 12.243396759033203,
      "learning_rate": 4.332569264890252e-07,
      "lm_loss": 5.4137,
      "loss": 1.3015,
      "step": 1926,
      "text_contrastive_loss": 0.7409,
      "train_positive_log_prob": -79.2809,
      "train_positive_token_accuracy": 0.0886,
      "train_positive_token_prob": 0.0326
    },
    {
      "contrastive_loss": 0.3488,
      "epoch": 4.349887133182844,
      "grad_norm": 12.974628448486328,
      "learning_rate": 4.3030748860800606e-07,
      "lm_loss": 5.4622,
      "loss": 1.2245,
      "step": 1927,
      "text_contrastive_loss": 0.6589,
      "train_positive_log_prob": -81.2073,
      "train_positive_token_accuracy": 0.0767,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3219,
      "epoch": 4.3521444695259595,
      "grad_norm": 13.554567337036133,
      "learning_rate": 4.273676728486925e-07,
      "lm_loss": 5.4657,
      "loss": 1.2173,
      "step": 1928,
      "text_contrastive_loss": 0.6976,
      "train_positive_log_prob": -81.0887,
      "train_positive_token_accuracy": 0.0821,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3091,
      "epoch": 4.354401805869075,
      "grad_norm": 11.172772407531738,
      "learning_rate": 4.244374854012734e-07,
      "lm_loss": 5.3847,
      "loss": 1.1814,
      "step": 1929,
      "text_contrastive_loss": 0.6675,
      "train_positive_log_prob": -79.6139,
      "train_positive_token_accuracy": 0.0819,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3764,
      "epoch": 4.356659142212189,
      "grad_norm": 12.820956230163574,
      "learning_rate": 4.215169324356666e-07,
      "lm_loss": 5.4759,
      "loss": 1.3309,
      "step": 1930,
      "text_contrastive_loss": 0.8139,
      "train_positive_log_prob": -80.7251,
      "train_positive_token_accuracy": 0.0829,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.2686,
      "epoch": 4.3589164785553045,
      "grad_norm": 12.839651107788086,
      "learning_rate": 4.186060201014991e-07,
      "lm_loss": 5.4411,
      "loss": 1.1247,
      "step": 1931,
      "text_contrastive_loss": 0.624,
      "train_positive_log_prob": -78.1693,
      "train_positive_token_accuracy": 0.0805,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3252,
      "epoch": 4.36117381489842,
      "grad_norm": 12.507498741149902,
      "learning_rate": 4.157047545281029e-07,
      "lm_loss": 5.4765,
      "loss": 1.2511,
      "step": 1932,
      "text_contrastive_loss": 0.7564,
      "train_positive_log_prob": -80.7387,
      "train_positive_token_accuracy": 0.0681,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.3636,
      "epoch": 4.363431151241535,
      "grad_norm": 14.36543083190918,
      "learning_rate": 4.1281314182449405e-07,
      "lm_loss": 5.4303,
      "loss": 1.3354,
      "step": 1933,
      "text_contrastive_loss": 0.8575,
      "train_positive_log_prob": -80.0147,
      "train_positive_token_accuracy": 0.0732,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.3823,
      "epoch": 4.3656884875846504,
      "grad_norm": 13.869869232177734,
      "learning_rate": 4.099311880793655e-07,
      "lm_loss": 5.3796,
      "loss": 1.3698,
      "step": 1934,
      "text_contrastive_loss": 0.8991,
      "train_positive_log_prob": -80.949,
      "train_positive_token_accuracy": 0.082,
      "train_positive_token_prob": 0.0325
    },
    {
      "contrastive_loss": 0.291,
      "epoch": 4.367945823927765,
      "grad_norm": 10.774118423461914,
      "learning_rate": 4.070588993610697e-07,
      "lm_loss": 5.5101,
      "loss": 1.1343,
      "step": 1935,
      "text_contrastive_loss": 0.5845,
      "train_positive_log_prob": -81.6393,
      "train_positive_token_accuracy": 0.0762,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3525,
      "epoch": 4.37020316027088,
      "grad_norm": 12.817048072814941,
      "learning_rate": 4.0419628171760927e-07,
      "lm_loss": 5.4295,
      "loss": 1.2563,
      "step": 1936,
      "text_contrastive_loss": 0.7219,
      "train_positive_log_prob": -78.9835,
      "train_positive_token_accuracy": 0.0784,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.329,
      "epoch": 4.3724604966139955,
      "grad_norm": 12.01680850982666,
      "learning_rate": 4.0134334117662375e-07,
      "lm_loss": 5.4157,
      "loss": 1.3076,
      "step": 1937,
      "text_contrastive_loss": 0.874,
      "train_positive_log_prob": -81.4031,
      "train_positive_token_accuracy": 0.0777,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3601,
      "epoch": 4.374717832957111,
      "grad_norm": 13.025857925415039,
      "learning_rate": 3.985000837453756e-07,
      "lm_loss": 5.491,
      "loss": 1.2922,
      "step": 1938,
      "text_contrastive_loss": 0.766,
      "train_positive_log_prob": -80.6333,
      "train_positive_token_accuracy": 0.0843,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3186,
      "epoch": 4.376975169300226,
      "grad_norm": 11.83201789855957,
      "learning_rate": 3.9566651541073586e-07,
      "lm_loss": 5.3517,
      "loss": 1.1773,
      "step": 1939,
      "text_contrastive_loss": 0.647,
      "train_positive_log_prob": -81.2197,
      "train_positive_token_accuracy": 0.0844,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.3609,
      "epoch": 4.3792325056433405,
      "grad_norm": 11.843232154846191,
      "learning_rate": 3.928426421391773e-07,
      "lm_loss": 5.4419,
      "loss": 1.2533,
      "step": 1940,
      "text_contrastive_loss": 0.6963,
      "train_positive_log_prob": -79.7574,
      "train_positive_token_accuracy": 0.0789,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3211,
      "epoch": 4.381489841986456,
      "grad_norm": 10.878660202026367,
      "learning_rate": 3.9002846987675704e-07,
      "lm_loss": 5.5271,
      "loss": 1.238,
      "step": 1941,
      "text_contrastive_loss": 0.7284,
      "train_positive_log_prob": -82.4716,
      "train_positive_token_accuracy": 0.077,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3773,
      "epoch": 4.383747178329571,
      "grad_norm": 12.85601806640625,
      "learning_rate": 3.872240045491055e-07,
      "lm_loss": 5.4526,
      "loss": 1.2866,
      "step": 1942,
      "text_contrastive_loss": 0.7282,
      "train_positive_log_prob": -80.0619,
      "train_positive_token_accuracy": 0.0771,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.2619,
      "epoch": 4.3860045146726865,
      "grad_norm": 10.641718864440918,
      "learning_rate": 3.8442925206141237e-07,
      "lm_loss": 5.4418,
      "loss": 1.1808,
      "step": 1943,
      "text_contrastive_loss": 0.7495,
      "train_positive_log_prob": -80.9498,
      "train_positive_token_accuracy": 0.0777,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3433,
      "epoch": 4.388261851015802,
      "grad_norm": 13.001375198364258,
      "learning_rate": 3.8164421829841756e-07,
      "lm_loss": 5.4389,
      "loss": 1.3035,
      "step": 1944,
      "text_contrastive_loss": 0.8326,
      "train_positive_log_prob": -81.3895,
      "train_positive_token_accuracy": 0.0733,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3319,
      "epoch": 4.390519187358916,
      "grad_norm": 12.554210662841797,
      "learning_rate": 3.7886890912439633e-07,
      "lm_loss": 5.4099,
      "loss": 1.2898,
      "step": 1945,
      "text_contrastive_loss": 0.8339,
      "train_positive_log_prob": -79.8156,
      "train_positive_token_accuracy": 0.0786,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.2732,
      "epoch": 4.3927765237020315,
      "grad_norm": 10.95527172088623,
      "learning_rate": 3.761033303831474e-07,
      "lm_loss": 5.3417,
      "loss": 1.1577,
      "step": 1946,
      "text_contrastive_loss": 0.7006,
      "train_positive_log_prob": -79.8084,
      "train_positive_token_accuracy": 0.081,
      "train_positive_token_prob": 0.0327
    },
    {
      "contrastive_loss": 0.3254,
      "epoch": 4.395033860045147,
      "grad_norm": 11.946934700012207,
      "learning_rate": 3.733474878979798e-07,
      "lm_loss": 5.5016,
      "loss": 1.2357,
      "step": 1947,
      "text_contrastive_loss": 0.7203,
      "train_positive_log_prob": -81.7702,
      "train_positive_token_accuracy": 0.0809,
      "train_positive_token_prob": 0.0325
    },
    {
      "contrastive_loss": 0.3162,
      "epoch": 4.397291196388262,
      "grad_norm": 11.75720500946045,
      "learning_rate": 3.706013874717024e-07,
      "lm_loss": 5.4124,
      "loss": 1.2045,
      "step": 1948,
      "text_contrastive_loss": 0.694,
      "train_positive_log_prob": -78.4789,
      "train_positive_token_accuracy": 0.0789,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3482,
      "epoch": 4.399548532731377,
      "grad_norm": 12.269476890563965,
      "learning_rate": 3.678650348866114e-07,
      "lm_loss": 5.4388,
      "loss": 1.2701,
      "step": 1949,
      "text_contrastive_loss": 0.756,
      "train_positive_log_prob": -81.4512,
      "train_positive_token_accuracy": 0.0725,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.3276,
      "epoch": 4.401805869074492,
      "grad_norm": 11.407697677612305,
      "learning_rate": 3.651384359044774e-07,
      "lm_loss": 5.4254,
      "loss": 1.2115,
      "step": 1950,
      "text_contrastive_loss": 0.6827,
      "train_positive_log_prob": -78.9622,
      "train_positive_token_accuracy": 0.0835,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.2941,
      "epoch": 4.404063205417607,
      "grad_norm": 11.566184997558594,
      "learning_rate": 3.6242159626653004e-07,
      "lm_loss": 5.2775,
      "loss": 1.2618,
      "step": 1951,
      "text_contrastive_loss": 0.8798,
      "train_positive_log_prob": -79.4109,
      "train_positive_token_accuracy": 0.08,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.3303,
      "epoch": 4.4063205417607225,
      "grad_norm": 12.965038299560547,
      "learning_rate": 3.597145216934556e-07,
      "lm_loss": 5.5454,
      "loss": 1.3134,
      "step": 1952,
      "text_contrastive_loss": 0.8572,
      "train_positive_log_prob": -83.4865,
      "train_positive_token_accuracy": 0.0758,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.2828,
      "epoch": 4.408577878103838,
      "grad_norm": 11.011374473571777,
      "learning_rate": 3.570172178853731e-07,
      "lm_loss": 5.4672,
      "loss": 1.2173,
      "step": 1953,
      "text_contrastive_loss": 0.7756,
      "train_positive_log_prob": -81.5175,
      "train_positive_token_accuracy": 0.0768,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3069,
      "epoch": 4.410835214446952,
      "grad_norm": 12.267878532409668,
      "learning_rate": 3.5432969052183186e-07,
      "lm_loss": 5.443,
      "loss": 1.2058,
      "step": 1954,
      "text_contrastive_loss": 0.7092,
      "train_positive_log_prob": -78.9866,
      "train_positive_token_accuracy": 0.0756,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.2954,
      "epoch": 4.413092550790068,
      "grad_norm": 11.222500801086426,
      "learning_rate": 3.516519452617922e-07,
      "lm_loss": 5.4162,
      "loss": 1.1945,
      "step": 1955,
      "text_contrastive_loss": 0.7149,
      "train_positive_log_prob": -79.4485,
      "train_positive_token_accuracy": 0.0786,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3573,
      "epoch": 4.415349887133183,
      "grad_norm": 12.776378631591797,
      "learning_rate": 3.4898398774361854e-07,
      "lm_loss": 5.2865,
      "loss": 1.2416,
      "step": 1956,
      "text_contrastive_loss": 0.7113,
      "train_positive_log_prob": -77.7679,
      "train_positive_token_accuracy": 0.0789,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.2655,
      "epoch": 4.417607223476298,
      "grad_norm": 12.210612297058105,
      "learning_rate": 3.463258235850653e-07,
      "lm_loss": 5.4574,
      "loss": 1.2067,
      "step": 1957,
      "text_contrastive_loss": 0.7909,
      "train_positive_log_prob": -82.1617,
      "train_positive_token_accuracy": 0.0776,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3802,
      "epoch": 4.4198645598194135,
      "grad_norm": 12.871219635009766,
      "learning_rate": 3.4367745838326807e-07,
      "lm_loss": 5.4506,
      "loss": 1.33,
      "step": 1958,
      "text_contrastive_loss": 0.8096,
      "train_positive_log_prob": -80.2472,
      "train_positive_token_accuracy": 0.078,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.3668,
      "epoch": 4.422121896162528,
      "grad_norm": 12.472192764282227,
      "learning_rate": 3.410388977147244e-07,
      "lm_loss": 5.3762,
      "loss": 1.2986,
      "step": 1959,
      "text_contrastive_loss": 0.7884,
      "train_positive_log_prob": -80.2454,
      "train_positive_token_accuracy": 0.0822,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.2747,
      "epoch": 4.424379232505643,
      "grad_norm": 11.645474433898926,
      "learning_rate": 3.3841014713529184e-07,
      "lm_loss": 5.4041,
      "loss": 1.1202,
      "step": 1960,
      "text_contrastive_loss": 0.6103,
      "train_positive_log_prob": -80.7221,
      "train_positive_token_accuracy": 0.074,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.427,
      "epoch": 4.426636568848759,
      "grad_norm": 13.484107971191406,
      "learning_rate": 3.357912121801682e-07,
      "lm_loss": 5.3172,
      "loss": 1.4158,
      "step": 1961,
      "text_contrastive_loss": 0.9143,
      "train_positive_log_prob": -77.3503,
      "train_positive_token_accuracy": 0.0771,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.3313,
      "epoch": 4.428893905191874,
      "grad_norm": 11.793492317199707,
      "learning_rate": 3.331820983638867e-07,
      "lm_loss": 5.3923,
      "loss": 1.2893,
      "step": 1962,
      "text_contrastive_loss": 0.8376,
      "train_positive_log_prob": -78.7351,
      "train_positive_token_accuracy": 0.0816,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3317,
      "epoch": 4.431151241534989,
      "grad_norm": 13.229374885559082,
      "learning_rate": 3.3058281118029553e-07,
      "lm_loss": 5.5145,
      "loss": 1.2835,
      "step": 1963,
      "text_contrastive_loss": 0.8007,
      "train_positive_log_prob": -83.6229,
      "train_positive_token_accuracy": 0.0768,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.3781,
      "epoch": 4.433408577878104,
      "grad_norm": 11.081595420837402,
      "learning_rate": 3.279933561025567e-07,
      "lm_loss": 5.4846,
      "loss": 1.3576,
      "step": 1964,
      "text_contrastive_loss": 0.8622,
      "train_positive_log_prob": -81.1751,
      "train_positive_token_accuracy": 0.0826,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.3986,
      "epoch": 4.435665914221219,
      "grad_norm": 13.550172805786133,
      "learning_rate": 3.254137385831263e-07,
      "lm_loss": 5.3798,
      "loss": 1.2674,
      "step": 1965,
      "text_contrastive_loss": 0.6616,
      "train_positive_log_prob": -79.9192,
      "train_positive_token_accuracy": 0.0763,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.3653,
      "epoch": 4.437923250564334,
      "grad_norm": 13.075263977050781,
      "learning_rate": 3.2284396405374787e-07,
      "lm_loss": 5.4491,
      "loss": 1.3857,
      "step": 1966,
      "text_contrastive_loss": 0.951,
      "train_positive_log_prob": -80.6276,
      "train_positive_token_accuracy": 0.0721,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3097,
      "epoch": 4.4401805869074495,
      "grad_norm": 11.862391471862793,
      "learning_rate": 3.202840379254374e-07,
      "lm_loss": 5.4522,
      "loss": 1.2188,
      "step": 1967,
      "text_contrastive_loss": 0.7277,
      "train_positive_log_prob": -81.4462,
      "train_positive_token_accuracy": 0.0863,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3162,
      "epoch": 4.442437923250564,
      "grad_norm": 12.69119930267334,
      "learning_rate": 3.177339655884737e-07,
      "lm_loss": 5.351,
      "loss": 1.2078,
      "step": 1968,
      "text_contrastive_loss": 0.713,
      "train_positive_log_prob": -77.9056,
      "train_positive_token_accuracy": 0.0746,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.4019,
      "epoch": 4.444695259593679,
      "grad_norm": 14.639025688171387,
      "learning_rate": 3.151937524123905e-07,
      "lm_loss": 5.3963,
      "loss": 1.374,
      "step": 1969,
      "text_contrastive_loss": 0.8648,
      "train_positive_log_prob": -80.1044,
      "train_positive_token_accuracy": 0.0795,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.312,
      "epoch": 4.446952595936795,
      "grad_norm": 12.351846694946289,
      "learning_rate": 3.1266340374595693e-07,
      "lm_loss": 5.3913,
      "loss": 1.2669,
      "step": 1970,
      "text_contrastive_loss": 0.8316,
      "train_positive_log_prob": -80.0297,
      "train_positive_token_accuracy": 0.0811,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.2739,
      "epoch": 4.44920993227991,
      "grad_norm": 11.822172164916992,
      "learning_rate": 3.1014292491717444e-07,
      "lm_loss": 5.4641,
      "loss": 1.1646,
      "step": 1971,
      "text_contrastive_loss": 0.6886,
      "train_positive_log_prob": -81.6376,
      "train_positive_token_accuracy": 0.0871,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.3555,
      "epoch": 4.451467268623025,
      "grad_norm": 13.001505851745605,
      "learning_rate": 3.076323212332605e-07,
      "lm_loss": 5.5178,
      "loss": 1.3104,
      "step": 1972,
      "text_contrastive_loss": 0.8062,
      "train_positive_log_prob": -82.5088,
      "train_positive_token_accuracy": 0.0825,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.2463,
      "epoch": 4.45372460496614,
      "grad_norm": 9.949920654296875,
      "learning_rate": 3.0513159798063906e-07,
      "lm_loss": 5.4626,
      "loss": 1.176,
      "step": 1973,
      "text_contrastive_loss": 0.767,
      "train_positive_log_prob": -81.6643,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.2975,
      "epoch": 4.455981941309255,
      "grad_norm": 11.72691535949707,
      "learning_rate": 3.026407604249315e-07,
      "lm_loss": 5.4633,
      "loss": 1.1925,
      "step": 1974,
      "text_contrastive_loss": 0.6973,
      "train_positive_log_prob": -81.5643,
      "train_positive_token_accuracy": 0.0816,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.3434,
      "epoch": 4.45823927765237,
      "grad_norm": 13.088930130004883,
      "learning_rate": 3.0015981381094073e-07,
      "lm_loss": 5.4003,
      "loss": 1.406,
      "step": 1975,
      "text_contrastive_loss": 1.0453,
      "train_positive_log_prob": -80.1802,
      "train_positive_token_accuracy": 0.0752,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3534,
      "epoch": 4.460496613995486,
      "grad_norm": 13.003207206726074,
      "learning_rate": 2.976887633626435e-07,
      "lm_loss": 5.408,
      "loss": 1.2628,
      "step": 1976,
      "text_contrastive_loss": 0.7372,
      "train_positive_log_prob": -80.3831,
      "train_positive_token_accuracy": 0.0781,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.2753,
      "epoch": 4.4627539503386,
      "grad_norm": 11.459562301635742,
      "learning_rate": 2.952276142831806e-07,
      "lm_loss": 5.3072,
      "loss": 1.1391,
      "step": 1977,
      "text_contrastive_loss": 0.6662,
      "train_positive_log_prob": -79.3733,
      "train_positive_token_accuracy": 0.0786,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3029,
      "epoch": 4.465011286681715,
      "grad_norm": 12.701268196105957,
      "learning_rate": 2.9277637175484376e-07,
      "lm_loss": 5.4284,
      "loss": 1.2047,
      "step": 1978,
      "text_contrastive_loss": 0.7179,
      "train_positive_log_prob": -78.9638,
      "train_positive_token_accuracy": 0.0795,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3661,
      "epoch": 4.467268623024831,
      "grad_norm": 12.158988952636719,
      "learning_rate": 2.9033504093906207e-07,
      "lm_loss": 5.3832,
      "loss": 1.2573,
      "step": 1979,
      "text_contrastive_loss": 0.7057,
      "train_positive_log_prob": -78.1847,
      "train_positive_token_accuracy": 0.0777,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.384,
      "epoch": 4.469525959367946,
      "grad_norm": 13.560571670532227,
      "learning_rate": 2.8790362697639685e-07,
      "lm_loss": 5.4802,
      "loss": 1.2471,
      "step": 1980,
      "text_contrastive_loss": 0.6301,
      "train_positive_log_prob": -79.8893,
      "train_positive_token_accuracy": 0.0766,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3292,
      "epoch": 4.471783295711061,
      "grad_norm": 12.4685697555542,
      "learning_rate": 2.854821349865289e-07,
      "lm_loss": 5.3318,
      "loss": 1.3252,
      "step": 1981,
      "text_contrastive_loss": 0.9256,
      "train_positive_log_prob": -79.1013,
      "train_positive_token_accuracy": 0.0887,
      "train_positive_token_prob": 0.0339
    },
    {
      "contrastive_loss": 0.344,
      "epoch": 4.474040632054176,
      "grad_norm": 11.924105644226074,
      "learning_rate": 2.8307057006824514e-07,
      "lm_loss": 5.4745,
      "loss": 1.3361,
      "step": 1982,
      "text_contrastive_loss": 0.8894,
      "train_positive_log_prob": -81.9041,
      "train_positive_token_accuracy": 0.0757,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.2922,
      "epoch": 4.476297968397291,
      "grad_norm": 12.074453353881836,
      "learning_rate": 2.806689372994292e-07,
      "lm_loss": 5.4906,
      "loss": 1.2187,
      "step": 1983,
      "text_contrastive_loss": 0.7548,
      "train_positive_log_prob": -82.8968,
      "train_positive_token_accuracy": 0.0735,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.4259,
      "epoch": 4.478555304740406,
      "grad_norm": 15.743043899536133,
      "learning_rate": 2.7827724173705273e-07,
      "lm_loss": 5.4123,
      "loss": 1.4059,
      "step": 1984,
      "text_contrastive_loss": 0.8776,
      "train_positive_log_prob": -78.8145,
      "train_positive_token_accuracy": 0.0848,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.4042,
      "epoch": 4.480812641083522,
      "grad_norm": 12.900703430175781,
      "learning_rate": 2.7589548841716274e-07,
      "lm_loss": 5.4043,
      "loss": 1.3865,
      "step": 1985,
      "text_contrastive_loss": 0.8838,
      "train_positive_log_prob": -79.5618,
      "train_positive_token_accuracy": 0.073,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.3141,
      "epoch": 4.483069977426637,
      "grad_norm": 12.590093612670898,
      "learning_rate": 2.735236823548715e-07,
      "lm_loss": 5.4271,
      "loss": 1.2328,
      "step": 1986,
      "text_contrastive_loss": 0.752,
      "train_positive_log_prob": -80.7912,
      "train_positive_token_accuracy": 0.0853,
      "train_positive_token_prob": 0.0325
    },
    {
      "contrastive_loss": 0.3985,
      "epoch": 4.485327313769751,
      "grad_norm": 13.033928871154785,
      "learning_rate": 2.711618285443457e-07,
      "lm_loss": 5.4622,
      "loss": 1.3368,
      "step": 1987,
      "text_contrastive_loss": 0.7842,
      "train_positive_log_prob": -80.5889,
      "train_positive_token_accuracy": 0.0789,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.3857,
      "epoch": 4.487584650112867,
      "grad_norm": 13.071440696716309,
      "learning_rate": 2.6880993195879614e-07,
      "lm_loss": 5.3629,
      "loss": 1.2775,
      "step": 1988,
      "text_contrastive_loss": 0.7109,
      "train_positive_log_prob": -78.0521,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.2822,
      "epoch": 4.489841986455982,
      "grad_norm": 14.284415245056152,
      "learning_rate": 2.6646799755046746e-07,
      "lm_loss": 5.4118,
      "loss": 1.1507,
      "step": 1989,
      "text_contrastive_loss": 0.6547,
      "train_positive_log_prob": -80.6014,
      "train_positive_token_accuracy": 0.0755,
      "train_positive_token_prob": 0.0296
    },
    {
      "contrastive_loss": 0.2255,
      "epoch": 4.492099322799097,
      "grad_norm": 9.530472755432129,
      "learning_rate": 2.64136030250628e-07,
      "lm_loss": 5.3235,
      "loss": 1.0414,
      "step": 1990,
      "text_contrastive_loss": 0.5671,
      "train_positive_log_prob": -76.9655,
      "train_positive_token_accuracy": 0.0778,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3283,
      "epoch": 4.494356659142213,
      "grad_norm": 12.200201988220215,
      "learning_rate": 2.618140349695575e-07,
      "lm_loss": 5.3967,
      "loss": 1.2619,
      "step": 1991,
      "text_contrastive_loss": 0.7879,
      "train_positive_log_prob": -79.5396,
      "train_positive_token_accuracy": 0.0771,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3181,
      "epoch": 4.496613995485327,
      "grad_norm": 13.046611785888672,
      "learning_rate": 2.595020165965401e-07,
      "lm_loss": 5.3985,
      "loss": 1.1363,
      "step": 1992,
      "text_contrastive_loss": 0.5566,
      "train_positive_log_prob": -79.5424,
      "train_positive_token_accuracy": 0.075,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.4152,
      "epoch": 4.498871331828442,
      "grad_norm": 15.92698860168457,
      "learning_rate": 2.571999799998509e-07,
      "lm_loss": 5.3889,
      "loss": 1.3561,
      "step": 1993,
      "text_contrastive_loss": 0.8039,
      "train_positive_log_prob": -79.6548,
      "train_positive_token_accuracy": 0.0798,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.4907,
      "epoch": 4.501128668171558,
      "grad_norm": 14.646669387817383,
      "learning_rate": 2.549079300267482e-07,
      "lm_loss": 5.404,
      "loss": 1.4439,
      "step": 1994,
      "text_contrastive_loss": 0.8257,
      "train_positive_log_prob": -80.0659,
      "train_positive_token_accuracy": 0.0825,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3904,
      "epoch": 4.503386004514673,
      "grad_norm": 13.377157211303711,
      "learning_rate": 2.526258715034602e-07,
      "lm_loss": 5.4566,
      "loss": 1.367,
      "step": 1995,
      "text_contrastive_loss": 0.8619,
      "train_positive_log_prob": -80.9489,
      "train_positive_token_accuracy": 0.0806,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3421,
      "epoch": 4.505643340857787,
      "grad_norm": 11.762893676757812,
      "learning_rate": 2.503538092351782e-07,
      "lm_loss": 5.4225,
      "loss": 1.2789,
      "step": 1996,
      "text_contrastive_loss": 0.7892,
      "train_positive_log_prob": -80.2756,
      "train_positive_token_accuracy": 0.0794,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.2713,
      "epoch": 4.507900677200903,
      "grad_norm": 11.286093711853027,
      "learning_rate": 2.480917480060441e-07,
      "lm_loss": 5.4995,
      "loss": 1.1857,
      "step": 1997,
      "text_contrastive_loss": 0.729,
      "train_positive_log_prob": -84.0377,
      "train_positive_token_accuracy": 0.0813,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3871,
      "epoch": 4.510158013544018,
      "grad_norm": 11.77099323272705,
      "learning_rate": 2.458396925791434e-07,
      "lm_loss": 5.5436,
      "loss": 1.3411,
      "step": 1998,
      "text_contrastive_loss": 0.7993,
      "train_positive_log_prob": -83.414,
      "train_positive_token_accuracy": 0.0719,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.2465,
      "epoch": 4.512415349887133,
      "grad_norm": 10.904062271118164,
      "learning_rate": 2.4359764769648907e-07,
      "lm_loss": 5.4694,
      "loss": 1.16,
      "step": 1999,
      "text_contrastive_loss": 0.7331,
      "train_positive_log_prob": -81.9734,
      "train_positive_token_accuracy": 0.0868,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3651,
      "epoch": 4.514672686230249,
      "grad_norm": 12.968461036682129,
      "learning_rate": 2.4136561807901916e-07,
      "lm_loss": 5.439,
      "loss": 1.3276,
      "step": 2000,
      "text_contrastive_loss": 0.8373,
      "train_positive_log_prob": -80.5832,
      "train_positive_token_accuracy": 0.0862,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.2988,
      "epoch": 4.516930022573363,
      "grad_norm": 11.34416675567627,
      "learning_rate": 2.391436084265814e-07,
      "lm_loss": 5.4296,
      "loss": 1.1813,
      "step": 2001,
      "text_contrastive_loss": 0.679,
      "train_positive_log_prob": -80.3034,
      "train_positive_token_accuracy": 0.0784,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.2805,
      "epoch": 4.519187358916478,
      "grad_norm": 10.901216506958008,
      "learning_rate": 2.3693162341792532e-07,
      "lm_loss": 5.3696,
      "loss": 1.1619,
      "step": 2002,
      "text_contrastive_loss": 0.6888,
      "train_positive_log_prob": -77.397,
      "train_positive_token_accuracy": 0.0783,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.3483,
      "epoch": 4.521444695259594,
      "grad_norm": 12.048800468444824,
      "learning_rate": 2.347296677106925e-07,
      "lm_loss": 5.4966,
      "loss": 1.3013,
      "step": 2003,
      "text_contrastive_loss": 0.8066,
      "train_positive_log_prob": -81.4393,
      "train_positive_token_accuracy": 0.0762,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3522,
      "epoch": 4.523702031602709,
      "grad_norm": 11.89638614654541,
      "learning_rate": 2.3253774594140633e-07,
      "lm_loss": 5.412,
      "loss": 1.2412,
      "step": 2004,
      "text_contrastive_loss": 0.6956,
      "train_positive_log_prob": -79.8289,
      "train_positive_token_accuracy": 0.0731,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3092,
      "epoch": 4.525959367945823,
      "grad_norm": 12.427459716796875,
      "learning_rate": 2.3035586272546207e-07,
      "lm_loss": 5.3236,
      "loss": 1.321,
      "step": 2005,
      "text_contrastive_loss": 0.9589,
      "train_positive_log_prob": -77.9007,
      "train_positive_token_accuracy": 0.0729,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.2784,
      "epoch": 4.528216704288939,
      "grad_norm": 11.306635856628418,
      "learning_rate": 2.2818402265711858e-07,
      "lm_loss": 5.5013,
      "loss": 1.1735,
      "step": 2006,
      "text_contrastive_loss": 0.6899,
      "train_positive_log_prob": -81.2113,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3845,
      "epoch": 4.530474040632054,
      "grad_norm": 15.276398658752441,
      "learning_rate": 2.2602223030948445e-07,
      "lm_loss": 5.381,
      "loss": 1.3115,
      "step": 2007,
      "text_contrastive_loss": 0.7777,
      "train_positive_log_prob": -81.9315,
      "train_positive_token_accuracy": 0.0893,
      "train_positive_token_prob": 0.0327
    },
    {
      "contrastive_loss": 0.2927,
      "epoch": 4.532731376975169,
      "grad_norm": 11.345015525817871,
      "learning_rate": 2.2387049023451458e-07,
      "lm_loss": 5.3351,
      "loss": 1.1545,
      "step": 2008,
      "text_contrastive_loss": 0.6567,
      "train_positive_log_prob": -78.8331,
      "train_positive_token_accuracy": 0.078,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.3535,
      "epoch": 4.534988713318285,
      "grad_norm": 15.586918830871582,
      "learning_rate": 2.2172880696299692e-07,
      "lm_loss": 5.3683,
      "loss": 1.2256,
      "step": 2009,
      "text_contrastive_loss": 0.6706,
      "train_positive_log_prob": -79.2257,
      "train_positive_token_accuracy": 0.0794,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.2647,
      "epoch": 4.5372460496614,
      "grad_norm": 10.706208229064941,
      "learning_rate": 2.1959718500454196e-07,
      "lm_loss": 5.5631,
      "loss": 1.1452,
      "step": 2010,
      "text_contrastive_loss": 0.6484,
      "train_positive_log_prob": -82.7724,
      "train_positive_token_accuracy": 0.073,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.3174,
      "epoch": 4.539503386004514,
      "grad_norm": 12.67294692993164,
      "learning_rate": 2.17475628847576e-07,
      "lm_loss": 5.387,
      "loss": 1.2475,
      "step": 2011,
      "text_contrastive_loss": 0.7827,
      "train_positive_log_prob": -79.5876,
      "train_positive_token_accuracy": 0.0812,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.2579,
      "epoch": 4.54176072234763,
      "grad_norm": 11.39964771270752,
      "learning_rate": 2.1536414295932896e-07,
      "lm_loss": 5.4511,
      "loss": 1.1641,
      "step": 2012,
      "text_contrastive_loss": 0.7222,
      "train_positive_log_prob": -79.8067,
      "train_positive_token_accuracy": 0.0803,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.4161,
      "epoch": 4.544018058690745,
      "grad_norm": 13.330033302307129,
      "learning_rate": 2.1326273178582822e-07,
      "lm_loss": 5.305,
      "loss": 1.4528,
      "step": 2013,
      "text_contrastive_loss": 1.0125,
      "train_positive_log_prob": -77.4401,
      "train_positive_token_accuracy": 0.0851,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.373,
      "epoch": 4.54627539503386,
      "grad_norm": 12.099190711975098,
      "learning_rate": 2.1117139975188716e-07,
      "lm_loss": 5.4084,
      "loss": 1.2817,
      "step": 2014,
      "text_contrastive_loss": 0.7357,
      "train_positive_log_prob": -80.1291,
      "train_positive_token_accuracy": 0.0757,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.2718,
      "epoch": 4.548532731376975,
      "grad_norm": 12.181597709655762,
      "learning_rate": 2.0909015126109488e-07,
      "lm_loss": 5.368,
      "loss": 1.1887,
      "step": 2015,
      "text_contrastive_loss": 0.7603,
      "train_positive_log_prob": -81.444,
      "train_positive_token_accuracy": 0.0775,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.2917,
      "epoch": 4.55079006772009,
      "grad_norm": 11.465213775634766,
      "learning_rate": 2.070189906958081e-07,
      "lm_loss": 5.4248,
      "loss": 1.1875,
      "step": 2016,
      "text_contrastive_loss": 0.7067,
      "train_positive_log_prob": -80.5051,
      "train_positive_token_accuracy": 0.0821,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.276,
      "epoch": 4.553047404063205,
      "grad_norm": 11.312572479248047,
      "learning_rate": 2.0495792241714386e-07,
      "lm_loss": 5.4656,
      "loss": 1.2279,
      "step": 2017,
      "text_contrastive_loss": 0.8106,
      "train_positive_log_prob": -80.32,
      "train_positive_token_accuracy": 0.0758,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.3549,
      "epoch": 4.555304740406321,
      "grad_norm": 14.06812858581543,
      "learning_rate": 2.029069507649678e-07,
      "lm_loss": 5.4875,
      "loss": 1.3419,
      "step": 2018,
      "text_contrastive_loss": 0.8765,
      "train_positive_log_prob": -81.2641,
      "train_positive_token_accuracy": 0.0707,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.2721,
      "epoch": 4.557562076749436,
      "grad_norm": 10.342501640319824,
      "learning_rate": 2.0086608005788376e-07,
      "lm_loss": 5.3842,
      "loss": 1.1507,
      "step": 2019,
      "text_contrastive_loss": 0.6803,
      "train_positive_log_prob": -78.0244,
      "train_positive_token_accuracy": 0.083,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.2837,
      "epoch": 4.5598194130925505,
      "grad_norm": 11.828228950500488,
      "learning_rate": 1.988353145932298e-07,
      "lm_loss": 5.4307,
      "loss": 1.1625,
      "step": 2020,
      "text_contrastive_loss": 0.6716,
      "train_positive_log_prob": -79.9561,
      "train_positive_token_accuracy": 0.0804,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.2986,
      "epoch": 4.562076749435666,
      "grad_norm": 12.166635513305664,
      "learning_rate": 1.9681465864706372e-07,
      "lm_loss": 5.4377,
      "loss": 1.2576,
      "step": 2021,
      "text_contrastive_loss": 0.8305,
      "train_positive_log_prob": -81.2903,
      "train_positive_token_accuracy": 0.0868,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3584,
      "epoch": 4.564334085778781,
      "grad_norm": 12.779674530029297,
      "learning_rate": 1.9480411647415708e-07,
      "lm_loss": 5.4291,
      "loss": 1.2428,
      "step": 2022,
      "text_contrastive_loss": 0.683,
      "train_positive_log_prob": -79.7797,
      "train_positive_token_accuracy": 0.0792,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.4095,
      "epoch": 4.566591422121896,
      "grad_norm": 12.529008865356445,
      "learning_rate": 1.9280369230798568e-07,
      "lm_loss": 5.365,
      "loss": 1.3624,
      "step": 2023,
      "text_contrastive_loss": 0.8327,
      "train_positive_log_prob": -76.9113,
      "train_positive_token_accuracy": 0.0815,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.2817,
      "epoch": 4.568848758465011,
      "grad_norm": 12.580851554870605,
      "learning_rate": 1.9081339036071956e-07,
      "lm_loss": 5.3876,
      "loss": 1.2247,
      "step": 2024,
      "text_contrastive_loss": 0.8087,
      "train_positive_log_prob": -79.8277,
      "train_positive_token_accuracy": 0.0754,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.2684,
      "epoch": 4.571106094808126,
      "grad_norm": 10.923868179321289,
      "learning_rate": 1.8883321482321583e-07,
      "lm_loss": 5.4357,
      "loss": 1.1561,
      "step": 2025,
      "text_contrastive_loss": 0.6884,
      "train_positive_log_prob": -79.1586,
      "train_positive_token_accuracy": 0.0821,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3664,
      "epoch": 4.573363431151241,
      "grad_norm": 13.49759578704834,
      "learning_rate": 1.8686316986500974e-07,
      "lm_loss": 5.4025,
      "loss": 1.3802,
      "step": 2026,
      "text_contrastive_loss": 0.9471,
      "train_positive_log_prob": -80.7852,
      "train_positive_token_accuracy": 0.0862,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3016,
      "epoch": 4.575620767494357,
      "grad_norm": 10.842415809631348,
      "learning_rate": 1.8490325963430368e-07,
      "lm_loss": 5.4726,
      "loss": 1.2499,
      "step": 2027,
      "text_contrastive_loss": 0.8021,
      "train_positive_log_prob": -79.3777,
      "train_positive_token_accuracy": 0.0749,
      "train_positive_token_prob": 0.0293
    },
    {
      "contrastive_loss": 0.3296,
      "epoch": 4.577878103837472,
      "grad_norm": 11.55229663848877,
      "learning_rate": 1.829534882579598e-07,
      "lm_loss": 5.5845,
      "loss": 1.2516,
      "step": 2028,
      "text_contrastive_loss": 0.7272,
      "train_positive_log_prob": -82.3522,
      "train_positive_token_accuracy": 0.076,
      "train_positive_token_prob": 0.0299
    },
    {
      "contrastive_loss": 0.3164,
      "epoch": 4.580135440180587,
      "grad_norm": 12.349048614501953,
      "learning_rate": 1.8101385984149343e-07,
      "lm_loss": 5.3823,
      "loss": 1.2898,
      "step": 2029,
      "text_contrastive_loss": 0.8703,
      "train_positive_log_prob": -79.4107,
      "train_positive_token_accuracy": 0.0776,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3504,
      "epoch": 4.582392776523702,
      "grad_norm": 12.200100898742676,
      "learning_rate": 1.7908437846906158e-07,
      "lm_loss": 5.435,
      "loss": 1.2723,
      "step": 2030,
      "text_contrastive_loss": 0.7567,
      "train_positive_log_prob": -81.249,
      "train_positive_token_accuracy": 0.082,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3954,
      "epoch": 4.584650112866817,
      "grad_norm": 13.013609886169434,
      "learning_rate": 1.7716504820345427e-07,
      "lm_loss": 5.3436,
      "loss": 1.3419,
      "step": 2031,
      "text_contrastive_loss": 0.8244,
      "train_positive_log_prob": -77.6288,
      "train_positive_token_accuracy": 0.0869,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.3222,
      "epoch": 4.586907449209932,
      "grad_norm": 11.47140121459961,
      "learning_rate": 1.752558730860876e-07,
      "lm_loss": 5.3549,
      "loss": 1.2048,
      "step": 2032,
      "text_contrastive_loss": 0.6943,
      "train_positive_log_prob": -77.5382,
      "train_positive_token_accuracy": 0.079,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.3463,
      "epoch": 4.589164785553048,
      "grad_norm": 13.19788646697998,
      "learning_rate": 1.733568571369948e-07,
      "lm_loss": 5.4858,
      "loss": 1.2803,
      "step": 2033,
      "text_contrastive_loss": 0.7707,
      "train_positive_log_prob": -80.9162,
      "train_positive_token_accuracy": 0.0787,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.3182,
      "epoch": 4.591422121896162,
      "grad_norm": 11.778922080993652,
      "learning_rate": 1.7146800435481837e-07,
      "lm_loss": 5.3939,
      "loss": 1.2261,
      "step": 2034,
      "text_contrastive_loss": 0.7371,
      "train_positive_log_prob": -79.3073,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.4292,
      "epoch": 4.5936794582392775,
      "grad_norm": 14.718167304992676,
      "learning_rate": 1.6958931871679908e-07,
      "lm_loss": 5.5364,
      "loss": 1.3682,
      "step": 2035,
      "text_contrastive_loss": 0.7706,
      "train_positive_log_prob": -83.2176,
      "train_positive_token_accuracy": 0.08,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3166,
      "epoch": 4.595936794582393,
      "grad_norm": 12.074480056762695,
      "learning_rate": 1.677208041787698e-07,
      "lm_loss": 5.4131,
      "loss": 1.1992,
      "step": 2036,
      "text_contrastive_loss": 0.6826,
      "train_positive_log_prob": -79.6951,
      "train_positive_token_accuracy": 0.0748,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3273,
      "epoch": 4.598194130925508,
      "grad_norm": 13.735128402709961,
      "learning_rate": 1.6586246467514833e-07,
      "lm_loss": 5.411,
      "loss": 1.305,
      "step": 2037,
      "text_contrastive_loss": 0.8731,
      "train_positive_log_prob": -81.145,
      "train_positive_token_accuracy": 0.0813,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.4123,
      "epoch": 4.600451467268623,
      "grad_norm": 13.067122459411621,
      "learning_rate": 1.6401430411892572e-07,
      "lm_loss": 5.3216,
      "loss": 1.3894,
      "step": 2038,
      "text_contrastive_loss": 0.8898,
      "train_positive_log_prob": -80.0312,
      "train_positive_token_accuracy": 0.0821,
      "train_positive_token_prob": 0.0331
    },
    {
      "contrastive_loss": 0.31,
      "epoch": 4.602708803611738,
      "grad_norm": 10.924623489379883,
      "learning_rate": 1.621763264016607e-07,
      "lm_loss": 5.4647,
      "loss": 1.1342,
      "step": 2039,
      "text_contrastive_loss": 0.5556,
      "train_positive_log_prob": -81.1827,
      "train_positive_token_accuracy": 0.0778,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3662,
      "epoch": 4.604966139954853,
      "grad_norm": 13.684930801391602,
      "learning_rate": 1.603485353934703e-07,
      "lm_loss": 5.4645,
      "loss": 1.3618,
      "step": 2040,
      "text_contrastive_loss": 0.8984,
      "train_positive_log_prob": -82.928,
      "train_positive_token_accuracy": 0.0855,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.2111,
      "epoch": 4.6072234762979685,
      "grad_norm": 10.00831127166748,
      "learning_rate": 1.5853093494302195e-07,
      "lm_loss": 5.4779,
      "loss": 1.058,
      "step": 2041,
      "text_contrastive_loss": 0.5982,
      "train_positive_log_prob": -80.8853,
      "train_positive_token_accuracy": 0.0775,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3747,
      "epoch": 4.609480812641084,
      "grad_norm": 14.141894340515137,
      "learning_rate": 1.567235288775265e-07,
      "lm_loss": 5.401,
      "loss": 1.319,
      "step": 2042,
      "text_contrastive_loss": 0.8084,
      "train_positive_log_prob": -78.6915,
      "train_positive_token_accuracy": 0.081,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.344,
      "epoch": 4.611738148984198,
      "grad_norm": 12.464301109313965,
      "learning_rate": 1.5492632100272686e-07,
      "lm_loss": 5.4333,
      "loss": 1.2704,
      "step": 2043,
      "text_contrastive_loss": 0.7661,
      "train_positive_log_prob": -82.3947,
      "train_positive_token_accuracy": 0.0873,
      "train_positive_token_prob": 0.033
    },
    {
      "contrastive_loss": 0.4377,
      "epoch": 4.6139954853273135,
      "grad_norm": 13.692405700683594,
      "learning_rate": 1.5313931510289482e-07,
      "lm_loss": 5.3693,
      "loss": 1.3814,
      "step": 2044,
      "text_contrastive_loss": 0.8136,
      "train_positive_log_prob": -78.9673,
      "train_positive_token_accuracy": 0.0845,
      "train_positive_token_prob": 0.0332
    },
    {
      "contrastive_loss": 0.2544,
      "epoch": 4.616252821670429,
      "grad_norm": 11.105146408081055,
      "learning_rate": 1.5136251494081822e-07,
      "lm_loss": 5.446,
      "loss": 1.1049,
      "step": 2045,
      "text_contrastive_loss": 0.6119,
      "train_positive_log_prob": -79.5606,
      "train_positive_token_accuracy": 0.0781,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.361,
      "epoch": 4.618510158013544,
      "grad_norm": 12.92881965637207,
      "learning_rate": 1.4959592425779768e-07,
      "lm_loss": 5.3528,
      "loss": 1.2643,
      "step": 2046,
      "text_contrastive_loss": 0.7361,
      "train_positive_log_prob": -79.4182,
      "train_positive_token_accuracy": 0.0798,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.3699,
      "epoch": 4.6207674943566595,
      "grad_norm": 14.304123878479004,
      "learning_rate": 1.4783954677363376e-07,
      "lm_loss": 5.4698,
      "loss": 1.316,
      "step": 2047,
      "text_contrastive_loss": 0.7984,
      "train_positive_log_prob": -82.6011,
      "train_positive_token_accuracy": 0.0755,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3806,
      "epoch": 4.623024830699774,
      "grad_norm": 12.955826759338379,
      "learning_rate": 1.4609338618662318e-07,
      "lm_loss": 5.4526,
      "loss": 1.3282,
      "step": 2048,
      "text_contrastive_loss": 0.8047,
      "train_positive_log_prob": -83.1533,
      "train_positive_token_accuracy": 0.0671,
      "train_positive_token_prob": 0.0295
    },
    {
      "contrastive_loss": 0.3194,
      "epoch": 4.625282167042889,
      "grad_norm": 12.944198608398438,
      "learning_rate": 1.4435744617354975e-07,
      "lm_loss": 5.4412,
      "loss": 1.3477,
      "step": 2049,
      "text_contrastive_loss": 0.9683,
      "train_positive_log_prob": -81.9415,
      "train_positive_token_accuracy": 0.0799,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.305,
      "epoch": 4.6275395033860045,
      "grad_norm": 11.911348342895508,
      "learning_rate": 1.4263173038967627e-07,
      "lm_loss": 5.5198,
      "loss": 1.2054,
      "step": 2050,
      "text_contrastive_loss": 0.697,
      "train_positive_log_prob": -81.1809,
      "train_positive_token_accuracy": 0.0722,
      "train_positive_token_prob": 0.0289
    },
    {
      "contrastive_loss": 0.2567,
      "epoch": 4.62979683972912,
      "grad_norm": 11.535606384277344,
      "learning_rate": 1.409162424687366e-07,
      "lm_loss": 5.3598,
      "loss": 1.1905,
      "step": 2051,
      "text_contrastive_loss": 0.7955,
      "train_positive_log_prob": -78.6422,
      "train_positive_token_accuracy": 0.0756,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.2979,
      "epoch": 4.632054176072235,
      "grad_norm": 13.395713806152344,
      "learning_rate": 1.3921098602292793e-07,
      "lm_loss": 5.3943,
      "loss": 1.2463,
      "step": 2052,
      "text_contrastive_loss": 0.8179,
      "train_positive_log_prob": -79.0636,
      "train_positive_token_accuracy": 0.0861,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.2936,
      "epoch": 4.6343115124153496,
      "grad_norm": 11.608037948608398,
      "learning_rate": 1.3751596464290529e-07,
      "lm_loss": 5.4302,
      "loss": 1.2308,
      "step": 2053,
      "text_contrastive_loss": 0.7883,
      "train_positive_log_prob": -79.7558,
      "train_positive_token_accuracy": 0.0734,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3458,
      "epoch": 4.636568848758465,
      "grad_norm": 13.384845733642578,
      "learning_rate": 1.358311818977709e-07,
      "lm_loss": 5.3795,
      "loss": 1.2662,
      "step": 2054,
      "text_contrastive_loss": 0.7648,
      "train_positive_log_prob": -77.6663,
      "train_positive_token_accuracy": 0.079,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3486,
      "epoch": 4.63882618510158,
      "grad_norm": 11.830622673034668,
      "learning_rate": 1.3415664133506812e-07,
      "lm_loss": 5.4747,
      "loss": 1.3308,
      "step": 2055,
      "text_contrastive_loss": 0.8694,
      "train_positive_log_prob": -79.8118,
      "train_positive_token_accuracy": 0.0758,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3462,
      "epoch": 4.6410835214446955,
      "grad_norm": 13.477181434631348,
      "learning_rate": 1.324923464807759e-07,
      "lm_loss": 5.4064,
      "loss": 1.3357,
      "step": 2056,
      "text_contrastive_loss": 0.8979,
      "train_positive_log_prob": -80.4606,
      "train_positive_token_accuracy": 0.0814,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.3742,
      "epoch": 4.643340857787811,
      "grad_norm": 12.482406616210938,
      "learning_rate": 1.308383008392977e-07,
      "lm_loss": 5.3617,
      "loss": 1.3185,
      "step": 2057,
      "text_contrastive_loss": 0.8163,
      "train_positive_log_prob": -78.4543,
      "train_positive_token_accuracy": 0.0723,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3072,
      "epoch": 4.645598194130925,
      "grad_norm": 11.806655883789062,
      "learning_rate": 1.2919450789345477e-07,
      "lm_loss": 5.4184,
      "loss": 1.2309,
      "step": 2058,
      "text_contrastive_loss": 0.7636,
      "train_positive_log_prob": -79.0748,
      "train_positive_token_accuracy": 0.0808,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.2766,
      "epoch": 4.6478555304740405,
      "grad_norm": 12.432291984558105,
      "learning_rate": 1.275609711044823e-07,
      "lm_loss": 5.342,
      "loss": 1.183,
      "step": 2059,
      "text_contrastive_loss": 0.7443,
      "train_positive_log_prob": -79.4607,
      "train_positive_token_accuracy": 0.0793,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.406,
      "epoch": 4.650112866817156,
      "grad_norm": 12.977935791015625,
      "learning_rate": 1.2593769391201827e-07,
      "lm_loss": 5.4356,
      "loss": 1.4555,
      "step": 2060,
      "text_contrastive_loss": 1.0119,
      "train_positive_log_prob": -80.8732,
      "train_positive_token_accuracy": 0.0848,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.3498,
      "epoch": 4.652370203160271,
      "grad_norm": 12.516545295715332,
      "learning_rate": 1.2432467973409857e-07,
      "lm_loss": 5.3364,
      "loss": 1.2904,
      "step": 2061,
      "text_contrastive_loss": 0.814,
      "train_positive_log_prob": -78.9008,
      "train_positive_token_accuracy": 0.0803,
      "train_positive_token_prob": 0.0325
    },
    {
      "contrastive_loss": 0.2692,
      "epoch": 4.654627539503386,
      "grad_norm": 10.78467082977295,
      "learning_rate": 1.2272193196714854e-07,
      "lm_loss": 5.5324,
      "loss": 1.1341,
      "step": 2062,
      "text_contrastive_loss": 0.6233,
      "train_positive_log_prob": -82.2738,
      "train_positive_token_accuracy": 0.0809,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3494,
      "epoch": 4.656884875846501,
      "grad_norm": 12.288996696472168,
      "learning_rate": 1.211294539859753e-07,
      "lm_loss": 5.4357,
      "loss": 1.2813,
      "step": 2063,
      "text_contrastive_loss": 0.7766,
      "train_positive_log_prob": -79.4139,
      "train_positive_token_accuracy": 0.0778,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3073,
      "epoch": 4.659142212189616,
      "grad_norm": 11.596695899963379,
      "learning_rate": 1.1954724914376215e-07,
      "lm_loss": 5.3737,
      "loss": 1.1777,
      "step": 2064,
      "text_contrastive_loss": 0.666,
      "train_positive_log_prob": -79.1985,
      "train_positive_token_accuracy": 0.0824,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3061,
      "epoch": 4.6613995485327315,
      "grad_norm": 12.038764953613281,
      "learning_rate": 1.1797532077206187e-07,
      "lm_loss": 5.4305,
      "loss": 1.211,
      "step": 2065,
      "text_contrastive_loss": 0.7238,
      "train_positive_log_prob": -81.8811,
      "train_positive_token_accuracy": 0.0764,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.4456,
      "epoch": 4.663656884875847,
      "grad_norm": 12.934012413024902,
      "learning_rate": 1.1641367218078736e-07,
      "lm_loss": 5.3544,
      "loss": 1.3961,
      "step": 2066,
      "text_contrastive_loss": 0.8301,
      "train_positive_log_prob": -77.4513,
      "train_positive_token_accuracy": 0.0864,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.3436,
      "epoch": 4.665914221218961,
      "grad_norm": 12.084171295166016,
      "learning_rate": 1.1486230665820552e-07,
      "lm_loss": 5.2328,
      "loss": 1.2445,
      "step": 2067,
      "text_contrastive_loss": 0.7552,
      "train_positive_log_prob": -76.1001,
      "train_positive_token_accuracy": 0.0857,
      "train_positive_token_prob": 0.0328
    },
    {
      "contrastive_loss": 0.4447,
      "epoch": 4.668171557562077,
      "grad_norm": 13.250980377197266,
      "learning_rate": 1.1332122747093277e-07,
      "lm_loss": 5.4104,
      "loss": 1.4142,
      "step": 2068,
      "text_contrastive_loss": 0.8569,
      "train_positive_log_prob": -83.3349,
      "train_positive_token_accuracy": 0.0834,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.3151,
      "epoch": 4.670428893905192,
      "grad_norm": 13.162932395935059,
      "learning_rate": 1.1179043786392507e-07,
      "lm_loss": 5.4408,
      "loss": 1.2327,
      "step": 2069,
      "text_contrastive_loss": 0.7472,
      "train_positive_log_prob": -83.866,
      "train_positive_token_accuracy": 0.0765,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.2952,
      "epoch": 4.672686230248307,
      "grad_norm": 11.531064987182617,
      "learning_rate": 1.1026994106047296e-07,
      "lm_loss": 5.4242,
      "loss": 1.2729,
      "step": 2070,
      "text_contrastive_loss": 0.8707,
      "train_positive_log_prob": -79.7545,
      "train_positive_token_accuracy": 0.0725,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.3958,
      "epoch": 4.674943566591422,
      "grad_norm": 13.006556510925293,
      "learning_rate": 1.0875974026219149e-07,
      "lm_loss": 5.3074,
      "loss": 1.3072,
      "step": 2071,
      "text_contrastive_loss": 0.7614,
      "train_positive_log_prob": -78.7218,
      "train_positive_token_accuracy": 0.088,
      "train_positive_token_prob": 0.0328
    },
    {
      "contrastive_loss": 0.3375,
      "epoch": 4.677200902934537,
      "grad_norm": 12.089827537536621,
      "learning_rate": 1.0725983864901978e-07,
      "lm_loss": 5.4113,
      "loss": 1.2594,
      "step": 2072,
      "text_contrastive_loss": 0.7614,
      "train_positive_log_prob": -80.4053,
      "train_positive_token_accuracy": 0.0828,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.3665,
      "epoch": 4.679458239277652,
      "grad_norm": 12.759051322937012,
      "learning_rate": 1.0577023937920816e-07,
      "lm_loss": 5.499,
      "loss": 1.3204,
      "step": 2073,
      "text_contrastive_loss": 0.808,
      "train_positive_log_prob": -82.7604,
      "train_positive_token_accuracy": 0.0771,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.3849,
      "epoch": 4.681715575620768,
      "grad_norm": 14.711612701416016,
      "learning_rate": 1.0429094558931485e-07,
      "lm_loss": 5.3991,
      "loss": 1.3684,
      "step": 2074,
      "text_contrastive_loss": 0.8873,
      "train_positive_log_prob": -79.6192,
      "train_positive_token_accuracy": 0.0872,
      "train_positive_token_prob": 0.0326
    },
    {
      "contrastive_loss": 0.3411,
      "epoch": 4.683972911963883,
      "grad_norm": 12.863508224487305,
      "learning_rate": 1.0282196039419823e-07,
      "lm_loss": 5.4376,
      "loss": 1.3317,
      "step": 2075,
      "text_contrastive_loss": 0.8936,
      "train_positive_log_prob": -80.5236,
      "train_positive_token_accuracy": 0.0744,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3127,
      "epoch": 4.686230248306998,
      "grad_norm": 10.788444519042969,
      "learning_rate": 1.0136328688700958e-07,
      "lm_loss": 5.4425,
      "loss": 1.2224,
      "step": 2076,
      "text_contrastive_loss": 0.731,
      "train_positive_log_prob": -79.3002,
      "train_positive_token_accuracy": 0.0718,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3667,
      "epoch": 4.688487584650113,
      "grad_norm": 13.088545799255371,
      "learning_rate": 9.99149281391898e-08,
      "lm_loss": 5.3919,
      "loss": 1.2321,
      "step": 2077,
      "text_contrastive_loss": 0.6524,
      "train_positive_log_prob": -78.898,
      "train_positive_token_accuracy": 0.0779,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3255,
      "epoch": 4.690744920993228,
      "grad_norm": 11.597572326660156,
      "learning_rate": 9.847688720045878e-08,
      "lm_loss": 5.4269,
      "loss": 1.2331,
      "step": 2078,
      "text_contrastive_loss": 0.73,
      "train_positive_log_prob": -79.6855,
      "train_positive_token_accuracy": 0.0843,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.2471,
      "epoch": 4.693002257336343,
      "grad_norm": 11.617135047912598,
      "learning_rate": 9.704916709881052e-08,
      "lm_loss": 5.4324,
      "loss": 1.145,
      "step": 2079,
      "text_contrastive_loss": 0.7092,
      "train_positive_log_prob": -79.9204,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3364,
      "epoch": 4.6952595936794586,
      "grad_norm": 11.719111442565918,
      "learning_rate": 9.5631770840508e-08,
      "lm_loss": 5.5484,
      "loss": 1.234,
      "step": 2080,
      "text_contrastive_loss": 0.6854,
      "train_positive_log_prob": -83.6129,
      "train_positive_token_accuracy": 0.0791,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.3692,
      "epoch": 4.697516930022573,
      "grad_norm": 12.207919120788574,
      "learning_rate": 9.422470141007667e-08,
      "lm_loss": 5.4526,
      "loss": 1.2597,
      "step": 2081,
      "text_contrastive_loss": 0.6903,
      "train_positive_log_prob": -82.4028,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.336,
      "epoch": 4.699774266365688,
      "grad_norm": 12.443987846374512,
      "learning_rate": 9.282796177029596e-08,
      "lm_loss": 5.4218,
      "loss": 1.2955,
      "step": 2082,
      "text_contrastive_loss": 0.8346,
      "train_positive_log_prob": -79.5021,
      "train_positive_token_accuracy": 0.0815,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.4101,
      "epoch": 4.702031602708804,
      "grad_norm": 12.711996078491211,
      "learning_rate": 9.144155486219442e-08,
      "lm_loss": 5.4367,
      "loss": 1.4459,
      "step": 2083,
      "text_contrastive_loss": 0.9843,
      "train_positive_log_prob": -80.0858,
      "train_positive_token_accuracy": 0.0785,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.2777,
      "epoch": 4.704288939051919,
      "grad_norm": 10.978333473205566,
      "learning_rate": 9.006548360504463e-08,
      "lm_loss": 5.4455,
      "loss": 1.1272,
      "step": 2084,
      "text_contrastive_loss": 0.61,
      "train_positive_log_prob": -80.8739,
      "train_positive_token_accuracy": 0.0784,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3009,
      "epoch": 4.706546275395034,
      "grad_norm": 12.159939765930176,
      "learning_rate": 8.869975089635552e-08,
      "lm_loss": 5.4226,
      "loss": 1.222,
      "step": 2085,
      "text_contrastive_loss": 0.7578,
      "train_positive_log_prob": -81.4469,
      "train_positive_token_accuracy": 0.0772,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3548,
      "epoch": 4.708803611738149,
      "grad_norm": 11.7377347946167,
      "learning_rate": 8.734435961186782e-08,
      "lm_loss": 5.4025,
      "loss": 1.2418,
      "step": 2086,
      "text_contrastive_loss": 0.6935,
      "train_positive_log_prob": -79.0716,
      "train_positive_token_accuracy": 0.0806,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3235,
      "epoch": 4.711060948081264,
      "grad_norm": 12.328231811523438,
      "learning_rate": 8.599931260554417e-08,
      "lm_loss": 5.5571,
      "loss": 1.2996,
      "step": 2087,
      "text_contrastive_loss": 0.8409,
      "train_positive_log_prob": -82.8615,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.2976,
      "epoch": 4.713318284424379,
      "grad_norm": 10.746467590332031,
      "learning_rate": 8.466461270956794e-08,
      "lm_loss": 5.4189,
      "loss": 1.2427,
      "step": 2088,
      "text_contrastive_loss": 0.8063,
      "train_positive_log_prob": -79.5882,
      "train_positive_token_accuracy": 0.0867,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.2718,
      "epoch": 4.715575620767495,
      "grad_norm": 11.545552253723145,
      "learning_rate": 8.334026273433659e-08,
      "lm_loss": 5.3835,
      "loss": 1.234,
      "step": 2089,
      "text_contrastive_loss": 0.8478,
      "train_positive_log_prob": -78.9033,
      "train_positive_token_accuracy": 0.0768,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.2311,
      "epoch": 4.717832957110609,
      "grad_norm": 10.946799278259277,
      "learning_rate": 8.202626546845172e-08,
      "lm_loss": 5.4353,
      "loss": 1.071,
      "step": 2090,
      "text_contrastive_loss": 0.5926,
      "train_positive_log_prob": -79.4662,
      "train_positive_token_accuracy": 0.078,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.4032,
      "epoch": 4.720090293453724,
      "grad_norm": 12.895364761352539,
      "learning_rate": 8.072262367871675e-08,
      "lm_loss": 5.3504,
      "loss": 1.3457,
      "step": 2091,
      "text_contrastive_loss": 0.8149,
      "train_positive_log_prob": -79.6706,
      "train_positive_token_accuracy": 0.0795,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.3089,
      "epoch": 4.72234762979684,
      "grad_norm": 13.007477760314941,
      "learning_rate": 7.942934011013037e-08,
      "lm_loss": 5.4861,
      "loss": 1.2169,
      "step": 2092,
      "text_contrastive_loss": 0.7188,
      "train_positive_log_prob": -82.4851,
      "train_positive_token_accuracy": 0.0787,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.2809,
      "epoch": 4.724604966139955,
      "grad_norm": 10.615347862243652,
      "learning_rate": 7.814641748588148e-08,
      "lm_loss": 5.2945,
      "loss": 1.1622,
      "step": 2093,
      "text_contrastive_loss": 0.7038,
      "train_positive_log_prob": -78.9861,
      "train_positive_token_accuracy": 0.0744,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.2786,
      "epoch": 4.72686230248307,
      "grad_norm": 13.077999114990234,
      "learning_rate": 7.687385850734086e-08,
      "lm_loss": 5.4131,
      "loss": 1.2135,
      "step": 2094,
      "text_contrastive_loss": 0.7872,
      "train_positive_log_prob": -80.4959,
      "train_positive_token_accuracy": 0.0757,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.312,
      "epoch": 4.729119638826186,
      "grad_norm": 13.22095775604248,
      "learning_rate": 7.561166585405789e-08,
      "lm_loss": 5.3892,
      "loss": 1.2536,
      "step": 2095,
      "text_contrastive_loss": 0.8053,
      "train_positive_log_prob": -77.1326,
      "train_positive_token_accuracy": 0.068,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.2928,
      "epoch": 4.7313769751693,
      "grad_norm": 12.82379150390625,
      "learning_rate": 7.435984218375436e-08,
      "lm_loss": 5.3739,
      "loss": 1.1897,
      "step": 2096,
      "text_contrastive_loss": 0.719,
      "train_positive_log_prob": -78.2753,
      "train_positive_token_accuracy": 0.0794,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3323,
      "epoch": 4.733634311512415,
      "grad_norm": 14.110048294067383,
      "learning_rate": 7.311839013231959e-08,
      "lm_loss": 5.4364,
      "loss": 1.2955,
      "step": 2097,
      "text_contrastive_loss": 0.8391,
      "train_positive_log_prob": -80.7377,
      "train_positive_token_accuracy": 0.0831,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.3956,
      "epoch": 4.735891647855531,
      "grad_norm": 14.666621208190918,
      "learning_rate": 7.188731231380253e-08,
      "lm_loss": 5.3727,
      "loss": 1.2443,
      "step": 2098,
      "text_contrastive_loss": 0.6229,
      "train_positive_log_prob": -79.4347,
      "train_positive_token_accuracy": 0.0748,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3946,
      "epoch": 4.738148984198646,
      "grad_norm": 12.903202056884766,
      "learning_rate": 7.066661132040853e-08,
      "lm_loss": 5.5108,
      "loss": 1.326,
      "step": 2099,
      "text_contrastive_loss": 0.7608,
      "train_positive_log_prob": -81.371,
      "train_positive_token_accuracy": 0.0825,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.2598,
      "epoch": 4.74040632054176,
      "grad_norm": 11.989819526672363,
      "learning_rate": 6.945628972249208e-08,
      "lm_loss": 5.4815,
      "loss": 1.1443,
      "step": 2100,
      "text_contrastive_loss": 0.6727,
      "train_positive_log_prob": -81.1732,
      "train_positive_token_accuracy": 0.0836,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.3638,
      "epoch": 4.742663656884876,
      "grad_norm": 12.796906471252441,
      "learning_rate": 6.825635006855458e-08,
      "lm_loss": 5.4179,
      "loss": 1.3187,
      "step": 2101,
      "text_contrastive_loss": 0.8262,
      "train_positive_log_prob": -79.9597,
      "train_positive_token_accuracy": 0.0815,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.3587,
      "epoch": 4.744920993227991,
      "grad_norm": 14.724745750427246,
      "learning_rate": 6.706679488523494e-08,
      "lm_loss": 5.451,
      "loss": 1.3276,
      "step": 2102,
      "text_contrastive_loss": 0.8476,
      "train_positive_log_prob": -80.2735,
      "train_positive_token_accuracy": 0.0846,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.2436,
      "epoch": 4.747178329571106,
      "grad_norm": 10.973912239074707,
      "learning_rate": 6.58876266773062e-08,
      "lm_loss": 5.3528,
      "loss": 1.1167,
      "step": 2103,
      "text_contrastive_loss": 0.6755,
      "train_positive_log_prob": -80.1458,
      "train_positive_token_accuracy": 0.0824,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.3384,
      "epoch": 4.749435665914222,
      "grad_norm": 12.817278861999512,
      "learning_rate": 6.471884792767169e-08,
      "lm_loss": 5.5092,
      "loss": 1.2561,
      "step": 2104,
      "text_contrastive_loss": 0.7336,
      "train_positive_log_prob": -81.477,
      "train_positive_token_accuracy": 0.0708,
      "train_positive_token_prob": 0.0298
    },
    {
      "contrastive_loss": 0.3605,
      "epoch": 4.751693002257336,
      "grad_norm": 12.139284133911133,
      "learning_rate": 6.356046109735614e-08,
      "lm_loss": 5.3043,
      "loss": 1.3008,
      "step": 2105,
      "text_contrastive_loss": 0.8197,
      "train_positive_log_prob": -78.9196,
      "train_positive_token_accuracy": 0.0869,
      "train_positive_token_prob": 0.0326
    },
    {
      "contrastive_loss": 0.3031,
      "epoch": 4.753950338600451,
      "grad_norm": 12.628264427185059,
      "learning_rate": 6.241246862550398e-08,
      "lm_loss": 5.3751,
      "loss": 1.2693,
      "step": 2106,
      "text_contrastive_loss": 0.8572,
      "train_positive_log_prob": -78.2553,
      "train_positive_token_accuracy": 0.0775,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.3356,
      "epoch": 4.756207674943567,
      "grad_norm": 13.342379570007324,
      "learning_rate": 6.127487292937328e-08,
      "lm_loss": 5.4735,
      "loss": 1.3297,
      "step": 2107,
      "text_contrastive_loss": 0.8935,
      "train_positive_log_prob": -79.8123,
      "train_positive_token_accuracy": 0.0766,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.2696,
      "epoch": 4.758465011286682,
      "grad_norm": 10.582463264465332,
      "learning_rate": 6.014767640432905e-08,
      "lm_loss": 5.4762,
      "loss": 1.164,
      "step": 2108,
      "text_contrastive_loss": 0.6935,
      "train_positive_log_prob": -82.9131,
      "train_positive_token_accuracy": 0.0872,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.4206,
      "epoch": 4.760722347629796,
      "grad_norm": 12.734881401062012,
      "learning_rate": 5.903088142384106e-08,
      "lm_loss": 5.3532,
      "loss": 1.3542,
      "step": 2109,
      "text_contrastive_loss": 0.7965,
      "train_positive_log_prob": -81.2604,
      "train_positive_token_accuracy": 0.0808,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.3928,
      "epoch": 4.762979683972912,
      "grad_norm": 14.01072883605957,
      "learning_rate": 5.7924490339474335e-08,
      "lm_loss": 5.5281,
      "loss": 1.3354,
      "step": 2110,
      "text_contrastive_loss": 0.7796,
      "train_positive_log_prob": -81.6016,
      "train_positive_token_accuracy": 0.0791,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3374,
      "epoch": 4.765237020316027,
      "grad_norm": 12.548912048339844,
      "learning_rate": 5.682850548089036e-08,
      "lm_loss": 5.4102,
      "loss": 1.2604,
      "step": 2111,
      "text_contrastive_loss": 0.7639,
      "train_positive_log_prob": -79.5031,
      "train_positive_token_accuracy": 0.0752,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3398,
      "epoch": 4.767494356659142,
      "grad_norm": 13.495406150817871,
      "learning_rate": 5.574292915583646e-08,
      "lm_loss": 5.3557,
      "loss": 1.2511,
      "step": 2112,
      "text_contrastive_loss": 0.7515,
      "train_positive_log_prob": -80.2072,
      "train_positive_token_accuracy": 0.0845,
      "train_positive_token_prob": 0.0327
    },
    {
      "contrastive_loss": 0.3384,
      "epoch": 4.769751693002258,
      "grad_norm": 11.788864135742188,
      "learning_rate": 5.46677636501447e-08,
      "lm_loss": 5.4145,
      "loss": 1.227,
      "step": 2113,
      "text_contrastive_loss": 0.6943,
      "train_positive_log_prob": -79.0759,
      "train_positive_token_accuracy": 0.0826,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.317,
      "epoch": 4.772009029345372,
      "grad_norm": 11.977039337158203,
      "learning_rate": 5.3603011227725265e-08,
      "lm_loss": 5.3856,
      "loss": 1.2106,
      "step": 2114,
      "text_contrastive_loss": 0.71,
      "train_positive_log_prob": -80.6362,
      "train_positive_token_accuracy": 0.0759,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.4162,
      "epoch": 4.774266365688487,
      "grad_norm": 15.057369232177734,
      "learning_rate": 5.2548674130561974e-08,
      "lm_loss": 5.3857,
      "loss": 1.4213,
      "step": 2115,
      "text_contrastive_loss": 0.9331,
      "train_positive_log_prob": -80.8506,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.3748,
      "epoch": 4.776523702031603,
      "grad_norm": 13.112998962402344,
      "learning_rate": 5.1504754578707294e-08,
      "lm_loss": 5.4176,
      "loss": 1.2763,
      "step": 2116,
      "text_contrastive_loss": 0.7194,
      "train_positive_log_prob": -81.0951,
      "train_positive_token_accuracy": 0.0814,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3005,
      "epoch": 4.778781038374718,
      "grad_norm": 12.008501052856445,
      "learning_rate": 5.047125477027959e-08,
      "lm_loss": 5.3666,
      "loss": 1.2459,
      "step": 2117,
      "text_contrastive_loss": 0.8174,
      "train_positive_log_prob": -80.3735,
      "train_positive_token_accuracy": 0.0816,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3083,
      "epoch": 4.781038374717833,
      "grad_norm": 13.533900260925293,
      "learning_rate": 4.944817688145642e-08,
      "lm_loss": 5.4471,
      "loss": 1.2277,
      "step": 2118,
      "text_contrastive_loss": 0.7494,
      "train_positive_log_prob": -80.7105,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.2717,
      "epoch": 4.783295711060948,
      "grad_norm": 13.106285095214844,
      "learning_rate": 4.843552306646904e-08,
      "lm_loss": 5.4763,
      "loss": 1.2196,
      "step": 2119,
      "text_contrastive_loss": 0.8005,
      "train_positive_log_prob": -78.9802,
      "train_positive_token_accuracy": 0.0764,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3135,
      "epoch": 4.785553047404063,
      "grad_norm": 12.778962135314941,
      "learning_rate": 4.743329545760122e-08,
      "lm_loss": 5.5184,
      "loss": 1.3231,
      "step": 2120,
      "text_contrastive_loss": 0.9156,
      "train_positive_log_prob": -80.8988,
      "train_positive_token_accuracy": 0.0677,
      "train_positive_token_prob": 0.029
    },
    {
      "contrastive_loss": 0.2959,
      "epoch": 4.787810383747178,
      "grad_norm": 11.397628784179688,
      "learning_rate": 4.644149616518212e-08,
      "lm_loss": 5.4751,
      "loss": 1.2509,
      "step": 2121,
      "text_contrastive_loss": 0.8151,
      "train_positive_log_prob": -83.5852,
      "train_positive_token_accuracy": 0.0798,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.339,
      "epoch": 4.790067720090294,
      "grad_norm": 13.37972640991211,
      "learning_rate": 4.5460127277582863e-08,
      "lm_loss": 5.4814,
      "loss": 1.2819,
      "step": 2122,
      "text_contrastive_loss": 0.7893,
      "train_positive_log_prob": -82.4566,
      "train_positive_token_accuracy": 0.0766,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.2208,
      "epoch": 4.792325056433409,
      "grad_norm": 11.886345863342285,
      "learning_rate": 4.448919086121217e-08,
      "lm_loss": 5.4168,
      "loss": 1.1237,
      "step": 2123,
      "text_contrastive_loss": 0.7226,
      "train_positive_log_prob": -79.4093,
      "train_positive_token_accuracy": 0.0855,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.4379,
      "epoch": 4.794582392776523,
      "grad_norm": 14.661003112792969,
      "learning_rate": 4.352868896051077e-08,
      "lm_loss": 5.2979,
      "loss": 1.417,
      "step": 2124,
      "text_contrastive_loss": 0.8986,
      "train_positive_log_prob": -76.606,
      "train_positive_token_accuracy": 0.0821,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3752,
      "epoch": 4.796839729119639,
      "grad_norm": 12.711482048034668,
      "learning_rate": 4.2578623597949174e-08,
      "lm_loss": 5.4417,
      "loss": 1.3594,
      "step": 2125,
      "text_contrastive_loss": 0.88,
      "train_positive_log_prob": -80.8598,
      "train_positive_token_accuracy": 0.0807,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.2863,
      "epoch": 4.799097065462754,
      "grad_norm": 11.192163467407227,
      "learning_rate": 4.163899677402161e-08,
      "lm_loss": 5.468,
      "loss": 1.2364,
      "step": 2126,
      "text_contrastive_loss": 0.8066,
      "train_positive_log_prob": -81.3618,
      "train_positive_token_accuracy": 0.0801,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3255,
      "epoch": 4.801354401805869,
      "grad_norm": 12.773097038269043,
      "learning_rate": 4.0709810467243204e-08,
      "lm_loss": 5.429,
      "loss": 1.2461,
      "step": 2127,
      "text_contrastive_loss": 0.7556,
      "train_positive_log_prob": -78.9775,
      "train_positive_token_accuracy": 0.0795,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.343,
      "epoch": 4.803611738148984,
      "grad_norm": 14.528846740722656,
      "learning_rate": 3.979106663414389e-08,
      "lm_loss": 5.523,
      "loss": 1.282,
      "step": 2128,
      "text_contrastive_loss": 0.7735,
      "train_positive_log_prob": -80.3841,
      "train_positive_token_accuracy": 0.0773,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.3483,
      "epoch": 4.805869074492099,
      "grad_norm": 12.990553855895996,
      "learning_rate": 3.8882767209266756e-08,
      "lm_loss": 5.3118,
      "loss": 1.299,
      "step": 2129,
      "text_contrastive_loss": 0.8389,
      "train_positive_log_prob": -78.0564,
      "train_positive_token_accuracy": 0.0803,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.3104,
      "epoch": 4.808126410835214,
      "grad_norm": 12.372488021850586,
      "learning_rate": 3.7984914105162474e-08,
      "lm_loss": 5.3832,
      "loss": 1.2436,
      "step": 2130,
      "text_contrastive_loss": 0.7897,
      "train_positive_log_prob": -78.5156,
      "train_positive_token_accuracy": 0.0798,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3918,
      "epoch": 4.81038374717833,
      "grad_norm": 13.191797256469727,
      "learning_rate": 3.709750921238486e-08,
      "lm_loss": 5.3872,
      "loss": 1.3841,
      "step": 2131,
      "text_contrastive_loss": 0.9073,
      "train_positive_log_prob": -79.9259,
      "train_positive_token_accuracy": 0.0798,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.2995,
      "epoch": 4.812641083521445,
      "grad_norm": 12.146777153015137,
      "learning_rate": 3.622055439948813e-08,
      "lm_loss": 5.4303,
      "loss": 1.2288,
      "step": 2132,
      "text_contrastive_loss": 0.7725,
      "train_positive_log_prob": -80.8785,
      "train_positive_token_accuracy": 0.0779,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.2753,
      "epoch": 4.8148984198645595,
      "grad_norm": 12.120311737060547,
      "learning_rate": 3.5354051513022405e-08,
      "lm_loss": 5.5964,
      "loss": 1.2464,
      "step": 2133,
      "text_contrastive_loss": 0.8229,
      "train_positive_log_prob": -81.5879,
      "train_positive_token_accuracy": 0.0734,
      "train_positive_token_prob": 0.0297
    },
    {
      "contrastive_loss": 0.4458,
      "epoch": 4.817155756207675,
      "grad_norm": 14.222434043884277,
      "learning_rate": 3.449800237753043e-08,
      "lm_loss": 5.5048,
      "loss": 1.4594,
      "step": 2134,
      "text_contrastive_loss": 0.9263,
      "train_positive_log_prob": -81.2377,
      "train_positive_token_accuracy": 0.0752,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.306,
      "epoch": 4.81941309255079,
      "grad_norm": 12.048348426818848,
      "learning_rate": 3.365240879554144e-08,
      "lm_loss": 5.3986,
      "loss": 1.2466,
      "step": 2135,
      "text_contrastive_loss": 0.8014,
      "train_positive_log_prob": -80.4526,
      "train_positive_token_accuracy": 0.0862,
      "train_positive_token_prob": 0.0327
    },
    {
      "contrastive_loss": 0.3533,
      "epoch": 4.821670428893905,
      "grad_norm": 12.44783878326416,
      "learning_rate": 3.281727254757061e-08,
      "lm_loss": 5.3986,
      "loss": 1.2914,
      "step": 2136,
      "text_contrastive_loss": 0.7965,
      "train_positive_log_prob": -81.5695,
      "train_positive_token_accuracy": 0.0778,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.2698,
      "epoch": 4.82392776523702,
      "grad_norm": 10.991819381713867,
      "learning_rate": 3.1992595392112966e-08,
      "lm_loss": 5.4381,
      "loss": 1.1362,
      "step": 2137,
      "text_contrastive_loss": 0.6451,
      "train_positive_log_prob": -78.2802,
      "train_positive_token_accuracy": 0.0746,
      "train_positive_token_prob": 0.03
    },
    {
      "contrastive_loss": 0.2979,
      "epoch": 4.826185101580135,
      "grad_norm": 11.42900562286377,
      "learning_rate": 3.117837906564114e-08,
      "lm_loss": 5.4505,
      "loss": 1.2333,
      "step": 2138,
      "text_contrastive_loss": 0.7807,
      "train_positive_log_prob": -80.6938,
      "train_positive_token_accuracy": 0.0781,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.201,
      "epoch": 4.8284424379232505,
      "grad_norm": 10.826726913452148,
      "learning_rate": 3.0374625282599826e-08,
      "lm_loss": 5.4096,
      "loss": 1.0695,
      "step": 2139,
      "text_contrastive_loss": 0.6552,
      "train_positive_log_prob": -80.4656,
      "train_positive_token_accuracy": 0.0861,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.4181,
      "epoch": 4.830699774266366,
      "grad_norm": 15.124938011169434,
      "learning_rate": 2.9581335735404672e-08,
      "lm_loss": 5.4617,
      "loss": 1.3917,
      "step": 2140,
      "text_contrastive_loss": 0.855,
      "train_positive_log_prob": -80.624,
      "train_positive_token_accuracy": 0.0836,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3782,
      "epoch": 4.832957110609481,
      "grad_norm": 14.185858726501465,
      "learning_rate": 2.8798512094436738e-08,
      "lm_loss": 5.3913,
      "loss": 1.3164,
      "step": 2141,
      "text_contrastive_loss": 0.7982,
      "train_positive_log_prob": -78.5765,
      "train_positive_token_accuracy": 0.0866,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3813,
      "epoch": 4.835214446952596,
      "grad_norm": 12.532388687133789,
      "learning_rate": 2.802615600804026e-08,
      "lm_loss": 5.3925,
      "loss": 1.3371,
      "step": 2142,
      "text_contrastive_loss": 0.8331,
      "train_positive_log_prob": -80.0637,
      "train_positive_token_accuracy": 0.0839,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.3296,
      "epoch": 4.837471783295711,
      "grad_norm": 12.764655113220215,
      "learning_rate": 2.7264269102517117e-08,
      "lm_loss": 5.5455,
      "loss": 1.2569,
      "step": 2143,
      "text_contrastive_loss": 0.7455,
      "train_positive_log_prob": -80.5822,
      "train_positive_token_accuracy": 0.0761,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3623,
      "epoch": 4.839729119638826,
      "grad_norm": 12.576608657836914,
      "learning_rate": 2.6512852982127357e-08,
      "lm_loss": 5.329,
      "loss": 1.3258,
      "step": 2144,
      "text_contrastive_loss": 0.8611,
      "train_positive_log_prob": -79.4155,
      "train_positive_token_accuracy": 0.0847,
      "train_positive_token_prob": 0.0325
    },
    {
      "contrastive_loss": 0.3862,
      "epoch": 4.841986455981941,
      "grad_norm": 13.27994441986084,
      "learning_rate": 2.577190922908035e-08,
      "lm_loss": 5.4388,
      "loss": 1.4206,
      "step": 2145,
      "text_contrastive_loss": 0.981,
      "train_positive_log_prob": -81.9209,
      "train_positive_token_accuracy": 0.082,
      "train_positive_token_prob": 0.0321
    },
    {
      "contrastive_loss": 0.3339,
      "epoch": 4.844243792325057,
      "grad_norm": 11.295732498168945,
      "learning_rate": 2.5041439403537537e-08,
      "lm_loss": 5.3908,
      "loss": 1.2231,
      "step": 2146,
      "text_contrastive_loss": 0.7002,
      "train_positive_log_prob": -80.1039,
      "train_positive_token_accuracy": 0.0812,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.325,
      "epoch": 4.846501128668171,
      "grad_norm": 13.89465618133545,
      "learning_rate": 2.4321445043603565e-08,
      "lm_loss": 5.3316,
      "loss": 1.2789,
      "step": 2147,
      "text_contrastive_loss": 0.8415,
      "train_positive_log_prob": -76.8932,
      "train_positive_token_accuracy": 0.0821,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.3286,
      "epoch": 4.8487584650112865,
      "grad_norm": 11.77916145324707,
      "learning_rate": 2.3611927665326275e-08,
      "lm_loss": 5.3282,
      "loss": 1.2164,
      "step": 2148,
      "text_contrastive_loss": 0.7099,
      "train_positive_log_prob": -79.9631,
      "train_positive_token_accuracy": 0.0807,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3907,
      "epoch": 4.851015801354402,
      "grad_norm": 13.596877098083496,
      "learning_rate": 2.291288876269393e-08,
      "lm_loss": 5.4148,
      "loss": 1.3509,
      "step": 2149,
      "text_contrastive_loss": 0.8375,
      "train_positive_log_prob": -79.6366,
      "train_positive_token_accuracy": 0.0794,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.3773,
      "epoch": 4.853273137697517,
      "grad_norm": 12.371967315673828,
      "learning_rate": 2.222432980762912e-08,
      "lm_loss": 5.4539,
      "loss": 1.3755,
      "step": 2150,
      "text_contrastive_loss": 0.9055,
      "train_positive_log_prob": -80.511,
      "train_positive_token_accuracy": 0.0725,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.361,
      "epoch": 4.855530474040632,
      "grad_norm": 14.236940383911133,
      "learning_rate": 2.1546252249988186e-08,
      "lm_loss": 5.4339,
      "loss": 1.2696,
      "step": 2151,
      "text_contrastive_loss": 0.7305,
      "train_positive_log_prob": -80.81,
      "train_positive_token_accuracy": 0.0841,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.272,
      "epoch": 4.857787810383747,
      "grad_norm": 11.388404846191406,
      "learning_rate": 2.087865751755791e-08,
      "lm_loss": 5.504,
      "loss": 1.1417,
      "step": 2152,
      "text_contrastive_loss": 0.6385,
      "train_positive_log_prob": -81.1409,
      "train_positive_token_accuracy": 0.0692,
      "train_positive_token_prob": 0.0301
    },
    {
      "contrastive_loss": 0.3098,
      "epoch": 4.860045146726862,
      "grad_norm": 12.139877319335938,
      "learning_rate": 2.0221547016051614e-08,
      "lm_loss": 5.3685,
      "loss": 1.1453,
      "step": 2153,
      "text_contrastive_loss": 0.5971,
      "train_positive_log_prob": -79.4589,
      "train_positive_token_accuracy": 0.08,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.3832,
      "epoch": 4.8623024830699775,
      "grad_norm": 13.00784683227539,
      "learning_rate": 1.957492212910639e-08,
      "lm_loss": 5.4133,
      "loss": 1.3761,
      "step": 2154,
      "text_contrastive_loss": 0.9032,
      "train_positive_log_prob": -80.165,
      "train_positive_token_accuracy": 0.0757,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3381,
      "epoch": 4.864559819413093,
      "grad_norm": 11.445502281188965,
      "learning_rate": 1.8938784218281435e-08,
      "lm_loss": 5.3975,
      "loss": 1.2394,
      "step": 2155,
      "text_contrastive_loss": 0.7231,
      "train_positive_log_prob": -80.9187,
      "train_positive_token_accuracy": 0.0743,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3416,
      "epoch": 4.866817155756207,
      "grad_norm": 12.394635200500488,
      "learning_rate": 1.8313134623051955e-08,
      "lm_loss": 5.476,
      "loss": 1.2625,
      "step": 2156,
      "text_contrastive_loss": 0.7466,
      "train_positive_log_prob": -80.7737,
      "train_positive_token_accuracy": 0.0802,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3451,
      "epoch": 4.8690744920993225,
      "grad_norm": 12.728734970092773,
      "learning_rate": 1.7697974660811357e-08,
      "lm_loss": 5.3634,
      "loss": 1.2057,
      "step": 2157,
      "text_contrastive_loss": 0.6484,
      "train_positive_log_prob": -80.9139,
      "train_positive_token_accuracy": 0.0798,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3577,
      "epoch": 4.871331828442438,
      "grad_norm": 13.499052047729492,
      "learning_rate": 1.7093305626864065e-08,
      "lm_loss": 5.3418,
      "loss": 1.2358,
      "step": 2158,
      "text_contrastive_loss": 0.6877,
      "train_positive_log_prob": -78.5233,
      "train_positive_token_accuracy": 0.0832,
      "train_positive_token_prob": 0.0327
    },
    {
      "contrastive_loss": 0.3278,
      "epoch": 4.873589164785553,
      "grad_norm": 12.652538299560547,
      "learning_rate": 1.6499128794423836e-08,
      "lm_loss": 5.3845,
      "loss": 1.2968,
      "step": 2159,
      "text_contrastive_loss": 0.861,
      "train_positive_log_prob": -80.3219,
      "train_positive_token_accuracy": 0.077,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.3964,
      "epoch": 4.8758465011286685,
      "grad_norm": 13.595474243164062,
      "learning_rate": 1.5915445414613208e-08,
      "lm_loss": 5.3284,
      "loss": 1.331,
      "step": 2160,
      "text_contrastive_loss": 0.8034,
      "train_positive_log_prob": -78.7615,
      "train_positive_token_accuracy": 0.0854,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3774,
      "epoch": 4.878103837471784,
      "grad_norm": 13.105132102966309,
      "learning_rate": 1.5342256716459058e-08,
      "lm_loss": 5.3624,
      "loss": 1.3081,
      "step": 2161,
      "text_contrastive_loss": 0.7889,
      "train_positive_log_prob": -80.5641,
      "train_positive_token_accuracy": 0.0793,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.2354,
      "epoch": 4.880361173814898,
      "grad_norm": 12.309341430664062,
      "learning_rate": 1.4779563906888172e-08,
      "lm_loss": 5.4188,
      "loss": 1.192,
      "step": 2162,
      "text_contrastive_loss": 0.8294,
      "train_positive_log_prob": -80.8815,
      "train_positive_token_accuracy": 0.0717,
      "train_positive_token_prob": 0.0294
    },
    {
      "contrastive_loss": 0.2926,
      "epoch": 4.8826185101580135,
      "grad_norm": 11.640186309814453,
      "learning_rate": 1.4227368170728894e-08,
      "lm_loss": 5.3122,
      "loss": 1.1701,
      "step": 2163,
      "text_contrastive_loss": 0.6926,
      "train_positive_log_prob": -76.9255,
      "train_positive_token_accuracy": 0.0782,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.4086,
      "epoch": 4.884875846501129,
      "grad_norm": 15.052828788757324,
      "learning_rate": 1.3685670670706697e-08,
      "lm_loss": 5.4157,
      "loss": 1.3768,
      "step": 2164,
      "text_contrastive_loss": 0.8532,
      "train_positive_log_prob": -79.8648,
      "train_positive_token_accuracy": 0.0723,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.3648,
      "epoch": 4.887133182844244,
      "grad_norm": 11.98923110961914,
      "learning_rate": 1.3154472547440289e-08,
      "lm_loss": 5.405,
      "loss": 1.3541,
      "step": 2165,
      "text_contrastive_loss": 0.8975,
      "train_positive_log_prob": -79.3221,
      "train_positive_token_accuracy": 0.0868,
      "train_positive_token_prob": 0.0331
    },
    {
      "contrastive_loss": 0.3488,
      "epoch": 4.889390519187359,
      "grad_norm": 12.5957612991333,
      "learning_rate": 1.2633774919441622e-08,
      "lm_loss": 5.3882,
      "loss": 1.2936,
      "step": 2166,
      "text_contrastive_loss": 0.812,
      "train_positive_log_prob": -79.3674,
      "train_positive_token_accuracy": 0.0711,
      "train_positive_token_prob": 0.0303
    },
    {
      "contrastive_loss": 0.3448,
      "epoch": 4.891647855530474,
      "grad_norm": 11.962646484375,
      "learning_rate": 1.2123578883110887e-08,
      "lm_loss": 5.4075,
      "loss": 1.2624,
      "step": 2167,
      "text_contrastive_loss": 0.7537,
      "train_positive_log_prob": -80.3766,
      "train_positive_token_accuracy": 0.0765,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.3378,
      "epoch": 4.893905191873589,
      "grad_norm": 11.774933815002441,
      "learning_rate": 1.1623885512737076e-08,
      "lm_loss": 5.3664,
      "loss": 1.2064,
      "step": 2168,
      "text_contrastive_loss": 0.6639,
      "train_positive_log_prob": -80.263,
      "train_positive_token_accuracy": 0.0846,
      "train_positive_token_prob": 0.0322
    },
    {
      "contrastive_loss": 0.2715,
      "epoch": 4.8961625282167045,
      "grad_norm": 11.338656425476074,
      "learning_rate": 1.1134695860493539e-08,
      "lm_loss": 5.2685,
      "loss": 1.1253,
      "step": 2169,
      "text_contrastive_loss": 0.6538,
      "train_positive_log_prob": -77.6493,
      "train_positive_token_accuracy": 0.0819,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.284,
      "epoch": 4.89841986455982,
      "grad_norm": 11.437028884887695,
      "learning_rate": 1.0656010956437979e-08,
      "lm_loss": 5.4169,
      "loss": 1.2548,
      "step": 2170,
      "text_contrastive_loss": 0.8582,
      "train_positive_log_prob": -80.4479,
      "train_positive_token_accuracy": 0.0786,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.3261,
      "epoch": 4.900677200902934,
      "grad_norm": 11.590265274047852,
      "learning_rate": 1.018783180850691e-08,
      "lm_loss": 5.4278,
      "loss": 1.1799,
      "step": 2171,
      "text_contrastive_loss": 0.6221,
      "train_positive_log_prob": -79.2501,
      "train_positive_token_accuracy": 0.0811,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.3522,
      "epoch": 4.9029345372460496,
      "grad_norm": 13.533720970153809,
      "learning_rate": 9.73015940251676e-09,
      "lm_loss": 5.388,
      "loss": 1.2409,
      "step": 2172,
      "text_contrastive_loss": 0.6997,
      "train_positive_log_prob": -79.0731,
      "train_positive_token_accuracy": 0.0745,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3091,
      "epoch": 4.905191873589165,
      "grad_norm": 12.223340034484863,
      "learning_rate": 9.282994702159986e-09,
      "lm_loss": 5.395,
      "loss": 1.2321,
      "step": 2173,
      "text_contrastive_loss": 0.767,
      "train_positive_log_prob": -79.5565,
      "train_positive_token_accuracy": 0.0821,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.3397,
      "epoch": 4.90744920993228,
      "grad_norm": 11.706966400146484,
      "learning_rate": 8.846338649005082e-09,
      "lm_loss": 5.4793,
      "loss": 1.2444,
      "step": 2174,
      "text_contrastive_loss": 0.7136,
      "train_positive_log_prob": -80.9685,
      "train_positive_token_accuracy": 0.079,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.313,
      "epoch": 4.909706546275395,
      "grad_norm": 12.395425796508789,
      "learning_rate": 8.42019216249046e-09,
      "lm_loss": 5.3893,
      "loss": 1.1799,
      "step": 2175,
      "text_contrastive_loss": 0.656,
      "train_positive_log_prob": -78.2505,
      "train_positive_token_accuracy": 0.0729,
      "train_positive_token_prob": 0.0312
    },
    {
      "contrastive_loss": 0.292,
      "epoch": 4.91196388261851,
      "grad_norm": 12.1083345413208,
      "learning_rate": 8.004556139927788e-09,
      "lm_loss": 5.3979,
      "loss": 1.1718,
      "step": 2176,
      "text_contrastive_loss": 0.6799,
      "train_positive_log_prob": -78.2929,
      "train_positive_token_accuracy": 0.0743,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3209,
      "epoch": 4.914221218961625,
      "grad_norm": 12.850380897521973,
      "learning_rate": 7.599431456495888e-09,
      "lm_loss": 5.3555,
      "loss": 1.2501,
      "step": 2177,
      "text_contrastive_loss": 0.7874,
      "train_positive_log_prob": -78.1134,
      "train_positive_token_accuracy": 0.0806,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.2447,
      "epoch": 4.9164785553047405,
      "grad_norm": 10.093464851379395,
      "learning_rate": 7.2048189652412784e-09,
      "lm_loss": 5.4659,
      "loss": 1.159,
      "step": 2178,
      "text_contrastive_loss": 0.7354,
      "train_positive_log_prob": -81.4749,
      "train_positive_token_accuracy": 0.0852,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.307,
      "epoch": 4.918735891647856,
      "grad_norm": 12.094605445861816,
      "learning_rate": 6.820719497074857e-09,
      "lm_loss": 5.4588,
      "loss": 1.2629,
      "step": 2179,
      "text_contrastive_loss": 0.8201,
      "train_positive_log_prob": -80.2418,
      "train_positive_token_accuracy": 0.0779,
      "train_positive_token_prob": 0.0313
    },
    {
      "contrastive_loss": 0.3124,
      "epoch": 4.92099322799097,
      "grad_norm": 11.375383377075195,
      "learning_rate": 6.447133860771893e-09,
      "lm_loss": 5.3883,
      "loss": 1.2174,
      "step": 2180,
      "text_contrastive_loss": 0.7324,
      "train_positive_log_prob": -79.6664,
      "train_positive_token_accuracy": 0.0803,
      "train_positive_token_prob": 0.032
    },
    {
      "contrastive_loss": 0.3418,
      "epoch": 4.923250564334086,
      "grad_norm": 12.987213134765625,
      "learning_rate": 6.084062842968696e-09,
      "lm_loss": 5.3302,
      "loss": 1.278,
      "step": 2181,
      "text_contrastive_loss": 0.8064,
      "train_positive_log_prob": -79.5401,
      "train_positive_token_accuracy": 0.0812,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.316,
      "epoch": 4.925507900677201,
      "grad_norm": 13.518106460571289,
      "learning_rate": 5.731507208160958e-09,
      "lm_loss": 5.55,
      "loss": 1.216,
      "step": 2182,
      "text_contrastive_loss": 0.6899,
      "train_positive_log_prob": -81.4133,
      "train_positive_token_accuracy": 0.0775,
      "train_positive_token_prob": 0.0307
    },
    {
      "contrastive_loss": 0.3942,
      "epoch": 4.927765237020316,
      "grad_norm": 14.615264892578125,
      "learning_rate": 5.389467698704298e-09,
      "lm_loss": 5.3942,
      "loss": 1.3891,
      "step": 2183,
      "text_contrastive_loss": 0.911,
      "train_positive_log_prob": -78.4682,
      "train_positive_token_accuracy": 0.083,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.3662,
      "epoch": 4.9300225733634315,
      "grad_norm": 11.848650932312012,
      "learning_rate": 5.057945034810385e-09,
      "lm_loss": 5.3746,
      "loss": 1.2872,
      "step": 2184,
      "text_contrastive_loss": 0.7672,
      "train_positive_log_prob": -79.4624,
      "train_positive_token_accuracy": 0.0808,
      "train_positive_token_prob": 0.0325
    },
    {
      "contrastive_loss": 0.2615,
      "epoch": 4.932279909706546,
      "grad_norm": 11.342528343200684,
      "learning_rate": 4.736939914545824e-09,
      "lm_loss": 5.363,
      "loss": 1.1611,
      "step": 2185,
      "text_contrastive_loss": 0.7268,
      "train_positive_log_prob": -76.8871,
      "train_positive_token_accuracy": 0.0787,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3441,
      "epoch": 4.934537246049661,
      "grad_norm": 14.007142066955566,
      "learning_rate": 4.4264530138310445e-09,
      "lm_loss": 5.4225,
      "loss": 1.2712,
      "step": 2186,
      "text_contrastive_loss": 0.7696,
      "train_positive_log_prob": -80.5107,
      "train_positive_token_accuracy": 0.0813,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3191,
      "epoch": 4.936794582392777,
      "grad_norm": 11.95272445678711,
      "learning_rate": 4.1264849864403044e-09,
      "lm_loss": 5.4757,
      "loss": 1.2197,
      "step": 2187,
      "text_contrastive_loss": 0.7061,
      "train_positive_log_prob": -79.8722,
      "train_positive_token_accuracy": 0.0755,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.3955,
      "epoch": 4.939051918735892,
      "grad_norm": 13.96945858001709,
      "learning_rate": 3.837036463997246e-09,
      "lm_loss": 5.4896,
      "loss": 1.3315,
      "step": 2188,
      "text_contrastive_loss": 0.774,
      "train_positive_log_prob": -83.0937,
      "train_positive_token_accuracy": 0.0823,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3248,
      "epoch": 4.941309255079007,
      "grad_norm": 12.298246383666992,
      "learning_rate": 3.558108055976006e-09,
      "lm_loss": 5.3952,
      "loss": 1.2783,
      "step": 2189,
      "text_contrastive_loss": 0.8278,
      "train_positive_log_prob": -78.2913,
      "train_positive_token_accuracy": 0.0781,
      "train_positive_token_prob": 0.0314
    },
    {
      "contrastive_loss": 0.2874,
      "epoch": 4.943566591422122,
      "grad_norm": 11.63664436340332,
      "learning_rate": 3.289700349698999e-09,
      "lm_loss": 5.4742,
      "loss": 1.1871,
      "step": 2190,
      "text_contrastive_loss": 0.7045,
      "train_positive_log_prob": -84.9146,
      "train_positive_token_accuracy": 0.0846,
      "train_positive_token_prob": 0.0329
    },
    {
      "contrastive_loss": 0.2688,
      "epoch": 4.945823927765237,
      "grad_norm": 12.425889015197754,
      "learning_rate": 3.0318139103363564e-09,
      "lm_loss": 5.352,
      "loss": 1.1751,
      "step": 2191,
      "text_contrastive_loss": 0.7422,
      "train_positive_log_prob": -77.8727,
      "train_positive_token_accuracy": 0.0821,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.3718,
      "epoch": 4.948081264108352,
      "grad_norm": 14.650064468383789,
      "learning_rate": 2.7844492809031567e-09,
      "lm_loss": 5.4311,
      "loss": 1.2509,
      "step": 2192,
      "text_contrastive_loss": 0.672,
      "train_positive_log_prob": -81.0791,
      "train_positive_token_accuracy": 0.0788,
      "train_positive_token_prob": 0.0319
    },
    {
      "contrastive_loss": 0.3667,
      "epoch": 4.950338600451468,
      "grad_norm": 16.653989791870117,
      "learning_rate": 2.547606982260531e-09,
      "lm_loss": 5.3663,
      "loss": 1.2801,
      "step": 2193,
      "text_contrastive_loss": 0.7535,
      "train_positive_log_prob": -82.4041,
      "train_positive_token_accuracy": 0.0768,
      "train_positive_token_prob": 0.0311
    },
    {
      "contrastive_loss": 0.3982,
      "epoch": 4.952595936794582,
      "grad_norm": 14.31432819366455,
      "learning_rate": 2.3212875131117805e-09,
      "lm_loss": 5.3636,
      "loss": 1.3578,
      "step": 2194,
      "text_contrastive_loss": 0.8466,
      "train_positive_log_prob": -77.1246,
      "train_positive_token_accuracy": 0.0753,
      "train_positive_token_prob": 0.031
    },
    {
      "contrastive_loss": 0.3434,
      "epoch": 4.954853273137697,
      "grad_norm": 11.791407585144043,
      "learning_rate": 2.1054913500051512e-09,
      "lm_loss": 5.3388,
      "loss": 1.2172,
      "step": 2195,
      "text_contrastive_loss": 0.6799,
      "train_positive_log_prob": -80.1854,
      "train_positive_token_accuracy": 0.0861,
      "train_positive_token_prob": 0.0327
    },
    {
      "contrastive_loss": 0.4087,
      "epoch": 4.957110609480813,
      "grad_norm": 14.255420684814453,
      "learning_rate": 1.9002189473288356e-09,
      "lm_loss": 5.4553,
      "loss": 1.3133,
      "step": 2196,
      "text_contrastive_loss": 0.7181,
      "train_positive_log_prob": -81.9416,
      "train_positive_token_accuracy": 0.0866,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.3213,
      "epoch": 4.959367945823928,
      "grad_norm": 10.890216827392578,
      "learning_rate": 1.7054707373126423e-09,
      "lm_loss": 5.441,
      "loss": 1.2117,
      "step": 2197,
      "text_contrastive_loss": 0.6927,
      "train_positive_log_prob": -79.7222,
      "train_positive_token_accuracy": 0.0744,
      "train_positive_token_prob": 0.0302
    },
    {
      "contrastive_loss": 0.2907,
      "epoch": 4.961625282167043,
      "grad_norm": 12.555000305175781,
      "learning_rate": 1.5212471300252163e-09,
      "lm_loss": 5.3561,
      "loss": 1.197,
      "step": 2198,
      "text_contrastive_loss": 0.7414,
      "train_positive_log_prob": -77.8124,
      "train_positive_token_accuracy": 0.0801,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.3132,
      "epoch": 4.963882618510158,
      "grad_norm": 11.874371528625488,
      "learning_rate": 1.347548513375707e-09,
      "lm_loss": 5.4174,
      "loss": 1.2188,
      "step": 2199,
      "text_contrastive_loss": 0.7279,
      "train_positive_log_prob": -78.0867,
      "train_positive_token_accuracy": 0.0748,
      "train_positive_token_prob": 0.0304
    },
    {
      "contrastive_loss": 0.3551,
      "epoch": 4.966139954853273,
      "grad_norm": 12.687294006347656,
      "learning_rate": 1.1843752531104368e-09,
      "lm_loss": 5.3605,
      "loss": 1.2945,
      "step": 2200,
      "text_contrastive_loss": 0.8069,
      "train_positive_log_prob": -81.1182,
      "train_positive_token_accuracy": 0.0787,
      "train_positive_token_prob": 0.0317
    },
    {
      "contrastive_loss": 0.425,
      "epoch": 4.968397291196388,
      "grad_norm": 15.35137939453125,
      "learning_rate": 1.0317276928134557e-09,
      "lm_loss": 5.3153,
      "loss": 1.3585,
      "step": 2201,
      "text_contrastive_loss": 0.804,
      "train_positive_log_prob": -78.1956,
      "train_positive_token_accuracy": 0.086,
      "train_positive_token_prob": 0.0323
    },
    {
      "contrastive_loss": 0.4068,
      "epoch": 4.970654627539504,
      "grad_norm": 13.613174438476562,
      "learning_rate": 8.896061539048762e-10,
      "lm_loss": 5.3597,
      "loss": 1.3727,
      "step": 2202,
      "text_contrastive_loss": 0.8598,
      "train_positive_log_prob": -78.7634,
      "train_positive_token_accuracy": 0.0811,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.2783,
      "epoch": 4.972911963882618,
      "grad_norm": 12.867480278015137,
      "learning_rate": 7.580109356419841e-10,
      "lm_loss": 5.4139,
      "loss": 1.1883,
      "step": 2203,
      "text_contrastive_loss": 0.7372,
      "train_positive_log_prob": -79.3215,
      "train_positive_token_accuracy": 0.0845,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.253,
      "epoch": 4.975169300225733,
      "grad_norm": 11.267635345458984,
      "learning_rate": 6.369423151164622e-10,
      "lm_loss": 5.4207,
      "loss": 1.186,
      "step": 2204,
      "text_contrastive_loss": 0.782,
      "train_positive_log_prob": -79.7659,
      "train_positive_token_accuracy": 0.0812,
      "train_positive_token_prob": 0.0305
    },
    {
      "contrastive_loss": 0.3076,
      "epoch": 4.977426636568849,
      "grad_norm": 12.308904647827148,
      "learning_rate": 5.264005472549461e-10,
      "lm_loss": 5.409,
      "loss": 1.2107,
      "step": 2205,
      "text_contrastive_loss": 0.7244,
      "train_positive_log_prob": -80.7564,
      "train_positive_token_accuracy": 0.078,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.238,
      "epoch": 4.979683972911964,
      "grad_norm": 12.098566055297852,
      "learning_rate": 4.2638586481846823e-10,
      "lm_loss": 5.44,
      "loss": 1.0985,
      "step": 2206,
      "text_contrastive_loss": 0.6332,
      "train_positive_log_prob": -79.9642,
      "train_positive_token_accuracy": 0.0776,
      "train_positive_token_prob": 0.0308
    },
    {
      "contrastive_loss": 0.3076,
      "epoch": 4.981941309255079,
      "grad_norm": 12.042253494262695,
      "learning_rate": 3.368984784024587e-10,
      "lm_loss": 5.3821,
      "loss": 1.208,
      "step": 2207,
      "text_contrastive_loss": 0.7244,
      "train_positive_log_prob": -79.5945,
      "train_positive_token_accuracy": 0.0729,
      "train_positive_token_prob": 0.0306
    },
    {
      "contrastive_loss": 0.3644,
      "epoch": 4.984198645598195,
      "grad_norm": 13.446281433105469,
      "learning_rate": 2.5793857643396924e-10,
      "lm_loss": 5.3932,
      "loss": 1.3001,
      "step": 2208,
      "text_contrastive_loss": 0.7927,
      "train_positive_log_prob": -79.8464,
      "train_positive_token_accuracy": 0.0718,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.2575,
      "epoch": 4.986455981941309,
      "grad_norm": 11.859148979187012,
      "learning_rate": 1.89506325175004e-10,
      "lm_loss": 5.3855,
      "loss": 1.1785,
      "step": 2209,
      "text_contrastive_loss": 0.7648,
      "train_positive_log_prob": -77.9696,
      "train_positive_token_accuracy": 0.0783,
      "train_positive_token_prob": 0.0324
    },
    {
      "contrastive_loss": 0.2738,
      "epoch": 4.988713318284424,
      "grad_norm": 11.753767967224121,
      "learning_rate": 1.316018687191889e-10,
      "lm_loss": 5.4317,
      "loss": 1.2152,
      "step": 2210,
      "text_contrastive_loss": 0.7965,
      "train_positive_log_prob": -81.3885,
      "train_positive_token_accuracy": 0.0795,
      "train_positive_token_prob": 0.0309
    },
    {
      "contrastive_loss": 0.3206,
      "epoch": 4.99097065462754,
      "grad_norm": 11.845379829406738,
      "learning_rate": 8.422532899121649e-11,
      "lm_loss": 5.4001,
      "loss": 1.2306,
      "step": 2211,
      "text_contrastive_loss": 0.74,
      "train_positive_log_prob": -78.0693,
      "train_positive_token_accuracy": 0.0792,
      "train_positive_token_prob": 0.0315
    },
    {
      "contrastive_loss": 0.327,
      "epoch": 4.993227990970655,
      "grad_norm": 13.591782569885254,
      "learning_rate": 4.737680575017667e-11,
      "lm_loss": 5.3766,
      "loss": 1.2574,
      "step": 2212,
      "text_contrastive_loss": 0.7853,
      "train_positive_log_prob": -78.3168,
      "train_positive_token_accuracy": 0.0885,
      "train_positive_token_prob": 0.0318
    },
    {
      "contrastive_loss": 0.3259,
      "epoch": 4.995485327313769,
      "grad_norm": 12.919146537780762,
      "learning_rate": 2.1056376585115723e-11,
      "lm_loss": 5.4108,
      "loss": 1.2339,
      "step": 2213,
      "text_contrastive_loss": 0.7337,
      "train_positive_log_prob": -79.6642,
      "train_positive_token_accuracy": 0.086,
      "train_positive_token_prob": 0.0327
    },
    {
      "contrastive_loss": 0.3853,
      "epoch": 4.997742663656885,
      "grad_norm": 14.124292373657227,
      "learning_rate": 5.2640969172568225e-12,
      "lm_loss": 5.4305,
      "loss": 1.389,
      "step": 2214,
      "text_contrastive_loss": 0.9214,
      "train_positive_log_prob": -77.827,
      "train_positive_token_accuracy": 0.0796,
      "train_positive_token_prob": 0.0316
    },
    {
      "contrastive_loss": 0.2102,
      "epoch": 5.0,
      "grad_norm": 13.013922691345215,
      "learning_rate": 0.0,
      "lm_loss": 5.4698,
      "loss": 1.0002,
      "step": 2215,
      "text_contrastive_loss": 0.486,
      "train_positive_log_prob": -82.2887,
      "train_positive_token_accuracy": 0.0793,
      "train_positive_token_prob": 0.0311
    },
    {
      "epoch": 5.0,
      "step": 2215,
      "total_flos": 1.487319842422784e+17,
      "train_loss": 1.4400290369718391,
      "train_runtime": 7589.3376,
      "train_samples_per_second": 74.636,
      "train_steps_per_second": 0.292
    }
  ],
  "logging_steps": 1,
  "max_steps": 2215,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.487319842422784e+17,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}