{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 100,
  "global_step": 4375,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 5.6899882705644975,
      "learning_rate": 4.5454545454545455e-06,
      "loss": 4.9489,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.313303192834888,
      "learning_rate": 9.090909090909091e-06,
      "loss": 5.0149,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.671634148022431,
      "learning_rate": 1.3636363636363635e-05,
      "loss": 4.704,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.83164039994204,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 5.2202,
      "step": 4
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.331876854370643,
      "learning_rate": 2.2727272727272726e-05,
      "loss": 4.6825,
      "step": 5
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.9782447915358237,
      "learning_rate": 2.727272727272727e-05,
      "loss": 4.4191,
      "step": 6
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.8447416541366186,
      "learning_rate": 3.1818181818181814e-05,
      "loss": 3.9869,
      "step": 7
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.1799015160513977,
      "learning_rate": 3.6363636363636364e-05,
      "loss": 3.7894,
      "step": 8
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3703769109341994,
      "learning_rate": 4.09090909090909e-05,
      "loss": 3.4733,
      "step": 9
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9303396464547136,
      "learning_rate": 4.545454545454545e-05,
      "loss": 3.2632,
      "step": 10
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1207409557996533,
      "learning_rate": 4.9999999999999996e-05,
      "loss": 3.1457,
      "step": 11
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9732831950233385,
      "learning_rate": 5.454545454545454e-05,
      "loss": 3.0448,
      "step": 12
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4457124204079697,
      "learning_rate": 5.909090909090908e-05,
      "loss": 2.9272,
      "step": 13
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0598932180566556,
      "learning_rate": 6.363636363636363e-05,
      "loss": 2.9642,
      "step": 14
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.107103680028823,
      "learning_rate": 6.818181818181817e-05,
      "loss": 3.0298,
      "step": 15
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3767381599969715,
      "learning_rate": 7.272727272727273e-05,
      "loss": 2.7237,
      "step": 16
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.117004905860884,
      "learning_rate": 7.727272727272726e-05,
      "loss": 2.9562,
      "step": 17
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.105622408222587,
      "learning_rate": 8.18181818181818e-05,
      "loss": 2.8269,
      "step": 18
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.082053097817895,
      "learning_rate": 8.636363636363636e-05,
      "loss": 2.7821,
      "step": 19
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9408580285208541,
      "learning_rate": 9.09090909090909e-05,
      "loss": 2.7491,
      "step": 20
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8190750865875447,
      "learning_rate": 9.545454545454545e-05,
      "loss": 2.6852,
      "step": 21
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8367821014312925,
      "learning_rate": 9.999999999999999e-05,
      "loss": 2.7615,
      "step": 22
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8405221570395635,
      "learning_rate": 0.00010454545454545455,
      "loss": 2.7083,
      "step": 23
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7218060245675572,
      "learning_rate": 0.00010909090909090908,
      "loss": 2.6305,
      "step": 24
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7530241549279189,
      "learning_rate": 0.00011363636363636362,
      "loss": 2.7175,
      "step": 25
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.69826707957787,
      "learning_rate": 0.00011818181818181817,
      "loss": 2.627,
      "step": 26
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7092844786400335,
      "learning_rate": 0.00012272727272727272,
      "loss": 2.573,
      "step": 27
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7308898894607077,
      "learning_rate": 0.00012727272727272725,
      "loss": 2.5391,
      "step": 28
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.83476706381957,
      "learning_rate": 0.0001318181818181818,
      "loss": 2.6457,
      "step": 29
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7316174354871663,
      "learning_rate": 0.00013636363636363634,
      "loss": 2.5442,
      "step": 30
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7729443438298704,
      "learning_rate": 0.0001409090909090909,
      "loss": 2.5913,
      "step": 31
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8265886407504855,
      "learning_rate": 0.00014545454545454546,
      "loss": 2.5899,
      "step": 32
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7849010418752537,
      "learning_rate": 0.00015,
      "loss": 2.3806,
      "step": 33
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.768706410561798,
      "learning_rate": 0.00015454545454545452,
      "loss": 2.4802,
      "step": 34
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.825827354040554,
      "learning_rate": 0.00015909090909090907,
      "loss": 2.5566,
      "step": 35
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7896766012089513,
      "learning_rate": 0.0001636363636363636,
      "loss": 2.474,
      "step": 36
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7923792808538245,
      "learning_rate": 0.00016818181818181816,
      "loss": 2.5262,
      "step": 37
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7689098715018539,
      "learning_rate": 0.00017272727272727272,
      "loss": 2.5691,
      "step": 38
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.767033618278646,
      "learning_rate": 0.00017727272727272728,
      "loss": 2.6176,
      "step": 39
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7899073964461848,
      "learning_rate": 0.0001818181818181818,
      "loss": 2.4692,
      "step": 40
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7878128380115905,
      "learning_rate": 0.00018636363636363634,
      "loss": 2.4306,
      "step": 41
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8628534337236539,
      "learning_rate": 0.0001909090909090909,
      "loss": 2.5159,
      "step": 42
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9159194978075866,
      "learning_rate": 0.00019545454545454543,
      "loss": 2.4547,
      "step": 43
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7425336506751342,
      "learning_rate": 0.00019999999999999998,
      "loss": 2.5254,
      "step": 44
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7918065706770165,
      "learning_rate": 0.0002045454545454545,
      "loss": 2.5238,
      "step": 45
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.93023321242069,
      "learning_rate": 0.0002090909090909091,
      "loss": 2.3934,
      "step": 46
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8399729995190708,
      "learning_rate": 0.00021363636363636363,
      "loss": 2.614,
      "step": 47
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7113366254982137,
      "learning_rate": 0.00021818181818181816,
      "loss": 2.6182,
      "step": 48
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7303674315076816,
      "learning_rate": 0.00022272727272727272,
      "loss": 2.3221,
      "step": 49
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7998959276662827,
      "learning_rate": 0.00022727272727272725,
      "loss": 2.4091,
      "step": 50
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.690092499300873,
      "learning_rate": 0.0002318181818181818,
      "loss": 2.4381,
      "step": 51
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7197381669679382,
      "learning_rate": 0.00023636363636363633,
      "loss": 2.3981,
      "step": 52
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7185287871537785,
      "learning_rate": 0.00024090909090909086,
      "loss": 2.3902,
      "step": 53
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7235518752495292,
      "learning_rate": 0.00024545454545454545,
      "loss": 2.4423,
      "step": 54
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7515473737137839,
      "learning_rate": 0.00025,
      "loss": 2.6013,
      "step": 55
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7910566072750824,
      "learning_rate": 0.0002545454545454545,
      "loss": 2.5372,
      "step": 56
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8130676760453753,
      "learning_rate": 0.00025909090909090907,
      "loss": 2.5049,
      "step": 57
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.662848917547693,
      "learning_rate": 0.0002636363636363636,
      "loss": 2.2621,
      "step": 58
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6759590128864548,
      "learning_rate": 0.0002681818181818181,
      "loss": 2.2966,
      "step": 59
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7912352849707737,
      "learning_rate": 0.0002727272727272727,
      "loss": 2.4878,
      "step": 60
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7232042345262207,
      "learning_rate": 0.00027727272727272724,
      "loss": 2.3753,
      "step": 61
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8730232315874268,
      "learning_rate": 0.0002818181818181818,
      "loss": 2.6525,
      "step": 62
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7376111939334709,
      "learning_rate": 0.00028636363636363636,
      "loss": 2.4162,
      "step": 63
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7655208092138928,
      "learning_rate": 0.0002909090909090909,
      "loss": 2.3876,
      "step": 64
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8566205692203729,
      "learning_rate": 0.0002954545454545454,
      "loss": 2.5353,
      "step": 65
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6917165870024561,
      "learning_rate": 0.0003,
      "loss": 2.416,
      "step": 66
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6849847639558945,
      "learning_rate": 0.00030454545454545453,
      "loss": 2.2864,
      "step": 67
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7429670439970971,
      "learning_rate": 0.00030909090909090903,
      "loss": 2.4441,
      "step": 68
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7350657256879124,
      "learning_rate": 0.0003136363636363636,
      "loss": 2.2988,
      "step": 69
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7440090830871278,
      "learning_rate": 0.00031818181818181815,
      "loss": 2.4653,
      "step": 70
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7421953307612521,
      "learning_rate": 0.00032272727272727265,
      "loss": 2.455,
      "step": 71
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8038207785921443,
      "learning_rate": 0.0003272727272727272,
      "loss": 2.5136,
      "step": 72
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7218637232393625,
      "learning_rate": 0.00033181818181818177,
      "loss": 2.4338,
      "step": 73
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6721445419819468,
      "learning_rate": 0.0003363636363636363,
      "loss": 2.3418,
      "step": 74
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7320452696928105,
      "learning_rate": 0.00034090909090909094,
      "loss": 2.2342,
      "step": 75
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7464232447318756,
      "learning_rate": 0.00034545454545454544,
      "loss": 2.4117,
      "step": 76
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.698605323531886,
      "learning_rate": 0.00035,
      "loss": 2.2924,
      "step": 77
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7436672683627857,
      "learning_rate": 0.00035454545454545455,
      "loss": 2.2388,
      "step": 78
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8574919811414031,
      "learning_rate": 0.00035909090909090906,
      "loss": 2.4663,
      "step": 79
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7245285718098012,
      "learning_rate": 0.0003636363636363636,
      "loss": 2.3387,
      "step": 80
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8435232682833903,
      "learning_rate": 0.00036818181818181817,
      "loss": 2.308,
      "step": 81
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8448947322536565,
      "learning_rate": 0.0003727272727272727,
      "loss": 2.3755,
      "step": 82
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7649424363813991,
      "learning_rate": 0.00037727272727272723,
      "loss": 2.3842,
      "step": 83
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9002529478560403,
      "learning_rate": 0.0003818181818181818,
      "loss": 2.3533,
      "step": 84
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7967338255202507,
      "learning_rate": 0.0003863636363636363,
      "loss": 2.3302,
      "step": 85
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8197712490677916,
      "learning_rate": 0.00039090909090909085,
      "loss": 2.4142,
      "step": 86
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7200667454850096,
      "learning_rate": 0.0003954545454545454,
      "loss": 2.3043,
      "step": 87
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6850455051407842,
      "learning_rate": 0.00039999999999999996,
      "loss": 2.4058,
      "step": 88
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8566059491406931,
      "learning_rate": 0.00040454545454545447,
      "loss": 2.4595,
      "step": 89
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7659359572413186,
      "learning_rate": 0.000409090909090909,
      "loss": 2.3925,
      "step": 90
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7499339845147437,
      "learning_rate": 0.00041363636363636364,
      "loss": 2.3421,
      "step": 91
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8045934207501154,
      "learning_rate": 0.0004181818181818182,
      "loss": 2.4736,
      "step": 92
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7926827791059288,
      "learning_rate": 0.0004227272727272727,
      "loss": 2.3923,
      "step": 93
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8539750535911281,
      "learning_rate": 0.00042727272727272726,
      "loss": 2.4064,
      "step": 94
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7812006103072849,
      "learning_rate": 0.0004318181818181818,
      "loss": 2.3544,
      "step": 95
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7508701708365426,
      "learning_rate": 0.0004363636363636363,
      "loss": 2.3625,
      "step": 96
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7882493317728478,
      "learning_rate": 0.0004409090909090909,
      "loss": 2.2912,
      "step": 97
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8640221239109133,
      "learning_rate": 0.00044545454545454543,
      "loss": 2.3642,
      "step": 98
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8562783193732348,
      "learning_rate": 0.00045,
      "loss": 2.3395,
      "step": 99
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8565402876896437,
      "learning_rate": 0.0004545454545454545,
      "loss": 2.3879,
      "step": 100
    },
    {
      "epoch": 0.11,
      "eval_blimp_filtered_avg": 0.7208955223880597,
      "eval_blimp_filtered_std": 0.004852166696436156,
      "step": 100
    },
    {
      "epoch": 0.11,
      "eval_blimp_supplement_avg": 0.8275862068965517,
      "eval_blimp_supplement_std": 0.01641366169189489,
      "step": 100
    },
    {
      "epoch": 0.11,
      "eval_vqa_filtered_avg": 0.62,
      "eval_vqa_filtered_std": 0.048783173121456316,
      "step": 100
    },
    {
      "epoch": 0.11,
      "eval_winoground_filtered_avg": 0.65,
      "eval_winoground_filtered_std": 0.0479372485441102,
      "step": 100
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.848722659509694,
      "learning_rate": 0.00045909090909090905,
      "loss": 2.4336,
      "step": 101
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7286363819256328,
      "learning_rate": 0.0004636363636363636,
      "loss": 2.3946,
      "step": 102
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7154719152168354,
      "learning_rate": 0.0004681818181818181,
      "loss": 2.2189,
      "step": 103
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8982880967041466,
      "learning_rate": 0.00047272727272727267,
      "loss": 2.5037,
      "step": 104
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8213734619436593,
      "learning_rate": 0.0004772727272727272,
      "loss": 2.3559,
      "step": 105
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8098967185804099,
      "learning_rate": 0.0004818181818181817,
      "loss": 2.3801,
      "step": 106
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7921099612603111,
      "learning_rate": 0.0004863636363636363,
      "loss": 2.5425,
      "step": 107
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7459522229159813,
      "learning_rate": 0.0004909090909090909,
      "loss": 2.3672,
      "step": 108
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8721611702885396,
      "learning_rate": 0.0004954545454545455,
      "loss": 2.3848,
      "step": 109
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8199871455498058,
      "learning_rate": 0.0005,
      "loss": 2.2948,
      "step": 110
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0244189637704555,
      "learning_rate": 0.0005045454545454546,
      "loss": 2.3687,
      "step": 111
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4451973525205113,
      "learning_rate": 0.000509090909090909,
      "loss": 2.4539,
      "step": 112
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.3106109637222665,
      "learning_rate": 0.0005136363636363636,
      "loss": 2.4334,
      "step": 113
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9303969656410228,
      "learning_rate": 0.0005181818181818181,
      "loss": 2.4525,
      "step": 114
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0560909211384562,
      "learning_rate": 0.0005227272727272727,
      "loss": 2.5307,
      "step": 115
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.52081284070849,
      "learning_rate": 0.0005272727272727272,
      "loss": 2.3618,
      "step": 116
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.689581310240795,
      "learning_rate": 0.0005318181818181818,
      "loss": 2.6617,
      "step": 117
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4789724723429967,
      "learning_rate": 0.0005363636363636363,
      "loss": 2.3608,
      "step": 118
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1005859781641265,
      "learning_rate": 0.0005409090909090908,
      "loss": 2.4549,
      "step": 119
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0860458560243311,
      "learning_rate": 0.0005454545454545454,
      "loss": 2.4579,
      "step": 120
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.274884810474263,
      "learning_rate": 0.0005499999999999999,
      "loss": 2.2827,
      "step": 121
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.445996022497982,
      "learning_rate": 0.0005545454545454545,
      "loss": 2.4756,
      "step": 122
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.571182718234623,
      "learning_rate": 0.000559090909090909,
      "loss": 2.5286,
      "step": 123
    },
    {
      "epoch": 0.14,
      "grad_norm": 11.756566387492118,
      "learning_rate": 0.0005636363636363636,
      "loss": 7.9232,
      "step": 124
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.407187647981287,
      "learning_rate": 0.0005681818181818182,
      "loss": 10.9687,
      "step": 125
    },
    {
      "epoch": 0.14,
      "grad_norm": 15.952245908057185,
      "learning_rate": 0.0005727272727272727,
      "loss": 7.9906,
      "step": 126
    },
    {
      "epoch": 0.15,
      "grad_norm": 41.70405282560889,
      "learning_rate": 0.0005772727272727273,
      "loss": 26.8464,
      "step": 127
    },
    {
      "epoch": 0.15,
      "grad_norm": 499.71355857812347,
      "learning_rate": 0.0005818181818181818,
      "loss": 18.4571,
      "step": 128
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.745040709099412,
      "learning_rate": 0.0005863636363636363,
      "loss": 7.4552,
      "step": 129
    },
    {
      "epoch": 0.15,
      "grad_norm": 24.194068208403817,
      "learning_rate": 0.0005909090909090908,
      "loss": 7.3717,
      "step": 130
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.874309551242373,
      "learning_rate": 0.0005954545454545454,
      "loss": 7.0362,
      "step": 131
    },
    {
      "epoch": 0.15,
      "grad_norm": 20.521117059610425,
      "learning_rate": 0.0006,
      "loss": 6.9502,
      "step": 132
    },
    {
      "epoch": 0.15,
      "grad_norm": 13.941044053590273,
      "learning_rate": 0.0005999999177672297,
      "loss": 6.6814,
      "step": 133
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.321148055548882,
      "learning_rate": 0.0005999996710689639,
      "loss": 6.4725,
      "step": 134
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.868322201286633,
      "learning_rate": 0.000599999259905338,
      "loss": 6.3314,
      "step": 135
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.508538090855016,
      "learning_rate": 0.0005999986842765774,
      "loss": 6.2596,
      "step": 136
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.668338435607754,
      "learning_rate": 0.0005999979441829973,
      "loss": 6.1353,
      "step": 137
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.5411530866911667,
      "learning_rate": 0.0005999970396250039,
      "loss": 6.2846,
      "step": 138
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.450068098080256,
      "learning_rate": 0.0005999959706030928,
      "loss": 6.0639,
      "step": 139
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.953357155459986,
      "learning_rate": 0.0005999947371178502,
      "loss": 6.0436,
      "step": 140
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.574898327668541,
      "learning_rate": 0.0005999933391699523,
      "loss": 6.1235,
      "step": 141
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.2293330805672427,
      "learning_rate": 0.0005999917767601654,
      "loss": 5.902,
      "step": 142
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.864875802496051,
      "learning_rate": 0.0005999900498893461,
      "loss": 5.934,
      "step": 143
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1441009762203374,
      "learning_rate": 0.0005999881585584412,
      "loss": 5.6018,
      "step": 144
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.4713219978500964,
      "learning_rate": 0.0005999861027684873,
      "loss": 5.8513,
      "step": 145
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.011490708621217,
      "learning_rate": 0.0005999838825206117,
      "loss": 5.7966,
      "step": 146
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.296766106768993,
      "learning_rate": 0.0005999814978160314,
      "loss": 5.7661,
      "step": 147
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.3156093211436315,
      "learning_rate": 0.0005999789486560538,
      "loss": 5.644,
      "step": 148
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.90596896447174,
      "learning_rate": 0.0005999762350420765,
      "loss": 5.6126,
      "step": 149
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.2025734276036415,
      "learning_rate": 0.000599973356975587,
      "loss": 5.6422,
      "step": 150
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.465258341112292,
      "learning_rate": 0.0005999703144581631,
      "loss": 5.694,
      "step": 151
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.837301787659427,
      "learning_rate": 0.0005999671074914729,
      "loss": 5.8244,
      "step": 152
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.565991330126649,
      "learning_rate": 0.0005999637360772745,
      "loss": 5.7922,
      "step": 153
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.999233509442854,
      "learning_rate": 0.000599960200217416,
      "loss": 5.5981,
      "step": 154
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.681485817945217,
      "learning_rate": 0.000599956499913836,
      "loss": 5.7056,
      "step": 155
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.716396656739732,
      "learning_rate": 0.000599952635168563,
      "loss": 5.6408,
      "step": 156
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.1723848744523653,
      "learning_rate": 0.0005999486059837158,
      "loss": 5.5681,
      "step": 157
    },
    {
      "epoch": 0.18,
      "grad_norm": 15.312953030922642,
      "learning_rate": 0.0005999444123615031,
      "loss": 5.6187,
      "step": 158
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.820601369170831,
      "learning_rate": 0.000599940054304224,
      "loss": 5.6417,
      "step": 159
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.755611579168663,
      "learning_rate": 0.0005999355318142679,
      "loss": 5.5611,
      "step": 160
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.65680311577697,
      "learning_rate": 0.0005999308448941138,
      "loss": 5.7147,
      "step": 161
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.732237027191075,
      "learning_rate": 0.0005999259935463312,
      "loss": 5.6456,
      "step": 162
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.022805262879111,
      "learning_rate": 0.00059992097777358,
      "loss": 5.6301,
      "step": 163
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.197891663547958,
      "learning_rate": 0.0005999157975786095,
      "loss": 5.3718,
      "step": 164
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.768254237746249,
      "learning_rate": 0.0005999104529642598,
      "loss": 5.4128,
      "step": 165
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.058083719937919,
      "learning_rate": 0.000599904943933461,
      "loss": 5.6592,
      "step": 166
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.92968975403688,
      "learning_rate": 0.0005998992704892331,
      "loss": 5.543,
      "step": 167
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.14093774557678,
      "learning_rate": 0.0005998934326346864,
      "loss": 5.5585,
      "step": 168
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.531956565793252,
      "learning_rate": 0.0005998874303730214,
      "loss": 5.578,
      "step": 169
    },
    {
      "epoch": 0.19,
      "grad_norm": 13.482819349209533,
      "learning_rate": 0.0005998812637075286,
      "loss": 5.6677,
      "step": 170
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.8332061660777974,
      "learning_rate": 0.0005998749326415888,
      "loss": 5.65,
      "step": 171
    },
    {
      "epoch": 0.2,
      "grad_norm": 11.173565402091388,
      "learning_rate": 0.0005998684371786726,
      "loss": 5.6192,
      "step": 172
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.294773442321036,
      "learning_rate": 0.000599861777322341,
      "loss": 5.5706,
      "step": 173
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.910379381554463,
      "learning_rate": 0.0005998549530762451,
      "loss": 5.577,
      "step": 174
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.332440046920278,
      "learning_rate": 0.0005998479644441262,
      "loss": 5.4371,
      "step": 175
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.3836159563843675,
      "learning_rate": 0.0005998408114298154,
      "loss": 5.6367,
      "step": 176
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.935743283041769,
      "learning_rate": 0.0005998334940372341,
      "loss": 5.6484,
      "step": 177
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.620663695889674,
      "learning_rate": 0.000599826012270394,
      "loss": 5.6219,
      "step": 178
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.683170166497655,
      "learning_rate": 0.0005998183661333966,
      "loss": 5.4882,
      "step": 179
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.464593588514576,
      "learning_rate": 0.0005998105556304337,
      "loss": 5.4765,
      "step": 180
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.591596863133778,
      "learning_rate": 0.0005998025807657872,
      "loss": 5.6047,
      "step": 181
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.205871708755827,
      "learning_rate": 0.000599794441543829,
      "loss": 5.5058,
      "step": 182
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.775968599966468,
      "learning_rate": 0.0005997861379690212,
      "loss": 5.5412,
      "step": 183
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.384345600715913,
      "learning_rate": 0.0005997776700459161,
      "loss": 5.5354,
      "step": 184
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.202020516512133,
      "learning_rate": 0.0005997690377791558,
      "loss": 5.3729,
      "step": 185
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.0705084062807435,
      "learning_rate": 0.0005997602411734726,
      "loss": 5.5032,
      "step": 186
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.3294094230501985,
      "learning_rate": 0.0005997512802336893,
      "loss": 5.4449,
      "step": 187
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.4692332837723323,
      "learning_rate": 0.0005997421549647179,
      "loss": 5.5083,
      "step": 188
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.7661575504906524,
      "learning_rate": 0.0005997328653715617,
      "loss": 5.4265,
      "step": 189
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.5537864998369266,
      "learning_rate": 0.000599723411459313,
      "loss": 5.5259,
      "step": 190
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.208257372185947,
      "learning_rate": 0.0005997137932331547,
      "loss": 5.4558,
      "step": 191
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.559454704818237,
      "learning_rate": 0.0005997040106983597,
      "loss": 5.5517,
      "step": 192
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.069151643297439,
      "learning_rate": 0.000599694063860291,
      "loss": 5.4612,
      "step": 193
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.081650485533267,
      "learning_rate": 0.0005996839527244015,
      "loss": 5.5635,
      "step": 194
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1955658961664455,
      "learning_rate": 0.0005996736772962345,
      "loss": 5.4068,
      "step": 195
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.769961288928332,
      "learning_rate": 0.000599663237581423,
      "loss": 5.5319,
      "step": 196
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9670928537268961,
      "learning_rate": 0.0005996526335856905,
      "loss": 5.4161,
      "step": 197
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9091224818951296,
      "learning_rate": 0.00059964186531485,
      "loss": 5.5261,
      "step": 198
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.431061434219896,
      "learning_rate": 0.0005996309327748052,
      "loss": 5.6589,
      "step": 199
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.9165221110391326,
      "learning_rate": 0.0005996198359715492,
      "loss": 5.492,
      "step": 200
    },
    {
      "epoch": 0.23,
      "eval_blimp_filtered_avg": 0.5583582089552239,
      "eval_blimp_filtered_std": 0.005392968902149062,
      "step": 200
    },
    {
      "epoch": 0.23,
      "eval_blimp_supplement_avg": 0.5,
      "eval_blimp_supplement_std": 0.022937146380016354,
      "step": 200
    },
    {
      "epoch": 0.23,
      "eval_vqa_filtered_avg": 0.15,
      "eval_vqa_filtered_std": 0.035887028128263734,
      "step": 200
    },
    {
      "epoch": 0.23,
      "eval_winoground_filtered_avg": 0.73,
      "eval_winoground_filtered_std": 0.044619604333847394,
      "step": 200
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.167737556078773,
      "learning_rate": 0.0005996085749111656,
      "loss": 5.5597,
      "step": 201
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.029631763384967,
      "learning_rate": 0.000599597149599828,
      "loss": 5.3916,
      "step": 202
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7214113929516655,
      "learning_rate": 0.0005995855600437998,
      "loss": 5.3554,
      "step": 203
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0998878694297851,
      "learning_rate": 0.0005995738062494348,
      "loss": 5.4597,
      "step": 204
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1019114322716383,
      "learning_rate": 0.0005995618882231765,
      "loss": 5.5064,
      "step": 205
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1189164881363165,
      "learning_rate": 0.0005995498059715586,
      "loss": 5.4009,
      "step": 206
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7410137110895191,
      "learning_rate": 0.0005995375595012048,
      "loss": 5.446,
      "step": 207
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9707199130395817,
      "learning_rate": 0.0005995251488188291,
      "loss": 5.4441,
      "step": 208
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.065245420986001,
      "learning_rate": 0.0005995125739312349,
      "loss": 5.2863,
      "step": 209
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1524204624051906,
      "learning_rate": 0.0005994998348453161,
      "loss": 5.4816,
      "step": 210
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0287862797008998,
      "learning_rate": 0.0005994869315680566,
      "loss": 5.4634,
      "step": 211
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.512372637512798,
      "learning_rate": 0.0005994738641065301,
      "loss": 5.564,
      "step": 212
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0995126895671135,
      "learning_rate": 0.0005994606324679005,
      "loss": 5.5937,
      "step": 213
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.7358722798358723,
      "learning_rate": 0.0005994472366594218,
      "loss": 5.4026,
      "step": 214
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4878559772883229,
      "learning_rate": 0.0005994336766884374,
      "loss": 5.5863,
      "step": 215
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0225436896988405,
      "learning_rate": 0.0005994199525623815,
      "loss": 5.3952,
      "step": 216
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.692949224495956,
      "learning_rate": 0.0005994060642887777,
      "loss": 5.4499,
      "step": 217
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6579508153083156,
      "learning_rate": 0.00059939201187524,
      "loss": 5.4049,
      "step": 218
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8678506895141744,
      "learning_rate": 0.000599377795329472,
      "loss": 5.4027,
      "step": 219
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4765692135287831,
      "learning_rate": 0.0005993634146592675,
      "loss": 5.318,
      "step": 220
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.9305110169647164,
      "learning_rate": 0.0005993488698725104,
      "loss": 5.5305,
      "step": 221
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6243146104972896,
      "learning_rate": 0.0005993341609771743,
      "loss": 5.3441,
      "step": 222
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2365997418732557,
      "learning_rate": 0.0005993192879813229,
      "loss": 5.6214,
      "step": 223
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.727127566853998,
      "learning_rate": 0.0005993042508931098,
      "loss": 5.4702,
      "step": 224
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6263868670636483,
      "learning_rate": 0.0005992890497207787,
      "loss": 5.5085,
      "step": 225
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1798141656432553,
      "learning_rate": 0.0005992736844726632,
      "loss": 5.5564,
      "step": 226
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9811010293115763,
      "learning_rate": 0.0005992581551571869,
      "loss": 5.3057,
      "step": 227
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2110016657394074,
      "learning_rate": 0.0005992424617828629,
      "loss": 5.3765,
      "step": 228
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9457725713764498,
      "learning_rate": 0.0005992266043582949,
      "loss": 5.3678,
      "step": 229
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7598753147142062,
      "learning_rate": 0.0005992105828921759,
      "loss": 5.3839,
      "step": 230
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9859802460704546,
      "learning_rate": 0.0005991943973932898,
      "loss": 5.5403,
      "step": 231
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5094405952936216,
      "learning_rate": 0.000599178047870509,
      "loss": 5.5175,
      "step": 232
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.782137157365417,
      "learning_rate": 0.0005991615343327973,
      "loss": 5.3597,
      "step": 233
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.424200586465043,
      "learning_rate": 0.0005991448567892072,
      "loss": 5.4816,
      "step": 234
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.8463539904688173,
      "learning_rate": 0.000599128015248882,
      "loss": 5.5235,
      "step": 235
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.15458706387072,
      "learning_rate": 0.0005991110097210545,
      "loss": 5.3415,
      "step": 236
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.295637360703834,
      "learning_rate": 0.0005990938402150473,
      "loss": 5.4644,
      "step": 237
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.4095194713035415,
      "learning_rate": 0.0005990765067402729,
      "loss": 5.5004,
      "step": 238
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.445918564956278,
      "learning_rate": 0.0005990590093062343,
      "loss": 5.6367,
      "step": 239
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8186308056122225,
      "learning_rate": 0.0005990413479225234,
      "loss": 5.4107,
      "step": 240
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0395943039589626,
      "learning_rate": 0.0005990235225988229,
      "loss": 5.6396,
      "step": 241
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.8121187859046297,
      "learning_rate": 0.0005990055333449047,
      "loss": 5.4573,
      "step": 242
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.947109922528538,
      "learning_rate": 0.0005989873801706311,
      "loss": 5.343,
      "step": 243
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.624504859892276,
      "learning_rate": 0.0005989690630859538,
      "loss": 5.4623,
      "step": 244
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.045752009774613,
      "learning_rate": 0.0005989505821009146,
      "loss": 5.2571,
      "step": 245
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5519833810335941,
      "learning_rate": 0.000598931937225645,
      "loss": 5.4051,
      "step": 246
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0673444247521937,
      "learning_rate": 0.0005989131284703668,
      "loss": 5.5218,
      "step": 247
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5078446622475545,
      "learning_rate": 0.000598894155845391,
      "loss": 5.4369,
      "step": 248
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.2762063605381666,
      "learning_rate": 0.0005988750193611188,
      "loss": 5.5545,
      "step": 249
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5359646005552658,
      "learning_rate": 0.0005988557190280415,
      "loss": 5.3746,
      "step": 250
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.216145337527657,
      "learning_rate": 0.0005988362548567394,
      "loss": 5.3403,
      "step": 251
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2244680077583707,
      "learning_rate": 0.0005988166268578834,
      "loss": 5.4123,
      "step": 252
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8514135451379858,
      "learning_rate": 0.0005987968350422339,
      "loss": 5.4508,
      "step": 253
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.06517833291948,
      "learning_rate": 0.000598776879420641,
      "loss": 5.6311,
      "step": 254
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.125027322594769,
      "learning_rate": 0.000598756760004045,
      "loss": 5.4056,
      "step": 255
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.120762877602598,
      "learning_rate": 0.0005987364768034756,
      "loss": 5.2515,
      "step": 256
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.0861484917793174,
      "learning_rate": 0.0005987160298300523,
      "loss": 5.4444,
      "step": 257
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.2344190471659515,
      "learning_rate": 0.0005986954190949846,
      "loss": 5.3986,
      "step": 258
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3498810977093658,
      "learning_rate": 0.0005986746446095719,
      "loss": 5.4916,
      "step": 259
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.9524087772184093,
      "learning_rate": 0.0005986537063852028,
      "loss": 5.3474,
      "step": 260
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5922022568819214,
      "learning_rate": 0.0005986326044333563,
      "loss": 5.4491,
      "step": 261
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.2020390053423644,
      "learning_rate": 0.0005986113387656007,
      "loss": 5.5313,
      "step": 262
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.693801444439002,
      "learning_rate": 0.0005985899093935943,
      "loss": 5.4478,
      "step": 263
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.8898661885810233,
      "learning_rate": 0.0005985683163290851,
      "loss": 5.579,
      "step": 264
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.5187766945426886,
      "learning_rate": 0.0005985465595839108,
      "loss": 5.3957,
      "step": 265
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9591204886163087,
      "learning_rate": 0.0005985246391699988,
      "loss": 5.3876,
      "step": 266
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.4113356157199224,
      "learning_rate": 0.0005985025550993664,
      "loss": 5.3935,
      "step": 267
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.6743713835932907,
      "learning_rate": 0.0005984803073841203,
      "loss": 5.4484,
      "step": 268
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.7460614224183266,
      "learning_rate": 0.0005984578960364573,
      "loss": 5.3542,
      "step": 269
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.2919084073262757,
      "learning_rate": 0.0005984353210686637,
      "loss": 5.4367,
      "step": 270
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.3335021215126193,
      "learning_rate": 0.0005984125824931154,
      "loss": 5.3813,
      "step": 271
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2888274358816194,
      "learning_rate": 0.0005983896803222781,
      "loss": 5.2022,
      "step": 272
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8398910237675967,
      "learning_rate": 0.0005983666145687073,
      "loss": 5.2159,
      "step": 273
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.76819729423109,
      "learning_rate": 0.000598343385245048,
      "loss": 5.344,
      "step": 274
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.275794378535884,
      "learning_rate": 0.000598319992364035,
      "loss": 5.2194,
      "step": 275
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.1574329818794626,
      "learning_rate": 0.0005982964359384927,
      "loss": 5.2684,
      "step": 276
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.387885482413438,
      "learning_rate": 0.0005982727159813352,
      "loss": 5.1345,
      "step": 277
    },
    {
      "epoch": 0.32,
      "grad_norm": 15.56019981927672,
      "learning_rate": 0.000598248832505566,
      "loss": 5.6096,
      "step": 278
    },
    {
      "epoch": 0.32,
      "grad_norm": 10.360953035424545,
      "learning_rate": 0.0005982247855242786,
      "loss": 5.8703,
      "step": 279
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.453270405337348,
      "learning_rate": 0.0005982005750506562,
      "loss": 6.0141,
      "step": 280
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.755505970047711,
      "learning_rate": 0.0005981762010979712,
      "loss": 5.7943,
      "step": 281
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.043611977119204,
      "learning_rate": 0.0005981516636795859,
      "loss": 5.7241,
      "step": 282
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.285257344656154,
      "learning_rate": 0.0005981269628089521,
      "loss": 5.6993,
      "step": 283
    },
    {
      "epoch": 0.32,
      "grad_norm": 11.32621885018214,
      "learning_rate": 0.0005981020984996114,
      "loss": 5.6036,
      "step": 284
    },
    {
      "epoch": 0.33,
      "grad_norm": 14.74265893660116,
      "learning_rate": 0.000598077070765195,
      "loss": 5.5839,
      "step": 285
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.291911457409011,
      "learning_rate": 0.0005980518796194231,
      "loss": 5.6241,
      "step": 286
    },
    {
      "epoch": 0.33,
      "grad_norm": 12.990504838977,
      "learning_rate": 0.0005980265250761064,
      "loss": 5.5118,
      "step": 287
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.397553973530215,
      "learning_rate": 0.0005980010071491445,
      "loss": 5.6128,
      "step": 288
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.489739675349318,
      "learning_rate": 0.0005979753258525269,
      "loss": 5.4221,
      "step": 289
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.087930876881718,
      "learning_rate": 0.0005979494812003325,
      "loss": 5.5251,
      "step": 290
    },
    {
      "epoch": 0.33,
      "grad_norm": 10.346962793473262,
      "learning_rate": 0.0005979234732067298,
      "loss": 5.6508,
      "step": 291
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1026848414481896,
      "learning_rate": 0.000597897301885977,
      "loss": 5.418,
      "step": 292
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.594176422850023,
      "learning_rate": 0.0005978709672524215,
      "loss": 5.452,
      "step": 293
    },
    {
      "epoch": 0.34,
      "grad_norm": 9.766287935711846,
      "learning_rate": 0.0005978444693205006,
      "loss": 5.4974,
      "step": 294
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.7445810793140284,
      "learning_rate": 0.000597817808104741,
      "loss": 5.4102,
      "step": 295
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.072418856913142,
      "learning_rate": 0.0005977909836197585,
      "loss": 5.4505,
      "step": 296
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.517669457012214,
      "learning_rate": 0.0005977639958802592,
      "loss": 5.4652,
      "step": 297
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.117130805350819,
      "learning_rate": 0.0005977368449010381,
      "loss": 5.5226,
      "step": 298
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8060157053815324,
      "learning_rate": 0.0005977095306969799,
      "loss": 5.3902,
      "step": 299
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.39666403418197,
      "learning_rate": 0.0005976820532830587,
      "loss": 5.3725,
      "step": 300
    },
    {
      "epoch": 0.34,
      "eval_blimp_filtered_avg": 0.5594029850746268,
      "eval_blimp_filtered_std": 0.005466712800431388,
      "step": 300
    },
    {
      "epoch": 0.34,
      "eval_blimp_supplement_avg": 0.5258620689655172,
      "eval_blimp_supplement_std": 0.02302704073531634,
      "step": 300
    },
    {
      "epoch": 0.34,
      "eval_vqa_filtered_avg": 0.14,
      "eval_vqa_filtered_std": 0.03487350880197772,
      "step": 300
    },
    {
      "epoch": 0.34,
      "eval_winoground_filtered_avg": 0.57,
      "eval_winoground_filtered_std": 0.04975698519562428,
      "step": 300
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.1158990392607375,
      "learning_rate": 0.0005976544126743383,
      "loss": 5.3367,
      "step": 301
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0154449703549338,
      "learning_rate": 0.0005976266088859715,
      "loss": 5.3467,
      "step": 302
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.686812167362636,
      "learning_rate": 0.0005975986419332011,
      "loss": 5.3398,
      "step": 303
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.509557450743787,
      "learning_rate": 0.0005975705118313589,
      "loss": 5.4016,
      "step": 304
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.4512671460471145,
      "learning_rate": 0.0005975422185958666,
      "loss": 5.3296,
      "step": 305
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9090727371809768,
      "learning_rate": 0.0005975137622422349,
      "loss": 5.2443,
      "step": 306
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.8644868008469704,
      "learning_rate": 0.0005974851427860639,
      "loss": 5.1167,
      "step": 307
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0056665540238834,
      "learning_rate": 0.0005974563602430437,
      "loss": 5.1504,
      "step": 308
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5508067504920053,
      "learning_rate": 0.0005974274146289533,
      "loss": 5.1995,
      "step": 309
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9266929387011655,
      "learning_rate": 0.0005973983059596612,
      "loss": 5.3255,
      "step": 310
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0464990054002434,
      "learning_rate": 0.0005973690342511252,
      "loss": 5.217,
      "step": 311
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5086874049787238,
      "learning_rate": 0.0005973395995193926,
      "loss": 5.2963,
      "step": 312
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.236425026117851,
      "learning_rate": 0.0005973100017806003,
      "loss": 5.3029,
      "step": 313
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8345242276237188,
      "learning_rate": 0.000597280241050974,
      "loss": 5.3068,
      "step": 314
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.387131216317475,
      "learning_rate": 0.0005972503173468295,
      "loss": 5.2659,
      "step": 315
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.9929778154271,
      "learning_rate": 0.0005972202306845711,
      "loss": 5.3761,
      "step": 316
    },
    {
      "epoch": 0.36,
      "grad_norm": 201.58557269872338,
      "learning_rate": 0.0005971899810806932,
      "loss": 5.3516,
      "step": 317
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5348126892692233,
      "learning_rate": 0.0005971595685517789,
      "loss": 5.1978,
      "step": 318
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.33872219585695,
      "learning_rate": 0.000597128993114501,
      "loss": 5.1429,
      "step": 319
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.042865971234179,
      "learning_rate": 0.0005970982547856215,
      "loss": 5.2872,
      "step": 320
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.8010061051085016,
      "learning_rate": 0.000597067353581992,
      "loss": 5.4718,
      "step": 321
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.1546143762495875,
      "learning_rate": 0.0005970362895205528,
      "loss": 5.3273,
      "step": 322
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.68635883224407,
      "learning_rate": 0.0005970050626183339,
      "loss": 5.4596,
      "step": 323
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.9089188336152296,
      "learning_rate": 0.0005969736728924544,
      "loss": 5.2995,
      "step": 324
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.381816442255853,
      "learning_rate": 0.0005969421203601228,
      "loss": 5.314,
      "step": 325
    },
    {
      "epoch": 0.37,
      "grad_norm": 12.997403626989605,
      "learning_rate": 0.0005969104050386367,
      "loss": 5.5858,
      "step": 326
    },
    {
      "epoch": 0.37,
      "grad_norm": 37.83285263097063,
      "learning_rate": 0.0005968785269453832,
      "loss": 5.4655,
      "step": 327
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.2981382083138504,
      "learning_rate": 0.0005968464860978384,
      "loss": 5.4473,
      "step": 328
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0912350924066583,
      "learning_rate": 0.0005968142825135674,
      "loss": 5.5116,
      "step": 329
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.909520922703617,
      "learning_rate": 0.0005967819162102252,
      "loss": 5.4709,
      "step": 330
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.664963702894739,
      "learning_rate": 0.0005967493872055553,
      "loss": 5.5243,
      "step": 331
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.4398470866271924,
      "learning_rate": 0.0005967166955173909,
      "loss": 5.3821,
      "step": 332
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.397848683946831,
      "learning_rate": 0.0005966838411636542,
      "loss": 5.5714,
      "step": 333
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.0323067559721197,
      "learning_rate": 0.0005966508241623563,
      "loss": 5.3786,
      "step": 334
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.075943496760952,
      "learning_rate": 0.0005966176445315981,
      "loss": 5.5032,
      "step": 335
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.031354517772589,
      "learning_rate": 0.0005965843022895691,
      "loss": 5.269,
      "step": 336
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.868262455582384,
      "learning_rate": 0.000596550797454548,
      "loss": 5.5177,
      "step": 337
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.605041610953078,
      "learning_rate": 0.000596517130044903,
      "loss": 5.3174,
      "step": 338
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8526656565631017,
      "learning_rate": 0.000596483300079091,
      "loss": 5.3418,
      "step": 339
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6520173949425077,
      "learning_rate": 0.0005964493075756584,
      "loss": 5.3879,
      "step": 340
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.141926894948423,
      "learning_rate": 0.0005964151525532403,
      "loss": 5.2671,
      "step": 341
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.404688978602847,
      "learning_rate": 0.0005963808350305615,
      "loss": 5.3187,
      "step": 342
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.6927056224349126,
      "learning_rate": 0.000596346355026435,
      "loss": 5.3977,
      "step": 343
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.6423084575806937,
      "learning_rate": 0.0005963117125597638,
      "loss": 5.4097,
      "step": 344
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1573791089115377,
      "learning_rate": 0.0005962769076495392,
      "loss": 5.2198,
      "step": 345
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.7931075806725527,
      "learning_rate": 0.0005962419403148421,
      "loss": 5.0811,
      "step": 346
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.4427168742098697,
      "learning_rate": 0.0005962068105748422,
      "loss": 5.1713,
      "step": 347
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2507327640090309,
      "learning_rate": 0.0005961715184487982,
      "loss": 5.1729,
      "step": 348
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2707946253594813,
      "learning_rate": 0.000596136063956058,
      "loss": 5.2439,
      "step": 349
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.295944116363648,
      "learning_rate": 0.0005961004471160584,
      "loss": 5.1512,
      "step": 350
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6712103533318206,
      "learning_rate": 0.0005960646679483252,
      "loss": 5.1529,
      "step": 351
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4711303228066783,
      "learning_rate": 0.0005960287264724732,
      "loss": 5.122,
      "step": 352
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.074882693360815,
      "learning_rate": 0.000595992622708206,
      "loss": 5.1299,
      "step": 353
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8127248924936334,
      "learning_rate": 0.0005959563566753166,
      "loss": 5.3082,
      "step": 354
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.374662595270538,
      "learning_rate": 0.0005959199283936866,
      "loss": 5.0728,
      "step": 355
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8806906988133945,
      "learning_rate": 0.0005958833378832867,
      "loss": 5.197,
      "step": 356
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0376267278197344,
      "learning_rate": 0.0005958465851641765,
      "loss": 5.1509,
      "step": 357
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8749449887306634,
      "learning_rate": 0.0005958096702565044,
      "loss": 5.1594,
      "step": 358
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8816726184283791,
      "learning_rate": 0.0005957725931805079,
      "loss": 5.1314,
      "step": 359
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2629463657867646,
      "learning_rate": 0.0005957353539565134,
      "loss": 5.0803,
      "step": 360
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7763445158930895,
      "learning_rate": 0.0005956979526049362,
      "loss": 5.2906,
      "step": 361
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9197127496341845,
      "learning_rate": 0.0005956603891462801,
      "loss": 5.0621,
      "step": 362
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1624221182677006,
      "learning_rate": 0.0005956226636011384,
      "loss": 5.142,
      "step": 363
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6854299500361504,
      "learning_rate": 0.0005955847759901928,
      "loss": 5.188,
      "step": 364
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7734036187759444,
      "learning_rate": 0.0005955467263342141,
      "loss": 5.0008,
      "step": 365
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0137525842490933,
      "learning_rate": 0.0005955085146540616,
      "loss": 5.104,
      "step": 366
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9554878057956477,
      "learning_rate": 0.0005954701409706838,
      "loss": 5.1863,
      "step": 367
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0250295336851294,
      "learning_rate": 0.0005954316053051179,
      "loss": 4.9032,
      "step": 368
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7310318298965819,
      "learning_rate": 0.0005953929076784897,
      "loss": 5.1252,
      "step": 369
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1490493756860594,
      "learning_rate": 0.0005953540481120141,
      "loss": 5.2095,
      "step": 370
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8557028871037503,
      "learning_rate": 0.0005953150266269947,
      "loss": 4.8182,
      "step": 371
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0434349376178245,
      "learning_rate": 0.0005952758432448236,
      "loss": 4.9799,
      "step": 372
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8955963805324878,
      "learning_rate": 0.0005952364979869819,
      "loss": 5.0193,
      "step": 373
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2224143509311085,
      "learning_rate": 0.0005951969908750397,
      "loss": 5.0098,
      "step": 374
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.259509324789556,
      "learning_rate": 0.000595157321930655,
      "loss": 5.0709,
      "step": 375
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.1567974766589004,
      "learning_rate": 0.0005951174911755754,
      "loss": 4.9933,
      "step": 376
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.2152692642440783,
      "learning_rate": 0.0005950774986316368,
      "loss": 5.1693,
      "step": 377
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8144612613972055,
      "learning_rate": 0.0005950373443207638,
      "loss": 4.9564,
      "step": 378
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3709706712775995,
      "learning_rate": 0.0005949970282649698,
      "loss": 4.8974,
      "step": 379
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9006209599805071,
      "learning_rate": 0.0005949565504863567,
      "loss": 5.1648,
      "step": 380
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6336451675308545,
      "learning_rate": 0.0005949159110071152,
      "loss": 5.1214,
      "step": 381
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6777576063871181,
      "learning_rate": 0.0005948751098495248,
      "loss": 4.9723,
      "step": 382
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.96408538708768,
      "learning_rate": 0.0005948341470359532,
      "loss": 5.0241,
      "step": 383
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0598865299320908,
      "learning_rate": 0.0005947930225888571,
      "loss": 4.9936,
      "step": 384
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3142792056542076,
      "learning_rate": 0.0005947517365307816,
      "loss": 5.0602,
      "step": 385
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9271718939336193,
      "learning_rate": 0.0005947102888843606,
      "loss": 4.9825,
      "step": 386
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1892966742310114,
      "learning_rate": 0.0005946686796723164,
      "loss": 5.1048,
      "step": 387
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9909909107414631,
      "learning_rate": 0.0005946269089174598,
      "loss": 4.9013,
      "step": 388
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7823938314555703,
      "learning_rate": 0.0005945849766426904,
      "loss": 4.8844,
      "step": 389
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2634648640303658,
      "learning_rate": 0.0005945428828709965,
      "loss": 5.0096,
      "step": 390
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7090472799200376,
      "learning_rate": 0.0005945006276254543,
      "loss": 5.1439,
      "step": 391
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8750037047061195,
      "learning_rate": 0.0005944582109292291,
      "loss": 5.0526,
      "step": 392
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5321473379163821,
      "learning_rate": 0.0005944156328055745,
      "loss": 4.9529,
      "step": 393
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1735041582282233,
      "learning_rate": 0.0005943728932778326,
      "loss": 4.9819,
      "step": 394
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1340206655506173,
      "learning_rate": 0.0005943299923694339,
      "loss": 5.0917,
      "step": 395
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.690818830374794,
      "learning_rate": 0.0005942869301038977,
      "loss": 5.1355,
      "step": 396
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9200426326800019,
      "learning_rate": 0.0005942437065048314,
      "loss": 5.0101,
      "step": 397
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9112682533831391,
      "learning_rate": 0.0005942003215959309,
      "loss": 5.1829,
      "step": 398
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2850557209928255,
      "learning_rate": 0.0005941567754009807,
      "loss": 5.2344,
      "step": 399
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5161343025925256,
      "learning_rate": 0.0005941130679438535,
      "loss": 5.1105,
      "step": 400
    },
    {
      "epoch": 0.46,
      "eval_blimp_filtered_avg": 0.4855223880597015,
      "eval_blimp_filtered_std": 0.005645894409064591,
      "step": 400
    },
    {
      "epoch": 0.46,
      "eval_blimp_supplement_avg": 0.4849137931034483,
      "eval_blimp_supplement_std": 0.022726419008559078,
      "step": 400
    },
    {
      "epoch": 0.46,
      "eval_vqa_filtered_avg": 0.34,
      "eval_vqa_filtered_std": 0.04760952285695236,
      "step": 400
    },
    {
      "epoch": 0.46,
      "eval_winoground_filtered_avg": 0.5,
      "eval_winoground_filtered_std": 0.050251890762960605,
      "step": 400
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5745987301659912,
      "learning_rate": 0.0005940691992485107,
      "loss": 5.0795,
      "step": 401
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8670176373799062,
      "learning_rate": 0.0005940251693390019,
      "loss": 4.9098,
      "step": 402
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8315657107530495,
      "learning_rate": 0.0005939809782394652,
      "loss": 4.9378,
      "step": 403
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9502158550245103,
      "learning_rate": 0.0005939366259741266,
      "loss": 4.8298,
      "step": 404
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6230339039493338,
      "learning_rate": 0.0005938921125673012,
      "loss": 5.1555,
      "step": 405
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8372063888595541,
      "learning_rate": 0.000593847438043392,
      "loss": 4.9752,
      "step": 406
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8947276246444711,
      "learning_rate": 0.0005938026024268903,
      "loss": 4.9903,
      "step": 407
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.30872518499831,
      "learning_rate": 0.000593757605742376,
      "loss": 5.1611,
      "step": 408
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7021113364896897,
      "learning_rate": 0.0005937124480145169,
      "loss": 4.8711,
      "step": 409
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6180190134793453,
      "learning_rate": 0.0005936671292680693,
      "loss": 4.9429,
      "step": 410
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8145336385971147,
      "learning_rate": 0.000593621649527878,
      "loss": 5.0408,
      "step": 411
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9413657734536987,
      "learning_rate": 0.0005935760088188756,
      "loss": 5.0782,
      "step": 412
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.396177954457218,
      "learning_rate": 0.0005935302071660833,
      "loss": 5.0698,
      "step": 413
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.396649026042867,
      "learning_rate": 0.0005934842445946104,
      "loss": 4.9115,
      "step": 414
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1462949756338208,
      "learning_rate": 0.0005934381211296544,
      "loss": 5.3148,
      "step": 415
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.6768886775268643,
      "learning_rate": 0.0005933918367965011,
      "loss": 4.9477,
      "step": 416
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0395456261469278,
      "learning_rate": 0.0005933453916205242,
      "loss": 4.9317,
      "step": 417
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.85315568690304,
      "learning_rate": 0.0005932987856271862,
      "loss": 4.9963,
      "step": 418
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8551974729633977,
      "learning_rate": 0.000593252018842037,
      "loss": 4.7788,
      "step": 419
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8686686531784708,
      "learning_rate": 0.0005932050912907153,
      "loss": 5.084,
      "step": 420
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9163273071238545,
      "learning_rate": 0.0005931580029989474,
      "loss": 5.0167,
      "step": 421
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9877793748773773,
      "learning_rate": 0.0005931107539925481,
      "loss": 5.0818,
      "step": 422
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0505100775641891,
      "learning_rate": 0.0005930633442974201,
      "loss": 4.8086,
      "step": 423
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8179493833075077,
      "learning_rate": 0.0005930157739395545,
      "loss": 4.9411,
      "step": 424
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7244350892354696,
      "learning_rate": 0.0005929680429450299,
      "loss": 5.0448,
      "step": 425
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0499001783554662,
      "learning_rate": 0.0005929201513400135,
      "loss": 4.8911,
      "step": 426
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8128710046308487,
      "learning_rate": 0.0005928720991507604,
      "loss": 4.7541,
      "step": 427
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9899594417674721,
      "learning_rate": 0.0005928238864036138,
      "loss": 4.9249,
      "step": 428
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8519475487475884,
      "learning_rate": 0.0005927755131250045,
      "loss": 4.9531,
      "step": 429
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.464984549099713,
      "learning_rate": 0.0005927269793414518,
      "loss": 4.9396,
      "step": 430
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.201953047130527,
      "learning_rate": 0.0005926782850795628,
      "loss": 4.734,
      "step": 431
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2717974382494048,
      "learning_rate": 0.0005926294303660327,
      "loss": 4.794,
      "step": 432
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.993575908219525,
      "learning_rate": 0.0005925804152276445,
      "loss": 5.0263,
      "step": 433
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3683592253935484,
      "learning_rate": 0.000592531239691269,
      "loss": 4.9283,
      "step": 434
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.153023199628829,
      "learning_rate": 0.0005924819037838655,
      "loss": 4.9314,
      "step": 435
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2501127620208763,
      "learning_rate": 0.0005924324075324805,
      "loss": 4.7887,
      "step": 436
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.3693157557954656,
      "learning_rate": 0.0005923827509642489,
      "loss": 4.8759,
      "step": 437
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.8303345750470963,
      "learning_rate": 0.0005923329341063935,
      "loss": 5.0531,
      "step": 438
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.880241403582436,
      "learning_rate": 0.0005922829569862245,
      "loss": 4.7967,
      "step": 439
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.4157370241388945,
      "learning_rate": 0.0005922328196311407,
      "loss": 4.9579,
      "step": 440
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4493882366755244,
      "learning_rate": 0.0005921825220686279,
      "loss": 4.9787,
      "step": 441
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.8163848850813746,
      "learning_rate": 0.0005921320643262605,
      "loss": 4.9163,
      "step": 442
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6892536001994958,
      "learning_rate": 0.0005920814464317,
      "loss": 4.9446,
      "step": 443
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.07121158587927,
      "learning_rate": 0.0005920306684126965,
      "loss": 4.896,
      "step": 444
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4684627849248248,
      "learning_rate": 0.0005919797302970871,
      "loss": 4.9125,
      "step": 445
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.9451580614618247,
      "learning_rate": 0.0005919286321127971,
      "loss": 4.8022,
      "step": 446
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8240641219923726,
      "learning_rate": 0.0005918773738878396,
      "loss": 5.0572,
      "step": 447
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.782360286052058,
      "learning_rate": 0.0005918259556503152,
      "loss": 4.9731,
      "step": 448
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.290655720816401,
      "learning_rate": 0.0005917743774284123,
      "loss": 4.8638,
      "step": 449
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6603345265998382,
      "learning_rate": 0.0005917226392504071,
      "loss": 4.998,
      "step": 450
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7725356772048324,
      "learning_rate": 0.0005916707411446634,
      "loss": 4.7678,
      "step": 451
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9364237803302594,
      "learning_rate": 0.0005916186831396327,
      "loss": 4.8794,
      "step": 452
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3183289526726105,
      "learning_rate": 0.0005915664652638541,
      "loss": 4.7109,
      "step": 453
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.5497725632182764,
      "learning_rate": 0.0005915140875459546,
      "loss": 5.0924,
      "step": 454
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.381288240314602,
      "learning_rate": 0.0005914615500146483,
      "loss": 5.0862,
      "step": 455
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4030847321270352,
      "learning_rate": 0.0005914088526987375,
      "loss": 4.919,
      "step": 456
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2577837464333326,
      "learning_rate": 0.0005913559956271119,
      "loss": 5.0596,
      "step": 457
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.1393888737298927,
      "learning_rate": 0.0005913029788287486,
      "loss": 4.6101,
      "step": 458
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7531951071642504,
      "learning_rate": 0.0005912498023327123,
      "loss": 4.6786,
      "step": 459
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2634790648468592,
      "learning_rate": 0.0005911964661681554,
      "loss": 4.8421,
      "step": 460
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4674237189664705,
      "learning_rate": 0.0005911429703643179,
      "loss": 4.8122,
      "step": 461
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.624924510258025,
      "learning_rate": 0.000591089314950527,
      "loss": 5.0403,
      "step": 462
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.414204484535188,
      "learning_rate": 0.0005910354999561977,
      "loss": 4.7091,
      "step": 463
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.443404140277432,
      "learning_rate": 0.0005909815254108323,
      "loss": 4.8525,
      "step": 464
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8264800002970221,
      "learning_rate": 0.0005909273913440208,
      "loss": 4.8763,
      "step": 465
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7602043521271793,
      "learning_rate": 0.0005908730977854403,
      "loss": 4.7583,
      "step": 466
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3402586010335686,
      "learning_rate": 0.0005908186447648557,
      "loss": 4.8228,
      "step": 467
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7136624019422253,
      "learning_rate": 0.000590764032312119,
      "loss": 4.9283,
      "step": 468
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.978101534272162,
      "learning_rate": 0.00059070926045717,
      "loss": 4.8035,
      "step": 469
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9530177933535502,
      "learning_rate": 0.0005906543292300352,
      "loss": 4.7629,
      "step": 470
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1346642108929814,
      "learning_rate": 0.0005905992386608293,
      "loss": 4.7692,
      "step": 471
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0308167631358847,
      "learning_rate": 0.0005905439887797538,
      "loss": 4.7912,
      "step": 472
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9875049162022476,
      "learning_rate": 0.0005904885796170979,
      "loss": 4.8697,
      "step": 473
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3418470115492953,
      "learning_rate": 0.0005904330112032377,
      "loss": 4.9477,
      "step": 474
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0480622368131916,
      "learning_rate": 0.0005903772835686369,
      "loss": 4.593,
      "step": 475
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.354536300576432,
      "learning_rate": 0.0005903213967438464,
      "loss": 4.7671,
      "step": 476
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1517031804785895,
      "learning_rate": 0.0005902653507595044,
      "loss": 4.7934,
      "step": 477
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.357036532682106,
      "learning_rate": 0.0005902091456463365,
      "loss": 4.7012,
      "step": 478
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6412632756399552,
      "learning_rate": 0.0005901527814351551,
      "loss": 4.6334,
      "step": 479
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.008841635849965,
      "learning_rate": 0.0005900962581568603,
      "loss": 4.8036,
      "step": 480
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1715832602810776,
      "learning_rate": 0.0005900395758424392,
      "loss": 4.8931,
      "step": 481
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1389431882251422,
      "learning_rate": 0.000589982734522966,
      "loss": 4.9751,
      "step": 482
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.529634724119913,
      "learning_rate": 0.0005899257342296021,
      "loss": 4.7829,
      "step": 483
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.7113162580036616,
      "learning_rate": 0.0005898685749935964,
      "loss": 4.7818,
      "step": 484
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.14422545664322,
      "learning_rate": 0.0005898112568462843,
      "loss": 4.8621,
      "step": 485
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.3386824191583413,
      "learning_rate": 0.0005897537798190889,
      "loss": 4.8773,
      "step": 486
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.2393840500097504,
      "learning_rate": 0.0005896961439435202,
      "loss": 4.9035,
      "step": 487
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.5573095920756312,
      "learning_rate": 0.000589638349251175,
      "loss": 4.691,
      "step": 488
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.9106696288016267,
      "learning_rate": 0.0005895803957737376,
      "loss": 4.7256,
      "step": 489
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0565024358512694,
      "learning_rate": 0.0005895222835429791,
      "loss": 4.8294,
      "step": 490
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.4590463691943576,
      "learning_rate": 0.0005894640125907579,
      "loss": 4.82,
      "step": 491
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3139952698871484,
      "learning_rate": 0.0005894055829490189,
      "loss": 4.6732,
      "step": 492
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4495280791730436,
      "learning_rate": 0.0005893469946497945,
      "loss": 4.7964,
      "step": 493
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5454952970252716,
      "learning_rate": 0.0005892882477252038,
      "loss": 4.841,
      "step": 494
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.130661382286492,
      "learning_rate": 0.0005892293422074531,
      "loss": 4.6413,
      "step": 495
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2142863255105913,
      "learning_rate": 0.0005891702781288353,
      "loss": 4.7086,
      "step": 496
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.949933694591531,
      "learning_rate": 0.0005891110555217306,
      "loss": 4.7153,
      "step": 497
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8936503977226508,
      "learning_rate": 0.0005890516744186058,
      "loss": 4.6979,
      "step": 498
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7925834791328705,
      "learning_rate": 0.0005889921348520146,
      "loss": 4.5879,
      "step": 499
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.008509942513787,
      "learning_rate": 0.000588932436854598,
      "loss": 4.6122,
      "step": 500
    },
    {
      "epoch": 0.57,
      "eval_blimp_filtered_avg": 0.5132835820895523,
      "eval_blimp_filtered_std": 0.0053635566345865944,
      "step": 500
    },
    {
      "epoch": 0.57,
      "eval_blimp_supplement_avg": 0.47413793103448276,
      "eval_blimp_supplement_std": 0.023246279862775738,
      "step": 500
    },
    {
      "epoch": 0.57,
      "eval_vqa_filtered_avg": 0.17,
      "eval_vqa_filtered_std": 0.03775251680686371,
      "step": 500
    },
    {
      "epoch": 0.57,
      "eval_winoground_filtered_avg": 0.55,
      "eval_winoground_filtered_std": 0.04999999999999999,
      "step": 500
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8171008064621451,
      "learning_rate": 0.0005888725804590833,
      "loss": 4.7365,
      "step": 501
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.838535759957165,
      "learning_rate": 0.000588812565698285,
      "loss": 5.02,
      "step": 502
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1205922192554914,
      "learning_rate": 0.0005887523926051042,
      "loss": 4.654,
      "step": 503
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1596528479611834,
      "learning_rate": 0.000588692061212529,
      "loss": 4.7663,
      "step": 504
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1993983992227317,
      "learning_rate": 0.0005886315715536341,
      "loss": 4.8463,
      "step": 505
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9718479619370148,
      "learning_rate": 0.0005885709236615812,
      "loss": 4.8378,
      "step": 506
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6810953901416547,
      "learning_rate": 0.0005885101175696185,
      "loss": 4.608,
      "step": 507
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9173504377743096,
      "learning_rate": 0.000588449153311081,
      "loss": 4.7006,
      "step": 508
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9626426105656214,
      "learning_rate": 0.0005883880309193903,
      "loss": 4.6051,
      "step": 509
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9126052009767455,
      "learning_rate": 0.0005883267504280551,
      "loss": 4.7994,
      "step": 510
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.563384748450241,
      "learning_rate": 0.0005882653118706704,
      "loss": 4.7805,
      "step": 511
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7646173077297718,
      "learning_rate": 0.0005882037152809179,
      "loss": 4.7342,
      "step": 512
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.315376183994279,
      "learning_rate": 0.000588141960692566,
      "loss": 4.7335,
      "step": 513
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.017408881181837,
      "learning_rate": 0.0005880800481394696,
      "loss": 4.8928,
      "step": 514
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5355368051852243,
      "learning_rate": 0.0005880179776555706,
      "loss": 4.5725,
      "step": 515
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6167280132290474,
      "learning_rate": 0.0005879557492748968,
      "loss": 4.5813,
      "step": 516
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.57654718341341,
      "learning_rate": 0.0005878933630315633,
      "loss": 4.696,
      "step": 517
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3657389484154014,
      "learning_rate": 0.0005878308189597712,
      "loss": 4.7612,
      "step": 518
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.051504319769276,
      "learning_rate": 0.0005877681170938084,
      "loss": 4.7385,
      "step": 519
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7348003082550534,
      "learning_rate": 0.000587705257468049,
      "loss": 4.6694,
      "step": 520
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.5503873834728132,
      "learning_rate": 0.0005876422401169542,
      "loss": 4.7072,
      "step": 521
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.2854321738868144,
      "learning_rate": 0.0005875790650750711,
      "loss": 4.7978,
      "step": 522
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.06854332718905,
      "learning_rate": 0.0005875157323770332,
      "loss": 4.6417,
      "step": 523
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.8834146160449166,
      "learning_rate": 0.0005874522420575609,
      "loss": 4.6135,
      "step": 524
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8619124332877415,
      "learning_rate": 0.0005873885941514608,
      "loss": 4.8998,
      "step": 525
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.139084953681298,
      "learning_rate": 0.0005873247886936257,
      "loss": 4.6173,
      "step": 526
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.3682004498947644,
      "learning_rate": 0.0005872608257190349,
      "loss": 4.8124,
      "step": 527
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.1919305740960056,
      "learning_rate": 0.0005871967052627542,
      "loss": 4.9145,
      "step": 528
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8074083573089332,
      "learning_rate": 0.0005871324273599357,
      "loss": 4.6684,
      "step": 529
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.760381800230039,
      "learning_rate": 0.0005870679920458176,
      "loss": 4.5479,
      "step": 530
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.7358811547781055,
      "learning_rate": 0.0005870033993557246,
      "loss": 4.4928,
      "step": 531
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.7177674714992777,
      "learning_rate": 0.0005869386493250674,
      "loss": 4.5819,
      "step": 532
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9799414964901767,
      "learning_rate": 0.0005868737419893435,
      "loss": 4.6163,
      "step": 533
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7802944079199146,
      "learning_rate": 0.0005868086773841361,
      "loss": 4.6801,
      "step": 534
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9483623847876221,
      "learning_rate": 0.0005867434555451148,
      "loss": 4.4441,
      "step": 535
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1427873743960972,
      "learning_rate": 0.0005866780765080355,
      "loss": 4.5677,
      "step": 536
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9405423787182783,
      "learning_rate": 0.00058661254030874,
      "loss": 4.6599,
      "step": 537
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2256930935071564,
      "learning_rate": 0.0005865468469831569,
      "loss": 4.6746,
      "step": 538
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.909741318211389,
      "learning_rate": 0.0005864809965672999,
      "loss": 4.4962,
      "step": 539
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5112603120659025,
      "learning_rate": 0.00058641498909727,
      "loss": 4.5661,
      "step": 540
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.969409632098589,
      "learning_rate": 0.0005863488246092533,
      "loss": 4.4646,
      "step": 541
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.770211682865135,
      "learning_rate": 0.0005862825031395225,
      "loss": 4.5369,
      "step": 542
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6040077016601832,
      "learning_rate": 0.0005862160247244363,
      "loss": 4.6617,
      "step": 543
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8940227446212239,
      "learning_rate": 0.0005861493894004394,
      "loss": 4.6954,
      "step": 544
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2910386809712782,
      "learning_rate": 0.0005860825972040626,
      "loss": 4.5581,
      "step": 545
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2131410991141303,
      "learning_rate": 0.0005860156481719224,
      "loss": 4.5131,
      "step": 546
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.120317683234146,
      "learning_rate": 0.0005859485423407216,
      "loss": 4.6343,
      "step": 547
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.310473575372834,
      "learning_rate": 0.0005858812797472489,
      "loss": 4.6142,
      "step": 548
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.273331360638896,
      "learning_rate": 0.0005858138604283789,
      "loss": 4.5779,
      "step": 549
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9570707889363734,
      "learning_rate": 0.000585746284421072,
      "loss": 4.5162,
      "step": 550
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.392164932188141,
      "learning_rate": 0.0005856785517623747,
      "loss": 4.6816,
      "step": 551
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2412222757338822,
      "learning_rate": 0.0005856106624894193,
      "loss": 4.7816,
      "step": 552
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8183044724648039,
      "learning_rate": 0.0005855426166394239,
      "loss": 4.2305,
      "step": 553
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.0650767565610204,
      "learning_rate": 0.0005854744142496926,
      "loss": 4.5111,
      "step": 554
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7537259644216572,
      "learning_rate": 0.000585406055357615,
      "loss": 4.6985,
      "step": 555
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4266868765331735,
      "learning_rate": 0.000585337540000667,
      "loss": 4.751,
      "step": 556
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.508195153365269,
      "learning_rate": 0.0005852688682164098,
      "loss": 4.3793,
      "step": 557
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.5526958022670714,
      "learning_rate": 0.0005852000400424905,
      "loss": 4.8661,
      "step": 558
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2189504667402664,
      "learning_rate": 0.0005851310555166422,
      "loss": 4.9176,
      "step": 559
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.539330463467029,
      "learning_rate": 0.0005850619146766832,
      "loss": 4.8204,
      "step": 560
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3164693020536904,
      "learning_rate": 0.0005849926175605179,
      "loss": 4.6477,
      "step": 561
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5578744816238042,
      "learning_rate": 0.0005849231642061363,
      "loss": 4.4022,
      "step": 562
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6624762082883593,
      "learning_rate": 0.0005848535546516141,
      "loss": 4.5103,
      "step": 563
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2319069138790335,
      "learning_rate": 0.0005847837889351124,
      "loss": 4.3334,
      "step": 564
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6544071013957493,
      "learning_rate": 0.000584713867094878,
      "loss": 4.5528,
      "step": 565
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.078503282807618,
      "learning_rate": 0.0005846437891692434,
      "loss": 4.6397,
      "step": 566
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.5058048319289474,
      "learning_rate": 0.0005845735551966269,
      "loss": 4.4403,
      "step": 567
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.134817404312362,
      "learning_rate": 0.0005845031652155316,
      "loss": 4.55,
      "step": 568
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0311693850083417,
      "learning_rate": 0.0005844326192645468,
      "loss": 4.6061,
      "step": 569
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6730180836450599,
      "learning_rate": 0.0005843619173823471,
      "loss": 4.6309,
      "step": 570
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7976482537109384,
      "learning_rate": 0.0005842910596076926,
      "loss": 4.4306,
      "step": 571
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2926770517049064,
      "learning_rate": 0.0005842200459794288,
      "loss": 4.247,
      "step": 572
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8722058193531201,
      "learning_rate": 0.0005841488765364868,
      "loss": 4.4775,
      "step": 573
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.579031044168635,
      "learning_rate": 0.0005840775513178827,
      "loss": 4.5176,
      "step": 574
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8003774524978253,
      "learning_rate": 0.0005840060703627186,
      "loss": 4.5182,
      "step": 575
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1941118664404586,
      "learning_rate": 0.0005839344337101816,
      "loss": 4.3376,
      "step": 576
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3887008061646828,
      "learning_rate": 0.0005838626413995441,
      "loss": 4.4326,
      "step": 577
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.3139035284049734,
      "learning_rate": 0.0005837906934701642,
      "loss": 4.3924,
      "step": 578
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.3568370387124467,
      "learning_rate": 0.0005837185899614848,
      "loss": 4.3549,
      "step": 579
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.2993200214867993,
      "learning_rate": 0.0005836463309130348,
      "loss": 4.6231,
      "step": 580
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6477160313473223,
      "learning_rate": 0.0005835739163644275,
      "loss": 4.6646,
      "step": 581
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.2185135423151476,
      "learning_rate": 0.0005835013463553622,
      "loss": 4.459,
      "step": 582
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4970352213018463,
      "learning_rate": 0.0005834286209256229,
      "loss": 4.5212,
      "step": 583
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3720074984260233,
      "learning_rate": 0.0005833557401150791,
      "loss": 4.6575,
      "step": 584
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.429644051071723,
      "learning_rate": 0.0005832827039636855,
      "loss": 4.5239,
      "step": 585
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8866397930135517,
      "learning_rate": 0.0005832095125114818,
      "loss": 4.6404,
      "step": 586
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.217479478831131,
      "learning_rate": 0.0005831361657985928,
      "loss": 4.5873,
      "step": 587
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3626623129430853,
      "learning_rate": 0.0005830626638652287,
      "loss": 4.3849,
      "step": 588
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.123611645060079,
      "learning_rate": 0.0005829890067516845,
      "loss": 4.5972,
      "step": 589
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.515740987022606,
      "learning_rate": 0.0005829151944983405,
      "loss": 4.5847,
      "step": 590
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.373894020911828,
      "learning_rate": 0.0005828412271456618,
      "loss": 4.4942,
      "step": 591
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.6168771308421395,
      "learning_rate": 0.0005827671047341988,
      "loss": 4.6232,
      "step": 592
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.4466098061821766,
      "learning_rate": 0.0005826928273045866,
      "loss": 4.734,
      "step": 593
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.5297440453794775,
      "learning_rate": 0.0005826183948975457,
      "loss": 4.6071,
      "step": 594
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.6786734055206285,
      "learning_rate": 0.0005825438075538812,
      "loss": 4.5363,
      "step": 595
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.378062849949167,
      "learning_rate": 0.0005824690653144833,
      "loss": 4.464,
      "step": 596
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5445799094739672,
      "learning_rate": 0.0005823941682203268,
      "loss": 4.3626,
      "step": 597
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9201647194414257,
      "learning_rate": 0.0005823191163124721,
      "loss": 4.4447,
      "step": 598
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4445228799635261,
      "learning_rate": 0.0005822439096320638,
      "loss": 4.6541,
      "step": 599
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.599566862053309,
      "learning_rate": 0.0005821685482203315,
      "loss": 4.4155,
      "step": 600
    },
    {
      "epoch": 0.69,
      "eval_blimp_filtered_avg": 0.5270149253731343,
      "eval_blimp_filtered_std": 0.005412543717709778,
      "step": 600
    },
    {
      "epoch": 0.69,
      "eval_blimp_supplement_avg": 0.5452586206896551,
      "eval_blimp_supplement_std": 0.023051305575903118,
      "step": 600
    },
    {
      "epoch": 0.69,
      "eval_vqa_filtered_avg": 0.35,
      "eval_vqa_filtered_std": 0.0479372485441102,
      "step": 600
    },
    {
      "epoch": 0.69,
      "eval_winoground_filtered_avg": 0.51,
      "eval_winoground_filtered_std": 0.05024183937956912,
      "step": 600
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.153028431537132,
      "learning_rate": 0.00058209303211859,
      "loss": 4.5192,
      "step": 601
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4375604090612792,
      "learning_rate": 0.0005820173613682385,
      "loss": 4.5397,
      "step": 602
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.18253603496043,
      "learning_rate": 0.0005819415360107609,
      "loss": 4.5341,
      "step": 603
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5037700705204315,
      "learning_rate": 0.0005818655560877264,
      "loss": 4.5133,
      "step": 604
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3452045210709744,
      "learning_rate": 0.0005817894216407885,
      "loss": 4.4875,
      "step": 605
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8609470942745091,
      "learning_rate": 0.0005817131327116853,
      "loss": 4.6046,
      "step": 606
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.520603707500873,
      "learning_rate": 0.0005816366893422399,
      "loss": 4.3918,
      "step": 607
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.707395452289135,
      "learning_rate": 0.00058156009157436,
      "loss": 4.6315,
      "step": 608
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.053411393029927,
      "learning_rate": 0.000581483339450038,
      "loss": 4.7913,
      "step": 609
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5492946681888873,
      "learning_rate": 0.0005814064330113509,
      "loss": 4.4569,
      "step": 610
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9012913476510545,
      "learning_rate": 0.0005813293723004598,
      "loss": 4.6044,
      "step": 611
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.106278887515256,
      "learning_rate": 0.0005812521573596112,
      "loss": 4.4048,
      "step": 612
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.2272607130511113,
      "learning_rate": 0.0005811747882311357,
      "loss": 4.4922,
      "step": 613
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.1886127550625463,
      "learning_rate": 0.0005810972649574484,
      "loss": 4.5624,
      "step": 614
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.8345762063578746,
      "learning_rate": 0.000581019587581049,
      "loss": 4.7792,
      "step": 615
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.7306034009254923,
      "learning_rate": 0.0005809417561445217,
      "loss": 4.6113,
      "step": 616
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.5817646888539425,
      "learning_rate": 0.0005808637706905351,
      "loss": 4.5571,
      "step": 617
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.9360810794666365,
      "learning_rate": 0.0005807856312618423,
      "loss": 4.688,
      "step": 618
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.6471596128698645,
      "learning_rate": 0.0005807073379012808,
      "loss": 4.6999,
      "step": 619
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.382956047893064,
      "learning_rate": 0.0005806288906517724,
      "loss": 4.5704,
      "step": 620
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.916998723576681,
      "learning_rate": 0.0005805502895563234,
      "loss": 4.6399,
      "step": 621
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.829786587257529,
      "learning_rate": 0.0005804715346580243,
      "loss": 4.6934,
      "step": 622
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8514127585254476,
      "learning_rate": 0.00058039262600005,
      "loss": 4.5135,
      "step": 623
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9313034622764944,
      "learning_rate": 0.0005803135636256599,
      "loss": 4.7101,
      "step": 624
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8799230152237083,
      "learning_rate": 0.000580234347578197,
      "loss": 4.4219,
      "step": 625
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.2563815904747493,
      "learning_rate": 0.0005801549779010894,
      "loss": 4.6391,
      "step": 626
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.2995755511664884,
      "learning_rate": 0.0005800754546378488,
      "loss": 4.496,
      "step": 627
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.3741097541216885,
      "learning_rate": 0.0005799957778320716,
      "loss": 4.3619,
      "step": 628
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.970973439588761,
      "learning_rate": 0.0005799159475274377,
      "loss": 4.6854,
      "step": 629
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.783998069759036,
      "learning_rate": 0.0005798359637677118,
      "loss": 4.6968,
      "step": 630
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.2706158269709262,
      "learning_rate": 0.0005797558265967425,
      "loss": 4.3752,
      "step": 631
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.543258477383476,
      "learning_rate": 0.0005796755360584624,
      "loss": 4.5127,
      "step": 632
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.2735701888155337,
      "learning_rate": 0.0005795950921968882,
      "loss": 4.3989,
      "step": 633
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1896237403404126,
      "learning_rate": 0.0005795144950561207,
      "loss": 4.4731,
      "step": 634
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.894490051112878,
      "learning_rate": 0.0005794337446803449,
      "loss": 4.4438,
      "step": 635
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6220094387318125,
      "learning_rate": 0.0005793528411138295,
      "loss": 4.3676,
      "step": 636
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.9351225765386078,
      "learning_rate": 0.0005792717844009275,
      "loss": 4.6176,
      "step": 637
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8508859886264898,
      "learning_rate": 0.0005791905745860756,
      "loss": 4.504,
      "step": 638
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.319166104294842,
      "learning_rate": 0.0005791092117137944,
      "loss": 4.5067,
      "step": 639
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7160917131649451,
      "learning_rate": 0.0005790276958286886,
      "loss": 4.5715,
      "step": 640
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.7815471157488063,
      "learning_rate": 0.0005789460269754467,
      "loss": 4.4707,
      "step": 641
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8560041999946482,
      "learning_rate": 0.0005788642051988413,
      "loss": 4.3444,
      "step": 642
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.6850493635169523,
      "learning_rate": 0.0005787822305437283,
      "loss": 4.6303,
      "step": 643
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2599905854419227,
      "learning_rate": 0.0005787001030550478,
      "loss": 4.6004,
      "step": 644
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.1079452933649114,
      "learning_rate": 0.0005786178227778236,
      "loss": 4.4919,
      "step": 645
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2731969815583444,
      "learning_rate": 0.0005785353897571635,
      "loss": 4.562,
      "step": 646
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9856625670721976,
      "learning_rate": 0.0005784528040382585,
      "loss": 4.6699,
      "step": 647
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7162728712097575,
      "learning_rate": 0.0005783700656663838,
      "loss": 4.5084,
      "step": 648
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.151568007916394,
      "learning_rate": 0.000578287174686898,
      "loss": 4.4242,
      "step": 649
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0549135843405837,
      "learning_rate": 0.0005782041311452436,
      "loss": 4.2738,
      "step": 650
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0730888488878283,
      "learning_rate": 0.0005781209350869464,
      "loss": 4.5284,
      "step": 651
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4718274032693577,
      "learning_rate": 0.0005780375865576161,
      "loss": 4.3566,
      "step": 652
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0415134553124974,
      "learning_rate": 0.0005779540856029458,
      "loss": 4.4505,
      "step": 653
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.4767026620698656,
      "learning_rate": 0.0005778704322687127,
      "loss": 4.5071,
      "step": 654
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5899963323412976,
      "learning_rate": 0.0005777866266007766,
      "loss": 4.4988,
      "step": 655
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.411916827894673,
      "learning_rate": 0.0005777026686450816,
      "loss": 4.4039,
      "step": 656
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0378019425329286,
      "learning_rate": 0.0005776185584476548,
      "loss": 4.4763,
      "step": 657
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.9649669942121437,
      "learning_rate": 0.0005775342960546071,
      "loss": 4.2827,
      "step": 658
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3561041938087637,
      "learning_rate": 0.0005774498815121327,
      "loss": 4.5832,
      "step": 659
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.2945572331672914,
      "learning_rate": 0.0005773653148665093,
      "loss": 4.5525,
      "step": 660
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.314855258008567,
      "learning_rate": 0.0005772805961640976,
      "loss": 4.6166,
      "step": 661
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.5718702502026827,
      "learning_rate": 0.0005771957254513422,
      "loss": 4.5738,
      "step": 662
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.41905268099026,
      "learning_rate": 0.0005771107027747708,
      "loss": 4.5421,
      "step": 663
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.766471664055671,
      "learning_rate": 0.0005770255281809943,
      "loss": 4.5963,
      "step": 664
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.299489263008332,
      "learning_rate": 0.000576940201716707,
      "loss": 4.5228,
      "step": 665
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1228439535348742,
      "learning_rate": 0.0005768547234286864,
      "loss": 4.6363,
      "step": 666
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.6356197606505996,
      "learning_rate": 0.0005767690933637936,
      "loss": 4.5081,
      "step": 667
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9680715852537185,
      "learning_rate": 0.000576683311568972,
      "loss": 4.538,
      "step": 668
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.888467494848188,
      "learning_rate": 0.0005765973780912493,
      "loss": 4.8447,
      "step": 669
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.138442266066419,
      "learning_rate": 0.0005765112929777355,
      "loss": 4.4692,
      "step": 670
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6728757287027467,
      "learning_rate": 0.0005764250562756242,
      "loss": 4.2637,
      "step": 671
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.3050192140866717,
      "learning_rate": 0.0005763386680321919,
      "loss": 4.6787,
      "step": 672
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.5816127137993963,
      "learning_rate": 0.0005762521282947983,
      "loss": 4.4092,
      "step": 673
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.353428573324596,
      "learning_rate": 0.0005761654371108858,
      "loss": 4.8245,
      "step": 674
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.4605311576050526,
      "learning_rate": 0.0005760785945279805,
      "loss": 4.5476,
      "step": 675
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.522482702851574,
      "learning_rate": 0.0005759916005936909,
      "loss": 4.5458,
      "step": 676
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9927883305313956,
      "learning_rate": 0.0005759044553557087,
      "loss": 4.4768,
      "step": 677
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3209147673977002,
      "learning_rate": 0.0005758171588618086,
      "loss": 4.3871,
      "step": 678
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9847510770126404,
      "learning_rate": 0.0005757297111598481,
      "loss": 4.567,
      "step": 679
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9235732603896354,
      "learning_rate": 0.0005756421122977676,
      "loss": 4.5036,
      "step": 680
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8214497724997958,
      "learning_rate": 0.0005755543623235905,
      "loss": 4.4471,
      "step": 681
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.172296730948008,
      "learning_rate": 0.0005754664612854229,
      "loss": 4.4955,
      "step": 682
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7239723242151066,
      "learning_rate": 0.0005753784092314538,
      "loss": 4.6132,
      "step": 683
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.865043112615227,
      "learning_rate": 0.0005752902062099548,
      "loss": 4.416,
      "step": 684
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9982553889175986,
      "learning_rate": 0.0005752018522692808,
      "loss": 4.3722,
      "step": 685
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.491077400324958,
      "learning_rate": 0.0005751133474578686,
      "loss": 4.4652,
      "step": 686
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1018609800900525,
      "learning_rate": 0.0005750246918242387,
      "loss": 4.36,
      "step": 687
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.8016664728383263,
      "learning_rate": 0.0005749358854169932,
      "loss": 4.6413,
      "step": 688
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.154897814606344,
      "learning_rate": 0.0005748469282848178,
      "loss": 4.6886,
      "step": 689
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.4299272344966587,
      "learning_rate": 0.0005747578204764802,
      "loss": 4.4376,
      "step": 690
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.933390787553071,
      "learning_rate": 0.0005746685620408312,
      "loss": 4.4609,
      "step": 691
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9915442022259469,
      "learning_rate": 0.0005745791530268037,
      "loss": 4.5932,
      "step": 692
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6728162714686678,
      "learning_rate": 0.0005744895934834135,
      "loss": 4.6454,
      "step": 693
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1689286642146983,
      "learning_rate": 0.0005743998834597588,
      "loss": 4.4759,
      "step": 694
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.6772331296898217,
      "learning_rate": 0.0005743100230050202,
      "loss": 4.4271,
      "step": 695
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8146385050830827,
      "learning_rate": 0.0005742200121684609,
      "loss": 4.5155,
      "step": 696
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.5340900372134594,
      "learning_rate": 0.0005741298509994265,
      "loss": 4.4396,
      "step": 697
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.997599201699876,
      "learning_rate": 0.000574039539547345,
      "loss": 4.4462,
      "step": 698
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.970713037666637,
      "learning_rate": 0.0005739490778617269,
      "loss": 4.4837,
      "step": 699
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.13766441551782,
      "learning_rate": 0.0005738584659921649,
      "loss": 4.4534,
      "step": 700
    },
    {
      "epoch": 0.8,
      "eval_blimp_filtered_avg": 0.553134328358209,
      "eval_blimp_filtered_std": 0.005455587933379622,
      "step": 700
    },
    {
      "epoch": 0.8,
      "eval_blimp_supplement_avg": 0.5344827586206896,
      "eval_blimp_supplement_std": 0.022812448704690593,
      "step": 700
    },
    {
      "epoch": 0.8,
      "eval_vqa_filtered_avg": 0.29,
      "eval_vqa_filtered_std": 0.04560480215720683,
      "step": 700
    },
    {
      "epoch": 0.8,
      "eval_winoground_filtered_avg": 0.54,
      "eval_winoground_filtered_std": 0.05009082659620333,
      "step": 700
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.532099053450028,
      "learning_rate": 0.0005737677039883341,
      "loss": 4.387,
      "step": 701
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8371789292717418,
      "learning_rate": 0.0005736767918999918,
      "loss": 4.5827,
      "step": 702
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.325423868832235,
      "learning_rate": 0.0005735857297769778,
      "loss": 4.3446,
      "step": 703
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9279613868833123,
      "learning_rate": 0.0005734945176692143,
      "loss": 4.3637,
      "step": 704
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.345789880027642,
      "learning_rate": 0.000573403155626705,
      "loss": 4.3231,
      "step": 705
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.891641206435242,
      "learning_rate": 0.0005733116436995363,
      "loss": 4.4166,
      "step": 706
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.7603105244784145,
      "learning_rate": 0.0005732199819378772,
      "loss": 4.4378,
      "step": 707
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.292011873698199,
      "learning_rate": 0.0005731281703919779,
      "loss": 4.3836,
      "step": 708
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.6493500158845205,
      "learning_rate": 0.0005730362091121712,
      "loss": 4.4418,
      "step": 709
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0544472718452804,
      "learning_rate": 0.0005729440981488723,
      "loss": 4.4469,
      "step": 710
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.022547840216597,
      "learning_rate": 0.000572851837552578,
      "loss": 4.604,
      "step": 711
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2369711109434218,
      "learning_rate": 0.0005727594273738669,
      "loss": 4.3847,
      "step": 712
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.297349688311103,
      "learning_rate": 0.0005726668676634005,
      "loss": 4.208,
      "step": 713
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.405706658596966,
      "learning_rate": 0.0005725741584719216,
      "loss": 4.3919,
      "step": 714
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.4577970425541227,
      "learning_rate": 0.0005724812998502548,
      "loss": 4.4645,
      "step": 715
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8851229245228893,
      "learning_rate": 0.0005723882918493073,
      "loss": 4.2623,
      "step": 716
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.1703966323707973,
      "learning_rate": 0.0005722951345200674,
      "loss": 4.3944,
      "step": 717
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0292912458059562,
      "learning_rate": 0.000572201827913606,
      "loss": 4.3766,
      "step": 718
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.0094306790503977,
      "learning_rate": 0.0005721083720810755,
      "loss": 4.5182,
      "step": 719
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1425603193754892,
      "learning_rate": 0.0005720147670737098,
      "loss": 4.5696,
      "step": 720
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3698665134131338,
      "learning_rate": 0.0005719210129428252,
      "loss": 4.0551,
      "step": 721
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8943244199712028,
      "learning_rate": 0.0005718271097398193,
      "loss": 4.4265,
      "step": 722
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.394845270112468,
      "learning_rate": 0.0005717330575161717,
      "loss": 4.1211,
      "step": 723
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7801541279360003,
      "learning_rate": 0.0005716388563234434,
      "loss": 4.4524,
      "step": 724
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.140343209570443,
      "learning_rate": 0.0005715445062132774,
      "loss": 4.4731,
      "step": 725
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1815931127587753,
      "learning_rate": 0.000571450007237398,
      "loss": 4.4272,
      "step": 726
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7310687566030418,
      "learning_rate": 0.0005713553594476114,
      "loss": 4.186,
      "step": 727
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2905058001831833,
      "learning_rate": 0.0005712605628958052,
      "loss": 4.4239,
      "step": 728
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.2222497163663566,
      "learning_rate": 0.0005711656176339488,
      "loss": 4.512,
      "step": 729
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.37575764752151,
      "learning_rate": 0.0005710705237140926,
      "loss": 4.4808,
      "step": 730
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.943089443595224,
      "learning_rate": 0.0005709752811883693,
      "loss": 4.604,
      "step": 731
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.6035895232186665,
      "learning_rate": 0.0005708798901089922,
      "loss": 4.3839,
      "step": 732
    },
    {
      "epoch": 0.84,
      "grad_norm": 9.848448183957649,
      "learning_rate": 0.0005707843505282566,
      "loss": 4.7996,
      "step": 733
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.7634232222221975,
      "learning_rate": 0.0005706886624985392,
      "loss": 4.8332,
      "step": 734
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.482064340844488,
      "learning_rate": 0.0005705928260722976,
      "loss": 5.0257,
      "step": 735
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.651911902393482,
      "learning_rate": 0.0005704968413020715,
      "loss": 4.8522,
      "step": 736
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8238237841287386,
      "learning_rate": 0.0005704007082404812,
      "loss": 4.8846,
      "step": 737
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3715742405050868,
      "learning_rate": 0.0005703044269402288,
      "loss": 4.7145,
      "step": 738
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.407277605814718,
      "learning_rate": 0.0005702079974540975,
      "loss": 4.6802,
      "step": 739
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.0453486683081543,
      "learning_rate": 0.0005701114198349517,
      "loss": 4.8464,
      "step": 740
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.724592874176681,
      "learning_rate": 0.0005700146941357369,
      "loss": 4.7677,
      "step": 741
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.673521662411844,
      "learning_rate": 0.0005699178204094801,
      "loss": 4.68,
      "step": 742
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.369231257596732,
      "learning_rate": 0.0005698207987092892,
      "loss": 4.6597,
      "step": 743
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.416722707502322,
      "learning_rate": 0.0005697236290883533,
      "loss": 4.8169,
      "step": 744
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.165726676004057,
      "learning_rate": 0.0005696263115999425,
      "loss": 4.6381,
      "step": 745
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6942329889251009,
      "learning_rate": 0.0005695288462974082,
      "loss": 4.4709,
      "step": 746
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.015745170207427,
      "learning_rate": 0.0005694312332341824,
      "loss": 4.5296,
      "step": 747
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.953013100441991,
      "learning_rate": 0.0005693334724637787,
      "loss": 4.5185,
      "step": 748
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.114676833833503,
      "learning_rate": 0.0005692355640397911,
      "loss": 4.7051,
      "step": 749
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.494260768573294,
      "learning_rate": 0.0005691375080158951,
      "loss": 4.3922,
      "step": 750
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.5618183225144366,
      "learning_rate": 0.0005690393044458466,
      "loss": 4.6168,
      "step": 751
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.019983992488592,
      "learning_rate": 0.0005689409533834825,
      "loss": 4.5672,
      "step": 752
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.017533357070491,
      "learning_rate": 0.000568842454882721,
      "loss": 4.7495,
      "step": 753
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.426341657146411,
      "learning_rate": 0.0005687438089975606,
      "loss": 4.5616,
      "step": 754
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.475375144051202,
      "learning_rate": 0.0005686450157820808,
      "loss": 4.5466,
      "step": 755
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.8075668304869077,
      "learning_rate": 0.0005685460752904419,
      "loss": 4.5189,
      "step": 756
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.4745007306094204,
      "learning_rate": 0.000568446987576885,
      "loss": 4.5952,
      "step": 757
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3430123682395663,
      "learning_rate": 0.0005683477526957315,
      "loss": 4.4737,
      "step": 758
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.3071858574030033,
      "learning_rate": 0.0005682483707013841,
      "loss": 4.6631,
      "step": 759
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.129151518153412,
      "learning_rate": 0.0005681488416483259,
      "loss": 4.4614,
      "step": 760
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.3496667672402167,
      "learning_rate": 0.0005680491655911202,
      "loss": 4.6929,
      "step": 761
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9927827409676904,
      "learning_rate": 0.0005679493425844116,
      "loss": 4.5659,
      "step": 762
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2389348480753737,
      "learning_rate": 0.0005678493726829247,
      "loss": 4.4976,
      "step": 763
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5569958291231294,
      "learning_rate": 0.000567749255941465,
      "loss": 4.6115,
      "step": 764
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9986881732505117,
      "learning_rate": 0.0005676489924149182,
      "loss": 4.62,
      "step": 765
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.632447485757524,
      "learning_rate": 0.0005675485821582508,
      "loss": 4.666,
      "step": 766
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2834395674626455,
      "learning_rate": 0.0005674480252265094,
      "loss": 4.5658,
      "step": 767
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8659965772429834,
      "learning_rate": 0.0005673473216748212,
      "loss": 4.4568,
      "step": 768
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7367966031530824,
      "learning_rate": 0.0005672464715583938,
      "loss": 4.4019,
      "step": 769
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0357248293931145,
      "learning_rate": 0.0005671454749325151,
      "loss": 4.3319,
      "step": 770
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4118371068199673,
      "learning_rate": 0.0005670443318525532,
      "loss": 4.5843,
      "step": 771
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6320678035539897,
      "learning_rate": 0.0005669430423739568,
      "loss": 4.3038,
      "step": 772
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.858698792622335,
      "learning_rate": 0.0005668416065522544,
      "loss": 4.2895,
      "step": 773
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.829095201647142,
      "learning_rate": 0.0005667400244430551,
      "loss": 4.6098,
      "step": 774
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.045990691717911,
      "learning_rate": 0.0005666382961020483,
      "loss": 4.4029,
      "step": 775
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8357215650205188,
      "learning_rate": 0.0005665364215850031,
      "loss": 4.4692,
      "step": 776
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4516546372915655,
      "learning_rate": 0.0005664344009477691,
      "loss": 4.3512,
      "step": 777
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1930442381589221,
      "learning_rate": 0.0005663322342462758,
      "loss": 4.5269,
      "step": 778
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.692403511162347,
      "learning_rate": 0.0005662299215365331,
      "loss": 4.3754,
      "step": 779
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.500655486702712,
      "learning_rate": 0.0005661274628746304,
      "loss": 4.4719,
      "step": 780
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4448450648992466,
      "learning_rate": 0.0005660248583167377,
      "loss": 4.7169,
      "step": 781
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0091761476074026,
      "learning_rate": 0.0005659221079191045,
      "loss": 4.5633,
      "step": 782
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.925721765807762,
      "learning_rate": 0.0005658192117380606,
      "loss": 4.5922,
      "step": 783
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.967872080142002,
      "learning_rate": 0.0005657161698300156,
      "loss": 4.5185,
      "step": 784
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3250771765072809,
      "learning_rate": 0.0005656129822514589,
      "loss": 4.3179,
      "step": 785
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2605820311888443,
      "learning_rate": 0.0005655096490589599,
      "loss": 4.2923,
      "step": 786
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6723937109250728,
      "learning_rate": 0.0005654061703091677,
      "loss": 4.1635,
      "step": 787
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1794030652620224,
      "learning_rate": 0.0005653025460588113,
      "loss": 4.3573,
      "step": 788
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4623138738350192,
      "learning_rate": 0.0005651987763646994,
      "loss": 4.4752,
      "step": 789
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0033312218999346,
      "learning_rate": 0.0005650948612837205,
      "loss": 4.5783,
      "step": 790
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.622095944806616,
      "learning_rate": 0.0005649908008728429,
      "loss": 4.124,
      "step": 791
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.4223052056056416,
      "learning_rate": 0.0005648865951891142,
      "loss": 4.3099,
      "step": 792
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.3198590554655696,
      "learning_rate": 0.0005647822442896619,
      "loss": 4.4748,
      "step": 793
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.783637270349249,
      "learning_rate": 0.000564677748231693,
      "loss": 4.4141,
      "step": 794
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.940965363632524,
      "learning_rate": 0.0005645731070724947,
      "loss": 4.3795,
      "step": 795
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.733091750642147,
      "learning_rate": 0.0005644683208694326,
      "loss": 4.514,
      "step": 796
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.0683662670866614,
      "learning_rate": 0.0005643633896799528,
      "loss": 4.5301,
      "step": 797
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.6429508219839413,
      "learning_rate": 0.0005642583135615803,
      "loss": 4.4977,
      "step": 798
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4669385770213428,
      "learning_rate": 0.00056415309257192,
      "loss": 4.5343,
      "step": 799
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5140112360435332,
      "learning_rate": 0.0005640477267686557,
      "loss": 4.47,
      "step": 800
    },
    {
      "epoch": 0.91,
      "eval_blimp_filtered_avg": 0.5392537313432836,
      "eval_blimp_filtered_std": 0.0055315890606076785,
      "step": 800
    },
    {
      "epoch": 0.91,
      "eval_blimp_supplement_avg": 0.5301724137931034,
      "eval_blimp_supplement_std": 0.02262080566474024,
      "step": 800
    },
    {
      "epoch": 0.91,
      "eval_vqa_filtered_avg": 0.17,
      "eval_vqa_filtered_std": 0.03775251680686371,
      "step": 800
    },
    {
      "epoch": 0.91,
      "eval_winoground_filtered_avg": 0.49,
      "eval_winoground_filtered_std": 0.05024183937956911,
      "step": 800
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.5179163214378097,
      "learning_rate": 0.0005639422162095511,
      "loss": 4.1751,
      "step": 801
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.6077674764452516,
      "learning_rate": 0.0005638365609524491,
      "loss": 4.3241,
      "step": 802
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.6254048476525464,
      "learning_rate": 0.0005637307610552715,
      "loss": 4.3363,
      "step": 803
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.6223143245718377,
      "learning_rate": 0.0005636248165760201,
      "loss": 4.3163,
      "step": 804
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.946814217523914,
      "learning_rate": 0.0005635187275727756,
      "loss": 4.334,
      "step": 805
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.851760967405565,
      "learning_rate": 0.0005634124941036977,
      "loss": 4.3922,
      "step": 806
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9260711573095337,
      "learning_rate": 0.0005633061162270258,
      "loss": 4.2261,
      "step": 807
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.0713108268202687,
      "learning_rate": 0.0005631995940010782,
      "loss": 4.6197,
      "step": 808
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.4059502725241124,
      "learning_rate": 0.0005630929274842521,
      "loss": 4.3439,
      "step": 809
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.410281311586372,
      "learning_rate": 0.0005629861167350243,
      "loss": 4.4831,
      "step": 810
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.7378744370954116,
      "learning_rate": 0.0005628791618119505,
      "loss": 4.1798,
      "step": 811
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2370648409165534,
      "learning_rate": 0.0005627720627736651,
      "loss": 4.4679,
      "step": 812
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.667516410670082,
      "learning_rate": 0.0005626648196788819,
      "loss": 4.398,
      "step": 813
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6302774429509728,
      "learning_rate": 0.0005625574325863936,
      "loss": 4.4626,
      "step": 814
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.721795133982151,
      "learning_rate": 0.0005624499015550716,
      "loss": 4.4726,
      "step": 815
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.759611590186062,
      "learning_rate": 0.0005623422266438666,
      "loss": 4.2706,
      "step": 816
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5953318563213739,
      "learning_rate": 0.000562234407911808,
      "loss": 4.4214,
      "step": 817
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5267615564582564,
      "learning_rate": 0.0005621264454180037,
      "loss": 4.5114,
      "step": 818
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.369872503239026,
      "learning_rate": 0.0005620183392216412,
      "loss": 4.1582,
      "step": 819
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2436422548990567,
      "learning_rate": 0.0005619100893819859,
      "loss": 4.3708,
      "step": 820
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.826957117750034,
      "learning_rate": 0.0005618016959583825,
      "loss": 4.456,
      "step": 821
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.0012567327677417,
      "learning_rate": 0.0005616931590102544,
      "loss": 4.5132,
      "step": 822
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.907955608971358,
      "learning_rate": 0.0005615844785971035,
      "loss": 4.3166,
      "step": 823
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.803442390900075,
      "learning_rate": 0.0005614756547785102,
      "loss": 4.303,
      "step": 824
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.4337066317283753,
      "learning_rate": 0.0005613666876141341,
      "loss": 4.6722,
      "step": 825
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.852406717139007,
      "learning_rate": 0.0005612575771637127,
      "loss": 4.3167,
      "step": 826
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8213680353396644,
      "learning_rate": 0.0005611483234870625,
      "loss": 4.3684,
      "step": 827
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8623478496037194,
      "learning_rate": 0.0005610389266440783,
      "loss": 4.4812,
      "step": 828
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.2353003239424747,
      "learning_rate": 0.0005609293866947336,
      "loss": 4.4897,
      "step": 829
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.964179452287191,
      "learning_rate": 0.0005608197036990802,
      "loss": 4.6249,
      "step": 830
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.4640227261806573,
      "learning_rate": 0.0005607098777172482,
      "loss": 4.3987,
      "step": 831
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.218959984083028,
      "learning_rate": 0.0005605999088094464,
      "loss": 4.6089,
      "step": 832
    },
    {
      "epoch": 0.95,
      "grad_norm": 18.624015607241194,
      "learning_rate": 0.0005604897970359617,
      "loss": 4.4225,
      "step": 833
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.594156337694963,
      "learning_rate": 0.0005603795424571593,
      "loss": 4.4438,
      "step": 834
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.9215600128270105,
      "learning_rate": 0.0005602691451334831,
      "loss": 4.658,
      "step": 835
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.5097119339315124,
      "learning_rate": 0.0005601586051254548,
      "loss": 4.6479,
      "step": 836
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.5698245992094395,
      "learning_rate": 0.0005600479224936743,
      "loss": 4.5435,
      "step": 837
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.470065444373569,
      "learning_rate": 0.00055993709729882,
      "loss": 4.6083,
      "step": 838
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.546118390346671,
      "learning_rate": 0.0005598261296016483,
      "loss": 4.4647,
      "step": 839
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.319978301056108,
      "learning_rate": 0.000559715019462994,
      "loss": 4.609,
      "step": 840
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.1188781892482322,
      "learning_rate": 0.0005596037669437692,
      "loss": 4.301,
      "step": 841
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.4374142383099775,
      "learning_rate": 0.0005594923721049651,
      "loss": 4.2956,
      "step": 842
    },
    {
      "epoch": 0.96,
      "grad_norm": 16.332017027734327,
      "learning_rate": 0.0005593808350076501,
      "loss": 4.5892,
      "step": 843
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.1534559813597,
      "learning_rate": 0.000559269155712971,
      "loss": 4.4758,
      "step": 844
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.0327205284902234,
      "learning_rate": 0.0005591573342821524,
      "loss": 4.8934,
      "step": 845
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.305301294735738,
      "learning_rate": 0.000559045370776497,
      "loss": 4.6047,
      "step": 846
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4493363946470963,
      "learning_rate": 0.000558933265257385,
      "loss": 4.6245,
      "step": 847
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5799028275658689,
      "learning_rate": 0.0005588210177862751,
      "loss": 4.5984,
      "step": 848
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4887415794623957,
      "learning_rate": 0.0005587086284247031,
      "loss": 4.7335,
      "step": 849
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7259266727108415,
      "learning_rate": 0.0005585960972342831,
      "loss": 4.6991,
      "step": 850
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5623247023516345,
      "learning_rate": 0.0005584834242767067,
      "loss": 4.4171,
      "step": 851
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2758342697998015,
      "learning_rate": 0.0005583706096137433,
      "loss": 4.2991,
      "step": 852
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.35947113452101,
      "learning_rate": 0.0005582576533072401,
      "loss": 4.44,
      "step": 853
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.9811034866921764,
      "learning_rate": 0.0005581445554191217,
      "loss": 4.7852,
      "step": 854
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.808423969727198,
      "learning_rate": 0.0005580313160113904,
      "loss": 4.7576,
      "step": 855
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.1381114754339956,
      "learning_rate": 0.0005579179351461263,
      "loss": 4.7348,
      "step": 856
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.12222792436702,
      "learning_rate": 0.0005578044128854868,
      "loss": 4.784,
      "step": 857
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.770492013846354,
      "learning_rate": 0.0005576907492917069,
      "loss": 4.6636,
      "step": 858
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.644363299562771,
      "learning_rate": 0.0005575769444270991,
      "loss": 4.7517,
      "step": 859
    },
    {
      "epoch": 0.98,
      "grad_norm": 20.397409079885954,
      "learning_rate": 0.0005574629983540534,
      "loss": 4.9617,
      "step": 860
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.305933257746408,
      "learning_rate": 0.0005573489111350369,
      "loss": 4.9096,
      "step": 861
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.3533563249940763,
      "learning_rate": 0.0005572346828325946,
      "loss": 4.7782,
      "step": 862
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.2372069206452614,
      "learning_rate": 0.0005571203135093484,
      "loss": 4.7903,
      "step": 863
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.3236152828187104,
      "learning_rate": 0.0005570058032279977,
      "loss": 4.7257,
      "step": 864
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.782334251720394,
      "learning_rate": 0.0005568911520513192,
      "loss": 4.749,
      "step": 865
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.204238082188978,
      "learning_rate": 0.0005567763600421668,
      "loss": 4.6186,
      "step": 866
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.421970105666145,
      "learning_rate": 0.0005566614272634715,
      "loss": 4.8092,
      "step": 867
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.152511146372104,
      "learning_rate": 0.0005565463537782417,
      "loss": 4.408,
      "step": 868
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.061902487254796,
      "learning_rate": 0.0005564311396495628,
      "loss": 4.7927,
      "step": 869
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.000914706693894,
      "learning_rate": 0.0005563157849405972,
      "loss": 4.8562,
      "step": 870
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.935337558021705,
      "learning_rate": 0.0005562002897145843,
      "loss": 4.8541,
      "step": 871
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.0460604942973273,
      "learning_rate": 0.0005560846540348412,
      "loss": 4.5184,
      "step": 872
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.514531003814646,
      "learning_rate": 0.0005559688779647613,
      "loss": 4.6261,
      "step": 873
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.355242705862316,
      "learning_rate": 0.0005558529615678151,
      "loss": 4.5235,
      "step": 874
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5685983195498576,
      "learning_rate": 0.0005557369049075501,
      "loss": 4.543,
      "step": 875
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1305602246095083,
      "learning_rate": 0.0005556207080475908,
      "loss": 4.7276,
      "step": 876
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.812624023379587,
      "learning_rate": 0.0005555043710516385,
      "loss": 4.3129,
      "step": 877
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.8368991001962853,
      "learning_rate": 0.0005553878939834711,
      "loss": 4.8652,
      "step": 878
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.675058269469499,
      "learning_rate": 0.0005552712769069436,
      "loss": 4.6468,
      "step": 879
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.940217263690691,
      "learning_rate": 0.0005551545198859878,
      "loss": 4.5284,
      "step": 880
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3380200472098052,
      "learning_rate": 0.0005550376229846118,
      "loss": 4.5162,
      "step": 881
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.7943092863661791,
      "learning_rate": 0.0005549205862669007,
      "loss": 4.5313,
      "step": 882
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3517825814193654,
      "learning_rate": 0.0005548034097970162,
      "loss": 4.4875,
      "step": 883
    },
    {
      "epoch": 1.01,
      "grad_norm": 3.5046003994965984,
      "learning_rate": 0.0005546860936391967,
      "loss": 4.4137,
      "step": 884
    },
    {
      "epoch": 1.01,
      "grad_norm": 2.1779626601970765,
      "learning_rate": 0.0005545686378577569,
      "loss": 4.5479,
      "step": 885
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.7561828405029052,
      "learning_rate": 0.0005544510425170885,
      "loss": 4.3525,
      "step": 886
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.2917986123962655,
      "learning_rate": 0.0005543333076816592,
      "loss": 4.8366,
      "step": 887
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.8030060265537868,
      "learning_rate": 0.0005542154334160134,
      "loss": 4.4555,
      "step": 888
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.045466852042441,
      "learning_rate": 0.0005540974197847722,
      "loss": 4.5348,
      "step": 889
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.9138919386184405,
      "learning_rate": 0.0005539792668526326,
      "loss": 4.5398,
      "step": 890
    },
    {
      "epoch": 1.02,
      "grad_norm": 2.1354413532255196,
      "learning_rate": 0.0005538609746843682,
      "loss": 4.434,
      "step": 891
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.2962485242921866,
      "learning_rate": 0.0005537425433448293,
      "loss": 4.7233,
      "step": 892
    },
    {
      "epoch": 1.02,
      "grad_norm": 2.702640807653589,
      "learning_rate": 0.0005536239728989417,
      "loss": 4.4054,
      "step": 893
    },
    {
      "epoch": 1.02,
      "grad_norm": 2.259799544711849,
      "learning_rate": 0.0005535052634117082,
      "loss": 4.5594,
      "step": 894
    },
    {
      "epoch": 1.02,
      "grad_norm": 3.49951362474867,
      "learning_rate": 0.0005533864149482074,
      "loss": 4.5235,
      "step": 895
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.7013173924549674,
      "learning_rate": 0.0005532674275735943,
      "loss": 4.5331,
      "step": 896
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.659914839210145,
      "learning_rate": 0.0005531483013531001,
      "loss": 4.3455,
      "step": 897
    },
    {
      "epoch": 1.03,
      "grad_norm": 3.7170247295286347,
      "learning_rate": 0.0005530290363520316,
      "loss": 4.7108,
      "step": 898
    },
    {
      "epoch": 1.03,
      "grad_norm": 3.3659989270849913,
      "learning_rate": 0.0005529096326357724,
      "loss": 4.3952,
      "step": 899
    },
    {
      "epoch": 1.03,
      "grad_norm": 2.3029658549539516,
      "learning_rate": 0.0005527900902697818,
      "loss": 4.4337,
      "step": 900
    },
    {
      "epoch": 1.03,
      "eval_blimp_filtered_avg": 0.5513432835820895,
      "eval_blimp_filtered_std": 0.005543231209120804,
      "step": 900
    },
    {
      "epoch": 1.03,
      "eval_blimp_supplement_avg": 0.5193965517241379,
      "eval_blimp_supplement_std": 0.02223243229499718,
      "step": 900
    },
    {
      "epoch": 1.03,
      "eval_vqa_filtered_avg": 0.35,
      "eval_vqa_filtered_std": 0.0479372485441102,
      "step": 900
    },
    {
      "epoch": 1.03,
      "eval_winoground_filtered_avg": 0.48,
      "eval_winoground_filtered_std": 0.05021167315686779,
      "step": 900
    },
    {
      "epoch": 1.03,
      "grad_norm": 30.208975867059706,
      "learning_rate": 0.000552670409319595,
      "loss": 4.7352,
      "step": 901
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.168300838753353,
      "learning_rate": 0.0005525505898508234,
      "loss": 5.5759,
      "step": 902
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.8935145281537624,
      "learning_rate": 0.0005524306319291543,
      "loss": 5.2075,
      "step": 903
    },
    {
      "epoch": 1.03,
      "grad_norm": 2.5855244713279215,
      "learning_rate": 0.0005523105356203507,
      "loss": 5.3973,
      "step": 904
    },
    {
      "epoch": 1.03,
      "grad_norm": 2.573845891636085,
      "learning_rate": 0.0005521903009902516,
      "loss": 5.4294,
      "step": 905
    },
    {
      "epoch": 1.04,
      "grad_norm": 2.6506058681087703,
      "learning_rate": 0.0005520699281047722,
      "loss": 5.3886,
      "step": 906
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.083833002217966,
      "learning_rate": 0.0005519494170299026,
      "loss": 5.1987,
      "step": 907
    },
    {
      "epoch": 1.04,
      "grad_norm": 2.337458107345143,
      "learning_rate": 0.0005518287678317096,
      "loss": 5.1824,
      "step": 908
    },
    {
      "epoch": 1.04,
      "grad_norm": 138.70356979361264,
      "learning_rate": 0.0005517079805763352,
      "loss": 5.2551,
      "step": 909
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.585243764953989,
      "learning_rate": 0.0005515870553299971,
      "loss": 5.0877,
      "step": 910
    },
    {
      "epoch": 1.04,
      "grad_norm": 8.135674924077264,
      "learning_rate": 0.000551465992158989,
      "loss": 5.5794,
      "step": 911
    },
    {
      "epoch": 1.04,
      "grad_norm": 3.389015482378525,
      "learning_rate": 0.0005513447911296796,
      "loss": 6.0232,
      "step": 912
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.849568905152137,
      "learning_rate": 0.0005512234523085139,
      "loss": 5.8414,
      "step": 913
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9990656949482055,
      "learning_rate": 0.0005511019757620119,
      "loss": 5.8587,
      "step": 914
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.345569885191008,
      "learning_rate": 0.0005509803615567693,
      "loss": 5.9227,
      "step": 915
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5618952641877304,
      "learning_rate": 0.0005508586097594573,
      "loss": 6.0792,
      "step": 916
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.3158770175624521,
      "learning_rate": 0.0005507367204368225,
      "loss": 5.7001,
      "step": 917
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.1731073465162067,
      "learning_rate": 0.0005506146936556868,
      "loss": 5.9324,
      "step": 918
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9540339883718265,
      "learning_rate": 0.0005504925294829476,
      "loss": 5.6842,
      "step": 919
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.430961627461695,
      "learning_rate": 0.0005503702279855775,
      "loss": 5.7338,
      "step": 920
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6821953616921737,
      "learning_rate": 0.0005502477892306244,
      "loss": 5.6171,
      "step": 921
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7754536364008806,
      "learning_rate": 0.0005501252132852117,
      "loss": 5.5628,
      "step": 922
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9602961477618711,
      "learning_rate": 0.0005500025002165375,
      "loss": 5.6171,
      "step": 923
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9916363444788016,
      "learning_rate": 0.0005498796500918757,
      "loss": 5.5605,
      "step": 924
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.2721100564937624,
      "learning_rate": 0.0005497566629785747,
      "loss": 5.5382,
      "step": 925
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.779774774092262,
      "learning_rate": 0.0005496335389440584,
      "loss": 5.4036,
      "step": 926
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.189913105183614,
      "learning_rate": 0.0005495102780558258,
      "loss": 5.4056,
      "step": 927
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.0134822929320961,
      "learning_rate": 0.0005493868803814507,
      "loss": 5.3807,
      "step": 928
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8704819303064539,
      "learning_rate": 0.000549263345988582,
      "loss": 5.5341,
      "step": 929
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8059298190144354,
      "learning_rate": 0.0005491396749449435,
      "loss": 5.4374,
      "step": 930
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.6904943204010585,
      "learning_rate": 0.0005490158673183339,
      "loss": 5.4564,
      "step": 931
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.5802006136391343,
      "learning_rate": 0.000548891923176627,
      "loss": 5.5058,
      "step": 932
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.5640010178765102,
      "learning_rate": 0.0005487678425877711,
      "loss": 5.5365,
      "step": 933
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.6536527164883787,
      "learning_rate": 0.0005486436256197896,
      "loss": 5.3437,
      "step": 934
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.6812923461974892,
      "learning_rate": 0.0005485192723407804,
      "loss": 5.5345,
      "step": 935
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.5803889600092739,
      "learning_rate": 0.0005483947828189164,
      "loss": 5.359,
      "step": 936
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.5620865526080171,
      "learning_rate": 0.0005482701571224449,
      "loss": 5.4258,
      "step": 937
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.5665152600534319,
      "learning_rate": 0.0005481453953196881,
      "loss": 5.4097,
      "step": 938
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.4554381218894331,
      "learning_rate": 0.0005480204974790429,
      "loss": 5.4437,
      "step": 939
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.4535771384583137,
      "learning_rate": 0.0005478954636689802,
      "loss": 5.5134,
      "step": 940
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.5198822494198634,
      "learning_rate": 0.0005477702939580462,
      "loss": 5.4077,
      "step": 941
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.6286392732786166,
      "learning_rate": 0.0005476449884148612,
      "loss": 5.5844,
      "step": 942
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.538645897339748,
      "learning_rate": 0.0005475195471081198,
      "loss": 5.2842,
      "step": 943
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.4448483241361265,
      "learning_rate": 0.0005473939701065914,
      "loss": 5.4075,
      "step": 944
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.4928477362971769,
      "learning_rate": 0.0005472682574791197,
      "loss": 5.3707,
      "step": 945
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.44884727838729455,
      "learning_rate": 0.0005471424092946224,
      "loss": 5.4251,
      "step": 946
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.45929253838373363,
      "learning_rate": 0.0005470164256220921,
      "loss": 5.4952,
      "step": 947
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.4236419429024913,
      "learning_rate": 0.0005468903065305954,
      "loss": 5.4036,
      "step": 948
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.5233565487235446,
      "learning_rate": 0.0005467640520892727,
      "loss": 5.3376,
      "step": 949
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.4329616277140555,
      "learning_rate": 0.0005466376623673395,
      "loss": 5.283,
      "step": 950
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.3667327889508278,
      "learning_rate": 0.0005465111374340847,
      "loss": 5.3586,
      "step": 951
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.4328317087537683,
      "learning_rate": 0.0005463844773588716,
      "loss": 5.461,
      "step": 952
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.46983453049818547,
      "learning_rate": 0.0005462576822111378,
      "loss": 5.324,
      "step": 953
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.4779465029832866,
      "learning_rate": 0.0005461307520603946,
      "loss": 5.3534,
      "step": 954
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.5113433793692967,
      "learning_rate": 0.0005460036869762274,
      "loss": 5.5193,
      "step": 955
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.39547383853989404,
      "learning_rate": 0.0005458764870282957,
      "loss": 5.4714,
      "step": 956
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.49252234865264993,
      "learning_rate": 0.000545749152286333,
      "loss": 5.3798,
      "step": 957
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.38514045882203557,
      "learning_rate": 0.0005456216828201463,
      "loss": 5.4192,
      "step": 958
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.4382990119053008,
      "learning_rate": 0.0005454940786996167,
      "loss": 5.5529,
      "step": 959
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.32111762981057274,
      "learning_rate": 0.0005453663399946994,
      "loss": 5.3956,
      "step": 960
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.47889088074544656,
      "learning_rate": 0.000545238466775423,
      "loss": 5.5267,
      "step": 961
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.4342888361795797,
      "learning_rate": 0.0005451104591118899,
      "loss": 5.2948,
      "step": 962
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.43507834937912576,
      "learning_rate": 0.0005449823170742763,
      "loss": 5.3297,
      "step": 963
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.4691778214607147,
      "learning_rate": 0.000544854040732832,
      "loss": 5.342,
      "step": 964
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.42373811765038427,
      "learning_rate": 0.0005447256301578807,
      "loss": 5.3019,
      "step": 965
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.47602448370485123,
      "learning_rate": 0.000544597085419819,
      "loss": 5.4233,
      "step": 966
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.4381774569483418,
      "learning_rate": 0.0005444684065891178,
      "loss": 5.3006,
      "step": 967
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.5762417968798564,
      "learning_rate": 0.0005443395937363213,
      "loss": 5.2971,
      "step": 968
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.38509568448221154,
      "learning_rate": 0.0005442106469320467,
      "loss": 5.4138,
      "step": 969
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.39923840991046794,
      "learning_rate": 0.0005440815662469855,
      "loss": 5.3991,
      "step": 970
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.4040077413261883,
      "learning_rate": 0.0005439523517519017,
      "loss": 5.4306,
      "step": 971
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.49684736549162845,
      "learning_rate": 0.0005438230035176333,
      "loss": 5.2718,
      "step": 972
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.39279382434163546,
      "learning_rate": 0.0005436935216150913,
      "loss": 5.3615,
      "step": 973
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.4561845712971142,
      "learning_rate": 0.0005435639061152602,
      "loss": 5.3471,
      "step": 974
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.4142674903649662,
      "learning_rate": 0.0005434341570891975,
      "loss": 5.2826,
      "step": 975
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.3966488199536707,
      "learning_rate": 0.000543304274608034,
      "loss": 5.222,
      "step": 976
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.3875496741056192,
      "learning_rate": 0.0005431742587429737,
      "loss": 5.373,
      "step": 977
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.37522111258147706,
      "learning_rate": 0.0005430441095652938,
      "loss": 5.3349,
      "step": 978
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.40874829397803397,
      "learning_rate": 0.0005429138271463443,
      "loss": 5.4291,
      "step": 979
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.4002631809269874,
      "learning_rate": 0.0005427834115575485,
      "loss": 5.1663,
      "step": 980
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.41485118376581664,
      "learning_rate": 0.0005426528628704026,
      "loss": 5.292,
      "step": 981
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.4224485995069427,
      "learning_rate": 0.000542522181156476,
      "loss": 5.3373,
      "step": 982
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.5036998860052495,
      "learning_rate": 0.0005423913664874105,
      "loss": 5.1369,
      "step": 983
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.6911664850605997,
      "learning_rate": 0.0005422604189349215,
      "loss": 5.0799,
      "step": 984
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.39056563564232344,
      "learning_rate": 0.0005421293385707966,
      "loss": 5.0372,
      "step": 985
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.440272074419684,
      "learning_rate": 0.0005419981254668966,
      "loss": 5.1056,
      "step": 986
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.5453022966537585,
      "learning_rate": 0.0005418667796951547,
      "loss": 5.0368,
      "step": 987
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7381615796782015,
      "learning_rate": 0.0005417353013275774,
      "loss": 5.283,
      "step": 988
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.4871024955019236,
      "learning_rate": 0.0005416036904362434,
      "loss": 5.1677,
      "step": 989
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.42346412380114135,
      "learning_rate": 0.0005414719470933042,
      "loss": 5.172,
      "step": 990
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.39785046165691595,
      "learning_rate": 0.0005413400713709841,
      "loss": 5.075,
      "step": 991
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.4597919559256249,
      "learning_rate": 0.0005412080633415796,
      "loss": 5.0261,
      "step": 992
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.48832401930097885,
      "learning_rate": 0.0005410759230774602,
      "loss": 5.162,
      "step": 993
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.44151080223688743,
      "learning_rate": 0.0005409436506510673,
      "loss": 5.2777,
      "step": 994
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.5070336368708293,
      "learning_rate": 0.0005408112461349154,
      "loss": 5.0851,
      "step": 995
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.4072520725175822,
      "learning_rate": 0.0005406787096015909,
      "loss": 5.2743,
      "step": 996
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.37469890432209424,
      "learning_rate": 0.0005405460411237527,
      "loss": 5.1918,
      "step": 997
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.5508067288261687,
      "learning_rate": 0.0005404132407741324,
      "loss": 5.1668,
      "step": 998
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.45907921678099606,
      "learning_rate": 0.0005402803086255334,
      "loss": 5.1898,
      "step": 999
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.43339870809685505,
      "learning_rate": 0.0005401472447508316,
      "loss": 5.1038,
      "step": 1000
    },
    {
      "epoch": 1.14,
      "eval_blimp_filtered_avg": 0.5517910447761194,
      "eval_blimp_filtered_std": 0.0054038062867259986,
      "step": 1000
    },
    {
      "epoch": 1.14,
      "eval_blimp_supplement_avg": 0.5021551724137931,
      "eval_blimp_supplement_std": 0.022915642200552703,
      "step": 1000
    },
    {
      "epoch": 1.14,
      "eval_vqa_filtered_avg": 0.32,
      "eval_vqa_filtered_std": 0.046882617226215034,
      "step": 1000
    },
    {
      "epoch": 1.14,
      "eval_winoground_filtered_avg": 0.62,
      "eval_winoground_filtered_std": 0.048783173121456316,
      "step": 1000
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.561893703164224,
      "learning_rate": 0.0005400140492229751,
      "loss": 5.098,
      "step": 1001
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0296193165838514,
      "learning_rate": 0.0005398807221149841,
      "loss": 5.2231,
      "step": 1002
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.3614293013467432,
      "learning_rate": 0.0005397472634999511,
      "loss": 5.3029,
      "step": 1003
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.3690829892552953,
      "learning_rate": 0.0005396136734510405,
      "loss": 5.1894,
      "step": 1004
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.48725399253717033,
      "learning_rate": 0.0005394799520414887,
      "loss": 5.2917,
      "step": 1005
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.4440608159355949,
      "learning_rate": 0.0005393460993446044,
      "loss": 5.099,
      "step": 1006
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.3287981011951608,
      "learning_rate": 0.0005392121154337681,
      "loss": 5.2243,
      "step": 1007
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.40426056123389564,
      "learning_rate": 0.0005390780003824322,
      "loss": 5.1458,
      "step": 1008
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.43324452593595014,
      "learning_rate": 0.0005389437542641211,
      "loss": 5.1089,
      "step": 1009
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.3934412888466023,
      "learning_rate": 0.000538809377152431,
      "loss": 5.249,
      "step": 1010
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.43515186525022376,
      "learning_rate": 0.00053867486912103,
      "loss": 5.189,
      "step": 1011
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.41632322727536497,
      "learning_rate": 0.0005385402302436577,
      "loss": 5.2967,
      "step": 1012
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.4233719150686152,
      "learning_rate": 0.0005384054605941257,
      "loss": 5.0676,
      "step": 1013
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.4138163754981368,
      "learning_rate": 0.0005382705602463172,
      "loss": 5.1771,
      "step": 1014
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.4344236088728919,
      "learning_rate": 0.0005381355292741872,
      "loss": 5.07,
      "step": 1015
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.3720724280629541,
      "learning_rate": 0.000538000367751762,
      "loss": 5.1999,
      "step": 1016
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.4388187206521687,
      "learning_rate": 0.0005378650757531398,
      "loss": 5.1968,
      "step": 1017
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.42228607254537337,
      "learning_rate": 0.0005377296533524899,
      "loss": 5.3091,
      "step": 1018
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.4223931884826277,
      "learning_rate": 0.0005375941006240536,
      "loss": 5.2244,
      "step": 1019
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.38467768367623273,
      "learning_rate": 0.0005374584176421434,
      "loss": 4.998,
      "step": 1020
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.5121740364456557,
      "learning_rate": 0.0005373226044811431,
      "loss": 5.0163,
      "step": 1021
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.4071237000591364,
      "learning_rate": 0.0005371866612155081,
      "loss": 5.067,
      "step": 1022
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.40043372174867475,
      "learning_rate": 0.0005370505879197648,
      "loss": 5.0916,
      "step": 1023
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.41700300758287395,
      "learning_rate": 0.0005369143846685113,
      "loss": 5.2139,
      "step": 1024
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.5689300660937899,
      "learning_rate": 0.0005367780515364169,
      "loss": 5.2711,
      "step": 1025
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.4435024728784905,
      "learning_rate": 0.0005366415885982215,
      "loss": 5.2629,
      "step": 1026
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.3848453877625384,
      "learning_rate": 0.0005365049959287369,
      "loss": 5.1593,
      "step": 1027
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.4351079424513521,
      "learning_rate": 0.0005363682736028456,
      "loss": 5.1141,
      "step": 1028
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.4106272769561721,
      "learning_rate": 0.0005362314216955014,
      "loss": 5.2094,
      "step": 1029
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.40084898571029265,
      "learning_rate": 0.0005360944402817289,
      "loss": 5.2634,
      "step": 1030
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.5247165125294693,
      "learning_rate": 0.0005359573294366241,
      "loss": 5.2283,
      "step": 1031
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.4449757409940425,
      "learning_rate": 0.0005358200892353535,
      "loss": 5.3638,
      "step": 1032
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.457256484491978,
      "learning_rate": 0.0005356827197531546,
      "loss": 5.3192,
      "step": 1033
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.39482283802731993,
      "learning_rate": 0.0005355452210653363,
      "loss": 5.1406,
      "step": 1034
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.416256348753842,
      "learning_rate": 0.0005354075932472775,
      "loss": 5.2381,
      "step": 1035
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.4058913011619471,
      "learning_rate": 0.0005352698363744285,
      "loss": 5.0604,
      "step": 1036
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.38800126651414524,
      "learning_rate": 0.0005351319505223101,
      "loss": 5.1878,
      "step": 1037
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.43945411448951976,
      "learning_rate": 0.0005349939357665139,
      "loss": 5.108,
      "step": 1038
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.41254712172825325,
      "learning_rate": 0.0005348557921827021,
      "loss": 5.1048,
      "step": 1039
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.43600766523554596,
      "learning_rate": 0.0005347175198466077,
      "loss": 5.2534,
      "step": 1040
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.5096982108250583,
      "learning_rate": 0.0005345791188340341,
      "loss": 5.2208,
      "step": 1041
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.3678708913414918,
      "learning_rate": 0.000534440589220855,
      "loss": 5.2111,
      "step": 1042
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.39416625832123875,
      "learning_rate": 0.0005343019310830153,
      "loss": 4.9403,
      "step": 1043
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.441812212371377,
      "learning_rate": 0.0005341631444965298,
      "loss": 5.0192,
      "step": 1044
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.4664726864030427,
      "learning_rate": 0.0005340242295374839,
      "loss": 5.1739,
      "step": 1045
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.5130144574780617,
      "learning_rate": 0.0005338851862820332,
      "loss": 5.1567,
      "step": 1046
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.47307547046601034,
      "learning_rate": 0.0005337460148064039,
      "loss": 5.3359,
      "step": 1047
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.4050663421380887,
      "learning_rate": 0.0005336067151868924,
      "loss": 5.3024,
      "step": 1048
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.4042550322657816,
      "learning_rate": 0.0005334672874998652,
      "loss": 5.152,
      "step": 1049
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.4328215366515683,
      "learning_rate": 0.0005333277318217592,
      "loss": 5.1895,
      "step": 1050
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.3710840385372714,
      "learning_rate": 0.0005331880482290815,
      "loss": 5.0731,
      "step": 1051
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.43238194363062804,
      "learning_rate": 0.0005330482367984092,
      "loss": 5.1334,
      "step": 1052
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.4769103042422726,
      "learning_rate": 0.0005329082976063893,
      "loss": 5.3516,
      "step": 1053
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.559016146728873,
      "learning_rate": 0.0005327682307297392,
      "loss": 5.3012,
      "step": 1054
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.48843851110696007,
      "learning_rate": 0.0005326280362452463,
      "loss": 5.0866,
      "step": 1055
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.40017829206814876,
      "learning_rate": 0.0005324877142297676,
      "loss": 5.0901,
      "step": 1056
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9415715783550416,
      "learning_rate": 0.0005323472647602302,
      "loss": 5.2464,
      "step": 1057
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.43328766131483776,
      "learning_rate": 0.0005322066879136312,
      "loss": 5.1957,
      "step": 1058
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7555414193547236,
      "learning_rate": 0.0005320659837670374,
      "loss": 5.1849,
      "step": 1059
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.6087953298003287,
      "learning_rate": 0.0005319251523975854,
      "loss": 5.3056,
      "step": 1060
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9355953042873021,
      "learning_rate": 0.0005317841938824815,
      "loss": 5.2214,
      "step": 1061
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.6018912899004875,
      "learning_rate": 0.0005316431082990019,
      "loss": 5.2228,
      "step": 1062
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7489310809591273,
      "learning_rate": 0.0005315018957244922,
      "loss": 5.3824,
      "step": 1063
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.594942303799756,
      "learning_rate": 0.0005313605562363678,
      "loss": 5.1788,
      "step": 1064
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.5581498152103808,
      "learning_rate": 0.0005312190899121135,
      "loss": 5.1229,
      "step": 1065
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.76101456135337,
      "learning_rate": 0.000531077496829284,
      "loss": 5.2446,
      "step": 1066
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.46515444374074355,
      "learning_rate": 0.000530935777065503,
      "loss": 5.2714,
      "step": 1067
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8552880602797566,
      "learning_rate": 0.000530793930698464,
      "loss": 5.0353,
      "step": 1068
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.4566651456802952,
      "learning_rate": 0.0005306519578059299,
      "loss": 5.0689,
      "step": 1069
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.8450954033538456,
      "learning_rate": 0.0005305098584657326,
      "loss": 5.1349,
      "step": 1070
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8344903563315349,
      "learning_rate": 0.0005303676327557738,
      "loss": 5.2492,
      "step": 1071
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0529220822832457,
      "learning_rate": 0.0005302252807540241,
      "loss": 5.1619,
      "step": 1072
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7847508089507222,
      "learning_rate": 0.0005300828025385237,
      "loss": 5.0921,
      "step": 1073
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.5765721281977121,
      "learning_rate": 0.0005299401981873817,
      "loss": 5.0825,
      "step": 1074
    },
    {
      "epoch": 1.23,
      "grad_norm": 2.8209609810367398,
      "learning_rate": 0.0005297974677787764,
      "loss": 5.1608,
      "step": 1075
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7479079624234641,
      "learning_rate": 0.0005296546113909552,
      "loss": 5.1184,
      "step": 1076
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.980141376444674,
      "learning_rate": 0.0005295116291022346,
      "loss": 5.3642,
      "step": 1077
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.1246465102398895,
      "learning_rate": 0.0005293685209910002,
      "loss": 5.179,
      "step": 1078
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8278087361805492,
      "learning_rate": 0.0005292252871357066,
      "loss": 5.2764,
      "step": 1079
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.5433776067573339,
      "learning_rate": 0.0005290819276148772,
      "loss": 5.1895,
      "step": 1080
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.776020564990142,
      "learning_rate": 0.0005289384425071044,
      "loss": 5.1853,
      "step": 1081
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.2909933279579868,
      "learning_rate": 0.000528794831891049,
      "loss": 5.0572,
      "step": 1082
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.065713103005686,
      "learning_rate": 0.0005286510958454414,
      "loss": 5.2095,
      "step": 1083
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.42218422264035027,
      "learning_rate": 0.0005285072344490803,
      "loss": 4.9899,
      "step": 1084
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.6041649032325777,
      "learning_rate": 0.000528363247780833,
      "loss": 5.1199,
      "step": 1085
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7779264492556021,
      "learning_rate": 0.0005282191359196357,
      "loss": 5.2199,
      "step": 1086
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7844203663727362,
      "learning_rate": 0.0005280748989444934,
      "loss": 5.1673,
      "step": 1087
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.5231115030213644,
      "learning_rate": 0.0005279305369344792,
      "loss": 5.05,
      "step": 1088
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.516594230724654,
      "learning_rate": 0.0005277860499687351,
      "loss": 5.2657,
      "step": 1089
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.5503100793018348,
      "learning_rate": 0.0005276414381264715,
      "loss": 5.1392,
      "step": 1090
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7698087590065096,
      "learning_rate": 0.0005274967014869675,
      "loss": 5.289,
      "step": 1091
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8246697311616605,
      "learning_rate": 0.0005273518401295702,
      "loss": 5.1584,
      "step": 1092
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.5362423575948931,
      "learning_rate": 0.0005272068541336952,
      "loss": 5.0175,
      "step": 1093
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.4853677333920353,
      "learning_rate": 0.0005270617435788268,
      "loss": 5.2061,
      "step": 1094
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.6987958342744579,
      "learning_rate": 0.0005269165085445169,
      "loss": 5.1275,
      "step": 1095
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7670909887700316,
      "learning_rate": 0.0005267711491103864,
      "loss": 4.9837,
      "step": 1096
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.46236244350157996,
      "learning_rate": 0.0005266256653561237,
      "loss": 5.1815,
      "step": 1097
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7110860064589077,
      "learning_rate": 0.0005264800573614859,
      "loss": 5.1434,
      "step": 1098
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9209883025218306,
      "learning_rate": 0.0005263343252062978,
      "loss": 5.0499,
      "step": 1099
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.546314212803252,
      "learning_rate": 0.0005261884689704527,
      "loss": 5.1397,
      "step": 1100
    },
    {
      "epoch": 1.26,
      "eval_blimp_filtered_avg": 0.5491044776119403,
      "eval_blimp_filtered_std": 0.005438878652849802,
      "step": 1100
    },
    {
      "epoch": 1.26,
      "eval_blimp_supplement_avg": 0.5193965517241379,
      "eval_blimp_supplement_std": 0.022485938205597543,
      "step": 1100
    },
    {
      "epoch": 1.26,
      "eval_vqa_filtered_avg": 0.33,
      "eval_vqa_filtered_std": 0.04725815626252606,
      "step": 1100
    },
    {
      "epoch": 1.26,
      "eval_winoground_filtered_avg": 0.47,
      "eval_winoground_filtered_std": 0.05016135580465919,
      "step": 1100
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7388797308847014,
      "learning_rate": 0.0005260424887339114,
      "loss": 5.1166,
      "step": 1101
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.551647769812749,
      "learning_rate": 0.0005258963845767034,
      "loss": 5.1336,
      "step": 1102
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7669349059020705,
      "learning_rate": 0.0005257501565789252,
      "loss": 4.9917,
      "step": 1103
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.054321445801503,
      "learning_rate": 0.000525603804820742,
      "loss": 4.9136,
      "step": 1104
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7449588210770962,
      "learning_rate": 0.0005254573293823864,
      "loss": 4.9924,
      "step": 1105
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.798373357755002,
      "learning_rate": 0.000525310730344159,
      "loss": 5.0163,
      "step": 1106
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.446993844530049,
      "learning_rate": 0.0005251640077864281,
      "loss": 5.0281,
      "step": 1107
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8892870322040823,
      "learning_rate": 0.0005250171617896297,
      "loss": 4.9492,
      "step": 1108
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.39901731637201354,
      "learning_rate": 0.0005248701924342675,
      "loss": 4.9201,
      "step": 1109
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.45972349421204234,
      "learning_rate": 0.0005247230998009127,
      "loss": 5.1685,
      "step": 1110
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9634401138068055,
      "learning_rate": 0.0005245758839702044,
      "loss": 5.0518,
      "step": 1111
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.440466459676811,
      "learning_rate": 0.000524428545022849,
      "loss": 5.0364,
      "step": 1112
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.7693820096440174,
      "learning_rate": 0.0005242810830396202,
      "loss": 5.1071,
      "step": 1113
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.1038071847876403,
      "learning_rate": 0.0005241334981013597,
      "loss": 5.0515,
      "step": 1114
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.1143026568331607,
      "learning_rate": 0.0005239857902889761,
      "loss": 4.9906,
      "step": 1115
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.563283524456605,
      "learning_rate": 0.0005238379596834455,
      "loss": 5.0492,
      "step": 1116
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6583808078563582,
      "learning_rate": 0.0005236900063658117,
      "loss": 5.237,
      "step": 1117
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8356491269305153,
      "learning_rate": 0.0005235419304171849,
      "loss": 5.1512,
      "step": 1118
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1297988268249588,
      "learning_rate": 0.0005233937319187435,
      "loss": 5.0435,
      "step": 1119
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.6528510026862676,
      "learning_rate": 0.0005232454109517326,
      "loss": 5.0446,
      "step": 1120
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.4819883250572987,
      "learning_rate": 0.0005230969675974644,
      "loss": 4.9617,
      "step": 1121
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.5205162404324648,
      "learning_rate": 0.0005229484019373182,
      "loss": 5.0766,
      "step": 1122
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9835362356962077,
      "learning_rate": 0.0005227997140527408,
      "loss": 4.9259,
      "step": 1123
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8178215553146111,
      "learning_rate": 0.0005226509040252451,
      "loss": 4.9575,
      "step": 1124
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0049795012814347,
      "learning_rate": 0.0005225019719364119,
      "loss": 5.0462,
      "step": 1125
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7621761099355314,
      "learning_rate": 0.0005223529178678884,
      "loss": 4.9663,
      "step": 1126
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8674700161516951,
      "learning_rate": 0.0005222037419013889,
      "loss": 4.9741,
      "step": 1127
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8917539173895458,
      "learning_rate": 0.0005220544441186942,
      "loss": 4.921,
      "step": 1128
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7421166579577743,
      "learning_rate": 0.0005219050246016523,
      "loss": 5.0439,
      "step": 1129
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8352406913970326,
      "learning_rate": 0.0005217554834321777,
      "loss": 4.9815,
      "step": 1130
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.510072262676206,
      "learning_rate": 0.0005216058206922515,
      "loss": 5.0424,
      "step": 1131
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8521222428339478,
      "learning_rate": 0.0005214560364639218,
      "loss": 4.9688,
      "step": 1132
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.4695137031523553,
      "learning_rate": 0.000521306130829303,
      "loss": 5.0004,
      "step": 1133
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9496825122600068,
      "learning_rate": 0.0005211561038705759,
      "loss": 5.0936,
      "step": 1134
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.6639662429669891,
      "learning_rate": 0.0005210059556699884,
      "loss": 4.9557,
      "step": 1135
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.6514236282093878,
      "learning_rate": 0.0005208556863098544,
      "loss": 4.7878,
      "step": 1136
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0286651488045517,
      "learning_rate": 0.0005207052958725542,
      "loss": 5.2284,
      "step": 1137
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.39673992022551086,
      "learning_rate": 0.0005205547844405348,
      "loss": 5.1993,
      "step": 1138
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0415379274760892,
      "learning_rate": 0.0005204041520963092,
      "loss": 4.9702,
      "step": 1139
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.4958305644905204,
      "learning_rate": 0.0005202533989224568,
      "loss": 5.2672,
      "step": 1140
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.4776426094183519,
      "learning_rate": 0.0005201025250016236,
      "loss": 4.9494,
      "step": 1141
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.6409779908552264,
      "learning_rate": 0.0005199515304165211,
      "loss": 4.9546,
      "step": 1142
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.5864650967387575,
      "learning_rate": 0.0005198004152499275,
      "loss": 5.0828,
      "step": 1143
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8344509195774621,
      "learning_rate": 0.0005196491795846868,
      "loss": 4.8471,
      "step": 1144
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.1317193855471057,
      "learning_rate": 0.0005194978235037094,
      "loss": 5.0192,
      "step": 1145
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.45619035158450266,
      "learning_rate": 0.0005193463470899713,
      "loss": 4.9606,
      "step": 1146
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8584507064489265,
      "learning_rate": 0.0005191947504265147,
      "loss": 5.1415,
      "step": 1147
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.4652728499512511,
      "learning_rate": 0.0005190430335964478,
      "loss": 4.9464,
      "step": 1148
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9712252813627654,
      "learning_rate": 0.0005188911966829445,
      "loss": 4.857,
      "step": 1149
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.6930626494864487,
      "learning_rate": 0.0005187392397692447,
      "loss": 4.8826,
      "step": 1150
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.573206615219214,
      "learning_rate": 0.0005185871629386537,
      "loss": 4.9485,
      "step": 1151
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.6511105787127845,
      "learning_rate": 0.000518434966274543,
      "loss": 5.0681,
      "step": 1152
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.5545478258381884,
      "learning_rate": 0.0005182826498603497,
      "loss": 5.0605,
      "step": 1153
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8252004102443666,
      "learning_rate": 0.0005181302137795764,
      "loss": 4.8447,
      "step": 1154
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.6473725755168579,
      "learning_rate": 0.0005179776581157914,
      "loss": 5.0475,
      "step": 1155
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.48383575322397915,
      "learning_rate": 0.0005178249829526286,
      "loss": 5.0081,
      "step": 1156
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.6016024293557599,
      "learning_rate": 0.0005176721883737871,
      "loss": 4.7762,
      "step": 1157
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.6948490952471152,
      "learning_rate": 0.000517519274463032,
      "loss": 4.932,
      "step": 1158
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7184347826556939,
      "learning_rate": 0.0005173662413041932,
      "loss": 5.0516,
      "step": 1159
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.6232110017312381,
      "learning_rate": 0.0005172130889811667,
      "loss": 4.8772,
      "step": 1160
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.13915488563697,
      "learning_rate": 0.0005170598175779131,
      "loss": 5.0593,
      "step": 1161
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.5733625966435614,
      "learning_rate": 0.0005169064271784589,
      "loss": 4.8745,
      "step": 1162
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.2201018034431876,
      "learning_rate": 0.0005167529178668952,
      "loss": 4.9017,
      "step": 1163
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7347307830627611,
      "learning_rate": 0.000516599289727379,
      "loss": 5.1694,
      "step": 1164
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.6475265633360185,
      "learning_rate": 0.0005164455428441318,
      "loss": 4.9257,
      "step": 1165
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.5842273018428894,
      "learning_rate": 0.0005162916773014406,
      "loss": 4.8271,
      "step": 1166
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.6407435764808377,
      "learning_rate": 0.0005161376931836573,
      "loss": 5.0074,
      "step": 1167
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8854846808554024,
      "learning_rate": 0.000515983590575199,
      "loss": 4.8804,
      "step": 1168
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.5298703149937839,
      "learning_rate": 0.0005158293695605474,
      "loss": 5.0125,
      "step": 1169
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0948542180653988,
      "learning_rate": 0.0005156750302242495,
      "loss": 4.908,
      "step": 1170
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.5636888837917247,
      "learning_rate": 0.0005155205726509167,
      "loss": 4.8616,
      "step": 1171
    },
    {
      "epoch": 1.34,
      "grad_norm": 2.1187701807239594,
      "learning_rate": 0.0005153659969252257,
      "loss": 5.1501,
      "step": 1172
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.7518600969040812,
      "learning_rate": 0.0005152113031319176,
      "loss": 4.9481,
      "step": 1173
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.3365198206110314,
      "learning_rate": 0.0005150564913557987,
      "loss": 4.9139,
      "step": 1174
    },
    {
      "epoch": 1.34,
      "grad_norm": 2.315331396497301,
      "learning_rate": 0.0005149015616817392,
      "loss": 4.9212,
      "step": 1175
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.5954872487348631,
      "learning_rate": 0.000514746514194675,
      "loss": 4.9505,
      "step": 1176
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0918409782012397,
      "learning_rate": 0.0005145913489796054,
      "loss": 4.9391,
      "step": 1177
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.41015411682616726,
      "learning_rate": 0.0005144360661215953,
      "loss": 4.9366,
      "step": 1178
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0424835235782717,
      "learning_rate": 0.0005142806657057735,
      "loss": 4.9783,
      "step": 1179
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.4191377513358202,
      "learning_rate": 0.0005141251478173332,
      "loss": 4.7821,
      "step": 1180
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.758495752798407,
      "learning_rate": 0.0005139695125415323,
      "loss": 4.7929,
      "step": 1181
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.4682771067503924,
      "learning_rate": 0.000513813759963693,
      "loss": 4.981,
      "step": 1182
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.5402319032702916,
      "learning_rate": 0.0005136578901692016,
      "loss": 4.9498,
      "step": 1183
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.5472569612557033,
      "learning_rate": 0.000513501903243509,
      "loss": 4.7101,
      "step": 1184
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.5823065367942866,
      "learning_rate": 0.0005133457992721299,
      "loss": 5.0122,
      "step": 1185
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.5158509450320617,
      "learning_rate": 0.0005131895783406435,
      "loss": 4.9891,
      "step": 1186
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.5200517975373709,
      "learning_rate": 0.000513033240534693,
      "loss": 4.8529,
      "step": 1187
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.5123793020328804,
      "learning_rate": 0.0005128767859399857,
      "loss": 4.9624,
      "step": 1188
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.40935990902343433,
      "learning_rate": 0.0005127202146422927,
      "loss": 4.8945,
      "step": 1189
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.658003655704528,
      "learning_rate": 0.0005125635267274496,
      "loss": 4.8389,
      "step": 1190
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.5595005191066111,
      "learning_rate": 0.0005124067222813552,
      "loss": 4.9355,
      "step": 1191
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.47186210492967656,
      "learning_rate": 0.000512249801389973,
      "loss": 4.7485,
      "step": 1192
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.4428419171738912,
      "learning_rate": 0.0005120927641393296,
      "loss": 4.7719,
      "step": 1193
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.4327517902953713,
      "learning_rate": 0.0005119356106155161,
      "loss": 4.9128,
      "step": 1194
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.5198701019397242,
      "learning_rate": 0.0005117783409046866,
      "loss": 5.0118,
      "step": 1195
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.5400972155378008,
      "learning_rate": 0.0005116209550930593,
      "loss": 5.0221,
      "step": 1196
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.6193886973040498,
      "learning_rate": 0.0005114634532669163,
      "loss": 4.992,
      "step": 1197
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.45108723732597616,
      "learning_rate": 0.0005113058355126027,
      "loss": 4.9304,
      "step": 1198
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9877624528823103,
      "learning_rate": 0.0005111481019165275,
      "loss": 4.8404,
      "step": 1199
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.41548693451522434,
      "learning_rate": 0.0005109902525651635,
      "loss": 4.888,
      "step": 1200
    },
    {
      "epoch": 1.37,
      "eval_blimp_filtered_avg": 0.5583582089552239,
      "eval_blimp_filtered_std": 0.00544020239367129,
      "step": 1200
    },
    {
      "epoch": 1.37,
      "eval_blimp_supplement_avg": 0.5172413793103449,
      "eval_blimp_supplement_std": 0.022523037230137665,
      "step": 1200
    },
    {
      "epoch": 1.37,
      "eval_vqa_filtered_avg": 0.27,
      "eval_vqa_filtered_std": 0.04461960433384741,
      "step": 1200
    },
    {
      "epoch": 1.37,
      "eval_winoground_filtered_avg": 0.51,
      "eval_winoground_filtered_std": 0.05024183937956913,
      "step": 1200
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9957865258578577,
      "learning_rate": 0.0005108322875450462,
      "loss": 4.8594,
      "step": 1201
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.6637565188705488,
      "learning_rate": 0.0005106742069427752,
      "loss": 5.0869,
      "step": 1202
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9651303579560007,
      "learning_rate": 0.0005105160108450131,
      "loss": 4.925,
      "step": 1203
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7578337233488569,
      "learning_rate": 0.0005103576993384859,
      "loss": 5.083,
      "step": 1204
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.6424627149000028,
      "learning_rate": 0.0005101992725099829,
      "loss": 5.014,
      "step": 1205
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7018920473198439,
      "learning_rate": 0.0005100407304463568,
      "loss": 4.9008,
      "step": 1206
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.6901004307853441,
      "learning_rate": 0.0005098820732345232,
      "loss": 4.8828,
      "step": 1207
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.339425812065133,
      "learning_rate": 0.0005097233009614606,
      "loss": 4.799,
      "step": 1208
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8226150980471355,
      "learning_rate": 0.0005095644137142113,
      "loss": 4.9402,
      "step": 1209
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.614965173080206,
      "learning_rate": 0.0005094054115798802,
      "loss": 4.8452,
      "step": 1210
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8738358304928479,
      "learning_rate": 0.0005092462946456348,
      "loss": 4.9363,
      "step": 1211
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.5182861711293806,
      "learning_rate": 0.0005090870629987064,
      "loss": 4.795,
      "step": 1212
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9607883807950071,
      "learning_rate": 0.0005089277167263886,
      "loss": 4.8792,
      "step": 1213
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.4876402318685862,
      "learning_rate": 0.0005087682559160378,
      "loss": 4.7971,
      "step": 1214
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0754644739650243,
      "learning_rate": 0.0005086086806550734,
      "loss": 4.8393,
      "step": 1215
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.5238109834341227,
      "learning_rate": 0.0005084489910309778,
      "loss": 4.9783,
      "step": 1216
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0266540298506723,
      "learning_rate": 0.0005082891871312955,
      "loss": 4.8608,
      "step": 1217
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7944673287759421,
      "learning_rate": 0.0005081292690436339,
      "loss": 4.9533,
      "step": 1218
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.5485454490627911,
      "learning_rate": 0.0005079692368556634,
      "loss": 4.6756,
      "step": 1219
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9326840098046364,
      "learning_rate": 0.0005078090906551162,
      "loss": 4.72,
      "step": 1220
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.47197634196124005,
      "learning_rate": 0.0005076488305297877,
      "loss": 4.7063,
      "step": 1221
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.4749222626088472,
      "learning_rate": 0.0005074884565675354,
      "loss": 4.9344,
      "step": 1222
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.431361413407069,
      "learning_rate": 0.0005073279688562792,
      "loss": 5.0848,
      "step": 1223
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7334212657567057,
      "learning_rate": 0.0005071673674840013,
      "loss": 4.948,
      "step": 1224
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.950154902108328,
      "learning_rate": 0.0005070066525387466,
      "loss": 4.955,
      "step": 1225
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7944280004669937,
      "learning_rate": 0.0005068458241086219,
      "loss": 4.9727,
      "step": 1226
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.2139408258615942,
      "learning_rate": 0.0005066848822817963,
      "loss": 5.0231,
      "step": 1227
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.5944887883002979,
      "learning_rate": 0.0005065238271465011,
      "loss": 4.7285,
      "step": 1228
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.6876835801730448,
      "learning_rate": 0.0005063626587910297,
      "loss": 4.9803,
      "step": 1229
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0795092472889753,
      "learning_rate": 0.0005062013773037374,
      "loss": 4.6979,
      "step": 1230
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8395141578376858,
      "learning_rate": 0.0005060399827730419,
      "loss": 5.0304,
      "step": 1231
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.47915198001552256,
      "learning_rate": 0.0005058784752874227,
      "loss": 4.9636,
      "step": 1232
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.7573407152077023,
      "learning_rate": 0.000505716854935421,
      "loss": 5.1032,
      "step": 1233
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0911197864536943,
      "learning_rate": 0.0005055551218056402,
      "loss": 4.8738,
      "step": 1234
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.2156853393203244,
      "learning_rate": 0.0005053932759867452,
      "loss": 4.8411,
      "step": 1235
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8730167999221258,
      "learning_rate": 0.0005052313175674632,
      "loss": 4.8779,
      "step": 1236
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1565281591282393,
      "learning_rate": 0.0005050692466365826,
      "loss": 4.9963,
      "step": 1237
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9156291269397416,
      "learning_rate": 0.0005049070632829536,
      "loss": 4.8102,
      "step": 1238
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1502784426219956,
      "learning_rate": 0.0005047447675954882,
      "loss": 4.8999,
      "step": 1239
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7639829592142456,
      "learning_rate": 0.00050458235966316,
      "loss": 4.7835,
      "step": 1240
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.6237974884676651,
      "learning_rate": 0.0005044198395750037,
      "loss": 4.956,
      "step": 1241
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.4339224680947942,
      "learning_rate": 0.0005042572074201163,
      "loss": 4.7663,
      "step": 1242
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9249375634005634,
      "learning_rate": 0.0005040944632876552,
      "loss": 4.8611,
      "step": 1243
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.128083878891834,
      "learning_rate": 0.0005039316072668402,
      "loss": 4.8098,
      "step": 1244
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0841392883389345,
      "learning_rate": 0.0005037686394469517,
      "loss": 4.9768,
      "step": 1245
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0073590112013175,
      "learning_rate": 0.0005036055599173318,
      "loss": 4.8392,
      "step": 1246
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9717709390098217,
      "learning_rate": 0.0005034423687673836,
      "loss": 4.8141,
      "step": 1247
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7780132618526815,
      "learning_rate": 0.0005032790660865718,
      "loss": 4.8616,
      "step": 1248
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8315162825271477,
      "learning_rate": 0.0005031156519644214,
      "loss": 4.6124,
      "step": 1249
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.4548498826413443,
      "learning_rate": 0.0005029521264905196,
      "loss": 4.9729,
      "step": 1250
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7893000678422016,
      "learning_rate": 0.0005027884897545139,
      "loss": 4.7712,
      "step": 1251
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.4082461493930724,
      "learning_rate": 0.0005026247418461128,
      "loss": 4.855,
      "step": 1252
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7963173493176894,
      "learning_rate": 0.0005024608828550861,
      "loss": 4.8843,
      "step": 1253
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.4481474948040786,
      "learning_rate": 0.0005022969128712641,
      "loss": 4.7978,
      "step": 1254
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.49120812105606754,
      "learning_rate": 0.0005021328319845385,
      "loss": 4.7432,
      "step": 1255
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.5851131275106991,
      "learning_rate": 0.0005019686402848614,
      "loss": 4.8114,
      "step": 1256
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.63304651258489,
      "learning_rate": 0.0005018043378622456,
      "loss": 4.8111,
      "step": 1257
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.6314754781660858,
      "learning_rate": 0.0005016399248067647,
      "loss": 4.7954,
      "step": 1258
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.5366770227908617,
      "learning_rate": 0.0005014754012085531,
      "loss": 4.9664,
      "step": 1259
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0991363018408031,
      "learning_rate": 0.0005013107671578056,
      "loss": 4.7331,
      "step": 1260
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7351901258274723,
      "learning_rate": 0.0005011460227447776,
      "loss": 4.6044,
      "step": 1261
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.4754451286872485,
      "learning_rate": 0.0005009811680597852,
      "loss": 4.8287,
      "step": 1262
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.2090020124165997,
      "learning_rate": 0.0005008162031932046,
      "loss": 4.6633,
      "step": 1263
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.367965450960771,
      "learning_rate": 0.0005006511282354727,
      "loss": 4.8497,
      "step": 1264
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.4029800100559322,
      "learning_rate": 0.0005004859432770865,
      "loss": 4.6876,
      "step": 1265
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.061469789627048,
      "learning_rate": 0.0005003206484086035,
      "loss": 4.7985,
      "step": 1266
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.5892286372984972,
      "learning_rate": 0.0005001552437206415,
      "loss": 4.7692,
      "step": 1267
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.5439740856881324,
      "learning_rate": 0.0004999897293038783,
      "loss": 4.6833,
      "step": 1268
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.492458645673748,
      "learning_rate": 0.0004998241052490519,
      "loss": 4.887,
      "step": 1269
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.45413512260815747,
      "learning_rate": 0.0004996583716469608,
      "loss": 4.6017,
      "step": 1270
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8283882911797816,
      "learning_rate": 0.0004994925285884627,
      "loss": 4.6879,
      "step": 1271
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.639335114117073,
      "learning_rate": 0.0004993265761644764,
      "loss": 4.6942,
      "step": 1272
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.5206270200592327,
      "learning_rate": 0.0004991605144659797,
      "loss": 4.8212,
      "step": 1273
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.6901427967060859,
      "learning_rate": 0.0004989943435840107,
      "loss": 4.6872,
      "step": 1274
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.6603345508535575,
      "learning_rate": 0.0004988280636096675,
      "loss": 4.6591,
      "step": 1275
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.5730093528645751,
      "learning_rate": 0.0004986616746341078,
      "loss": 4.6447,
      "step": 1276
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.42362682469383894,
      "learning_rate": 0.000498495176748549,
      "loss": 4.6906,
      "step": 1277
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.48937732940455675,
      "learning_rate": 0.0004983285700442684,
      "loss": 4.7235,
      "step": 1278
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.6884060486082092,
      "learning_rate": 0.000498161854612603,
      "loss": 4.7805,
      "step": 1279
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.5310802121161438,
      "learning_rate": 0.000497995030544949,
      "loss": 4.741,
      "step": 1280
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1227830939190784,
      "learning_rate": 0.0004978280979327628,
      "loss": 4.7948,
      "step": 1281
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.6429077618547205,
      "learning_rate": 0.0004976610568675596,
      "loss": 4.7321,
      "step": 1282
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.3059947716068132,
      "learning_rate": 0.0004974939074409145,
      "loss": 4.7174,
      "step": 1283
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.801050480124198,
      "learning_rate": 0.0004973266497444621,
      "loss": 4.6705,
      "step": 1284
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8958140741169052,
      "learning_rate": 0.0004971592838698959,
      "loss": 4.7151,
      "step": 1285
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7629280237440986,
      "learning_rate": 0.000496991809908969,
      "loss": 4.5902,
      "step": 1286
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.6556233287369142,
      "learning_rate": 0.0004968242279534938,
      "loss": 4.9567,
      "step": 1287
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.6905479871514387,
      "learning_rate": 0.0004966565380953419,
      "loss": 4.7847,
      "step": 1288
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.070199058361432,
      "learning_rate": 0.0004964887404264438,
      "loss": 4.6834,
      "step": 1289
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.6174604576519214,
      "learning_rate": 0.0004963208350387893,
      "loss": 4.8873,
      "step": 1290
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1293149455453142,
      "learning_rate": 0.0004961528220244273,
      "loss": 4.5958,
      "step": 1291
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4572752181086612,
      "learning_rate": 0.0004959847014754657,
      "loss": 4.6508,
      "step": 1292
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.45409888145265287,
      "learning_rate": 0.0004958164734840712,
      "loss": 4.7024,
      "step": 1293
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4156967579984714,
      "learning_rate": 0.0004956481381424695,
      "loss": 4.6617,
      "step": 1294
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8419720982766232,
      "learning_rate": 0.0004954796955429451,
      "loss": 4.7783,
      "step": 1295
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.3810440242487838,
      "learning_rate": 0.0004953111457778415,
      "loss": 4.8309,
      "step": 1296
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.221250539264744,
      "learning_rate": 0.0004951424889395607,
      "loss": 4.6704,
      "step": 1297
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5138682629424352,
      "learning_rate": 0.0004949737251205633,
      "loss": 4.83,
      "step": 1298
    },
    {
      "epoch": 1.48,
      "grad_norm": 2.204955815774822,
      "learning_rate": 0.0004948048544133691,
      "loss": 4.7474,
      "step": 1299
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.4067880260243774,
      "learning_rate": 0.0004946358769105559,
      "loss": 4.7254,
      "step": 1300
    },
    {
      "epoch": 1.49,
      "eval_blimp_filtered_avg": 0.5467164179104478,
      "eval_blimp_filtered_std": 0.005442104706995175,
      "step": 1300
    },
    {
      "epoch": 1.49,
      "eval_blimp_supplement_avg": 0.540948275862069,
      "eval_blimp_supplement_std": 0.022892130922599736,
      "step": 1300
    },
    {
      "epoch": 1.49,
      "eval_vqa_filtered_avg": 0.15,
      "eval_vqa_filtered_std": 0.035887028128263734,
      "step": 1300
    },
    {
      "epoch": 1.49,
      "eval_winoground_filtered_avg": 0.52,
      "eval_winoground_filtered_std": 0.05021167315686779,
      "step": 1300
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.3858859649800603,
      "learning_rate": 0.0004944667927047604,
      "loss": 4.7761,
      "step": 1301
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.5719051975111162,
      "learning_rate": 0.0004942976018886776,
      "loss": 4.7397,
      "step": 1302
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.7301528252563851,
      "learning_rate": 0.000494128304555061,
      "loss": 4.6252,
      "step": 1303
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.5916035548349112,
      "learning_rate": 0.0004939589007967228,
      "loss": 4.6803,
      "step": 1304
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.4469993774576138,
      "learning_rate": 0.0004937893907065329,
      "loss": 4.7683,
      "step": 1305
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.6450585381532522,
      "learning_rate": 0.00049361977437742,
      "loss": 4.9302,
      "step": 1306
    },
    {
      "epoch": 1.49,
      "grad_norm": 2.378127853025137,
      "learning_rate": 0.000493450051902371,
      "loss": 4.9909,
      "step": 1307
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.5976587289671206,
      "learning_rate": 0.0004932802233744308,
      "loss": 4.6942,
      "step": 1308
    },
    {
      "epoch": 1.5,
      "grad_norm": 2.447640477599913,
      "learning_rate": 0.0004931102888867026,
      "loss": 4.803,
      "step": 1309
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1908911280457426,
      "learning_rate": 0.0004929402485323476,
      "loss": 4.9085,
      "step": 1310
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.3934998226557072,
      "learning_rate": 0.0004927701024045849,
      "loss": 4.8782,
      "step": 1311
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.5446495349752738,
      "learning_rate": 0.0004925998505966921,
      "loss": 4.6614,
      "step": 1312
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.3846799533531913,
      "learning_rate": 0.0004924294932020039,
      "loss": 4.8579,
      "step": 1313
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.4293905552806972,
      "learning_rate": 0.0004922590303139139,
      "loss": 4.8049,
      "step": 1314
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.304156679243488,
      "learning_rate": 0.0004920884620258726,
      "loss": 4.5957,
      "step": 1315
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.5510272422892712,
      "learning_rate": 0.0004919177884313891,
      "loss": 4.875,
      "step": 1316
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.367319332203774,
      "learning_rate": 0.0004917470096240293,
      "loss": 4.8574,
      "step": 1317
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.47503688442215297,
      "learning_rate": 0.0004915761256974177,
      "loss": 4.7461,
      "step": 1318
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1431459999872369,
      "learning_rate": 0.0004914051367452358,
      "loss": 4.8458,
      "step": 1319
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.437267377825884,
      "learning_rate": 0.000491234042861223,
      "loss": 4.725,
      "step": 1320
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8747912964496404,
      "learning_rate": 0.0004910628441391761,
      "loss": 4.7648,
      "step": 1321
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8521175772691112,
      "learning_rate": 0.0004908915406729494,
      "loss": 4.6996,
      "step": 1322
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.053915445675989,
      "learning_rate": 0.0004907201325564547,
      "loss": 4.6883,
      "step": 1323
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.6778016050191935,
      "learning_rate": 0.0004905486198836609,
      "loss": 4.751,
      "step": 1324
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.6869392776255355,
      "learning_rate": 0.0004903770027485947,
      "loss": 4.8649,
      "step": 1325
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.74329793020045,
      "learning_rate": 0.0004902052812453395,
      "loss": 4.8407,
      "step": 1326
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0688334380602857,
      "learning_rate": 0.0004900334554680364,
      "loss": 4.8187,
      "step": 1327
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.6038245273867472,
      "learning_rate": 0.0004898615255108834,
      "loss": 4.6202,
      "step": 1328
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5453462263854487,
      "learning_rate": 0.0004896894914681356,
      "loss": 4.6939,
      "step": 1329
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.44444864107388815,
      "learning_rate": 0.0004895173534341053,
      "loss": 4.6116,
      "step": 1330
    },
    {
      "epoch": 1.52,
      "grad_norm": 2.036988184508873,
      "learning_rate": 0.0004893451115031618,
      "loss": 4.7477,
      "step": 1331
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0882518685638196,
      "learning_rate": 0.0004891727657697312,
      "loss": 4.6446,
      "step": 1332
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0628123204130204,
      "learning_rate": 0.0004890003163282968,
      "loss": 4.769,
      "step": 1333
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.6698103215131185,
      "learning_rate": 0.0004888277632733983,
      "loss": 4.3971,
      "step": 1334
    },
    {
      "epoch": 1.53,
      "grad_norm": 2.6262767320960543,
      "learning_rate": 0.0004886551066996328,
      "loss": 4.6933,
      "step": 1335
    },
    {
      "epoch": 1.53,
      "grad_norm": 3.155644456002479,
      "learning_rate": 0.0004884823467016535,
      "loss": 5.0961,
      "step": 1336
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9896598010878288,
      "learning_rate": 0.0004883094833741708,
      "loss": 4.9325,
      "step": 1337
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.7628983777222198,
      "learning_rate": 0.0004881365168119515,
      "loss": 4.8234,
      "step": 1338
    },
    {
      "epoch": 1.53,
      "grad_norm": 2.2339061368749555,
      "learning_rate": 0.00048796344710981917,
      "loss": 4.68,
      "step": 1339
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.8439691112891878,
      "learning_rate": 0.0004877902743626537,
      "loss": 4.9194,
      "step": 1340
    },
    {
      "epoch": 1.53,
      "grad_norm": 2.2315756389591717,
      "learning_rate": 0.0004876169986653915,
      "loss": 5.1379,
      "step": 1341
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7000102601826791,
      "learning_rate": 0.00048744362011302587,
      "loss": 4.8732,
      "step": 1342
    },
    {
      "epoch": 1.53,
      "grad_norm": 3.2558031191071217,
      "learning_rate": 0.0004872701388006057,
      "loss": 4.7404,
      "step": 1343
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9742743382110485,
      "learning_rate": 0.00048709655482323695,
      "loss": 4.7395,
      "step": 1344
    },
    {
      "epoch": 1.54,
      "grad_norm": 2.555355960934867,
      "learning_rate": 0.00048692286827608144,
      "loss": 4.8071,
      "step": 1345
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.876899528784007,
      "learning_rate": 0.00048674907925435744,
      "loss": 4.5844,
      "step": 1346
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.88872568308708,
      "learning_rate": 0.0004865751878533391,
      "loss": 5.1362,
      "step": 1347
    },
    {
      "epoch": 1.54,
      "grad_norm": 2.356039397813764,
      "learning_rate": 0.0004864011941683571,
      "loss": 4.9453,
      "step": 1348
    },
    {
      "epoch": 1.54,
      "grad_norm": 2.4333170809079694,
      "learning_rate": 0.0004862270982947979,
      "loss": 5.1346,
      "step": 1349
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0516138116267888,
      "learning_rate": 0.0004860529003281041,
      "loss": 5.0118,
      "step": 1350
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.4618361571732472,
      "learning_rate": 0.00048587860036377424,
      "loss": 5.2146,
      "step": 1351
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.356003512472729,
      "learning_rate": 0.00048570419849736275,
      "loss": 5.0642,
      "step": 1352
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0395091003090329,
      "learning_rate": 0.00048552969482447997,
      "loss": 5.0751,
      "step": 1353
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.767687223085626,
      "learning_rate": 0.00048535508944079204,
      "loss": 5.2233,
      "step": 1354
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.6389836781127758,
      "learning_rate": 0.0004851803824420209,
      "loss": 4.976,
      "step": 1355
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.6035947210613768,
      "learning_rate": 0.0004850055739239441,
      "loss": 4.8781,
      "step": 1356
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.2327314573399948,
      "learning_rate": 0.00048483066398239484,
      "loss": 5.0509,
      "step": 1357
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0548462389875854,
      "learning_rate": 0.0004846556527132621,
      "loss": 4.9994,
      "step": 1358
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7759647235778956,
      "learning_rate": 0.0004844805402124902,
      "loss": 4.8225,
      "step": 1359
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.6492476032985707,
      "learning_rate": 0.0004843053265760792,
      "loss": 4.9365,
      "step": 1360
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1652259501079139,
      "learning_rate": 0.00048413001190008426,
      "loss": 4.9432,
      "step": 1361
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8770466793294488,
      "learning_rate": 0.0004839545962806161,
      "loss": 4.9574,
      "step": 1362
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.742162330247459,
      "learning_rate": 0.0004837790798138411,
      "loss": 4.8017,
      "step": 1363
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9097192443886456,
      "learning_rate": 0.00048360346259598033,
      "loss": 5.009,
      "step": 1364
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.6557750179809073,
      "learning_rate": 0.00048342774472331056,
      "loss": 4.7991,
      "step": 1365
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9866823245433447,
      "learning_rate": 0.00048325192629216333,
      "loss": 5.0082,
      "step": 1366
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.89149319123167,
      "learning_rate": 0.0004830760073989259,
      "loss": 4.8187,
      "step": 1367
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.6895872759546015,
      "learning_rate": 0.00048289998814003997,
      "loss": 4.8978,
      "step": 1368
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7464817420797909,
      "learning_rate": 0.00048272386861200275,
      "loss": 4.9383,
      "step": 1369
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7182235230206421,
      "learning_rate": 0.0004825476489113661,
      "loss": 4.9526,
      "step": 1370
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.464471966305937,
      "learning_rate": 0.000482371329134737,
      "loss": 4.8158,
      "step": 1371
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8729558374903325,
      "learning_rate": 0.00048219490937877706,
      "loss": 4.6788,
      "step": 1372
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.5584509883951888,
      "learning_rate": 0.0004820183897402029,
      "loss": 4.7441,
      "step": 1373
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.48636883545966814,
      "learning_rate": 0.00048184177031578604,
      "loss": 4.7471,
      "step": 1374
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.6509337508930622,
      "learning_rate": 0.00048166505120235224,
      "loss": 4.8365,
      "step": 1375
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.44559332057989465,
      "learning_rate": 0.0004814882324967824,
      "loss": 4.8401,
      "step": 1376
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.5311355603205702,
      "learning_rate": 0.0004813113142960116,
      "loss": 4.7636,
      "step": 1377
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.48411680752815783,
      "learning_rate": 0.0004811342966970298,
      "loss": 4.6858,
      "step": 1378
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.3805785910338646,
      "learning_rate": 0.0004809571797968812,
      "loss": 4.6444,
      "step": 1379
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7448640827646127,
      "learning_rate": 0.0004807799636926648,
      "loss": 4.8185,
      "step": 1380
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.5174703093227337,
      "learning_rate": 0.00048060264848153345,
      "loss": 4.6472,
      "step": 1381
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.44953261244096343,
      "learning_rate": 0.00048042523426069486,
      "loss": 4.6392,
      "step": 1382
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.49875075243037065,
      "learning_rate": 0.0004802477211274106,
      "loss": 4.6904,
      "step": 1383
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.45890908478755144,
      "learning_rate": 0.0004800701091789968,
      "loss": 4.6999,
      "step": 1384
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.4795545764168666,
      "learning_rate": 0.00047989239851282343,
      "loss": 4.7719,
      "step": 1385
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.4164235961384302,
      "learning_rate": 0.00047971458922631494,
      "loss": 4.8652,
      "step": 1386
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.5952719571367953,
      "learning_rate": 0.00047953668141694953,
      "loss": 4.6569,
      "step": 1387
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.47208228718784045,
      "learning_rate": 0.0004793586751822596,
      "loss": 4.9877,
      "step": 1388
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.4928132344293389,
      "learning_rate": 0.00047918057061983145,
      "loss": 4.7921,
      "step": 1389
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.5765504772480727,
      "learning_rate": 0.00047900236782730536,
      "loss": 4.6866,
      "step": 1390
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.3994470284200598,
      "learning_rate": 0.0004788240669023752,
      "loss": 4.7505,
      "step": 1391
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.4111779963764068,
      "learning_rate": 0.00047864566794278917,
      "loss": 4.7897,
      "step": 1392
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.880224665976422,
      "learning_rate": 0.0004784671710463486,
      "loss": 4.7381,
      "step": 1393
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.6544814334197376,
      "learning_rate": 0.00047828857631090866,
      "loss": 4.8309,
      "step": 1394
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8779247759855048,
      "learning_rate": 0.00047810988383437873,
      "loss": 4.5432,
      "step": 1395
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.5696889940774883,
      "learning_rate": 0.00047793109371472095,
      "loss": 4.8037,
      "step": 1396
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7494605259976475,
      "learning_rate": 0.0004777522060499515,
      "loss": 4.5989,
      "step": 1397
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.405849873533486,
      "learning_rate": 0.00047757322093813987,
      "loss": 4.6336,
      "step": 1398
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8550914982094713,
      "learning_rate": 0.00047739413847740903,
      "loss": 4.6241,
      "step": 1399
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.5558949521132047,
      "learning_rate": 0.0004772149587659353,
      "loss": 4.6385,
      "step": 1400
    },
    {
      "epoch": 1.6,
      "eval_blimp_filtered_avg": 0.5486567164179105,
      "eval_blimp_filtered_std": 0.005423902078597133,
      "step": 1400
    },
    {
      "epoch": 1.6,
      "eval_blimp_supplement_avg": 0.4978448275862069,
      "eval_blimp_supplement_std": 0.02262365955039507,
      "step": 1400
    },
    {
      "epoch": 1.6,
      "eval_vqa_filtered_avg": 0.12,
      "eval_vqa_filtered_std": 0.03265986323710906,
      "step": 1400
    },
    {
      "epoch": 1.6,
      "eval_winoground_filtered_avg": 0.55,
      "eval_winoground_filtered_std": 0.05,
      "step": 1400
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8336343032944394,
      "learning_rate": 0.00047703568190194827,
      "loss": 4.6168,
      "step": 1401
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.3977989934017465,
      "learning_rate": 0.0004768563079837308,
      "loss": 4.6656,
      "step": 1402
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.48833622544829103,
      "learning_rate": 0.00047667683710961913,
      "loss": 4.6639,
      "step": 1403
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.6201376658756721,
      "learning_rate": 0.0004764972693780023,
      "loss": 4.6015,
      "step": 1404
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.45060823558163343,
      "learning_rate": 0.0004763176048873229,
      "loss": 4.465,
      "step": 1405
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.5513200057615996,
      "learning_rate": 0.0004761378437360761,
      "loss": 4.6264,
      "step": 1406
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.5303636884422888,
      "learning_rate": 0.0004759579860228103,
      "loss": 4.7561,
      "step": 1407
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.5828719715564574,
      "learning_rate": 0.00047577803184612697,
      "loss": 4.7605,
      "step": 1408
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.491684759345481,
      "learning_rate": 0.0004755979813046803,
      "loss": 4.5292,
      "step": 1409
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.5700256073475649,
      "learning_rate": 0.00047541783449717715,
      "loss": 4.5218,
      "step": 1410
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.690719867302228,
      "learning_rate": 0.00047523759152237746,
      "loss": 4.7592,
      "step": 1411
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.4783347732188243,
      "learning_rate": 0.0004750572524790938,
      "loss": 4.5417,
      "step": 1412
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.311904722278894,
      "learning_rate": 0.0004748768174661912,
      "loss": 4.7346,
      "step": 1413
    },
    {
      "epoch": 1.62,
      "grad_norm": 2.039716183102799,
      "learning_rate": 0.0004746962865825876,
      "loss": 4.5815,
      "step": 1414
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7763142112437567,
      "learning_rate": 0.0004745156599272533,
      "loss": 4.5282,
      "step": 1415
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2327429361222049,
      "learning_rate": 0.00047433493759921124,
      "loss": 4.522,
      "step": 1416
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.022801721757132,
      "learning_rate": 0.00047415411969753673,
      "loss": 4.4127,
      "step": 1417
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.5485097704381883,
      "learning_rate": 0.0004739732063213574,
      "loss": 4.5386,
      "step": 1418
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.4633553754141256,
      "learning_rate": 0.0004737921975698533,
      "loss": 4.6599,
      "step": 1419
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0682156168501629,
      "learning_rate": 0.0004736110935422568,
      "loss": 4.6262,
      "step": 1420
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.5875852940501651,
      "learning_rate": 0.0004734298943378526,
      "loss": 4.4744,
      "step": 1421
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7184393812421046,
      "learning_rate": 0.0004732486000559773,
      "loss": 4.6543,
      "step": 1422
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.459657755517521,
      "learning_rate": 0.00047306721079601983,
      "loss": 4.5357,
      "step": 1423
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7069916733163126,
      "learning_rate": 0.0004728857266574211,
      "loss": 4.486,
      "step": 1424
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.42824071906555994,
      "learning_rate": 0.0004727041477396741,
      "loss": 4.4179,
      "step": 1425
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.6996728013593154,
      "learning_rate": 0.00047252247414232367,
      "loss": 4.6625,
      "step": 1426
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.6613682009533245,
      "learning_rate": 0.0004723407059649668,
      "loss": 4.4985,
      "step": 1427
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.5157407781149183,
      "learning_rate": 0.0004721588433072519,
      "loss": 4.5895,
      "step": 1428
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7048944820456728,
      "learning_rate": 0.0004719768862688798,
      "loss": 4.7225,
      "step": 1429
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.6262316373051795,
      "learning_rate": 0.0004717948349496023,
      "loss": 4.5735,
      "step": 1430
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8522926897856793,
      "learning_rate": 0.0004716126894492236,
      "loss": 4.583,
      "step": 1431
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.953880313702566,
      "learning_rate": 0.0004714304498675991,
      "loss": 4.7722,
      "step": 1432
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7744020269228191,
      "learning_rate": 0.000471248116304636,
      "loss": 4.6082,
      "step": 1433
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1055975363378021,
      "learning_rate": 0.00047106568886029276,
      "loss": 4.5719,
      "step": 1434
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7480413036131542,
      "learning_rate": 0.0004708831676345796,
      "loss": 4.6206,
      "step": 1435
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.408364565913358,
      "learning_rate": 0.00047070055272755795,
      "loss": 4.4276,
      "step": 1436
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.21661965567842,
      "learning_rate": 0.00047051784423934087,
      "loss": 4.6314,
      "step": 1437
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7983588404581601,
      "learning_rate": 0.00047033504227009216,
      "loss": 4.5995,
      "step": 1438
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1885033725376346,
      "learning_rate": 0.0004701521469200277,
      "loss": 4.6225,
      "step": 1439
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7010293270107962,
      "learning_rate": 0.0004699691582894137,
      "loss": 4.7033,
      "step": 1440
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8026210036924896,
      "learning_rate": 0.0004697860764785681,
      "loss": 4.6279,
      "step": 1441
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.292836579976248,
      "learning_rate": 0.0004696029015878596,
      "loss": 4.625,
      "step": 1442
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0391875969159525,
      "learning_rate": 0.0004694196337177082,
      "loss": 4.4795,
      "step": 1443
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.6960164897456129,
      "learning_rate": 0.0004692362729685847,
      "loss": 4.5121,
      "step": 1444
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.271940700352897,
      "learning_rate": 0.00046905281944101084,
      "loss": 4.6049,
      "step": 1445
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7326496801015598,
      "learning_rate": 0.0004688692732355592,
      "loss": 4.6007,
      "step": 1446
    },
    {
      "epoch": 1.65,
      "grad_norm": 2.1452438217738163,
      "learning_rate": 0.00046868563445285315,
      "loss": 4.6451,
      "step": 1447
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2998642561112748,
      "learning_rate": 0.00046850190319356705,
      "loss": 4.5305,
      "step": 1448
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.4626702907678346,
      "learning_rate": 0.00046831807955842556,
      "loss": 4.6695,
      "step": 1449
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.6534669079528632,
      "learning_rate": 0.00046813416364820435,
      "loss": 4.6443,
      "step": 1450
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0688731454564804,
      "learning_rate": 0.0004679501555637294,
      "loss": 4.5762,
      "step": 1451
    },
    {
      "epoch": 1.66,
      "grad_norm": 2.09981490315287,
      "learning_rate": 0.0004677660554058774,
      "loss": 4.4893,
      "step": 1452
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.5100514339572236,
      "learning_rate": 0.00046758186327557544,
      "loss": 4.6868,
      "step": 1453
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.2492471971514083,
      "learning_rate": 0.00046739757927380106,
      "loss": 4.3353,
      "step": 1454
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.5469293110300977,
      "learning_rate": 0.0004672132035015822,
      "loss": 4.5055,
      "step": 1455
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.2579870590866056,
      "learning_rate": 0.00046702873605999694,
      "loss": 4.5368,
      "step": 1456
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.5958553573610544,
      "learning_rate": 0.00046684417705017384,
      "loss": 4.4653,
      "step": 1457
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2281957190731483,
      "learning_rate": 0.0004666595265732916,
      "loss": 4.7814,
      "step": 1458
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.4900740490343557,
      "learning_rate": 0.0004664747847305789,
      "loss": 4.4939,
      "step": 1459
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0841072960419629,
      "learning_rate": 0.00046628995162331465,
      "loss": 4.499,
      "step": 1460
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.5404246112370104,
      "learning_rate": 0.0004661050273528278,
      "loss": 4.4651,
      "step": 1461
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.5472302297342864,
      "learning_rate": 0.00046592001202049727,
      "loss": 4.551,
      "step": 1462
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.5579220926830852,
      "learning_rate": 0.0004657349057277518,
      "loss": 4.5228,
      "step": 1463
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8373877029205464,
      "learning_rate": 0.00046554970857607027,
      "loss": 4.4877,
      "step": 1464
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.6552871235740625,
      "learning_rate": 0.000465364420666981,
      "loss": 4.385,
      "step": 1465
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.5533089422614667,
      "learning_rate": 0.00046517904210206226,
      "loss": 4.4582,
      "step": 1466
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.5272674926590186,
      "learning_rate": 0.00046499357298294204,
      "loss": 4.528,
      "step": 1467
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9381489017405934,
      "learning_rate": 0.00046480801341129794,
      "loss": 4.448,
      "step": 1468
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0140125859524247,
      "learning_rate": 0.00046462236348885716,
      "loss": 4.4023,
      "step": 1469
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7254247415004822,
      "learning_rate": 0.00046443662331739643,
      "loss": 4.546,
      "step": 1470
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.5818023262302461,
      "learning_rate": 0.00046425079299874185,
      "loss": 4.5079,
      "step": 1471
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8144083944954458,
      "learning_rate": 0.00046406487263476917,
      "loss": 4.4881,
      "step": 1472
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.6603133548722864,
      "learning_rate": 0.00046387886232740326,
      "loss": 4.4152,
      "step": 1473
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1950135868753649,
      "learning_rate": 0.0004636927621786184,
      "loss": 4.4233,
      "step": 1474
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.426835645225259,
      "learning_rate": 0.0004635065722904381,
      "loss": 4.388,
      "step": 1475
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.098018527034672,
      "learning_rate": 0.0004633202927649352,
      "loss": 4.5237,
      "step": 1476
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0111653860246692,
      "learning_rate": 0.00046313392370423155,
      "loss": 4.3329,
      "step": 1477
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.5425928466560297,
      "learning_rate": 0.0004629474652104981,
      "loss": 4.3738,
      "step": 1478
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.44346157691869187,
      "learning_rate": 0.00046276091738595485,
      "loss": 4.5344,
      "step": 1479
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.6908299364714499,
      "learning_rate": 0.0004625742803328708,
      "loss": 4.2441,
      "step": 1480
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8386324132653891,
      "learning_rate": 0.00046238755415356367,
      "loss": 4.5414,
      "step": 1481
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7794293168957607,
      "learning_rate": 0.0004622007389504003,
      "loss": 4.4319,
      "step": 1482
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.47208493229831217,
      "learning_rate": 0.00046201383482579627,
      "loss": 4.3757,
      "step": 1483
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0444789947063238,
      "learning_rate": 0.00046182684188221583,
      "loss": 4.5969,
      "step": 1484
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.333137218371951,
      "learning_rate": 0.000461639760222172,
      "loss": 4.5076,
      "step": 1485
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.5584816850865542,
      "learning_rate": 0.0004614525899482265,
      "loss": 4.4563,
      "step": 1486
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.976026380170844,
      "learning_rate": 0.00046126533116298927,
      "loss": 4.6841,
      "step": 1487
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.9504723397362387,
      "learning_rate": 0.0004610779839691192,
      "loss": 4.5912,
      "step": 1488
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1938994740564937,
      "learning_rate": 0.00046089054846932344,
      "loss": 4.4707,
      "step": 1489
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.6005844219105226,
      "learning_rate": 0.0004607030247663577,
      "loss": 4.4557,
      "step": 1490
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9278027705042341,
      "learning_rate": 0.00046051541296302575,
      "loss": 4.5477,
      "step": 1491
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.5144290433547679,
      "learning_rate": 0.00046032771316218,
      "loss": 4.5463,
      "step": 1492
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8503685862660372,
      "learning_rate": 0.0004601399254667208,
      "loss": 4.5529,
      "step": 1493
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8315915348155466,
      "learning_rate": 0.0004599520499795971,
      "loss": 4.3742,
      "step": 1494
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.6319480784344171,
      "learning_rate": 0.0004597640868038054,
      "loss": 4.4062,
      "step": 1495
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7561568061366876,
      "learning_rate": 0.00045957603604239076,
      "loss": 4.6252,
      "step": 1496
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.5497514378646382,
      "learning_rate": 0.00045938789779844595,
      "loss": 4.324,
      "step": 1497
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.6226330913796483,
      "learning_rate": 0.00045919967217511204,
      "loss": 4.4532,
      "step": 1498
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.48237361176527244,
      "learning_rate": 0.00045901135927557755,
      "loss": 4.467,
      "step": 1499
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9372067902547772,
      "learning_rate": 0.0004588229592030793,
      "loss": 4.4392,
      "step": 1500
    },
    {
      "epoch": 1.71,
      "eval_blimp_filtered_avg": 0.5313432835820896,
      "eval_blimp_filtered_std": 0.005503755937770046,
      "step": 1500
    },
    {
      "epoch": 1.71,
      "eval_blimp_supplement_avg": 0.5280172413793104,
      "eval_blimp_supplement_std": 0.02282582338755408,
      "step": 1500
    },
    {
      "epoch": 1.71,
      "eval_vqa_filtered_avg": 0.32,
      "eval_vqa_filtered_std": 0.04688261722621503,
      "step": 1500
    },
    {
      "epoch": 1.71,
      "eval_winoground_filtered_avg": 0.49,
      "eval_winoground_filtered_std": 0.05024183937956912,
      "step": 1500
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.654145061622795,
      "learning_rate": 0.0004586344720609016,
      "loss": 4.6077,
      "step": 1501
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.526227217844653,
      "learning_rate": 0.0004584458979523766,
      "loss": 4.5942,
      "step": 1502
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.5786092814506665,
      "learning_rate": 0.0004582572369808841,
      "loss": 4.5841,
      "step": 1503
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.4992249834821379,
      "learning_rate": 0.0004580684892498514,
      "loss": 4.689,
      "step": 1504
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7093856690445237,
      "learning_rate": 0.00045787965486275377,
      "loss": 4.5217,
      "step": 1505
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.949826681657814,
      "learning_rate": 0.00045769073392311344,
      "loss": 4.3849,
      "step": 1506
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.5645535110299973,
      "learning_rate": 0.0004575017265345005,
      "loss": 4.6325,
      "step": 1507
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1033980958579388,
      "learning_rate": 0.00045731263280053227,
      "loss": 4.5902,
      "step": 1508
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9327253236530959,
      "learning_rate": 0.00045712345282487344,
      "loss": 4.5289,
      "step": 1509
    },
    {
      "epoch": 1.73,
      "grad_norm": 2.1436300111871147,
      "learning_rate": 0.000456934186711236,
      "loss": 4.4449,
      "step": 1510
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.5817931763904416,
      "learning_rate": 0.000456744834563379,
      "loss": 4.5626,
      "step": 1511
    },
    {
      "epoch": 1.73,
      "grad_norm": 2.1372299532619667,
      "learning_rate": 0.0004565553964851087,
      "loss": 4.6252,
      "step": 1512
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.3881038760533755,
      "learning_rate": 0.000456365872580279,
      "loss": 4.499,
      "step": 1513
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0589549281892254,
      "learning_rate": 0.00045617626295279,
      "loss": 4.3902,
      "step": 1514
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7382290860503711,
      "learning_rate": 0.0004559865677065892,
      "loss": 4.6045,
      "step": 1515
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.423513580311941,
      "learning_rate": 0.00045579678694567124,
      "loss": 4.5644,
      "step": 1516
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.8946698768063079,
      "learning_rate": 0.00045560692077407743,
      "loss": 4.5081,
      "step": 1517
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.4862550637998149,
      "learning_rate": 0.0004554169692958958,
      "loss": 4.5511,
      "step": 1518
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0756857163605837,
      "learning_rate": 0.00045522693261526115,
      "loss": 4.4308,
      "step": 1519
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5185065081935016,
      "learning_rate": 0.00045503681083635544,
      "loss": 4.3477,
      "step": 1520
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9903672820407596,
      "learning_rate": 0.0004548466040634066,
      "loss": 4.5014,
      "step": 1521
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0489641368864815,
      "learning_rate": 0.0004546563124006898,
      "loss": 4.4527,
      "step": 1522
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.533081796064669,
      "learning_rate": 0.0004544659359525261,
      "loss": 4.6019,
      "step": 1523
    },
    {
      "epoch": 1.74,
      "grad_norm": 2.1273933593012666,
      "learning_rate": 0.00045427547482328363,
      "loss": 4.655,
      "step": 1524
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.8852063385845723,
      "learning_rate": 0.0004540849291173766,
      "loss": 4.3868,
      "step": 1525
    },
    {
      "epoch": 1.74,
      "grad_norm": 2.109922658905615,
      "learning_rate": 0.00045389429893926574,
      "loss": 4.2372,
      "step": 1526
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9360131719277137,
      "learning_rate": 0.000453703584393458,
      "loss": 4.5194,
      "step": 1527
    },
    {
      "epoch": 1.75,
      "grad_norm": 2.764261241191548,
      "learning_rate": 0.0004535127855845067,
      "loss": 4.4637,
      "step": 1528
    },
    {
      "epoch": 1.75,
      "grad_norm": 2.0557846456704674,
      "learning_rate": 0.00045332190261701115,
      "loss": 4.6643,
      "step": 1529
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.5941633237900894,
      "learning_rate": 0.000453130935595617,
      "loss": 4.6425,
      "step": 1530
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.846811508636747,
      "learning_rate": 0.0004529398846250158,
      "loss": 4.4481,
      "step": 1531
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.452209922559483,
      "learning_rate": 0.00045274874980994536,
      "loss": 4.5801,
      "step": 1532
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.8002040359388272,
      "learning_rate": 0.0004525575312551892,
      "loss": 4.5893,
      "step": 1533
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.5718974262300127,
      "learning_rate": 0.0004523662290655769,
      "loss": 4.6487,
      "step": 1534
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.7654485645598723,
      "learning_rate": 0.0004521748433459839,
      "loss": 4.6022,
      "step": 1535
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0322585322051914,
      "learning_rate": 0.00045198337420133126,
      "loss": 4.5261,
      "step": 1536
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3814834071902213,
      "learning_rate": 0.00045179182173658606,
      "loss": 4.425,
      "step": 1537
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.689913139266974,
      "learning_rate": 0.00045160018605676084,
      "loss": 4.3825,
      "step": 1538
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.6607967865290028,
      "learning_rate": 0.0004514084672669137,
      "loss": 4.6327,
      "step": 1539
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.6807913903453778,
      "learning_rate": 0.0004512166654721486,
      "loss": 4.489,
      "step": 1540
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.6542090792430012,
      "learning_rate": 0.00045102478077761477,
      "loss": 4.4205,
      "step": 1541
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.824480276696534,
      "learning_rate": 0.0004508328132885069,
      "loss": 4.5661,
      "step": 1542
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.4592661596487153,
      "learning_rate": 0.0004506407631100651,
      "loss": 4.4651,
      "step": 1543
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7493008324865154,
      "learning_rate": 0.0004504486303475749,
      "loss": 4.8073,
      "step": 1544
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.5991501039749012,
      "learning_rate": 0.00045025641510636704,
      "loss": 4.4707,
      "step": 1545
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1285217323236216,
      "learning_rate": 0.00045006411749181724,
      "loss": 4.6226,
      "step": 1546
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.46591175013868713,
      "learning_rate": 0.000449871737609347,
      "loss": 4.5421,
      "step": 1547
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9424459584122187,
      "learning_rate": 0.0004496792755644221,
      "loss": 4.3357,
      "step": 1548
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.666702380037904,
      "learning_rate": 0.0004494867314625542,
      "loss": 4.6317,
      "step": 1549
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.5083531514679692,
      "learning_rate": 0.00044929410540929917,
      "loss": 4.535,
      "step": 1550
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.4824459398322389,
      "learning_rate": 0.00044910139751025836,
      "loss": 4.6558,
      "step": 1551
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.5733738462978704,
      "learning_rate": 0.00044890860787107783,
      "loss": 4.5481,
      "step": 1552
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.5570605605961345,
      "learning_rate": 0.0004487157365974482,
      "loss": 4.6373,
      "step": 1553
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.6037547014433808,
      "learning_rate": 0.00044852278379510534,
      "loss": 4.4533,
      "step": 1554
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.5466371607699255,
      "learning_rate": 0.0004483297495698294,
      "loss": 4.2645,
      "step": 1555
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.6836285309049864,
      "learning_rate": 0.00044813663402744537,
      "loss": 4.3832,
      "step": 1556
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7440476140350695,
      "learning_rate": 0.00044794343727382264,
      "loss": 4.5324,
      "step": 1557
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.5972126201520713,
      "learning_rate": 0.0004477501594148753,
      "loss": 4.383,
      "step": 1558
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.5664749444139975,
      "learning_rate": 0.0004475568005565619,
      "loss": 4.2492,
      "step": 1559
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.4720967638929448,
      "learning_rate": 0.0004473633608048853,
      "loss": 4.4472,
      "step": 1560
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.45218504991240904,
      "learning_rate": 0.00044716984026589275,
      "loss": 4.3195,
      "step": 1561
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.5311894508382072,
      "learning_rate": 0.00044697623904567564,
      "loss": 4.4659,
      "step": 1562
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7004978387249767,
      "learning_rate": 0.00044678255725037,
      "loss": 4.4841,
      "step": 1563
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.44525585533389095,
      "learning_rate": 0.00044658879498615556,
      "loss": 4.3947,
      "step": 1564
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8513477205579004,
      "learning_rate": 0.0004463949523592564,
      "loss": 4.427,
      "step": 1565
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.6941682015123531,
      "learning_rate": 0.0004462010294759408,
      "loss": 4.5005,
      "step": 1566
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7912810140554617,
      "learning_rate": 0.00044600702644252065,
      "loss": 4.5903,
      "step": 1567
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7956141515129387,
      "learning_rate": 0.00044581294336535216,
      "loss": 4.3732,
      "step": 1568
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.6347753028153407,
      "learning_rate": 0.00044561878035083506,
      "loss": 4.4276,
      "step": 1569
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8131680034655744,
      "learning_rate": 0.00044542453750541326,
      "loss": 4.4113,
      "step": 1570
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.5531160249916168,
      "learning_rate": 0.00044523021493557424,
      "loss": 4.4114,
      "step": 1571
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.6709155154628429,
      "learning_rate": 0.00044503581274784926,
      "loss": 4.6583,
      "step": 1572
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.5006304854746432,
      "learning_rate": 0.00044484133104881306,
      "loss": 4.4336,
      "step": 1573
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.553156417242004,
      "learning_rate": 0.00044464676994508424,
      "loss": 4.527,
      "step": 1574
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.5218126318756796,
      "learning_rate": 0.0004444521295433247,
      "loss": 4.5587,
      "step": 1575
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.6502480420225046,
      "learning_rate": 0.00044425740995023997,
      "loss": 4.6106,
      "step": 1576
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.42275079682782113,
      "learning_rate": 0.0004440626112725789,
      "loss": 4.471,
      "step": 1577
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.5336234341063297,
      "learning_rate": 0.00044386773361713365,
      "loss": 4.4296,
      "step": 1578
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.8768322447889828,
      "learning_rate": 0.00044367277709073995,
      "loss": 4.4361,
      "step": 1579
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.4953182098422311,
      "learning_rate": 0.0004434777418002763,
      "loss": 4.6317,
      "step": 1580
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.6246272115043877,
      "learning_rate": 0.00044328262785266485,
      "loss": 4.4747,
      "step": 1581
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.5755007673529107,
      "learning_rate": 0.0004430874353548706,
      "loss": 4.4884,
      "step": 1582
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7282181717800094,
      "learning_rate": 0.0004428921644139017,
      "loss": 4.4085,
      "step": 1583
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.40345437354128066,
      "learning_rate": 0.00044269681513680925,
      "loss": 4.5598,
      "step": 1584
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.49080899691349517,
      "learning_rate": 0.00044250138763068735,
      "loss": 4.4885,
      "step": 1585
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.5832158170699938,
      "learning_rate": 0.0004423058820026729,
      "loss": 4.4993,
      "step": 1586
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7176815465231735,
      "learning_rate": 0.0004421102983599459,
      "loss": 4.5301,
      "step": 1587
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.44961417394655306,
      "learning_rate": 0.00044191463680972877,
      "loss": 4.5264,
      "step": 1588
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2205992151088976,
      "learning_rate": 0.0004417188974592867,
      "loss": 4.5153,
      "step": 1589
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2555779788918737,
      "learning_rate": 0.00044152308041592775,
      "loss": 4.3809,
      "step": 1590
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0206842307813042,
      "learning_rate": 0.0004413271857870024,
      "loss": 4.4599,
      "step": 1591
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2798291072883572,
      "learning_rate": 0.0004411312136799038,
      "loss": 4.6609,
      "step": 1592
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9014588228099107,
      "learning_rate": 0.00044093516420206725,
      "loss": 4.3682,
      "step": 1593
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.6508098069080454,
      "learning_rate": 0.0004407390374609709,
      "loss": 4.4755,
      "step": 1594
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2627251952775134,
      "learning_rate": 0.000440542833564135,
      "loss": 4.2805,
      "step": 1595
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.5785295793103927,
      "learning_rate": 0.0004403465526191221,
      "loss": 4.3623,
      "step": 1596
    },
    {
      "epoch": 1.83,
      "grad_norm": 2.4499139141361086,
      "learning_rate": 0.00044015019473353707,
      "loss": 4.4762,
      "step": 1597
    },
    {
      "epoch": 1.83,
      "grad_norm": 2.8318889060683925,
      "learning_rate": 0.000439953760015027,
      "loss": 4.3095,
      "step": 1598
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.8192322592907868,
      "learning_rate": 0.0004397572485712809,
      "loss": 4.3611,
      "step": 1599
    },
    {
      "epoch": 1.83,
      "grad_norm": 2.340115734256017,
      "learning_rate": 0.00043956066051003,
      "loss": 4.4017,
      "step": 1600
    },
    {
      "epoch": 1.83,
      "eval_blimp_filtered_avg": 0.5214925373134328,
      "eval_blimp_filtered_std": 0.0055171088893015184,
      "step": 1600
    },
    {
      "epoch": 1.83,
      "eval_blimp_supplement_avg": 0.540948275862069,
      "eval_blimp_supplement_std": 0.022895976379159827,
      "step": 1600
    },
    {
      "epoch": 1.83,
      "eval_vqa_filtered_avg": 0.3,
      "eval_vqa_filtered_std": 0.046056618647183814,
      "step": 1600
    },
    {
      "epoch": 1.83,
      "eval_winoground_filtered_avg": 0.53,
      "eval_winoground_filtered_std": 0.0501613558046592,
      "step": 1600
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1062711573457815,
      "learning_rate": 0.0004393639959390475,
      "loss": 4.378,
      "step": 1601
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.7572324970881712,
      "learning_rate": 0.00043916725496614874,
      "loss": 4.4632,
      "step": 1602
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1753154767198906,
      "learning_rate": 0.00043897043769919054,
      "loss": 4.4543,
      "step": 1603
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.6986362591812874,
      "learning_rate": 0.00043877354424607176,
      "loss": 4.3068,
      "step": 1604
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.6796840040972915,
      "learning_rate": 0.00043857657471473314,
      "loss": 4.6881,
      "step": 1605
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9760170048085693,
      "learning_rate": 0.00043837952921315694,
      "loss": 4.4336,
      "step": 1606
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.788095595336318,
      "learning_rate": 0.0004381824078493673,
      "loss": 4.5834,
      "step": 1607
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.5240881010696307,
      "learning_rate": 0.0004379852107314295,
      "loss": 4.7793,
      "step": 1608
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.6800203301251186,
      "learning_rate": 0.00043778793796745086,
      "loss": 4.441,
      "step": 1609
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.8861323881164404,
      "learning_rate": 0.0004375905896655799,
      "loss": 4.5062,
      "step": 1610
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7560387351358162,
      "learning_rate": 0.00043739316593400667,
      "loss": 4.4551,
      "step": 1611
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.5964493157160932,
      "learning_rate": 0.0004371956668809623,
      "loss": 4.5077,
      "step": 1612
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.139274966933268,
      "learning_rate": 0.0004369980926147197,
      "loss": 4.5886,
      "step": 1613
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7724423684446049,
      "learning_rate": 0.00043680044324359243,
      "loss": 4.5777,
      "step": 1614
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.8352952048050188,
      "learning_rate": 0.00043660271887593585,
      "loss": 4.4798,
      "step": 1615
    },
    {
      "epoch": 1.85,
      "grad_norm": 2.1520712907073194,
      "learning_rate": 0.0004364049196201458,
      "loss": 4.4537,
      "step": 1616
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7571214660564398,
      "learning_rate": 0.0004362070455846597,
      "loss": 4.3287,
      "step": 1617
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.8137930679825731,
      "learning_rate": 0.00043600909687795564,
      "loss": 4.5292,
      "step": 1618
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3744957516032565,
      "learning_rate": 0.0004358110736085528,
      "loss": 4.4856,
      "step": 1619
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9668678109511869,
      "learning_rate": 0.00043561297588501124,
      "loss": 4.4445,
      "step": 1620
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.6740960450757343,
      "learning_rate": 0.0004354148038159317,
      "loss": 4.3517,
      "step": 1621
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.8289741952368377,
      "learning_rate": 0.0004352165575099558,
      "loss": 4.3395,
      "step": 1622
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3421842419217718,
      "learning_rate": 0.0004350182370757659,
      "loss": 4.4447,
      "step": 1623
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5527515320007677,
      "learning_rate": 0.00043481984262208465,
      "loss": 4.5282,
      "step": 1624
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.8512202419733643,
      "learning_rate": 0.00043462137425767596,
      "loss": 4.2819,
      "step": 1625
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.8913513992739794,
      "learning_rate": 0.00043442283209134364,
      "loss": 4.6412,
      "step": 1626
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.5874133955296026,
      "learning_rate": 0.00043422421623193214,
      "loss": 4.5435,
      "step": 1627
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0951708805144085,
      "learning_rate": 0.0004340255267883264,
      "loss": 4.7743,
      "step": 1628
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.8670436598001132,
      "learning_rate": 0.00043382676386945164,
      "loss": 4.3273,
      "step": 1629
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3655586090725098,
      "learning_rate": 0.00043362792758427335,
      "loss": 4.5056,
      "step": 1630
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.6214217589923374,
      "learning_rate": 0.00043342901804179726,
      "loss": 4.4349,
      "step": 1631
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1083160195572712,
      "learning_rate": 0.00043323003535106924,
      "loss": 4.3371,
      "step": 1632
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0672355165574567,
      "learning_rate": 0.0004330309796211752,
      "loss": 4.2983,
      "step": 1633
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7224957724859657,
      "learning_rate": 0.0004328318509612414,
      "loss": 4.5616,
      "step": 1634
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0695952625647405,
      "learning_rate": 0.00043263264948043364,
      "loss": 4.4534,
      "step": 1635
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.6704887843305182,
      "learning_rate": 0.0004324333752879579,
      "loss": 4.5169,
      "step": 1636
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7873972733995961,
      "learning_rate": 0.00043223402849306005,
      "loss": 4.4162,
      "step": 1637
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.4120683543190572,
      "learning_rate": 0.00043203460920502565,
      "loss": 4.5158,
      "step": 1638
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9772876591209761,
      "learning_rate": 0.0004318351175331799,
      "loss": 4.4812,
      "step": 1639
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.6957044394239216,
      "learning_rate": 0.00043163555358688796,
      "loss": 4.3274,
      "step": 1640
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.5408209088733036,
      "learning_rate": 0.00043143591747555444,
      "loss": 4.5473,
      "step": 1641
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.8075507081405351,
      "learning_rate": 0.0004312362093086236,
      "loss": 4.2734,
      "step": 1642
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.775619692545571,
      "learning_rate": 0.000431036429195579,
      "loss": 4.4186,
      "step": 1643
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.5877722406983326,
      "learning_rate": 0.0004308365772459439,
      "loss": 4.6196,
      "step": 1644
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7497860339814169,
      "learning_rate": 0.0004306366535692809,
      "loss": 4.541,
      "step": 1645
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.5783856027556545,
      "learning_rate": 0.00043043665827519165,
      "loss": 4.5663,
      "step": 1646
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.6646240536949809,
      "learning_rate": 0.0004302365914733174,
      "loss": 4.499,
      "step": 1647
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1692365510955416,
      "learning_rate": 0.0004300364532733384,
      "loss": 4.5122,
      "step": 1648
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.125916644396735,
      "learning_rate": 0.00042983624378497426,
      "loss": 4.2775,
      "step": 1649
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.6241479816132806,
      "learning_rate": 0.00042963596311798325,
      "loss": 4.4907,
      "step": 1650
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2027929384749094,
      "learning_rate": 0.00042943561138216314,
      "loss": 4.4543,
      "step": 1651
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.5950079789723692,
      "learning_rate": 0.00042923518868735045,
      "loss": 4.2052,
      "step": 1652
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4473557972125584,
      "learning_rate": 0.00042903469514342054,
      "loss": 4.512,
      "step": 1653
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7118452392677886,
      "learning_rate": 0.0004288341308602877,
      "loss": 4.4998,
      "step": 1654
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0007710009356714,
      "learning_rate": 0.0004286334959479048,
      "loss": 4.5244,
      "step": 1655
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0316587186171344,
      "learning_rate": 0.00042843279051626385,
      "loss": 4.3448,
      "step": 1656
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0275663885997315,
      "learning_rate": 0.0004282320146753953,
      "loss": 4.3363,
      "step": 1657
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.6308786939732394,
      "learning_rate": 0.0004280311685353679,
      "loss": 4.3691,
      "step": 1658
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.5516393458016108,
      "learning_rate": 0.00042783025220628946,
      "loss": 4.6687,
      "step": 1659
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.519094490004298,
      "learning_rate": 0.0004276292657983059,
      "loss": 4.3569,
      "step": 1660
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7778047105828223,
      "learning_rate": 0.0004274282094216017,
      "loss": 4.4711,
      "step": 1661
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7222897603470187,
      "learning_rate": 0.00042722708318639974,
      "loss": 4.4056,
      "step": 1662
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.6740884088950804,
      "learning_rate": 0.00042702588720296104,
      "loss": 4.4864,
      "step": 1663
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7510745543541218,
      "learning_rate": 0.0004268246215815851,
      "loss": 4.453,
      "step": 1664
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7319593364986465,
      "learning_rate": 0.00042662328643260924,
      "loss": 4.4867,
      "step": 1665
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.5005295564207787,
      "learning_rate": 0.00042642188186640934,
      "loss": 4.5488,
      "step": 1666
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.6856624232920483,
      "learning_rate": 0.0004262204079933988,
      "loss": 4.4724,
      "step": 1667
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.5112606223014141,
      "learning_rate": 0.00042601886492402965,
      "loss": 4.3625,
      "step": 1668
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.5208898593574679,
      "learning_rate": 0.00042581725276879135,
      "loss": 4.3936,
      "step": 1669
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.5433240621282674,
      "learning_rate": 0.00042561557163821144,
      "loss": 4.2148,
      "step": 1670
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.8918250753718338,
      "learning_rate": 0.0004254138216428552,
      "loss": 4.2593,
      "step": 1671
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2435816516270333,
      "learning_rate": 0.0004252120028933258,
      "loss": 4.3805,
      "step": 1672
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0098729821994417,
      "learning_rate": 0.0004250101155002639,
      "loss": 4.4043,
      "step": 1673
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.992161613470479,
      "learning_rate": 0.0004248081595743479,
      "loss": 4.5608,
      "step": 1674
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.4171298117993861,
      "learning_rate": 0.0004246061352262938,
      "loss": 4.3392,
      "step": 1675
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9508658021202646,
      "learning_rate": 0.0004244040425668552,
      "loss": 4.4749,
      "step": 1676
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.731792602020031,
      "learning_rate": 0.0004242018817068228,
      "loss": 4.3818,
      "step": 1677
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1319227514405912,
      "learning_rate": 0.0004239996527570251,
      "loss": 4.4845,
      "step": 1678
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9317245734123896,
      "learning_rate": 0.00042379735582832764,
      "loss": 4.429,
      "step": 1679
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.6242143530598611,
      "learning_rate": 0.00042359499103163334,
      "loss": 4.4413,
      "step": 1680
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7172844527427172,
      "learning_rate": 0.00042339255847788246,
      "loss": 4.3425,
      "step": 1681
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9794042652786109,
      "learning_rate": 0.000423190058278052,
      "loss": 4.4374,
      "step": 1682
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.108826880671933,
      "learning_rate": 0.0004229874905431566,
      "loss": 4.5155,
      "step": 1683
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.5743233084639326,
      "learning_rate": 0.0004227848553842474,
      "loss": 4.3519,
      "step": 1684
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7692965815613436,
      "learning_rate": 0.00042258215291241294,
      "loss": 4.2709,
      "step": 1685
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5353250114366237,
      "learning_rate": 0.0004223793832387783,
      "loss": 4.399,
      "step": 1686
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0149592077859573,
      "learning_rate": 0.0004221765464745056,
      "loss": 4.5445,
      "step": 1687
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.5389667455725036,
      "learning_rate": 0.00042197364273079374,
      "loss": 4.3855,
      "step": 1688
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7273276135371077,
      "learning_rate": 0.0004217706721188782,
      "loss": 4.3124,
      "step": 1689
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7948143531830061,
      "learning_rate": 0.00042156763475003137,
      "loss": 4.3479,
      "step": 1690
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.8465187287578428,
      "learning_rate": 0.00042136453073556197,
      "loss": 4.3771,
      "step": 1691
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9693227388160204,
      "learning_rate": 0.0004211613601868154,
      "loss": 4.6063,
      "step": 1692
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.788445395513107,
      "learning_rate": 0.0004209581232151735,
      "loss": 4.1839,
      "step": 1693
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7342286878837484,
      "learning_rate": 0.00042075481993205445,
      "loss": 4.3877,
      "step": 1694
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6714318116683777,
      "learning_rate": 0.0004205514504489131,
      "loss": 4.4136,
      "step": 1695
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.9178084493793184,
      "learning_rate": 0.00042034801487724025,
      "loss": 4.4047,
      "step": 1696
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2141523527405822,
      "learning_rate": 0.00042014451332856293,
      "loss": 4.3602,
      "step": 1697
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.8311193323040716,
      "learning_rate": 0.00041994094591444453,
      "loss": 4.2386,
      "step": 1698
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.9629229422004671,
      "learning_rate": 0.0004197373127464844,
      "loss": 4.4829,
      "step": 1699
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.488589131529667,
      "learning_rate": 0.0004195336139363182,
      "loss": 4.5186,
      "step": 1700
    },
    {
      "epoch": 1.94,
      "eval_blimp_filtered_avg": 0.5177611940298508,
      "eval_blimp_filtered_std": 0.005552093716543691,
      "step": 1700
    },
    {
      "epoch": 1.94,
      "eval_blimp_supplement_avg": 0.5452586206896551,
      "eval_blimp_supplement_std": 0.02300868182607275,
      "step": 1700
    },
    {
      "epoch": 1.94,
      "eval_vqa_filtered_avg": 0.33,
      "eval_vqa_filtered_std": 0.04725815626252606,
      "step": 1700
    },
    {
      "epoch": 1.94,
      "eval_winoground_filtered_avg": 0.5,
      "eval_winoground_filtered_std": 0.050251890762960605,
      "step": 1700
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7300250838856243,
      "learning_rate": 0.00041932984959561713,
      "loss": 4.4523,
      "step": 1701
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.618073334619606,
      "learning_rate": 0.00041912601983608863,
      "loss": 4.3033,
      "step": 1702
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4032577676168612,
      "learning_rate": 0.00041892212476947607,
      "loss": 4.4134,
      "step": 1703
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9148487490119513,
      "learning_rate": 0.00041871816450755845,
      "loss": 4.3107,
      "step": 1704
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0026347803407174,
      "learning_rate": 0.0004185141391621504,
      "loss": 4.273,
      "step": 1705
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1976773350505738,
      "learning_rate": 0.0004183100488451026,
      "loss": 4.5786,
      "step": 1706
    },
    {
      "epoch": 1.95,
      "grad_norm": 2.3566242182390615,
      "learning_rate": 0.00041810589366830087,
      "loss": 4.4315,
      "step": 1707
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9167315515058958,
      "learning_rate": 0.0004179016737436672,
      "loss": 4.3555,
      "step": 1708
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.6255470256391695,
      "learning_rate": 0.0004176973891831583,
      "loss": 4.2553,
      "step": 1709
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3376835619454706,
      "learning_rate": 0.0004174930400987671,
      "loss": 4.3643,
      "step": 1710
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.6461831683423362,
      "learning_rate": 0.0004172886266025213,
      "loss": 4.3821,
      "step": 1711
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.488299611786309,
      "learning_rate": 0.00041708414880648435,
      "loss": 4.3476,
      "step": 1712
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.6603762085036762,
      "learning_rate": 0.00041687960682275463,
      "loss": 4.2867,
      "step": 1713
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2553519366708934,
      "learning_rate": 0.0004166750007634658,
      "loss": 4.4367,
      "step": 1714
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7761759439308343,
      "learning_rate": 0.00041647033074078687,
      "loss": 4.4742,
      "step": 1715
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.675467029815893,
      "learning_rate": 0.00041626559686692153,
      "loss": 4.2413,
      "step": 1716
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.8284579231336491,
      "learning_rate": 0.00041606079925410886,
      "loss": 4.2623,
      "step": 1717
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7269783456016601,
      "learning_rate": 0.0004158559380146223,
      "loss": 4.4271,
      "step": 1718
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.8173792912053426,
      "learning_rate": 0.0004156510132607711,
      "loss": 4.3246,
      "step": 1719
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.9097719074472587,
      "learning_rate": 0.0004154460251048985,
      "loss": 4.5922,
      "step": 1720
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.529644444988264,
      "learning_rate": 0.0004152409736593827,
      "loss": 4.4221,
      "step": 1721
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7221716303188545,
      "learning_rate": 0.0004150358590366369,
      "loss": 4.2605,
      "step": 1722
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.9174787508302422,
      "learning_rate": 0.0004148306813491087,
      "loss": 4.3734,
      "step": 1723
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.6158856805895577,
      "learning_rate": 0.00041462544070928017,
      "loss": 4.3793,
      "step": 1724
    },
    {
      "epoch": 1.97,
      "grad_norm": 2.2094503159869308,
      "learning_rate": 0.0004144201372296681,
      "loss": 4.4607,
      "step": 1725
    },
    {
      "epoch": 1.97,
      "grad_norm": 2.024960978003341,
      "learning_rate": 0.0004142147710228236,
      "loss": 4.4571,
      "step": 1726
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1468928219261922,
      "learning_rate": 0.00041400934220133233,
      "loss": 4.443,
      "step": 1727
    },
    {
      "epoch": 1.97,
      "grad_norm": 2.014696256779845,
      "learning_rate": 0.00041380385087781403,
      "loss": 4.3176,
      "step": 1728
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2621236752226246,
      "learning_rate": 0.000413598297164923,
      "loss": 4.402,
      "step": 1729
    },
    {
      "epoch": 1.98,
      "grad_norm": 2.3179207177525445,
      "learning_rate": 0.00041339268117534736,
      "loss": 4.3529,
      "step": 1730
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3836473571134773,
      "learning_rate": 0.00041318700302180966,
      "loss": 4.4947,
      "step": 1731
    },
    {
      "epoch": 1.98,
      "grad_norm": 3.407274018527,
      "learning_rate": 0.0004129812628170667,
      "loss": 4.449,
      "step": 1732
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.6947471370419148,
      "learning_rate": 0.0004127754606739088,
      "loss": 4.5125,
      "step": 1733
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.5929383599010065,
      "learning_rate": 0.0004125695967051605,
      "loss": 4.3996,
      "step": 1734
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.6965220077427591,
      "learning_rate": 0.00041236367102368037,
      "loss": 4.3799,
      "step": 1735
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7655269554544015,
      "learning_rate": 0.0004121576837423606,
      "loss": 4.4077,
      "step": 1736
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1175446043421966,
      "learning_rate": 0.00041195163497412723,
      "loss": 4.4055,
      "step": 1737
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4566760564025911,
      "learning_rate": 0.00041174552483193993,
      "loss": 4.4222,
      "step": 1738
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4581941908818814,
      "learning_rate": 0.00041153935342879214,
      "loss": 4.374,
      "step": 1739
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5184799360393917,
      "learning_rate": 0.000411333120877711,
      "loss": 4.402,
      "step": 1740
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.5027711746244671,
      "learning_rate": 0.00041112682729175663,
      "loss": 4.1992,
      "step": 1741
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.8530257331050496,
      "learning_rate": 0.0004109204727840234,
      "loss": 4.2583,
      "step": 1742
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.5710299656445248,
      "learning_rate": 0.00041071405746763835,
      "loss": 4.5542,
      "step": 1743
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.7936214798249622,
      "learning_rate": 0.0004105075814557623,
      "loss": 4.4091,
      "step": 1744
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.7464092300216617,
      "learning_rate": 0.0004103010448615892,
      "loss": 4.2143,
      "step": 1745
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.621981372302032,
      "learning_rate": 0.0004100944477983462,
      "loss": 4.3101,
      "step": 1746
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.8856327944858391,
      "learning_rate": 0.00040988779037929364,
      "loss": 4.2575,
      "step": 1747
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.5404639675732871,
      "learning_rate": 0.0004096810727177249,
      "loss": 4.4514,
      "step": 1748
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4846447606399151,
      "learning_rate": 0.0004094742949269666,
      "loss": 4.3881,
      "step": 1749
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1916002205291676,
      "learning_rate": 0.00040926745712037784,
      "loss": 4.5727,
      "step": 1750
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.6793960794289209,
      "learning_rate": 0.00040906055941135116,
      "loss": 4.3721,
      "step": 1751
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.872012429068228,
      "learning_rate": 0.00040885360191331166,
      "loss": 4.4496,
      "step": 1752
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0494357633900915,
      "learning_rate": 0.00040864658473971724,
      "loss": 4.3193,
      "step": 1753
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.8626468099760267,
      "learning_rate": 0.00040843950800405844,
      "loss": 4.4595,
      "step": 1754
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.5098361943562533,
      "learning_rate": 0.00040823237181985867,
      "loss": 4.4016,
      "step": 1755
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5154094787466759,
      "learning_rate": 0.00040802517630067383,
      "loss": 4.4414,
      "step": 1756
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.4639249290006274,
      "learning_rate": 0.0004078179215600922,
      "loss": 4.4489,
      "step": 1757
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.172193060316531,
      "learning_rate": 0.0004076106077117348,
      "loss": 4.4148,
      "step": 1758
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1144647373015413,
      "learning_rate": 0.00040740323486925493,
      "loss": 4.4443,
      "step": 1759
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5604760003574236,
      "learning_rate": 0.00040719580314633807,
      "loss": 4.498,
      "step": 1760
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.9305536586387498,
      "learning_rate": 0.0004069883126567022,
      "loss": 4.2186,
      "step": 1761
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3316756240173309,
      "learning_rate": 0.00040678076351409736,
      "loss": 4.3583,
      "step": 1762
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.5015316637154318,
      "learning_rate": 0.0004065731558323059,
      "loss": 4.3843,
      "step": 1763
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.6919509052401066,
      "learning_rate": 0.00040636548972514217,
      "loss": 4.4375,
      "step": 1764
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.5899623439957696,
      "learning_rate": 0.0004061577653064526,
      "loss": 4.3863,
      "step": 1765
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.9256181552407693,
      "learning_rate": 0.0004059499826901155,
      "loss": 4.4612,
      "step": 1766
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.7401326536428897,
      "learning_rate": 0.0004057421419900412,
      "loss": 4.3706,
      "step": 1767
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.6909642052377967,
      "learning_rate": 0.0004055342433201717,
      "loss": 4.3979,
      "step": 1768
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.6845071193232816,
      "learning_rate": 0.000405326286794481,
      "loss": 4.4335,
      "step": 1769
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.5459638238364051,
      "learning_rate": 0.00040511827252697467,
      "loss": 4.4919,
      "step": 1770
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.7967376620133855,
      "learning_rate": 0.00040491020063168994,
      "loss": 4.2999,
      "step": 1771
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.8004048381600135,
      "learning_rate": 0.0004047020712226957,
      "loss": 4.4307,
      "step": 1772
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.8287935906479291,
      "learning_rate": 0.0004044938844140924,
      "loss": 4.3612,
      "step": 1773
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.9836174616291818,
      "learning_rate": 0.0004042856403200118,
      "loss": 4.5172,
      "step": 1774
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.5205850853546091,
      "learning_rate": 0.00040407733905461715,
      "loss": 4.3342,
      "step": 1775
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.650782700890393,
      "learning_rate": 0.00040386898073210305,
      "loss": 4.2359,
      "step": 1776
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.565982894072005,
      "learning_rate": 0.0004036605654666954,
      "loss": 4.3739,
      "step": 1777
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.4863634811990835,
      "learning_rate": 0.0004034520933726514,
      "loss": 4.4819,
      "step": 1778
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.9509882863250791,
      "learning_rate": 0.0004032435645642591,
      "loss": 4.4461,
      "step": 1779
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.691904850770592,
      "learning_rate": 0.00040303497915583804,
      "loss": 4.3363,
      "step": 1780
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.8285217994520352,
      "learning_rate": 0.00040282633726173856,
      "loss": 4.2729,
      "step": 1781
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0438603557221944,
      "learning_rate": 0.0004026176389963419,
      "loss": 4.3537,
      "step": 1782
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.4947396635819412,
      "learning_rate": 0.00040240888447406033,
      "loss": 4.4694,
      "step": 1783
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.6256779128679641,
      "learning_rate": 0.00040220007380933686,
      "loss": 4.4824,
      "step": 1784
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.8283134875351507,
      "learning_rate": 0.00040199120711664565,
      "loss": 4.4153,
      "step": 1785
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.7592506693934623,
      "learning_rate": 0.000401782284510491,
      "loss": 4.5871,
      "step": 1786
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.4838933637138438,
      "learning_rate": 0.0004015733061054083,
      "loss": 4.5312,
      "step": 1787
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.5259381115834612,
      "learning_rate": 0.0004013642720159631,
      "loss": 4.4037,
      "step": 1788
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.43242940515886147,
      "learning_rate": 0.0004011551823567521,
      "loss": 4.303,
      "step": 1789
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.9883664804878028,
      "learning_rate": 0.0004009460372424018,
      "loss": 4.4398,
      "step": 1790
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.6022556328322494,
      "learning_rate": 0.00040073683678756974,
      "loss": 4.4385,
      "step": 1791
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.553223105640268,
      "learning_rate": 0.0004005275811069432,
      "loss": 4.3427,
      "step": 1792
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.6100627113656589,
      "learning_rate": 0.0004003182703152402,
      "loss": 4.436,
      "step": 1793
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.7707356538582438,
      "learning_rate": 0.00040010890452720854,
      "loss": 4.4925,
      "step": 1794
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.44880951159965,
      "learning_rate": 0.00039989948385762666,
      "loss": 4.5008,
      "step": 1795
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.48781824236574567,
      "learning_rate": 0.0003996900084213027,
      "loss": 4.3456,
      "step": 1796
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.5701972776370563,
      "learning_rate": 0.0003994804783330751,
      "loss": 4.2546,
      "step": 1797
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.9212027993139891,
      "learning_rate": 0.00039927089370781193,
      "loss": 4.4039,
      "step": 1798
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.6057753769984237,
      "learning_rate": 0.0003990612546604114,
      "loss": 4.3214,
      "step": 1799
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.9917820280240645,
      "learning_rate": 0.0003988515613058016,
      "loss": 4.417,
      "step": 1800
    },
    {
      "epoch": 2.06,
      "eval_blimp_filtered_avg": 0.5180597014925373,
      "eval_blimp_filtered_std": 0.0055886917359090175,
      "step": 1800
    },
    {
      "epoch": 2.06,
      "eval_blimp_supplement_avg": 0.5517241379310345,
      "eval_blimp_supplement_std": 0.02311303070191751,
      "step": 1800
    },
    {
      "epoch": 2.06,
      "eval_vqa_filtered_avg": 0.34,
      "eval_vqa_filtered_std": 0.047609522856952365,
      "step": 1800
    },
    {
      "epoch": 2.06,
      "eval_winoground_filtered_avg": 0.48,
      "eval_winoground_filtered_std": 0.05021167315686779,
      "step": 1800
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0449728097343656,
      "learning_rate": 0.0003986418137589403,
      "loss": 4.3009,
      "step": 1801
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.5545312366554692,
      "learning_rate": 0.000398432012134815,
      "loss": 4.2492,
      "step": 1802
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3840083468460764,
      "learning_rate": 0.0003982221565484427,
      "loss": 4.3108,
      "step": 1803
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.339426742976476,
      "learning_rate": 0.0003980122471148701,
      "loss": 4.3457,
      "step": 1804
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.6107240803312746,
      "learning_rate": 0.0003978022839491736,
      "loss": 4.3249,
      "step": 1805
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.6606070708023982,
      "learning_rate": 0.00039759226716645877,
      "loss": 4.3759,
      "step": 1806
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.8520228054668548,
      "learning_rate": 0.00039738219688186064,
      "loss": 4.3245,
      "step": 1807
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2459207869462223,
      "learning_rate": 0.0003971720732105437,
      "loss": 4.4886,
      "step": 1808
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.6505091954033541,
      "learning_rate": 0.0003969618962677016,
      "loss": 4.3171,
      "step": 1809
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4576826547423836,
      "learning_rate": 0.0003967516661685573,
      "loss": 4.3231,
      "step": 1810
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.939585752353907,
      "learning_rate": 0.0003965413830283627,
      "loss": 4.391,
      "step": 1811
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.3260532380080146,
      "learning_rate": 0.0003963310469623989,
      "loss": 4.2982,
      "step": 1812
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.8834472439386788,
      "learning_rate": 0.00039612065808597614,
      "loss": 4.2209,
      "step": 1813
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.9738929503949598,
      "learning_rate": 0.00039591021651443347,
      "loss": 4.2313,
      "step": 1814
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.8016675042380728,
      "learning_rate": 0.00039569972236313866,
      "loss": 4.2145,
      "step": 1815
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3837154419679747,
      "learning_rate": 0.0003954891757474887,
      "loss": 4.4249,
      "step": 1816
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.5619410145955279,
      "learning_rate": 0.000395278576782909,
      "loss": 4.2743,
      "step": 1817
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.0269468865765687,
      "learning_rate": 0.000395067925584854,
      "loss": 4.4762,
      "step": 1818
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2245096716925368,
      "learning_rate": 0.0003948572222688063,
      "loss": 4.2368,
      "step": 1819
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1358501906530618,
      "learning_rate": 0.0003946464669502775,
      "loss": 4.205,
      "step": 1820
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.5126446303714004,
      "learning_rate": 0.00039443565974480767,
      "loss": 4.2779,
      "step": 1821
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.0523584443418934,
      "learning_rate": 0.00039422480076796494,
      "loss": 4.3968,
      "step": 1822
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.645138812274915,
      "learning_rate": 0.0003940138901353463,
      "loss": 4.3991,
      "step": 1823
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.6389423193212534,
      "learning_rate": 0.00039380292796257674,
      "loss": 4.4694,
      "step": 1824
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.724346591718879,
      "learning_rate": 0.0003935919143653098,
      "loss": 4.3319,
      "step": 1825
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.7907655729551181,
      "learning_rate": 0.00039338084945922684,
      "loss": 4.4086,
      "step": 1826
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.030241278632822,
      "learning_rate": 0.0003931697333600376,
      "loss": 4.3546,
      "step": 1827
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.8623956694180515,
      "learning_rate": 0.0003929585661834798,
      "loss": 4.3129,
      "step": 1828
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2165064977901894,
      "learning_rate": 0.0003927473480453193,
      "loss": 4.4959,
      "step": 1829
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.0019290359651514,
      "learning_rate": 0.00039253607906134967,
      "loss": 4.4233,
      "step": 1830
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3072799717712242,
      "learning_rate": 0.0003923247593473925,
      "loss": 4.304,
      "step": 1831
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.0400281684940977,
      "learning_rate": 0.00039211338901929706,
      "loss": 4.4678,
      "step": 1832
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.6092118713716793,
      "learning_rate": 0.00039190196819294075,
      "loss": 4.3624,
      "step": 1833
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.5798191804952015,
      "learning_rate": 0.0003916904969842281,
      "loss": 4.4589,
      "step": 1834
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.6506000561530033,
      "learning_rate": 0.0003914789755090916,
      "loss": 4.2735,
      "step": 1835
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.1831595181097803,
      "learning_rate": 0.00039126740388349126,
      "loss": 4.4347,
      "step": 1836
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.5289697997865335,
      "learning_rate": 0.00039105578222341453,
      "loss": 4.4403,
      "step": 1837
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.0147280453085443,
      "learning_rate": 0.00039084411064487645,
      "loss": 4.4506,
      "step": 1838
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.5598534355086349,
      "learning_rate": 0.000390632389263919,
      "loss": 4.3747,
      "step": 1839
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.836513424692338,
      "learning_rate": 0.00039042061819661195,
      "loss": 4.2392,
      "step": 1840
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.5892461694388671,
      "learning_rate": 0.0003902087975590521,
      "loss": 4.3193,
      "step": 1841
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2386409270869203,
      "learning_rate": 0.0003899969274673634,
      "loss": 4.2106,
      "step": 1842
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.8003772093939621,
      "learning_rate": 0.000389785008037697,
      "loss": 4.1398,
      "step": 1843
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1164739756975015,
      "learning_rate": 0.000389573039386231,
      "loss": 4.2384,
      "step": 1844
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.5637190388740277,
      "learning_rate": 0.00038936102162917053,
      "loss": 4.3483,
      "step": 1845
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.8625411023635501,
      "learning_rate": 0.0003891489548827476,
      "loss": 4.1983,
      "step": 1846
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.5319855441958788,
      "learning_rate": 0.0003889368392632212,
      "loss": 4.2883,
      "step": 1847
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.6819321824673902,
      "learning_rate": 0.000388724674886877,
      "loss": 4.2565,
      "step": 1848
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.569093534872474,
      "learning_rate": 0.0003885124618700274,
      "loss": 4.3668,
      "step": 1849
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.7346841149769192,
      "learning_rate": 0.00038830020032901147,
      "loss": 4.3794,
      "step": 1850
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.090468755953588,
      "learning_rate": 0.000388087890380195,
      "loss": 4.3192,
      "step": 1851
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.6452335895493556,
      "learning_rate": 0.0003878755321399702,
      "loss": 4.2989,
      "step": 1852
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.6382796458995724,
      "learning_rate": 0.00038766312572475575,
      "loss": 4.2366,
      "step": 1853
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.6688957044362543,
      "learning_rate": 0.0003874506712509967,
      "loss": 4.3003,
      "step": 1854
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.024630627076105,
      "learning_rate": 0.00038723816883516463,
      "loss": 4.2687,
      "step": 1855
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.5303406699603224,
      "learning_rate": 0.0003870256185937573,
      "loss": 4.2146,
      "step": 1856
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.046553373277083,
      "learning_rate": 0.0003868130206432986,
      "loss": 4.4741,
      "step": 1857
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.5178050947537223,
      "learning_rate": 0.0003866003751003386,
      "loss": 4.3701,
      "step": 1858
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.7577686674214523,
      "learning_rate": 0.00038638768208145374,
      "loss": 4.2662,
      "step": 1859
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3456568053127247,
      "learning_rate": 0.000386174941703246,
      "loss": 4.3344,
      "step": 1860
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.847863919004829,
      "learning_rate": 0.0003859621540823438,
      "loss": 4.4171,
      "step": 1861
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.117724752105106,
      "learning_rate": 0.000385749319335401,
      "loss": 4.2398,
      "step": 1862
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.6255588774099449,
      "learning_rate": 0.00038553643757909786,
      "loss": 4.3058,
      "step": 1863
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.8527020243574898,
      "learning_rate": 0.0003853235089301398,
      "loss": 4.2551,
      "step": 1864
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.179348689486761,
      "learning_rate": 0.0003851105335052585,
      "loss": 4.4247,
      "step": 1865
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.6950032175360856,
      "learning_rate": 0.00038489751142121065,
      "loss": 4.1867,
      "step": 1866
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3516421885910455,
      "learning_rate": 0.00038468444279477924,
      "loss": 4.2827,
      "step": 1867
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.9861483577316386,
      "learning_rate": 0.00038447132774277227,
      "loss": 4.3759,
      "step": 1868
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.6639698920242082,
      "learning_rate": 0.0003842581663820234,
      "loss": 4.3996,
      "step": 1869
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.7467008123060245,
      "learning_rate": 0.00038404495882939153,
      "loss": 4.4333,
      "step": 1870
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.0172595251619347,
      "learning_rate": 0.0003838317052017612,
      "loss": 4.3245,
      "step": 1871
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.5695280618302556,
      "learning_rate": 0.0003836184056160417,
      "loss": 4.3191,
      "step": 1872
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.5309567551676843,
      "learning_rate": 0.000383405060189168,
      "loss": 4.4582,
      "step": 1873
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2663917207564606,
      "learning_rate": 0.00038319166903809983,
      "loss": 4.5093,
      "step": 1874
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.562629003857273,
      "learning_rate": 0.0003829782322798224,
      "loss": 4.4028,
      "step": 1875
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4049115481085637,
      "learning_rate": 0.0003827647500313456,
      "loss": 4.2516,
      "step": 1876
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.46218488290562715,
      "learning_rate": 0.00038255122240970417,
      "loss": 4.3072,
      "step": 1877
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.3873334639259354,
      "learning_rate": 0.000382337649531958,
      "loss": 4.4592,
      "step": 1878
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.217042444620816,
      "learning_rate": 0.0003821240315151918,
      "loss": 4.2833,
      "step": 1879
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.6582655093268786,
      "learning_rate": 0.0003819103684765148,
      "loss": 4.2842,
      "step": 1880
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3972401654050137,
      "learning_rate": 0.00038169666053306114,
      "loss": 4.1703,
      "step": 1881
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.7467223298150372,
      "learning_rate": 0.0003814829078019892,
      "loss": 4.4403,
      "step": 1882
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.0411155739528064,
      "learning_rate": 0.00038126911040048243,
      "loss": 4.2929,
      "step": 1883
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.7678010807756125,
      "learning_rate": 0.0003810552684457484,
      "loss": 4.2307,
      "step": 1884
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4847062489648213,
      "learning_rate": 0.0003808413820550192,
      "loss": 4.179,
      "step": 1885
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1084009441128622,
      "learning_rate": 0.0003806274513455514,
      "loss": 4.3166,
      "step": 1886
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.7340223164359629,
      "learning_rate": 0.00038041347643462577,
      "loss": 4.3272,
      "step": 1887
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.8556201502140945,
      "learning_rate": 0.00038019945743954713,
      "loss": 4.3661,
      "step": 1888
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.9375531938463024,
      "learning_rate": 0.0003799853944776447,
      "loss": 4.2137,
      "step": 1889
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.7251919281605783,
      "learning_rate": 0.0003797712876662719,
      "loss": 4.433,
      "step": 1890
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3056054653112923,
      "learning_rate": 0.00037955713712280604,
      "loss": 4.1496,
      "step": 1891
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.916591949066931,
      "learning_rate": 0.0003793429429646483,
      "loss": 4.1207,
      "step": 1892
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.288326928281593,
      "learning_rate": 0.00037912870530922374,
      "loss": 4.4123,
      "step": 1893
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5115048686310228,
      "learning_rate": 0.00037891442427398156,
      "loss": 4.5052,
      "step": 1894
    },
    {
      "epoch": 2.17,
      "grad_norm": 3.173971963082131,
      "learning_rate": 0.0003787000999763947,
      "loss": 4.4974,
      "step": 1895
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.6209881081686413,
      "learning_rate": 0.0003784857325339595,
      "loss": 4.0736,
      "step": 1896
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.6068529181484643,
      "learning_rate": 0.0003782713220641962,
      "loss": 4.5044,
      "step": 1897
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1037728457052378,
      "learning_rate": 0.0003780568686846486,
      "loss": 4.4919,
      "step": 1898
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.37521295134448,
      "learning_rate": 0.00037784237251288407,
      "loss": 4.2654,
      "step": 1899
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4949886192277226,
      "learning_rate": 0.0003776278336664932,
      "loss": 4.2359,
      "step": 1900
    },
    {
      "epoch": 2.17,
      "eval_blimp_filtered_avg": 0.5161194029850746,
      "eval_blimp_filtered_std": 0.005657380882384819,
      "step": 1900
    },
    {
      "epoch": 2.17,
      "eval_blimp_supplement_avg": 0.5129310344827587,
      "eval_blimp_supplement_std": 0.02313771269549358,
      "step": 1900
    },
    {
      "epoch": 2.17,
      "eval_vqa_filtered_avg": 0.35,
      "eval_vqa_filtered_std": 0.0479372485441102,
      "step": 1900
    },
    {
      "epoch": 2.17,
      "eval_winoground_filtered_avg": 0.47,
      "eval_winoground_filtered_std": 0.05016135580465919,
      "step": 1900
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.5820969312267368,
      "learning_rate": 0.00037741325226309025,
      "loss": 4.321,
      "step": 1901
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.0901489784564493,
      "learning_rate": 0.0003771986284203128,
      "loss": 4.2728,
      "step": 1902
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.7353463222580834,
      "learning_rate": 0.00037698396225582147,
      "loss": 4.5326,
      "step": 1903
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.58336009380215,
      "learning_rate": 0.00037676925388730023,
      "loss": 4.4527,
      "step": 1904
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.772569497686848,
      "learning_rate": 0.0003765545034324561,
      "loss": 4.2108,
      "step": 1905
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1086024571479756,
      "learning_rate": 0.00037633971100901946,
      "loss": 4.3602,
      "step": 1906
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.7145970974797617,
      "learning_rate": 0.0003761248767347435,
      "loss": 4.1881,
      "step": 1907
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.8056646561604979,
      "learning_rate": 0.000375910000727404,
      "loss": 4.2723,
      "step": 1908
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.910407155458624,
      "learning_rate": 0.00037569508310480017,
      "loss": 4.2565,
      "step": 1909
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3524974809425307,
      "learning_rate": 0.0003754801239847539,
      "loss": 4.4411,
      "step": 1910
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.915177277794558,
      "learning_rate": 0.0003752651234851096,
      "loss": 4.4339,
      "step": 1911
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.6259194071439245,
      "learning_rate": 0.0003750500817237347,
      "loss": 4.258,
      "step": 1912
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.0877094530218336,
      "learning_rate": 0.0003748349988185188,
      "loss": 4.4552,
      "step": 1913
    },
    {
      "epoch": 2.19,
      "grad_norm": 3.3150051661743203,
      "learning_rate": 0.0003746198748873745,
      "loss": 4.5314,
      "step": 1914
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.8936618888231032,
      "learning_rate": 0.0003744047100482367,
      "loss": 4.2391,
      "step": 1915
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.8156435653029486,
      "learning_rate": 0.00037418950441906265,
      "loss": 4.3262,
      "step": 1916
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.827652116930843,
      "learning_rate": 0.00037397425811783213,
      "loss": 4.1682,
      "step": 1917
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.069305715030624,
      "learning_rate": 0.0003737589712625472,
      "loss": 4.2621,
      "step": 1918
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.6345891548158528,
      "learning_rate": 0.00037354364397123197,
      "loss": 4.3089,
      "step": 1919
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.1556614565244177,
      "learning_rate": 0.0003733282763619328,
      "loss": 4.4017,
      "step": 1920
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1481279313257724,
      "learning_rate": 0.00037311286855271834,
      "loss": 4.2642,
      "step": 1921
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.6249412811920643,
      "learning_rate": 0.00037289742066167907,
      "loss": 4.301,
      "step": 1922
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.8142456024049323,
      "learning_rate": 0.0003726819328069275,
      "loss": 4.4713,
      "step": 1923
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.8963025194147003,
      "learning_rate": 0.00037246640510659796,
      "loss": 4.184,
      "step": 1924
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.955004418327837,
      "learning_rate": 0.00037225083767884683,
      "loss": 4.326,
      "step": 1925
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1953897342261859,
      "learning_rate": 0.00037203523064185213,
      "loss": 4.2717,
      "step": 1926
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3617322372185952,
      "learning_rate": 0.0003718195841138137,
      "loss": 4.5012,
      "step": 1927
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.746900102699365,
      "learning_rate": 0.00037160389821295277,
      "loss": 4.4317,
      "step": 1928
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.8559967701882609,
      "learning_rate": 0.0003713881730575125,
      "loss": 4.3187,
      "step": 1929
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1399301083678148,
      "learning_rate": 0.00037117240876575715,
      "loss": 4.4031,
      "step": 1930
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.8227259263083517,
      "learning_rate": 0.0003709566054559731,
      "loss": 4.273,
      "step": 1931
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.5822931495425235,
      "learning_rate": 0.0003707407632464674,
      "loss": 4.2315,
      "step": 1932
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9058795621979164,
      "learning_rate": 0.00037052488225556874,
      "loss": 4.1084,
      "step": 1933
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.373272974683091,
      "learning_rate": 0.0003703089626016272,
      "loss": 4.3785,
      "step": 1934
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.7772662032031363,
      "learning_rate": 0.00037009300440301386,
      "loss": 4.2644,
      "step": 1935
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2646756676713584,
      "learning_rate": 0.00036987700777812093,
      "loss": 4.1889,
      "step": 1936
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.6197336335870793,
      "learning_rate": 0.00036966097284536195,
      "loss": 4.2526,
      "step": 1937
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.8004531813768027,
      "learning_rate": 0.00036944489972317104,
      "loss": 4.2407,
      "step": 1938
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.5476796485163309,
      "learning_rate": 0.0003692287885300037,
      "loss": 4.1308,
      "step": 1939
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.7083165043678129,
      "learning_rate": 0.00036901263938433573,
      "loss": 4.1768,
      "step": 1940
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.7104264981580629,
      "learning_rate": 0.0003687964524046644,
      "loss": 4.2019,
      "step": 1941
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.6861222249003641,
      "learning_rate": 0.00036858022770950725,
      "loss": 4.4594,
      "step": 1942
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.5694346035106297,
      "learning_rate": 0.00036836396541740273,
      "loss": 4.429,
      "step": 1943
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.327718001828511,
      "learning_rate": 0.00036814766564690975,
      "loss": 4.2689,
      "step": 1944
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.5119791783496837,
      "learning_rate": 0.00036793132851660796,
      "loss": 4.2603,
      "step": 1945
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.8128235488407769,
      "learning_rate": 0.00036771495414509724,
      "loss": 4.1576,
      "step": 1946
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.5913145594958984,
      "learning_rate": 0.0003674985426509981,
      "loss": 4.2332,
      "step": 1947
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.5207294336822073,
      "learning_rate": 0.00036728209415295125,
      "loss": 4.1698,
      "step": 1948
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.6458615980432652,
      "learning_rate": 0.00036706560876961786,
      "loss": 4.2807,
      "step": 1949
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.6757926080366181,
      "learning_rate": 0.0003668490866196791,
      "loss": 4.3613,
      "step": 1950
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.5708256359523067,
      "learning_rate": 0.0003666325278218364,
      "loss": 4.4347,
      "step": 1951
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.249263629449511,
      "learning_rate": 0.00036641593249481144,
      "loss": 4.3893,
      "step": 1952
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9876521414430748,
      "learning_rate": 0.0003661993007573456,
      "loss": 4.2624,
      "step": 1953
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.788202432230404,
      "learning_rate": 0.0003659826327282006,
      "loss": 4.2957,
      "step": 1954
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.370327915307975,
      "learning_rate": 0.00036576592852615765,
      "loss": 4.4122,
      "step": 1955
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1757210412032089,
      "learning_rate": 0.0003655491882700181,
      "loss": 4.1165,
      "step": 1956
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.8528593409148948,
      "learning_rate": 0.00036533241207860296,
      "loss": 4.1394,
      "step": 1957
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.260786243595356,
      "learning_rate": 0.00036511560007075296,
      "loss": 4.3379,
      "step": 1958
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.9210336585158017,
      "learning_rate": 0.00036489875236532835,
      "loss": 4.2524,
      "step": 1959
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.5068193582258416,
      "learning_rate": 0.0003646818690812092,
      "loss": 4.2391,
      "step": 1960
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.0753184504675808,
      "learning_rate": 0.00036446495033729475,
      "loss": 4.4959,
      "step": 1961
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3537620783234052,
      "learning_rate": 0.00036424799625250394,
      "loss": 4.1611,
      "step": 1962
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.5568171824219594,
      "learning_rate": 0.00036403100694577513,
      "loss": 4.3755,
      "step": 1963
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4209395900691182,
      "learning_rate": 0.0003638139825360657,
      "loss": 4.4317,
      "step": 1964
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.0481892760580325,
      "learning_rate": 0.00036359692314235254,
      "loss": 4.1986,
      "step": 1965
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.6713210023040883,
      "learning_rate": 0.0003633798288836316,
      "loss": 4.302,
      "step": 1966
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.986233318460866,
      "learning_rate": 0.00036316269987891783,
      "loss": 4.2121,
      "step": 1967
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.9097523119314848,
      "learning_rate": 0.00036294553624724554,
      "loss": 4.173,
      "step": 1968
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.8761538711462373,
      "learning_rate": 0.00036272833810766777,
      "loss": 4.0156,
      "step": 1969
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.9389458630694194,
      "learning_rate": 0.00036251110557925653,
      "loss": 4.3884,
      "step": 1970
    },
    {
      "epoch": 2.25,
      "grad_norm": 3.5865913904359346,
      "learning_rate": 0.00036229383878110276,
      "loss": 4.5273,
      "step": 1971
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.9757630038370871,
      "learning_rate": 0.00036207653783231603,
      "loss": 4.3175,
      "step": 1972
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.5048807949752656,
      "learning_rate": 0.0003618592028520249,
      "loss": 4.2563,
      "step": 1973
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5768652553963096,
      "learning_rate": 0.0003616418339593763,
      "loss": 4.2402,
      "step": 1974
    },
    {
      "epoch": 2.26,
      "grad_norm": 3.100278801880563,
      "learning_rate": 0.0003614244312735358,
      "loss": 4.3517,
      "step": 1975
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.0491672597350368,
      "learning_rate": 0.00036120699491368777,
      "loss": 4.2304,
      "step": 1976
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.024384941169139,
      "learning_rate": 0.00036098952499903473,
      "loss": 4.444,
      "step": 1977
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.768458783703319,
      "learning_rate": 0.0003607720216487977,
      "loss": 4.5618,
      "step": 1978
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.750316020921413,
      "learning_rate": 0.0003605544849822159,
      "loss": 4.3365,
      "step": 1979
    },
    {
      "epoch": 2.26,
      "grad_norm": 3.8992631162026465,
      "learning_rate": 0.00036033691511854724,
      "loss": 4.2631,
      "step": 1980
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4199882641157622,
      "learning_rate": 0.00036011931217706734,
      "loss": 4.333,
      "step": 1981
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.1468013352365176,
      "learning_rate": 0.00035990167627707024,
      "loss": 4.4306,
      "step": 1982
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.0580504763610508,
      "learning_rate": 0.00035968400753786777,
      "loss": 4.4036,
      "step": 1983
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.9848472814343332,
      "learning_rate": 0.00035946630607879026,
      "loss": 4.2871,
      "step": 1984
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.508705835087846,
      "learning_rate": 0.0003592485720191854,
      "loss": 4.2564,
      "step": 1985
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3557176853602833,
      "learning_rate": 0.00035903080547841924,
      "loss": 4.2564,
      "step": 1986
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2079872770352276,
      "learning_rate": 0.00035881300657587513,
      "loss": 4.2786,
      "step": 1987
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1348203234397871,
      "learning_rate": 0.0003585951754309548,
      "loss": 4.264,
      "step": 1988
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3662149553607998,
      "learning_rate": 0.000358377312163077,
      "loss": 4.2526,
      "step": 1989
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.6185461414346178,
      "learning_rate": 0.00035815941689167866,
      "loss": 4.1711,
      "step": 1990
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.2528369414675007,
      "learning_rate": 0.0003579414897362138,
      "loss": 4.329,
      "step": 1991
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.538266406416692,
      "learning_rate": 0.00035772353081615416,
      "loss": 4.3181,
      "step": 1992
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.844320560985542,
      "learning_rate": 0.00035750554025098895,
      "loss": 4.4685,
      "step": 1993
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1949789412577927,
      "learning_rate": 0.00035728751816022444,
      "loss": 4.3234,
      "step": 1994
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.5230718564159114,
      "learning_rate": 0.0003570694646633845,
      "loss": 4.291,
      "step": 1995
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2895626638097109,
      "learning_rate": 0.0003568513798800101,
      "loss": 4.2317,
      "step": 1996
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2685763777791328,
      "learning_rate": 0.0003566332639296592,
      "loss": 4.3211,
      "step": 1997
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.424884858269157,
      "learning_rate": 0.00035641511693190715,
      "loss": 4.2813,
      "step": 1998
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.9950928632674216,
      "learning_rate": 0.000356196939006346,
      "loss": 4.3262,
      "step": 1999
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.0404976954183174,
      "learning_rate": 0.00035597873027258515,
      "loss": 4.2821,
      "step": 2000
    },
    {
      "epoch": 2.29,
      "eval_blimp_filtered_avg": 0.5262686567164179,
      "eval_blimp_filtered_std": 0.005585973329074082,
      "step": 2000
    },
    {
      "epoch": 2.29,
      "eval_blimp_supplement_avg": 0.5732758620689655,
      "eval_blimp_supplement_std": 0.02289001564612328,
      "step": 2000
    },
    {
      "epoch": 2.29,
      "eval_vqa_filtered_avg": 0.35,
      "eval_vqa_filtered_std": 0.0479372485441102,
      "step": 2000
    },
    {
      "epoch": 2.29,
      "eval_winoground_filtered_avg": 0.5,
      "eval_winoground_filtered_std": 0.050251890762960605,
      "step": 2000
    },
    {
      "epoch": 2.29,
      "grad_norm": 5.945998840036045,
      "learning_rate": 0.00035576049085025047,
      "loss": 4.3944,
      "step": 2001
    },
    {
      "epoch": 2.29,
      "grad_norm": 3.325152607339408,
      "learning_rate": 0.0003555422208589849,
      "loss": 4.3731,
      "step": 2002
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.7014394659757326,
      "learning_rate": 0.000355323920418448,
      "loss": 4.3447,
      "step": 2003
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.3452969000796395,
      "learning_rate": 0.0003551055896483162,
      "loss": 4.4419,
      "step": 2004
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.878275280046811,
      "learning_rate": 0.0003548872286682826,
      "loss": 4.4025,
      "step": 2005
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.44421632041388,
      "learning_rate": 0.00035466883759805644,
      "loss": 4.4146,
      "step": 2006
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.8710743348412628,
      "learning_rate": 0.0003544504165573638,
      "loss": 4.454,
      "step": 2007
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.1615288451290926,
      "learning_rate": 0.00035423196566594726,
      "loss": 4.3533,
      "step": 2008
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.1253801064628823,
      "learning_rate": 0.0003540134850435656,
      "loss": 4.4062,
      "step": 2009
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9004705450455097,
      "learning_rate": 0.0003537949748099938,
      "loss": 4.4734,
      "step": 2010
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3047356638265686,
      "learning_rate": 0.00035357643508502343,
      "loss": 4.5541,
      "step": 2011
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.8401669844041402,
      "learning_rate": 0.00035335786598846177,
      "loss": 4.3171,
      "step": 2012
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.7368863093662565,
      "learning_rate": 0.00035313926764013264,
      "loss": 4.2369,
      "step": 2013
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.147659838817981,
      "learning_rate": 0.00035292064015987536,
      "loss": 4.2651,
      "step": 2014
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9213533850711806,
      "learning_rate": 0.000352701983667546,
      "loss": 4.5075,
      "step": 2015
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.1374542468623616,
      "learning_rate": 0.0003524832982830158,
      "loss": 4.2059,
      "step": 2016
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.5738557515061314,
      "learning_rate": 0.0003522645841261722,
      "loss": 4.3067,
      "step": 2017
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2020113557378003,
      "learning_rate": 0.0003520458413169183,
      "loss": 4.2509,
      "step": 2018
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.9203762210171189,
      "learning_rate": 0.00035182706997517304,
      "loss": 4.5718,
      "step": 2019
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.8921145514980305,
      "learning_rate": 0.0003516082702208709,
      "loss": 4.3716,
      "step": 2020
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.6362117934549254,
      "learning_rate": 0.0003513894421739618,
      "loss": 4.2583,
      "step": 2021
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.055670495154233,
      "learning_rate": 0.00035117058595441144,
      "loss": 4.1751,
      "step": 2022
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.6940826143919365,
      "learning_rate": 0.00035095170168220085,
      "loss": 4.4344,
      "step": 2023
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.1923763045152749,
      "learning_rate": 0.00035073278947732637,
      "loss": 4.5912,
      "step": 2024
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.1321928326535502,
      "learning_rate": 0.0003505138494597998,
      "loss": 4.4396,
      "step": 2025
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3559095209723704,
      "learning_rate": 0.0003502948817496479,
      "loss": 4.1998,
      "step": 2026
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3546591263336438,
      "learning_rate": 0.0003500758864669131,
      "loss": 4.302,
      "step": 2027
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5204312070809398,
      "learning_rate": 0.0003498568637316525,
      "loss": 4.1786,
      "step": 2028
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.9016302206450255,
      "learning_rate": 0.0003496378136639384,
      "loss": 4.3215,
      "step": 2029
    },
    {
      "epoch": 2.32,
      "grad_norm": 3.0267062261501403,
      "learning_rate": 0.0003494187363838581,
      "loss": 4.524,
      "step": 2030
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2371216524586623,
      "learning_rate": 0.0003491996320115138,
      "loss": 4.3414,
      "step": 2031
    },
    {
      "epoch": 2.32,
      "grad_norm": 3.237086108649337,
      "learning_rate": 0.00034898050066702263,
      "loss": 4.5709,
      "step": 2032
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.935468666445416,
      "learning_rate": 0.0003487613424705164,
      "loss": 4.3614,
      "step": 2033
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.5961412957220067,
      "learning_rate": 0.00034854215754214163,
      "loss": 4.2368,
      "step": 2034
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.888503913086469,
      "learning_rate": 0.0003483229460020597,
      "loss": 4.3581,
      "step": 2035
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.7336219948015488,
      "learning_rate": 0.0003481037079704462,
      "loss": 4.3256,
      "step": 2036
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.7357181597401315,
      "learning_rate": 0.0003478844435674917,
      "loss": 4.5346,
      "step": 2037
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.3057051491823373,
      "learning_rate": 0.00034766515291340073,
      "loss": 4.2294,
      "step": 2038
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.221037250606619,
      "learning_rate": 0.0003474458361283928,
      "loss": 4.3276,
      "step": 2039
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4980208600000442,
      "learning_rate": 0.00034722649333270125,
      "loss": 4.099,
      "step": 2040
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.6782516188174156,
      "learning_rate": 0.00034700712464657375,
      "loss": 4.3065,
      "step": 2041
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.647976231915306,
      "learning_rate": 0.00034678773019027244,
      "loss": 4.4367,
      "step": 2042
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.1384975327780065,
      "learning_rate": 0.00034656831008407333,
      "loss": 4.3238,
      "step": 2043
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.6587566614298829,
      "learning_rate": 0.00034634886444826656,
      "loss": 4.2638,
      "step": 2044
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.0483363718211718,
      "learning_rate": 0.00034612939340315633,
      "loss": 4.206,
      "step": 2045
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.8146452213085433,
      "learning_rate": 0.0003459098970690607,
      "loss": 4.2934,
      "step": 2046
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4575552758300554,
      "learning_rate": 0.00034569037556631167,
      "loss": 4.4479,
      "step": 2047
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.6097205146382844,
      "learning_rate": 0.0003454708290152548,
      "loss": 4.4297,
      "step": 2048
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.2416012832580923,
      "learning_rate": 0.0003452512575362498,
      "loss": 4.4094,
      "step": 2049
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.6247764502966586,
      "learning_rate": 0.0003450316612496696,
      "loss": 4.4184,
      "step": 2050
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.9558155045077503,
      "learning_rate": 0.00034481204027590105,
      "loss": 4.4383,
      "step": 2051
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.817541657081176,
      "learning_rate": 0.00034459239473534446,
      "loss": 4.1508,
      "step": 2052
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5822298759451072,
      "learning_rate": 0.0003443727247484134,
      "loss": 4.5307,
      "step": 2053
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6360847704370725,
      "learning_rate": 0.00034415303043553537,
      "loss": 4.2026,
      "step": 2054
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.5822928185260654,
      "learning_rate": 0.00034393331191715046,
      "loss": 4.2437,
      "step": 2055
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.012139310296186,
      "learning_rate": 0.0003437135693137127,
      "loss": 4.4965,
      "step": 2056
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.013341996856864,
      "learning_rate": 0.0003434938027456887,
      "loss": 4.3383,
      "step": 2057
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.877136738561602,
      "learning_rate": 0.000343274012333559,
      "loss": 4.266,
      "step": 2058
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.8404633545866598,
      "learning_rate": 0.0003430541981978164,
      "loss": 4.384,
      "step": 2059
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.1630072917249445,
      "learning_rate": 0.0003428343604589673,
      "loss": 4.456,
      "step": 2060
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.2859663350041606,
      "learning_rate": 0.00034261449923753053,
      "loss": 4.692,
      "step": 2061
    },
    {
      "epoch": 2.36,
      "grad_norm": 3.616893376145364,
      "learning_rate": 0.0003423946146540385,
      "loss": 4.5318,
      "step": 2062
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4606122379041813,
      "learning_rate": 0.00034217470682903556,
      "loss": 4.3132,
      "step": 2063
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.506174124411359,
      "learning_rate": 0.0003419547758830796,
      "loss": 4.489,
      "step": 2064
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.8098339563369505,
      "learning_rate": 0.00034173482193674046,
      "loss": 4.4841,
      "step": 2065
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.592640341951285,
      "learning_rate": 0.0003415148451106013,
      "loss": 4.2266,
      "step": 2066
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.9374334977527548,
      "learning_rate": 0.0003412948455252573,
      "loss": 4.322,
      "step": 2067
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.022863282643399,
      "learning_rate": 0.0003410748233013163,
      "loss": 4.3931,
      "step": 2068
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.826713062280081,
      "learning_rate": 0.0003408547785593986,
      "loss": 4.275,
      "step": 2069
    },
    {
      "epoch": 2.37,
      "grad_norm": 3.2849605069189574,
      "learning_rate": 0.0003406347114201368,
      "loss": 4.3879,
      "step": 2070
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.8779037976742416,
      "learning_rate": 0.0003404146220041758,
      "loss": 4.3096,
      "step": 2071
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.8216772367562108,
      "learning_rate": 0.0003401945104321725,
      "loss": 4.1529,
      "step": 2072
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3522536622379977,
      "learning_rate": 0.0003399743768247963,
      "loss": 4.237,
      "step": 2073
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.061632019276528,
      "learning_rate": 0.0003397542213027284,
      "loss": 4.399,
      "step": 2074
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.7650810140939595,
      "learning_rate": 0.0003395340439866623,
      "loss": 4.3029,
      "step": 2075
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.9884145636490744,
      "learning_rate": 0.000339313844997303,
      "loss": 4.2263,
      "step": 2076
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.9462477831585521,
      "learning_rate": 0.00033909362445536795,
      "loss": 4.2402,
      "step": 2077
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3889386826701697,
      "learning_rate": 0.0003388733824815859,
      "loss": 4.2577,
      "step": 2078
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.1715712716663387,
      "learning_rate": 0.0003386531191966977,
      "loss": 4.426,
      "step": 2079
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2911118654801799,
      "learning_rate": 0.0003384328347214557,
      "loss": 4.2857,
      "step": 2080
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.688596967688997,
      "learning_rate": 0.0003382125291766238,
      "loss": 4.3336,
      "step": 2081
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.7028085704812395,
      "learning_rate": 0.00033799220268297773,
      "loss": 4.1727,
      "step": 2082
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5606640292027352,
      "learning_rate": 0.0003377718553613046,
      "loss": 4.5283,
      "step": 2083
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5995840061181728,
      "learning_rate": 0.00033755148733240263,
      "loss": 4.4106,
      "step": 2084
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.072740181271018,
      "learning_rate": 0.0003373310987170818,
      "loss": 4.4641,
      "step": 2085
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9747347984798919,
      "learning_rate": 0.0003371106896361633,
      "loss": 4.2459,
      "step": 2086
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.264381319036501,
      "learning_rate": 0.00033689026021047926,
      "loss": 4.345,
      "step": 2087
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.1683204462778978,
      "learning_rate": 0.00033666981056087336,
      "loss": 4.2796,
      "step": 2088
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.1412289658585866,
      "learning_rate": 0.0003364493408082,
      "loss": 4.3462,
      "step": 2089
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.9921328967977208,
      "learning_rate": 0.0003362288510733249,
      "loss": 4.1843,
      "step": 2090
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.6071874418461483,
      "learning_rate": 0.0003360083414771245,
      "loss": 4.4001,
      "step": 2091
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3315871535438608,
      "learning_rate": 0.00033578781214048636,
      "loss": 4.1977,
      "step": 2092
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.3286616698695477,
      "learning_rate": 0.0003355672631843086,
      "loss": 4.2894,
      "step": 2093
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.5377001062467163,
      "learning_rate": 0.0003353466947295003,
      "loss": 4.0858,
      "step": 2094
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.474559361089454,
      "learning_rate": 0.00033512610689698117,
      "loss": 4.3401,
      "step": 2095
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.8175377242376495,
      "learning_rate": 0.00033490549980768144,
      "loss": 4.1963,
      "step": 2096
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.951475067059975,
      "learning_rate": 0.00033468487358254216,
      "loss": 4.3098,
      "step": 2097
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.218287126185963,
      "learning_rate": 0.0003344642283425146,
      "loss": 4.2653,
      "step": 2098
    },
    {
      "epoch": 2.4,
      "grad_norm": 5.888154726808916,
      "learning_rate": 0.0003342435642085605,
      "loss": 4.5174,
      "step": 2099
    },
    {
      "epoch": 2.4,
      "grad_norm": 5.0807510741464315,
      "learning_rate": 0.0003340228813016521,
      "loss": 4.4391,
      "step": 2100
    },
    {
      "epoch": 2.4,
      "eval_blimp_filtered_avg": 0.5288059701492537,
      "eval_blimp_filtered_std": 0.005531426344079286,
      "step": 2100
    },
    {
      "epoch": 2.4,
      "eval_blimp_supplement_avg": 0.5086206896551724,
      "eval_blimp_supplement_std": 0.022537039417513526,
      "step": 2100
    },
    {
      "epoch": 2.4,
      "eval_vqa_filtered_avg": 0.36,
      "eval_vqa_filtered_std": 0.04824181513244218,
      "step": 2100
    },
    {
      "epoch": 2.4,
      "eval_winoground_filtered_avg": 0.51,
      "eval_winoground_filtered_std": 0.05024183937956912,
      "step": 2100
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.4702287633316455,
      "learning_rate": 0.0003338021797427718,
      "loss": 4.319,
      "step": 2101
    },
    {
      "epoch": 2.4,
      "grad_norm": 6.132327663173627,
      "learning_rate": 0.0003335814596529124,
      "loss": 4.4507,
      "step": 2102
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.639737031001395,
      "learning_rate": 0.0003333607211530765,
      "loss": 4.4528,
      "step": 2103
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.869568203535451,
      "learning_rate": 0.00033313996436427706,
      "loss": 4.3915,
      "step": 2104
    },
    {
      "epoch": 2.41,
      "grad_norm": 3.5555764617915218,
      "learning_rate": 0.00033291918940753733,
      "loss": 4.5243,
      "step": 2105
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.6478901907461402,
      "learning_rate": 0.0003326983964038899,
      "loss": 4.3477,
      "step": 2106
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.509279748641807,
      "learning_rate": 0.00033247758547437766,
      "loss": 4.3939,
      "step": 2107
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.8128713245805477,
      "learning_rate": 0.00033225675674005323,
      "loss": 4.1904,
      "step": 2108
    },
    {
      "epoch": 2.41,
      "grad_norm": 3.6886622153993964,
      "learning_rate": 0.00033203591032197915,
      "loss": 4.3686,
      "step": 2109
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5393696559696803,
      "learning_rate": 0.0003318150463412272,
      "loss": 4.4038,
      "step": 2110
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.0441385921930375,
      "learning_rate": 0.0003315941649188794,
      "loss": 4.39,
      "step": 2111
    },
    {
      "epoch": 2.41,
      "grad_norm": 3.650374379334776,
      "learning_rate": 0.00033137326617602684,
      "loss": 4.4701,
      "step": 2112
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.7872597783197107,
      "learning_rate": 0.0003311523502337704,
      "loss": 4.1471,
      "step": 2113
    },
    {
      "epoch": 2.42,
      "grad_norm": 3.1487838100333367,
      "learning_rate": 0.0003309314172132202,
      "loss": 4.3344,
      "step": 2114
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.3063435308502935,
      "learning_rate": 0.0003307104672354958,
      "loss": 4.456,
      "step": 2115
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.1161791010514834,
      "learning_rate": 0.0003304895004217261,
      "loss": 4.2653,
      "step": 2116
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.4142665959262977,
      "learning_rate": 0.0003302685168930492,
      "loss": 4.457,
      "step": 2117
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.6967440730606428,
      "learning_rate": 0.0003300475167706123,
      "loss": 4.0471,
      "step": 2118
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.312932638723793,
      "learning_rate": 0.0003298265001755718,
      "loss": 4.3997,
      "step": 2119
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.7607761055910163,
      "learning_rate": 0.00032960546722909294,
      "loss": 4.2669,
      "step": 2120
    },
    {
      "epoch": 2.42,
      "grad_norm": 8.274198737154466,
      "learning_rate": 0.0003293844180523503,
      "loss": 5.108,
      "step": 2121
    },
    {
      "epoch": 2.43,
      "grad_norm": 5.655252625282623,
      "learning_rate": 0.00032916335276652703,
      "loss": 4.664,
      "step": 2122
    },
    {
      "epoch": 2.43,
      "grad_norm": 3.5245707123847945,
      "learning_rate": 0.0003289422714928151,
      "loss": 4.7528,
      "step": 2123
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.331277846511171,
      "learning_rate": 0.00032872117435241545,
      "loss": 4.6241,
      "step": 2124
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.890652502540131,
      "learning_rate": 0.0003285000614665375,
      "loss": 4.9838,
      "step": 2125
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6956966277234777,
      "learning_rate": 0.0003282789329563996,
      "loss": 4.7498,
      "step": 2126
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.1082287484685287,
      "learning_rate": 0.0003280577889432284,
      "loss": 4.596,
      "step": 2127
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3020297373927001,
      "learning_rate": 0.0003278366295482591,
      "loss": 4.7916,
      "step": 2128
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.7851203989761089,
      "learning_rate": 0.0003276154548927353,
      "loss": 4.7432,
      "step": 2129
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.1385184540832236,
      "learning_rate": 0.0003273942650979092,
      "loss": 4.6066,
      "step": 2130
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.373740175633821,
      "learning_rate": 0.00032717306028504095,
      "loss": 4.8065,
      "step": 2131
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.9101432415882775,
      "learning_rate": 0.0003269518405753993,
      "loss": 4.6091,
      "step": 2132
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.8910493646524733,
      "learning_rate": 0.0003267306060902609,
      "loss": 4.6262,
      "step": 2133
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5586138721525074,
      "learning_rate": 0.00032650935695091056,
      "loss": 4.5887,
      "step": 2134
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.1091828102231855,
      "learning_rate": 0.000326288093278641,
      "loss": 4.5476,
      "step": 2135
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7272225760692963,
      "learning_rate": 0.00032606681519475335,
      "loss": 4.6667,
      "step": 2136
    },
    {
      "epoch": 2.44,
      "grad_norm": 3.366410812641328,
      "learning_rate": 0.0003258455228205562,
      "loss": 4.7115,
      "step": 2137
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7996883862360058,
      "learning_rate": 0.00032562421627736613,
      "loss": 4.4369,
      "step": 2138
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5531571195220526,
      "learning_rate": 0.0003254028956865073,
      "loss": 4.3716,
      "step": 2139
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.3183097340067484,
      "learning_rate": 0.00032518156116931204,
      "loss": 4.5631,
      "step": 2140
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.26479498633631,
      "learning_rate": 0.00032496021284711987,
      "loss": 4.5874,
      "step": 2141
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.0625356774885313,
      "learning_rate": 0.000324738850841278,
      "loss": 4.5055,
      "step": 2142
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.1089949405040678,
      "learning_rate": 0.00032451747527314123,
      "loss": 4.56,
      "step": 2143
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.953285581630864,
      "learning_rate": 0.0003242960862640717,
      "loss": 4.5574,
      "step": 2144
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.1041663084556672,
      "learning_rate": 0.00032407468393543903,
      "loss": 4.5456,
      "step": 2145
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.42304304205384,
      "learning_rate": 0.00032385326840861996,
      "loss": 4.5343,
      "step": 2146
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.470983964809429,
      "learning_rate": 0.00032363183980499866,
      "loss": 4.5649,
      "step": 2147
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.1466249138533289,
      "learning_rate": 0.00032341039824596626,
      "loss": 4.5266,
      "step": 2148
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.889864992126179,
      "learning_rate": 0.0003231889438529213,
      "loss": 4.4206,
      "step": 2149
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.032203993825711,
      "learning_rate": 0.00032296747674726895,
      "loss": 4.2761,
      "step": 2150
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.7774473849767566,
      "learning_rate": 0.00032274599705042165,
      "loss": 4.2463,
      "step": 2151
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.2466082496018474,
      "learning_rate": 0.00032252450488379875,
      "loss": 4.4972,
      "step": 2152
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6969391965348628,
      "learning_rate": 0.0003223030003688263,
      "loss": 4.5117,
      "step": 2153
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2238753623515313,
      "learning_rate": 0.0003220814836269371,
      "loss": 4.5193,
      "step": 2154
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.5551645975756583,
      "learning_rate": 0.00032185995477957075,
      "loss": 4.4527,
      "step": 2155
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.0916470991071967,
      "learning_rate": 0.0003216384139481735,
      "loss": 4.4665,
      "step": 2156
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.9132242940534141,
      "learning_rate": 0.0003214168612541981,
      "loss": 4.3322,
      "step": 2157
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.8995121921835332,
      "learning_rate": 0.0003211952968191038,
      "loss": 4.3415,
      "step": 2158
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.0949373159669995,
      "learning_rate": 0.00032097372076435635,
      "loss": 4.1976,
      "step": 2159
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.8777176642543518,
      "learning_rate": 0.00032075213321142783,
      "loss": 4.4238,
      "step": 2160
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.2927080568614344,
      "learning_rate": 0.00032053053428179667,
      "loss": 4.6023,
      "step": 2161
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.0130556060030205,
      "learning_rate": 0.00032030892409694734,
      "loss": 4.2252,
      "step": 2162
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.8784144258398178,
      "learning_rate": 0.00032008730277837083,
      "loss": 4.5588,
      "step": 2163
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.8822153552017953,
      "learning_rate": 0.00031986567044756396,
      "loss": 4.4684,
      "step": 2164
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.7763324998164362,
      "learning_rate": 0.00031964402722602965,
      "loss": 4.3765,
      "step": 2165
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.6757496395307732,
      "learning_rate": 0.0003194223732352768,
      "loss": 4.1134,
      "step": 2166
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7161414040768455,
      "learning_rate": 0.0003192007085968202,
      "loss": 4.3379,
      "step": 2167
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.9792547502785244,
      "learning_rate": 0.0003189790334321807,
      "loss": 4.3109,
      "step": 2168
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.6235680072004474,
      "learning_rate": 0.00031875734786288436,
      "loss": 4.3398,
      "step": 2169
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.1369774628459963,
      "learning_rate": 0.00031853565201046363,
      "loss": 4.4931,
      "step": 2170
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.0451450847769115,
      "learning_rate": 0.00031831394599645606,
      "loss": 4.4587,
      "step": 2171
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.1372854242545163,
      "learning_rate": 0.0003180922299424052,
      "loss": 4.4317,
      "step": 2172
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.7165175615987772,
      "learning_rate": 0.00031787050396985965,
      "loss": 4.4702,
      "step": 2173
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.7593508164417825,
      "learning_rate": 0.0003176487682003737,
      "loss": 4.315,
      "step": 2174
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.7147793222473221,
      "learning_rate": 0.00031742702275550723,
      "loss": 4.475,
      "step": 2175
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.8087601859267176,
      "learning_rate": 0.00031720526775682494,
      "loss": 4.1864,
      "step": 2176
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.359881403152142,
      "learning_rate": 0.0003169835033258972,
      "loss": 4.365,
      "step": 2177
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5854663866430123,
      "learning_rate": 0.00031676172958429916,
      "loss": 4.2412,
      "step": 2178
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.9417179104411515,
      "learning_rate": 0.0003165399466536115,
      "loss": 4.3947,
      "step": 2179
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.96477900716057,
      "learning_rate": 0.00031631815465541955,
      "loss": 4.4546,
      "step": 2180
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.6963270614724165,
      "learning_rate": 0.00031609635371131396,
      "loss": 4.4116,
      "step": 2181
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.290415040944062,
      "learning_rate": 0.00031587454394288987,
      "loss": 4.4453,
      "step": 2182
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5534969938751992,
      "learning_rate": 0.00031565272547174765,
      "loss": 4.2669,
      "step": 2183
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.259650263339209,
      "learning_rate": 0.0003154308984194922,
      "loss": 4.3684,
      "step": 2184
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.156932969795427,
      "learning_rate": 0.00031520906290773333,
      "loss": 4.4865,
      "step": 2185
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.9591340804816504,
      "learning_rate": 0.00031498721905808523,
      "loss": 4.4401,
      "step": 2186
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.0239763053563462,
      "learning_rate": 0.0003147653669921669,
      "loss": 4.3722,
      "step": 2187
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.8913263511878871,
      "learning_rate": 0.00031454350683160167,
      "loss": 4.3291,
      "step": 2188
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.072686650128036,
      "learning_rate": 0.0003143216386980174,
      "loss": 4.6129,
      "step": 2189
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.5280147900542698,
      "learning_rate": 0.0003140997627130463,
      "loss": 4.3233,
      "step": 2190
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3157124403149179,
      "learning_rate": 0.00031387787899832487,
      "loss": 4.2192,
      "step": 2191
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.8332603987957956,
      "learning_rate": 0.0003136559876754939,
      "loss": 4.3905,
      "step": 2192
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.940803043047535,
      "learning_rate": 0.0003134340888661983,
      "loss": 4.3358,
      "step": 2193
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2604112487708798,
      "learning_rate": 0.00031321218269208693,
      "loss": 4.1901,
      "step": 2194
    },
    {
      "epoch": 2.51,
      "grad_norm": 5.321321063440286,
      "learning_rate": 0.00031299026927481314,
      "loss": 4.43,
      "step": 2195
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.6846007339144923,
      "learning_rate": 0.0003127683487360338,
      "loss": 4.2143,
      "step": 2196
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.5946532702718637,
      "learning_rate": 0.0003125464211974098,
      "loss": 4.2905,
      "step": 2197
    },
    {
      "epoch": 2.51,
      "grad_norm": 3.4543648926416655,
      "learning_rate": 0.00031232448678060606,
      "loss": 4.3033,
      "step": 2198
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.5003184335551039,
      "learning_rate": 0.000312102545607291,
      "loss": 4.3346,
      "step": 2199
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6895045016514099,
      "learning_rate": 0.00031188059779913707,
      "loss": 4.2035,
      "step": 2200
    },
    {
      "epoch": 2.51,
      "eval_blimp_filtered_avg": 0.5288059701492537,
      "eval_blimp_filtered_std": 0.005435360900795471,
      "step": 2200
    },
    {
      "epoch": 2.51,
      "eval_blimp_supplement_avg": 0.5172413793103449,
      "eval_blimp_supplement_std": 0.022874433495946925,
      "step": 2200
    },
    {
      "epoch": 2.51,
      "eval_vqa_filtered_avg": 0.35,
      "eval_vqa_filtered_std": 0.0479372485441102,
      "step": 2200
    },
    {
      "epoch": 2.51,
      "eval_winoground_filtered_avg": 0.49,
      "eval_winoground_filtered_std": 0.05024183937956912,
      "step": 2200
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.3209570366329744,
      "learning_rate": 0.00031165864347781983,
      "loss": 4.3946,
      "step": 2201
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3459952245884348,
      "learning_rate": 0.000311436682765019,
      "loss": 4.3824,
      "step": 2202
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.9727966252921615,
      "learning_rate": 0.0003112147157824174,
      "loss": 4.3544,
      "step": 2203
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.4600080752103253,
      "learning_rate": 0.0003109927426517016,
      "loss": 4.393,
      "step": 2204
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.7644956125317075,
      "learning_rate": 0.00031077076349456107,
      "loss": 4.327,
      "step": 2205
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4802207567157337,
      "learning_rate": 0.0003105487784326893,
      "loss": 4.3617,
      "step": 2206
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.2045241291362014,
      "learning_rate": 0.00031032678758778225,
      "loss": 4.5737,
      "step": 2207
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.085373145762028,
      "learning_rate": 0.00031010479108153964,
      "loss": 4.2875,
      "step": 2208
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.7149033060207812,
      "learning_rate": 0.0003098827890356637,
      "loss": 4.3896,
      "step": 2209
    },
    {
      "epoch": 2.53,
      "grad_norm": 3.447788438106106,
      "learning_rate": 0.00030966078157186047,
      "loss": 4.3652,
      "step": 2210
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2359919873591187,
      "learning_rate": 0.0003094387688118384,
      "loss": 4.3119,
      "step": 2211
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.8927472061604494,
      "learning_rate": 0.00030921675087730886,
      "loss": 4.5134,
      "step": 2212
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.4089725296561126,
      "learning_rate": 0.00030899472788998623,
      "loss": 4.2828,
      "step": 2213
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.7358133305135421,
      "learning_rate": 0.0003087726999715877,
      "loss": 4.3742,
      "step": 2214
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.8483383066374621,
      "learning_rate": 0.0003085506672438331,
      "loss": 4.3095,
      "step": 2215
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.888578228305651,
      "learning_rate": 0.0003083286298284447,
      "loss": 4.1725,
      "step": 2216
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.671181932320894,
      "learning_rate": 0.0003081065878471477,
      "loss": 4.2494,
      "step": 2217
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3288839725642714,
      "learning_rate": 0.0003078845414216694,
      "loss": 4.415,
      "step": 2218
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.139424986936041,
      "learning_rate": 0.00030766249067374,
      "loss": 4.2287,
      "step": 2219
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.4580496699950736,
      "learning_rate": 0.0003074404357250916,
      "loss": 4.3068,
      "step": 2220
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.9850679353102665,
      "learning_rate": 0.000307218376697459,
      "loss": 4.1687,
      "step": 2221
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.9362252973349048,
      "learning_rate": 0.00030699631371257894,
      "loss": 4.4689,
      "step": 2222
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.250279081020961,
      "learning_rate": 0.00030677424689219056,
      "loss": 4.4239,
      "step": 2223
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.540751604055826,
      "learning_rate": 0.00030655217635803487,
      "loss": 4.4128,
      "step": 2224
    },
    {
      "epoch": 2.54,
      "grad_norm": 5.158894822222172,
      "learning_rate": 0.00030633010223185504,
      "loss": 4.504,
      "step": 2225
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.624796486895582,
      "learning_rate": 0.00030610802463539635,
      "loss": 4.4197,
      "step": 2226
    },
    {
      "epoch": 2.55,
      "grad_norm": 5.438886645806711,
      "learning_rate": 0.00030588594369040565,
      "loss": 4.5508,
      "step": 2227
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.93942618697539,
      "learning_rate": 0.000305663859518632,
      "loss": 4.7087,
      "step": 2228
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.2739636397143284,
      "learning_rate": 0.0003054417722418259,
      "loss": 4.512,
      "step": 2229
    },
    {
      "epoch": 2.55,
      "grad_norm": 3.6609241250958147,
      "learning_rate": 0.00030521968198173987,
      "loss": 4.6358,
      "step": 2230
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.582776370814491,
      "learning_rate": 0.00030499758886012765,
      "loss": 4.7484,
      "step": 2231
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.4414297139026595,
      "learning_rate": 0.000304775492998745,
      "loss": 4.5734,
      "step": 2232
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.173766589208741,
      "learning_rate": 0.0003045533945193488,
      "loss": 4.3983,
      "step": 2233
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.9696876191556565,
      "learning_rate": 0.00030433129354369773,
      "loss": 4.4891,
      "step": 2234
    },
    {
      "epoch": 2.55,
      "grad_norm": 3.338010937453103,
      "learning_rate": 0.0003041091901935515,
      "loss": 4.5843,
      "step": 2235
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.997217390466014,
      "learning_rate": 0.0003038870845906713,
      "loss": 4.5074,
      "step": 2236
    },
    {
      "epoch": 2.56,
      "grad_norm": 5.188210538499853,
      "learning_rate": 0.00030366497685681954,
      "loss": 4.5048,
      "step": 2237
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6740658443714023,
      "learning_rate": 0.0003034428671137599,
      "loss": 4.4301,
      "step": 2238
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.329003769751257,
      "learning_rate": 0.0003032207554832568,
      "loss": 4.3007,
      "step": 2239
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.7048098814772614,
      "learning_rate": 0.00030299864208707607,
      "loss": 4.1912,
      "step": 2240
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4070091656123644,
      "learning_rate": 0.0003027765270469844,
      "loss": 4.2864,
      "step": 2241
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.9162101303481792,
      "learning_rate": 0.0003025544104847493,
      "loss": 4.3008,
      "step": 2242
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.046456942046342,
      "learning_rate": 0.00030233229252213926,
      "loss": 4.4101,
      "step": 2243
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.7535899146283395,
      "learning_rate": 0.00030211017328092327,
      "loss": 4.4258,
      "step": 2244
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.7116533731339875,
      "learning_rate": 0.00030188805288287136,
      "loss": 4.1992,
      "step": 2245
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5018265346131665,
      "learning_rate": 0.00030166593144975403,
      "loss": 4.3136,
      "step": 2246
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.9684224100150691,
      "learning_rate": 0.0003014438091033423,
      "loss": 4.491,
      "step": 2247
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.080287473275391,
      "learning_rate": 0.00030122168596540764,
      "loss": 4.2714,
      "step": 2248
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5536016897440224,
      "learning_rate": 0.0003009995621577224,
      "loss": 4.471,
      "step": 2249
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.1398120180410132,
      "learning_rate": 0.00030077743780205856,
      "loss": 4.1575,
      "step": 2250
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.0505921432368766,
      "learning_rate": 0.00030055531302018907,
      "loss": 4.505,
      "step": 2251
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.6875402043445167,
      "learning_rate": 0.00030033318793388656,
      "loss": 4.3958,
      "step": 2252
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.7150643426274241,
      "learning_rate": 0.0003001110626649244,
      "loss": 4.4882,
      "step": 2253
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.380782414167924,
      "learning_rate": 0.00029988893733507566,
      "loss": 4.3245,
      "step": 2254
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.7020007314942747,
      "learning_rate": 0.00029966681206611344,
      "loss": 4.312,
      "step": 2255
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.5236301090997395,
      "learning_rate": 0.00029944468697981093,
      "loss": 4.2915,
      "step": 2256
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4904266813120155,
      "learning_rate": 0.0002992225621979414,
      "loss": 4.3707,
      "step": 2257
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.9724120414575723,
      "learning_rate": 0.0002990004378422777,
      "loss": 4.3381,
      "step": 2258
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.0816068780481227,
      "learning_rate": 0.0002987783140345923,
      "loss": 4.3347,
      "step": 2259
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.4504179408684841,
      "learning_rate": 0.0002985561908966577,
      "loss": 4.4011,
      "step": 2260
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.7973584630116908,
      "learning_rate": 0.0002983340685502459,
      "loss": 4.2082,
      "step": 2261
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.309066969335737,
      "learning_rate": 0.0002981119471171287,
      "loss": 4.4639,
      "step": 2262
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.251558124574615,
      "learning_rate": 0.00029788982671907673,
      "loss": 4.3506,
      "step": 2263
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.9136989308039597,
      "learning_rate": 0.0002976677074778608,
      "loss": 4.266,
      "step": 2264
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.9386342621317563,
      "learning_rate": 0.00029744558951525066,
      "loss": 4.3484,
      "step": 2265
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.1251874111219151,
      "learning_rate": 0.00029722347295301556,
      "loss": 4.2438,
      "step": 2266
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.2944831528535317,
      "learning_rate": 0.00029700135791292393,
      "loss": 4.3637,
      "step": 2267
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.355721183286784,
      "learning_rate": 0.0002967792445167432,
      "loss": 4.3078,
      "step": 2268
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.6449674217044504,
      "learning_rate": 0.0002965571328862401,
      "loss": 4.4297,
      "step": 2269
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.6430677757973748,
      "learning_rate": 0.00029633502314318035,
      "loss": 4.2022,
      "step": 2270
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.0557078403562836,
      "learning_rate": 0.00029611291540932865,
      "loss": 4.4227,
      "step": 2271
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.6065317747631467,
      "learning_rate": 0.0002958908098064485,
      "loss": 4.212,
      "step": 2272
    },
    {
      "epoch": 2.6,
      "grad_norm": 5.447514734150979,
      "learning_rate": 0.0002956687064563022,
      "loss": 4.1951,
      "step": 2273
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.7166608530222605,
      "learning_rate": 0.0002954466054806511,
      "loss": 4.2941,
      "step": 2274
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.859944420184991,
      "learning_rate": 0.000295224507001255,
      "loss": 4.2357,
      "step": 2275
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.442382217026208,
      "learning_rate": 0.0002950024111398723,
      "loss": 4.214,
      "step": 2276
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.6262753871302342,
      "learning_rate": 0.00029478031801826013,
      "loss": 4.3396,
      "step": 2277
    },
    {
      "epoch": 2.6,
      "grad_norm": 5.409394165865939,
      "learning_rate": 0.000294558227758174,
      "loss": 4.5918,
      "step": 2278
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.2352726590298537,
      "learning_rate": 0.000294336140481368,
      "loss": 4.2783,
      "step": 2279
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.3111333987220926,
      "learning_rate": 0.0002941140563095943,
      "loss": 4.4927,
      "step": 2280
    },
    {
      "epoch": 2.61,
      "grad_norm": 3.09885958819297,
      "learning_rate": 0.00029389197536460365,
      "loss": 4.4625,
      "step": 2281
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.975683469560857,
      "learning_rate": 0.0002936698977681449,
      "loss": 4.1609,
      "step": 2282
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.7578148621399394,
      "learning_rate": 0.0002934478236419651,
      "loss": 4.2658,
      "step": 2283
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.9413044418629747,
      "learning_rate": 0.00029322575310780944,
      "loss": 4.3923,
      "step": 2284
    },
    {
      "epoch": 2.61,
      "grad_norm": 3.108926660752368,
      "learning_rate": 0.000293003686287421,
      "loss": 4.4338,
      "step": 2285
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3066627303412668,
      "learning_rate": 0.00029278162330254097,
      "loss": 4.3493,
      "step": 2286
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3567465151123534,
      "learning_rate": 0.00029255956427490834,
      "loss": 4.4,
      "step": 2287
    },
    {
      "epoch": 2.61,
      "grad_norm": 3.7553266799879896,
      "learning_rate": 0.00029233750932626003,
      "loss": 4.4734,
      "step": 2288
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.1273898421543045,
      "learning_rate": 0.0002921154585783306,
      "loss": 4.3404,
      "step": 2289
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.0927583620782526,
      "learning_rate": 0.0002918934121528523,
      "loss": 4.2912,
      "step": 2290
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6898422140242562,
      "learning_rate": 0.00029167137017155527,
      "loss": 4.4021,
      "step": 2291
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.0696671970440566,
      "learning_rate": 0.00029144933275616696,
      "loss": 4.4117,
      "step": 2292
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.9557313765926126,
      "learning_rate": 0.0002912273000284123,
      "loss": 4.4116,
      "step": 2293
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5563010420761698,
      "learning_rate": 0.00029100527211001377,
      "loss": 4.3476,
      "step": 2294
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.9426540689483062,
      "learning_rate": 0.0002907832491226911,
      "loss": 4.4253,
      "step": 2295
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.5994127975902223,
      "learning_rate": 0.00029056123118816166,
      "loss": 4.2409,
      "step": 2296
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5083264269189016,
      "learning_rate": 0.00029033921842813953,
      "loss": 4.2943,
      "step": 2297
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3448793667983174,
      "learning_rate": 0.0002901172109643362,
      "loss": 3.9476,
      "step": 2298
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.5688283858563615,
      "learning_rate": 0.00028989520891846036,
      "loss": 4.2914,
      "step": 2299
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3115752912377703,
      "learning_rate": 0.00028967321241221765,
      "loss": 4.4191,
      "step": 2300
    },
    {
      "epoch": 2.63,
      "eval_blimp_filtered_avg": 0.5271641791044777,
      "eval_blimp_filtered_std": 0.0055033061926466975,
      "step": 2300
    },
    {
      "epoch": 2.63,
      "eval_blimp_supplement_avg": 0.5043103448275862,
      "eval_blimp_supplement_std": 0.022841245982437135,
      "step": 2300
    },
    {
      "epoch": 2.63,
      "eval_vqa_filtered_avg": 0.35,
      "eval_vqa_filtered_std": 0.0479372485441102,
      "step": 2300
    },
    {
      "epoch": 2.63,
      "eval_winoground_filtered_avg": 0.49,
      "eval_winoground_filtered_std": 0.05024183937956912,
      "step": 2300
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4079782126170108,
      "learning_rate": 0.00028945122156731075,
      "loss": 4.1759,
      "step": 2301
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.0672890614094868,
      "learning_rate": 0.0002892292365054389,
      "loss": 4.2851,
      "step": 2302
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.8273065827177049,
      "learning_rate": 0.00028900725734829836,
      "loss": 4.3457,
      "step": 2303
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.27879900863289,
      "learning_rate": 0.00028878528421758253,
      "loss": 4.3369,
      "step": 2304
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.9729041182523845,
      "learning_rate": 0.00028856331723498105,
      "loss": 4.4518,
      "step": 2305
    },
    {
      "epoch": 2.64,
      "grad_norm": 5.8485080558022755,
      "learning_rate": 0.00028834135652218017,
      "loss": 4.4518,
      "step": 2306
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.696541393972968,
      "learning_rate": 0.000288119402200863,
      "loss": 4.09,
      "step": 2307
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.7052006660951307,
      "learning_rate": 0.00028789745439270886,
      "loss": 4.3882,
      "step": 2308
    },
    {
      "epoch": 2.64,
      "grad_norm": 5.640204300057213,
      "learning_rate": 0.000287675513219394,
      "loss": 4.5902,
      "step": 2309
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4050318150262908,
      "learning_rate": 0.0002874535788025902,
      "loss": 4.4239,
      "step": 2310
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4049010323756423,
      "learning_rate": 0.0002872316512639662,
      "loss": 4.4407,
      "step": 2311
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.71760796655439,
      "learning_rate": 0.0002870097307251868,
      "loss": 4.3448,
      "step": 2312
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.357070726389828,
      "learning_rate": 0.0002867878173079131,
      "loss": 4.4178,
      "step": 2313
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.9962240860109267,
      "learning_rate": 0.0002865659111338018,
      "loss": 4.0915,
      "step": 2314
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.2333832928155484,
      "learning_rate": 0.0002863440123245061,
      "loss": 4.3536,
      "step": 2315
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.612066502389392,
      "learning_rate": 0.0002861221210016751,
      "loss": 4.4801,
      "step": 2316
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4404549516657346,
      "learning_rate": 0.00028590023728695366,
      "loss": 4.6146,
      "step": 2317
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.974081918749349,
      "learning_rate": 0.0002856783613019826,
      "loss": 4.291,
      "step": 2318
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.7135896416775674,
      "learning_rate": 0.00028545649316839833,
      "loss": 4.2509,
      "step": 2319
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4725686269649927,
      "learning_rate": 0.00028523463300783304,
      "loss": 4.3603,
      "step": 2320
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.1269754694963419,
      "learning_rate": 0.00028501278094191467,
      "loss": 4.3943,
      "step": 2321
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.9147364453050987,
      "learning_rate": 0.0002847909370922667,
      "loss": 4.3609,
      "step": 2322
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.8150969047583511,
      "learning_rate": 0.00028456910158050776,
      "loss": 4.3304,
      "step": 2323
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.7857459651465912,
      "learning_rate": 0.00028434727452825235,
      "loss": 4.3625,
      "step": 2324
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.9030219005323503,
      "learning_rate": 0.0002841254560571101,
      "loss": 4.2372,
      "step": 2325
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.1078058058514557,
      "learning_rate": 0.0002839036462886861,
      "loss": 4.0541,
      "step": 2326
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.6707546803745114,
      "learning_rate": 0.0002836818453445804,
      "loss": 4.3139,
      "step": 2327
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.9819440745730001,
      "learning_rate": 0.0002834600533463885,
      "loss": 4.2085,
      "step": 2328
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.7426465597479164,
      "learning_rate": 0.0002832382704157008,
      "loss": 4.3612,
      "step": 2329
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.255295869223125,
      "learning_rate": 0.0002830164966741028,
      "loss": 4.1483,
      "step": 2330
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.5220371518609976,
      "learning_rate": 0.000282794732243175,
      "loss": 4.1273,
      "step": 2331
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.7753217141279873,
      "learning_rate": 0.00028257297724449277,
      "loss": 4.1221,
      "step": 2332
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6283196173403756,
      "learning_rate": 0.0002823512317996262,
      "loss": 4.3878,
      "step": 2333
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.112398148856668,
      "learning_rate": 0.00028212949603014035,
      "loss": 4.3286,
      "step": 2334
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.0207375487858874,
      "learning_rate": 0.00028190777005759486,
      "loss": 4.2133,
      "step": 2335
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.1668499718240484,
      "learning_rate": 0.0002816860540035439,
      "loss": 4.3273,
      "step": 2336
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.328319795023674,
      "learning_rate": 0.00028146434798953637,
      "loss": 4.3533,
      "step": 2337
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.1403670790659244,
      "learning_rate": 0.0002812426521371156,
      "loss": 4.3453,
      "step": 2338
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.6182615651649437,
      "learning_rate": 0.00028102096656781937,
      "loss": 4.2018,
      "step": 2339
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.2379013745070362,
      "learning_rate": 0.0002807992914031797,
      "loss": 4.2253,
      "step": 2340
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.32228929142265,
      "learning_rate": 0.0002805776267647232,
      "loss": 4.2342,
      "step": 2341
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.8389277281365515,
      "learning_rate": 0.0002803559727739703,
      "loss": 4.1933,
      "step": 2342
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.063081859360418,
      "learning_rate": 0.00028013432955243604,
      "loss": 4.5445,
      "step": 2343
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.2760985773935203,
      "learning_rate": 0.00027991269722162917,
      "loss": 4.2203,
      "step": 2344
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4864970732377776,
      "learning_rate": 0.0002796910759030526,
      "loss": 4.3592,
      "step": 2345
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.4606536184524214,
      "learning_rate": 0.0002794694657182033,
      "loss": 4.2188,
      "step": 2346
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.262518679012878,
      "learning_rate": 0.00027924786678857217,
      "loss": 4.2866,
      "step": 2347
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.423756465643589,
      "learning_rate": 0.00027902627923564365,
      "loss": 4.2946,
      "step": 2348
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5763320717818459,
      "learning_rate": 0.00027880470318089617,
      "loss": 4.1811,
      "step": 2349
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4726902693921242,
      "learning_rate": 0.00027858313874580185,
      "loss": 4.2424,
      "step": 2350
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.5870611920388359,
      "learning_rate": 0.00027836158605182656,
      "loss": 4.2,
      "step": 2351
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.7465765263609636,
      "learning_rate": 0.0002781400452204293,
      "loss": 4.35,
      "step": 2352
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9933353619229455,
      "learning_rate": 0.0002779185163730629,
      "loss": 4.4204,
      "step": 2353
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.6985505623737489,
      "learning_rate": 0.0002776969996311737,
      "loss": 4.2703,
      "step": 2354
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.0380363971616795,
      "learning_rate": 0.0002774754951162012,
      "loss": 4.3091,
      "step": 2355
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.1629785769881646,
      "learning_rate": 0.0002772540029495784,
      "loss": 4.5021,
      "step": 2356
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.5476361888916004,
      "learning_rate": 0.00027703252325273105,
      "loss": 4.3604,
      "step": 2357
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.8130295054536668,
      "learning_rate": 0.0002768110561470787,
      "loss": 4.2637,
      "step": 2358
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.825002961380643,
      "learning_rate": 0.0002765896017540337,
      "loss": 4.4267,
      "step": 2359
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.673164769585789,
      "learning_rate": 0.00027636816019500145,
      "loss": 4.4508,
      "step": 2360
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.5538523670227395,
      "learning_rate": 0.00027614673159138,
      "loss": 4.4611,
      "step": 2361
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3400379169374679,
      "learning_rate": 0.0002759253160645609,
      "loss": 4.3602,
      "step": 2362
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.0523736306199836,
      "learning_rate": 0.0002757039137359282,
      "loss": 4.2375,
      "step": 2363
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.8380415195633117,
      "learning_rate": 0.00027548252472685877,
      "loss": 4.3231,
      "step": 2364
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4902812522394382,
      "learning_rate": 0.00027526114915872195,
      "loss": 4.4165,
      "step": 2365
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.8208638457827007,
      "learning_rate": 0.0002750397871528801,
      "loss": 4.3429,
      "step": 2366
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3130379975055768,
      "learning_rate": 0.0002748184388306879,
      "loss": 4.1567,
      "step": 2367
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.9066970511291104,
      "learning_rate": 0.00027459710431349264,
      "loss": 4.2985,
      "step": 2368
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.7184326269758523,
      "learning_rate": 0.0002743757837226339,
      "loss": 4.2431,
      "step": 2369
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3206239079736553,
      "learning_rate": 0.00027415447717944383,
      "loss": 4.2916,
      "step": 2370
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.3247156205440564,
      "learning_rate": 0.0002739331848052466,
      "loss": 4.3808,
      "step": 2371
    },
    {
      "epoch": 2.71,
      "grad_norm": 6.425953120992067,
      "learning_rate": 0.0002737119067213589,
      "loss": 4.5812,
      "step": 2372
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.824791270227781,
      "learning_rate": 0.00027349064304908955,
      "loss": 4.4341,
      "step": 2373
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.2426533258009074,
      "learning_rate": 0.00027326939390973917,
      "loss": 4.2938,
      "step": 2374
    },
    {
      "epoch": 2.71,
      "grad_norm": 5.425335011885779,
      "learning_rate": 0.0002730481594246007,
      "loss": 4.7417,
      "step": 2375
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.8348985576551495,
      "learning_rate": 0.000272826939714959,
      "loss": 4.5723,
      "step": 2376
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.393893246030523,
      "learning_rate": 0.0002726057349020908,
      "loss": 4.3729,
      "step": 2377
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.1301721066781862,
      "learning_rate": 0.0002723845451072647,
      "loss": 4.55,
      "step": 2378
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.0540487935948053,
      "learning_rate": 0.00027216337045174097,
      "loss": 4.6937,
      "step": 2379
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.778671978797922,
      "learning_rate": 0.0002719422110567716,
      "loss": 4.6864,
      "step": 2380
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.3553341591383727,
      "learning_rate": 0.00027172106704360043,
      "loss": 4.5567,
      "step": 2381
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.7544938243284398,
      "learning_rate": 0.00027149993853346245,
      "loss": 4.1503,
      "step": 2382
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.131178932833919,
      "learning_rate": 0.00027127882564758455,
      "loss": 4.4901,
      "step": 2383
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.351914126894973,
      "learning_rate": 0.00027105772850718487,
      "loss": 4.4113,
      "step": 2384
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.931524743641572,
      "learning_rate": 0.00027083664723347303,
      "loss": 4.5625,
      "step": 2385
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6062676130825801,
      "learning_rate": 0.00027061558194764967,
      "loss": 4.2058,
      "step": 2386
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4106702526256814,
      "learning_rate": 0.000270394532770907,
      "loss": 4.3231,
      "step": 2387
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.672709758489334,
      "learning_rate": 0.00027017349982442823,
      "loss": 4.4051,
      "step": 2388
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.5507906352031786,
      "learning_rate": 0.0002699524832293876,
      "loss": 4.3307,
      "step": 2389
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.1536149680023255,
      "learning_rate": 0.0002697314831069508,
      "loss": 4.2658,
      "step": 2390
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3790333653791862,
      "learning_rate": 0.0002695104995782739,
      "loss": 4.2069,
      "step": 2391
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.0087300257882044,
      "learning_rate": 0.0002692895327645042,
      "loss": 4.297,
      "step": 2392
    },
    {
      "epoch": 2.73,
      "grad_norm": 7.632478819434307,
      "learning_rate": 0.00026906858278677974,
      "loss": 4.168,
      "step": 2393
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.626695969170174,
      "learning_rate": 0.00026884764976622965,
      "loss": 4.6988,
      "step": 2394
    },
    {
      "epoch": 2.74,
      "grad_norm": 8.288464344088812,
      "learning_rate": 0.00026862673382397316,
      "loss": 4.7142,
      "step": 2395
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.622592165531978,
      "learning_rate": 0.0002684058350811206,
      "loss": 4.4142,
      "step": 2396
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4028413928306576,
      "learning_rate": 0.00026818495365877264,
      "loss": 4.3487,
      "step": 2397
    },
    {
      "epoch": 2.74,
      "grad_norm": 5.664336156744327,
      "learning_rate": 0.00026796408967802096,
      "loss": 4.5118,
      "step": 2398
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.850895326149334,
      "learning_rate": 0.0002677432432599467,
      "loss": 4.6287,
      "step": 2399
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.5905521515725654,
      "learning_rate": 0.0002675224145256223,
      "loss": 4.4235,
      "step": 2400
    },
    {
      "epoch": 2.74,
      "eval_blimp_filtered_avg": 0.5262686567164179,
      "eval_blimp_filtered_std": 0.005558351812312166,
      "step": 2400
    },
    {
      "epoch": 2.74,
      "eval_blimp_supplement_avg": 0.5280172413793104,
      "eval_blimp_supplement_std": 0.022810969291901215,
      "step": 2400
    },
    {
      "epoch": 2.74,
      "eval_vqa_filtered_avg": 0.36,
      "eval_vqa_filtered_std": 0.04824181513244218,
      "step": 2400
    },
    {
      "epoch": 2.74,
      "eval_winoground_filtered_avg": 0.55,
      "eval_winoground_filtered_std": 0.05,
      "step": 2400
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.99824979014869,
      "learning_rate": 0.00026730160359611006,
      "loss": 4.1203,
      "step": 2401
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.1707531964064097,
      "learning_rate": 0.0002670808105924627,
      "loss": 4.347,
      "step": 2402
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.2177646473190618,
      "learning_rate": 0.0002668600356357229,
      "loss": 4.6087,
      "step": 2403
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.3545682828580268,
      "learning_rate": 0.0002666392788469235,
      "loss": 4.5883,
      "step": 2404
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.334560206009573,
      "learning_rate": 0.0002664185403470876,
      "loss": 4.4897,
      "step": 2405
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.770936159004711,
      "learning_rate": 0.0002661978202572281,
      "loss": 4.1992,
      "step": 2406
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.7368464459301234,
      "learning_rate": 0.00026597711869834795,
      "loss": 4.1563,
      "step": 2407
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.8833880067523006,
      "learning_rate": 0.00026575643579143947,
      "loss": 4.3527,
      "step": 2408
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.8571622653674449,
      "learning_rate": 0.00026553577165748537,
      "loss": 4.3097,
      "step": 2409
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.1727395919291579,
      "learning_rate": 0.0002653151264174578,
      "loss": 4.4953,
      "step": 2410
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.6886746921895335,
      "learning_rate": 0.0002650945001923185,
      "loss": 4.2927,
      "step": 2411
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.373311796584587,
      "learning_rate": 0.00026487389310301884,
      "loss": 4.1877,
      "step": 2412
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4562892930366158,
      "learning_rate": 0.0002646533052704997,
      "loss": 4.6184,
      "step": 2413
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.8639457733764047,
      "learning_rate": 0.00026443273681569135,
      "loss": 4.2612,
      "step": 2414
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.7818623149849582,
      "learning_rate": 0.0002642121878595137,
      "loss": 4.6863,
      "step": 2415
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.8231654771778978,
      "learning_rate": 0.00026399165852287543,
      "loss": 4.3108,
      "step": 2416
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.716871118734228,
      "learning_rate": 0.0002637711489266751,
      "loss": 4.1962,
      "step": 2417
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.9088942810638283,
      "learning_rate": 0.0002635506591917999,
      "loss": 4.4559,
      "step": 2418
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4820473127158955,
      "learning_rate": 0.00026333018943912665,
      "loss": 4.2945,
      "step": 2419
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9019388966757095,
      "learning_rate": 0.0002631097397895207,
      "loss": 4.2484,
      "step": 2420
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.7424138103174842,
      "learning_rate": 0.00026288931036383666,
      "loss": 4.1204,
      "step": 2421
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.331060428321565,
      "learning_rate": 0.0002626689012829181,
      "loss": 4.2375,
      "step": 2422
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9913597042339909,
      "learning_rate": 0.0002624485126675973,
      "loss": 4.4649,
      "step": 2423
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.5529213745470625,
      "learning_rate": 0.0002622281446386954,
      "loss": 4.3446,
      "step": 2424
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.8480828302347831,
      "learning_rate": 0.0002620077973170222,
      "loss": 4.3045,
      "step": 2425
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.0283122369122324,
      "learning_rate": 0.00026178747082337614,
      "loss": 4.0179,
      "step": 2426
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.049479911725009,
      "learning_rate": 0.0002615671652785443,
      "loss": 4.4121,
      "step": 2427
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.603630834514023,
      "learning_rate": 0.0002613468808033023,
      "loss": 4.2332,
      "step": 2428
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.9977057004564149,
      "learning_rate": 0.0002611266175184141,
      "loss": 4.2668,
      "step": 2429
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.0664215053262873,
      "learning_rate": 0.000260906375544632,
      "loss": 4.3656,
      "step": 2430
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.6881169721941325,
      "learning_rate": 0.0002606861550026969,
      "loss": 4.2123,
      "step": 2431
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3204197837360618,
      "learning_rate": 0.0002604659560133377,
      "loss": 4.1416,
      "step": 2432
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.2574998432535367,
      "learning_rate": 0.0002602457786972716,
      "loss": 4.2291,
      "step": 2433
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.5554443425465179,
      "learning_rate": 0.0002600256231752037,
      "loss": 4.1614,
      "step": 2434
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.6407737169777484,
      "learning_rate": 0.00025980548956782746,
      "loss": 4.3533,
      "step": 2435
    },
    {
      "epoch": 2.78,
      "grad_norm": 7.827846926145454,
      "learning_rate": 0.00025958537799582426,
      "loss": 4.2396,
      "step": 2436
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.478601082166011,
      "learning_rate": 0.00025936528857986314,
      "loss": 4.5495,
      "step": 2437
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.4535065890655,
      "learning_rate": 0.00025914522144060133,
      "loss": 4.5321,
      "step": 2438
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.2453028569480487,
      "learning_rate": 0.00025892517669868364,
      "loss": 4.626,
      "step": 2439
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.1789937223459397,
      "learning_rate": 0.0002587051544747426,
      "loss": 4.4243,
      "step": 2440
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7028958252683373,
      "learning_rate": 0.0002584851548893987,
      "loss": 4.4861,
      "step": 2441
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.5051429208959295,
      "learning_rate": 0.0002582651780632595,
      "loss": 4.2952,
      "step": 2442
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.2636528777860927,
      "learning_rate": 0.0002580452241169204,
      "loss": 4.2502,
      "step": 2443
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.221441870911716,
      "learning_rate": 0.00025782529317096433,
      "loss": 4.3699,
      "step": 2444
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.6412057015144468,
      "learning_rate": 0.00025760538534596157,
      "loss": 4.2175,
      "step": 2445
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.2124456085358624,
      "learning_rate": 0.0002573855007624694,
      "loss": 4.368,
      "step": 2446
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.3088602647083825,
      "learning_rate": 0.00025716563954103266,
      "loss": 4.4485,
      "step": 2447
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.170421206099158,
      "learning_rate": 0.0002569458018021835,
      "loss": 4.4699,
      "step": 2448
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.8369327027412495,
      "learning_rate": 0.00025672598766644106,
      "loss": 4.3763,
      "step": 2449
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.938364076776348,
      "learning_rate": 0.0002565061972543113,
      "loss": 4.4062,
      "step": 2450
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.8373194593740894,
      "learning_rate": 0.0002562864306862873,
      "loss": 4.1043,
      "step": 2451
    },
    {
      "epoch": 2.8,
      "grad_norm": 11.677280012197496,
      "learning_rate": 0.00025606668808284943,
      "loss": 4.3996,
      "step": 2452
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.2518797258684893,
      "learning_rate": 0.0002558469695644647,
      "loss": 4.4345,
      "step": 2453
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.347827295640945,
      "learning_rate": 0.0002556272752515865,
      "loss": 4.6806,
      "step": 2454
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.2791294899446446,
      "learning_rate": 0.0002554076052646555,
      "loss": 4.783,
      "step": 2455
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.0825124315075594,
      "learning_rate": 0.0002551879597240989,
      "loss": 4.7617,
      "step": 2456
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.2801136521118432,
      "learning_rate": 0.00025496833875033047,
      "loss": 4.6163,
      "step": 2457
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.8408972045096217,
      "learning_rate": 0.0002547487424637502,
      "loss": 4.8371,
      "step": 2458
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3390657051676296,
      "learning_rate": 0.00025452917098474516,
      "loss": 4.6053,
      "step": 2459
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.2518553489021658,
      "learning_rate": 0.0002543096244336884,
      "loss": 4.5719,
      "step": 2460
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.2547823688556723,
      "learning_rate": 0.0002540901029309392,
      "loss": 4.7789,
      "step": 2461
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4466436775581233,
      "learning_rate": 0.00025387060659684367,
      "loss": 4.6624,
      "step": 2462
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.7510828498938955,
      "learning_rate": 0.0002536511355517334,
      "loss": 4.5877,
      "step": 2463
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5852186617651742,
      "learning_rate": 0.0002534316899159266,
      "loss": 4.6253,
      "step": 2464
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.1523198590744084,
      "learning_rate": 0.0002532122698097275,
      "loss": 4.6131,
      "step": 2465
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.068884267196014,
      "learning_rate": 0.00025299287535342625,
      "loss": 4.5747,
      "step": 2466
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.265330236177896,
      "learning_rate": 0.00025277350666729875,
      "loss": 4.6157,
      "step": 2467
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.8685165177456566,
      "learning_rate": 0.00025255416387160714,
      "loss": 4.5911,
      "step": 2468
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4462807825674788,
      "learning_rate": 0.0002523348470865991,
      "loss": 4.6345,
      "step": 2469
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.7974338514807957,
      "learning_rate": 0.0002521155564325083,
      "loss": 4.4022,
      "step": 2470
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4327295852652384,
      "learning_rate": 0.00025189629202955374,
      "loss": 4.5031,
      "step": 2471
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.874255317510061,
      "learning_rate": 0.00025167705399794035,
      "loss": 4.4118,
      "step": 2472
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.9919097691232142,
      "learning_rate": 0.0002514578424578583,
      "loss": 4.5023,
      "step": 2473
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.037123683953898,
      "learning_rate": 0.00025123865752948364,
      "loss": 4.4696,
      "step": 2474
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.9323962611367328,
      "learning_rate": 0.0002510194993329774,
      "loss": 4.4952,
      "step": 2475
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.6230990706496498,
      "learning_rate": 0.0002508003679884862,
      "loss": 4.2457,
      "step": 2476
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3375277448145781,
      "learning_rate": 0.00025058126361614193,
      "loss": 4.3794,
      "step": 2477
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.1494561526114837,
      "learning_rate": 0.0002503621863360616,
      "loss": 4.3924,
      "step": 2478
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.824463389480698,
      "learning_rate": 0.00025014313626834755,
      "loss": 4.23,
      "step": 2479
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.2055468441140331,
      "learning_rate": 0.0002499241135330869,
      "loss": 4.4901,
      "step": 2480
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.8412227814348342,
      "learning_rate": 0.000249705118250352,
      "loss": 4.2681,
      "step": 2481
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.5708481463921478,
      "learning_rate": 0.00024948615054020025,
      "loss": 4.2648,
      "step": 2482
    },
    {
      "epoch": 2.84,
      "grad_norm": 6.758569382516103,
      "learning_rate": 0.00024926721052267363,
      "loss": 4.244,
      "step": 2483
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.811911989934368,
      "learning_rate": 0.00024904829831779915,
      "loss": 4.6532,
      "step": 2484
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.351155928990301,
      "learning_rate": 0.00024882941404558856,
      "loss": 4.4838,
      "step": 2485
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.6520006368457554,
      "learning_rate": 0.00024861055782603815,
      "loss": 4.3348,
      "step": 2486
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4801960946219987,
      "learning_rate": 0.00024839172977912915,
      "loss": 4.6513,
      "step": 2487
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5520826948168076,
      "learning_rate": 0.00024817293002482696,
      "loss": 4.5324,
      "step": 2488
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.1523930399562032,
      "learning_rate": 0.00024795415868308164,
      "loss": 4.4049,
      "step": 2489
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.7278089850771241,
      "learning_rate": 0.00024773541587382774,
      "loss": 4.4099,
      "step": 2490
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.8520056265238279,
      "learning_rate": 0.00024751670171698425,
      "loss": 4.4215,
      "step": 2491
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.0690514262090678,
      "learning_rate": 0.000247298016332454,
      "loss": 4.4654,
      "step": 2492
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.6582093377940849,
      "learning_rate": 0.0002470793598401246,
      "loss": 4.3856,
      "step": 2493
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.4650402367175494,
      "learning_rate": 0.00024686073235986736,
      "loss": 4.4173,
      "step": 2494
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.5550051698566519,
      "learning_rate": 0.00024664213401153813,
      "loss": 4.498,
      "step": 2495
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.8802403572982475,
      "learning_rate": 0.0002464235649149766,
      "loss": 4.1077,
      "step": 2496
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.6562823599233069,
      "learning_rate": 0.00024620502519000616,
      "loss": 4.3583,
      "step": 2497
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.48291473954310066,
      "learning_rate": 0.00024598651495643433,
      "loss": 4.359,
      "step": 2498
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.8845710443897843,
      "learning_rate": 0.00024576803433405263,
      "loss": 4.2643,
      "step": 2499
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.8606982636574948,
      "learning_rate": 0.00024554958344263623,
      "loss": 4.4885,
      "step": 2500
    },
    {
      "epoch": 2.86,
      "eval_blimp_filtered_avg": 0.5313432835820896,
      "eval_blimp_filtered_std": 0.005502651952302882,
      "step": 2500
    },
    {
      "epoch": 2.86,
      "eval_blimp_supplement_avg": 0.540948275862069,
      "eval_blimp_supplement_std": 0.022776400906345202,
      "step": 2500
    },
    {
      "epoch": 2.86,
      "eval_vqa_filtered_avg": 0.35,
      "eval_vqa_filtered_std": 0.0479372485441102,
      "step": 2500
    },
    {
      "epoch": 2.86,
      "eval_winoground_filtered_avg": 0.52,
      "eval_winoground_filtered_std": 0.05021167315686779,
      "step": 2500
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.5651469465697178,
      "learning_rate": 0.00024533116240194356,
      "loss": 4.1912,
      "step": 2501
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.14749751983357,
      "learning_rate": 0.0002451127713317174,
      "loss": 4.229,
      "step": 2502
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.8241128709779351,
      "learning_rate": 0.0002448944103516836,
      "loss": 4.3691,
      "step": 2503
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.5204256856361168,
      "learning_rate": 0.000244676079581552,
      "loss": 4.1866,
      "step": 2504
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.9138386358067094,
      "learning_rate": 0.0002444577791410151,
      "loss": 4.1387,
      "step": 2505
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.7633904432965017,
      "learning_rate": 0.00024423950914974953,
      "loss": 4.3597,
      "step": 2506
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.5535881430660151,
      "learning_rate": 0.00024402126972741483,
      "loss": 4.3361,
      "step": 2507
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.9553212589826067,
      "learning_rate": 0.000243803060993654,
      "loss": 4.2876,
      "step": 2508
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.001626313172623,
      "learning_rate": 0.00024358488306809288,
      "loss": 4.3321,
      "step": 2509
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.8562308593409345,
      "learning_rate": 0.00024336673607034076,
      "loss": 4.2753,
      "step": 2510
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.7068258144551033,
      "learning_rate": 0.0002431486201199899,
      "loss": 4.2083,
      "step": 2511
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.7381055658211995,
      "learning_rate": 0.00024293053533661543,
      "loss": 4.1509,
      "step": 2512
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.6004570795659198,
      "learning_rate": 0.00024271248183977556,
      "loss": 4.2049,
      "step": 2513
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.0525446924899813,
      "learning_rate": 0.00024249445974901108,
      "loss": 4.3442,
      "step": 2514
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.5799693014198636,
      "learning_rate": 0.00024227646918384576,
      "loss": 4.2701,
      "step": 2515
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.4880977986877435,
      "learning_rate": 0.00024205851026378613,
      "loss": 4.1782,
      "step": 2516
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.5859847907237523,
      "learning_rate": 0.00024184058310832134,
      "loss": 4.361,
      "step": 2517
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.7466025025315636,
      "learning_rate": 0.00024162268783692296,
      "loss": 4.193,
      "step": 2518
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.9549652036442368,
      "learning_rate": 0.0002414048245690452,
      "loss": 4.1619,
      "step": 2519
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.4468170790229295,
      "learning_rate": 0.00024118699342412479,
      "loss": 4.281,
      "step": 2520
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.8571379079232043,
      "learning_rate": 0.00024096919452158087,
      "loss": 4.339,
      "step": 2521
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.6737576755736234,
      "learning_rate": 0.00024075142798081457,
      "loss": 4.3447,
      "step": 2522
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.8879965708655246,
      "learning_rate": 0.00024053369392120974,
      "loss": 4.2534,
      "step": 2523
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.8479273253744606,
      "learning_rate": 0.00024031599246213212,
      "loss": 4.2303,
      "step": 2524
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.4698391717351646,
      "learning_rate": 0.00024009832372292984,
      "loss": 4.1858,
      "step": 2525
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.2901717077375932,
      "learning_rate": 0.00023988068782293266,
      "loss": 4.1041,
      "step": 2526
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.7083022794606397,
      "learning_rate": 0.00023966308488145274,
      "loss": 4.1434,
      "step": 2527
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.7011069805650739,
      "learning_rate": 0.000239445515017784,
      "loss": 4.2788,
      "step": 2528
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.1421536150407554,
      "learning_rate": 0.0002392279783512023,
      "loss": 4.2421,
      "step": 2529
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.5875615755101561,
      "learning_rate": 0.0002390104750009653,
      "loss": 4.1006,
      "step": 2530
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.1268722681188232,
      "learning_rate": 0.0002387930050863122,
      "loss": 4.398,
      "step": 2531
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.6222772437799192,
      "learning_rate": 0.00023857556872646413,
      "loss": 4.1596,
      "step": 2532
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.6338559472716655,
      "learning_rate": 0.00023835816604062367,
      "loss": 4.2389,
      "step": 2533
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.0666248549531778,
      "learning_rate": 0.00023814079714797512,
      "loss": 4.2561,
      "step": 2534
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.5946667373598951,
      "learning_rate": 0.00023792346216768394,
      "loss": 4.266,
      "step": 2535
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.5122066332603631,
      "learning_rate": 0.00023770616121889724,
      "loss": 4.2613,
      "step": 2536
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.6377780125525387,
      "learning_rate": 0.0002374888944207434,
      "loss": 4.1513,
      "step": 2537
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.6721856454702817,
      "learning_rate": 0.00023727166189233228,
      "loss": 4.2296,
      "step": 2538
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.5696060569492103,
      "learning_rate": 0.00023705446375275448,
      "loss": 4.2158,
      "step": 2539
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.773353550327232,
      "learning_rate": 0.00023683730012108217,
      "loss": 4.1747,
      "step": 2540
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5065550956301166,
      "learning_rate": 0.00023662017111636838,
      "loss": 4.245,
      "step": 2541
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.0783950898470078,
      "learning_rate": 0.0002364030768576475,
      "loss": 4.2991,
      "step": 2542
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.263051871743404,
      "learning_rate": 0.0002361860174639343,
      "loss": 4.1776,
      "step": 2543
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.6999807348176544,
      "learning_rate": 0.00023596899305422487,
      "loss": 4.4206,
      "step": 2544
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.614169177116863,
      "learning_rate": 0.00023575200374749593,
      "loss": 4.1526,
      "step": 2545
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9961232909655852,
      "learning_rate": 0.0002355350496627053,
      "loss": 4.3485,
      "step": 2546
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.5658671883088607,
      "learning_rate": 0.0002353181309187909,
      "loss": 4.3091,
      "step": 2547
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.7423701182339253,
      "learning_rate": 0.0002351012476346716,
      "loss": 4.4892,
      "step": 2548
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.038336311941712,
      "learning_rate": 0.00023488439992924704,
      "loss": 4.2157,
      "step": 2549
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.6920914073037782,
      "learning_rate": 0.00023466758792139696,
      "loss": 4.2824,
      "step": 2550
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.7356411953829696,
      "learning_rate": 0.00023445081172998193,
      "loss": 4.2419,
      "step": 2551
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.6502002009118366,
      "learning_rate": 0.00023423407147384233,
      "loss": 4.1811,
      "step": 2552
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.6672266588624904,
      "learning_rate": 0.00023401736727179936,
      "loss": 4.1375,
      "step": 2553
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.577132639609706,
      "learning_rate": 0.00023380069924265433,
      "loss": 4.1765,
      "step": 2554
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.262287404435852,
      "learning_rate": 0.00023358406750518865,
      "loss": 4.3358,
      "step": 2555
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.5813001231482584,
      "learning_rate": 0.00023336747217816354,
      "loss": 4.0951,
      "step": 2556
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.8142413040340654,
      "learning_rate": 0.00023315091338032087,
      "loss": 4.2364,
      "step": 2557
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.6536407916309178,
      "learning_rate": 0.0002329343912303821,
      "loss": 4.3941,
      "step": 2558
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.8187149026569829,
      "learning_rate": 0.00023271790584704873,
      "loss": 4.2477,
      "step": 2559
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.51777212786076,
      "learning_rate": 0.00023250145734900185,
      "loss": 4.1844,
      "step": 2560
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.7599364493390554,
      "learning_rate": 0.00023228504585490268,
      "loss": 3.9695,
      "step": 2561
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.008857826506059,
      "learning_rate": 0.000232068671483392,
      "loss": 4.2541,
      "step": 2562
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.9374779660914048,
      "learning_rate": 0.0002318523343530902,
      "loss": 4.3179,
      "step": 2563
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.806498926719277,
      "learning_rate": 0.00023163603458259725,
      "loss": 4.2933,
      "step": 2564
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.0956353073645593,
      "learning_rate": 0.00023141977229049272,
      "loss": 4.5247,
      "step": 2565
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.9079718685868076,
      "learning_rate": 0.0002312035475953356,
      "loss": 4.2378,
      "step": 2566
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.46063949434249796,
      "learning_rate": 0.00023098736061566422,
      "loss": 4.042,
      "step": 2567
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3294150048550546,
      "learning_rate": 0.0002307712114699964,
      "loss": 4.0306,
      "step": 2568
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3650402565694655,
      "learning_rate": 0.0002305551002768289,
      "loss": 4.2429,
      "step": 2569
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.6161399296902454,
      "learning_rate": 0.00023033902715463803,
      "loss": 4.2256,
      "step": 2570
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.9331703347458324,
      "learning_rate": 0.00023012299222187896,
      "loss": 4.3089,
      "step": 2571
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.146545007594941,
      "learning_rate": 0.00022990699559698614,
      "loss": 4.0443,
      "step": 2572
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.1751121481909939,
      "learning_rate": 0.0002296910373983728,
      "loss": 4.2889,
      "step": 2573
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.9203205725349273,
      "learning_rate": 0.00022947511774443126,
      "loss": 4.2464,
      "step": 2574
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.6017249705186978,
      "learning_rate": 0.00022925923675353263,
      "loss": 4.2374,
      "step": 2575
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6875214362325415,
      "learning_rate": 0.00022904339454402693,
      "loss": 4.2171,
      "step": 2576
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3721523367939834,
      "learning_rate": 0.00022882759123424277,
      "loss": 4.2348,
      "step": 2577
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.6422949860422037,
      "learning_rate": 0.00022861182694248753,
      "loss": 4.0803,
      "step": 2578
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.286854686785068,
      "learning_rate": 0.0002283961017870472,
      "loss": 4.3155,
      "step": 2579
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.2035783891425007,
      "learning_rate": 0.00022818041588618636,
      "loss": 4.2553,
      "step": 2580
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.699275702023369,
      "learning_rate": 0.00022796476935814784,
      "loss": 4.0383,
      "step": 2581
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3587234727586166,
      "learning_rate": 0.00022774916232115312,
      "loss": 4.2491,
      "step": 2582
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.501453398333668,
      "learning_rate": 0.000227533594893402,
      "loss": 4.3552,
      "step": 2583
    },
    {
      "epoch": 2.95,
      "grad_norm": 12.560714816003086,
      "learning_rate": 0.00022731806719307246,
      "loss": 4.3116,
      "step": 2584
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.302987805636268,
      "learning_rate": 0.00022710257933832094,
      "loss": 4.6409,
      "step": 2585
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.523318009111459,
      "learning_rate": 0.00022688713144728163,
      "loss": 4.4539,
      "step": 2586
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.447460117349026,
      "learning_rate": 0.00022667172363806718,
      "loss": 4.6503,
      "step": 2587
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.9855463787455052,
      "learning_rate": 0.000226456356028768,
      "loss": 4.7454,
      "step": 2588
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.529293789336613,
      "learning_rate": 0.00022624102873745287,
      "loss": 4.4759,
      "step": 2589
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.334558072473303,
      "learning_rate": 0.00022602574188216787,
      "loss": 4.3868,
      "step": 2590
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.9607808878570054,
      "learning_rate": 0.00022581049558093735,
      "loss": 4.3096,
      "step": 2591
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.168540936643506,
      "learning_rate": 0.00022559528995176324,
      "loss": 4.3386,
      "step": 2592
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.1514301601909844,
      "learning_rate": 0.00022538012511262554,
      "loss": 4.3142,
      "step": 2593
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.0339426434520007,
      "learning_rate": 0.00022516500118148125,
      "loss": 4.3733,
      "step": 2594
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6850221192288715,
      "learning_rate": 0.0002249499182762653,
      "loss": 4.2383,
      "step": 2595
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3541074559284516,
      "learning_rate": 0.0002247348765148903,
      "loss": 4.2689,
      "step": 2596
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.872179566716467,
      "learning_rate": 0.00022451987601524614,
      "loss": 4.453,
      "step": 2597
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4706715133882597,
      "learning_rate": 0.00022430491689519986,
      "loss": 4.033,
      "step": 2598
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.1450198748905875,
      "learning_rate": 0.00022408999927259597,
      "loss": 4.3444,
      "step": 2599
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.0701641251360345,
      "learning_rate": 0.00022387512326525654,
      "loss": 4.2313,
      "step": 2600
    },
    {
      "epoch": 2.97,
      "eval_blimp_filtered_avg": 0.5455223880597015,
      "eval_blimp_filtered_std": 0.005564077161698807,
      "step": 2600
    },
    {
      "epoch": 2.97,
      "eval_blimp_supplement_avg": 0.5387931034482759,
      "eval_blimp_supplement_std": 0.0227509399822816,
      "step": 2600
    },
    {
      "epoch": 2.97,
      "eval_vqa_filtered_avg": 0.37,
      "eval_vqa_filtered_std": 0.048523658709391,
      "step": 2600
    },
    {
      "epoch": 2.97,
      "eval_winoground_filtered_avg": 0.48,
      "eval_winoground_filtered_std": 0.05021167315686779,
      "step": 2600
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.8727229357878289,
      "learning_rate": 0.0002236602889909804,
      "loss": 4.5116,
      "step": 2601
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.9244721871122329,
      "learning_rate": 0.00022344549656754388,
      "loss": 4.1678,
      "step": 2602
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.9385567240041308,
      "learning_rate": 0.0002232307461126998,
      "loss": 3.9222,
      "step": 2603
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5224162654985514,
      "learning_rate": 0.00022301603774417856,
      "loss": 4.228,
      "step": 2604
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.0765719130203972,
      "learning_rate": 0.0002228013715796872,
      "loss": 4.1071,
      "step": 2605
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.7413854552588149,
      "learning_rate": 0.0002225867477369097,
      "loss": 4.2614,
      "step": 2606
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.1346633351640534,
      "learning_rate": 0.0002223721663335068,
      "loss": 4.1651,
      "step": 2607
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.0560600309340842,
      "learning_rate": 0.00022215762748711593,
      "loss": 3.9808,
      "step": 2608
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.1497443992074179,
      "learning_rate": 0.00022194313131535133,
      "loss": 4.3269,
      "step": 2609
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4722261049234469,
      "learning_rate": 0.00022172867793580374,
      "loss": 4.2138,
      "step": 2610
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5910058756181615,
      "learning_rate": 0.00022151426746604044,
      "loss": 4.3151,
      "step": 2611
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.8856523009462958,
      "learning_rate": 0.00022129990002360523,
      "loss": 4.2509,
      "step": 2612
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5227832889412158,
      "learning_rate": 0.0002210855757260183,
      "loss": 4.257,
      "step": 2613
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4809934748197426,
      "learning_rate": 0.00022087129469077624,
      "loss": 4.2507,
      "step": 2614
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.194483785462596,
      "learning_rate": 0.00022065705703535175,
      "loss": 4.2309,
      "step": 2615
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.901556622663321,
      "learning_rate": 0.00022044286287719393,
      "loss": 4.4746,
      "step": 2616
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.9241563786779986,
      "learning_rate": 0.00022022871233372797,
      "loss": 4.2084,
      "step": 2617
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3324927865244371,
      "learning_rate": 0.00022001460552235517,
      "loss": 4.3767,
      "step": 2618
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.8992921493918756,
      "learning_rate": 0.00021980054256045292,
      "loss": 4.3004,
      "step": 2619
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.1697682318310916,
      "learning_rate": 0.00021958652356537426,
      "loss": 4.3022,
      "step": 2620
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4984140062763884,
      "learning_rate": 0.00021937254865444855,
      "loss": 4.2635,
      "step": 2621
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.7280864180126836,
      "learning_rate": 0.00021915861794498073,
      "loss": 4.446,
      "step": 2622
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.0087830658007348,
      "learning_rate": 0.00021894473155425166,
      "loss": 4.2718,
      "step": 2623
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9629618155233188,
      "learning_rate": 0.0002187308895995176,
      "loss": 4.2258,
      "step": 2624
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.7353616869265273,
      "learning_rate": 0.00021851709219801078,
      "loss": 4.2871,
      "step": 2625
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.7495729052643275,
      "learning_rate": 0.0002183033394669389,
      "loss": 4.0758,
      "step": 2626
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.7158592969252013,
      "learning_rate": 0.00021808963152348518,
      "loss": 3.9756,
      "step": 2627
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.5638799552630359,
      "learning_rate": 0.00021787596848480816,
      "loss": 4.2823,
      "step": 2628
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.7890001682828334,
      "learning_rate": 0.00021766235046804193,
      "loss": 4.0262,
      "step": 2629
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.4942836086737641,
      "learning_rate": 0.00021744877759029583,
      "loss": 3.9777,
      "step": 2630
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.8135541961657559,
      "learning_rate": 0.00021723524996865448,
      "loss": 4.1506,
      "step": 2631
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.1664684631955407,
      "learning_rate": 0.00021702176772017757,
      "loss": 4.2646,
      "step": 2632
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.8293682398811614,
      "learning_rate": 0.00021680833096190009,
      "loss": 4.0576,
      "step": 2633
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.4927010229853246,
      "learning_rate": 0.00021659493981083201,
      "loss": 4.0795,
      "step": 2634
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.6424839655986803,
      "learning_rate": 0.0002163815943839583,
      "loss": 4.2375,
      "step": 2635
    },
    {
      "epoch": 3.01,
      "grad_norm": 2.108226246900263,
      "learning_rate": 0.0002161682947982389,
      "loss": 4.2053,
      "step": 2636
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.792232717440733,
      "learning_rate": 0.00021595504117060844,
      "loss": 4.2079,
      "step": 2637
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.2774698437159082,
      "learning_rate": 0.00021574183361797652,
      "loss": 4.2522,
      "step": 2638
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.4542567917906226,
      "learning_rate": 0.00021552867225722763,
      "loss": 4.1341,
      "step": 2639
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.6800197798831682,
      "learning_rate": 0.0002153155572052208,
      "loss": 4.3572,
      "step": 2640
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.4070728368563141,
      "learning_rate": 0.00021510248857878938,
      "loss": 4.0916,
      "step": 2641
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.3117404605045788,
      "learning_rate": 0.00021488946649474153,
      "loss": 4.3946,
      "step": 2642
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.7358254102278927,
      "learning_rate": 0.0002146764910698601,
      "loss": 4.1967,
      "step": 2643
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.214865216623838,
      "learning_rate": 0.0002144635624209022,
      "loss": 4.0662,
      "step": 2644
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.6759751989953697,
      "learning_rate": 0.00021425068066459896,
      "loss": 4.2703,
      "step": 2645
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.6881551204194081,
      "learning_rate": 0.00021403784591765616,
      "loss": 4.0712,
      "step": 2646
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.2131652979686258,
      "learning_rate": 0.00021382505829675388,
      "loss": 4.3338,
      "step": 2647
    },
    {
      "epoch": 3.03,
      "grad_norm": 2.039979036730362,
      "learning_rate": 0.00021361231791854635,
      "loss": 4.3175,
      "step": 2648
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.8060956063803566,
      "learning_rate": 0.00021339962489966136,
      "loss": 4.2945,
      "step": 2649
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.0289587912320695,
      "learning_rate": 0.0002131869793567014,
      "loss": 4.1901,
      "step": 2650
    },
    {
      "epoch": 3.03,
      "grad_norm": 2.1665868182128447,
      "learning_rate": 0.00021297438140624266,
      "loss": 4.1961,
      "step": 2651
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.3611300667508444,
      "learning_rate": 0.00021276183116483542,
      "loss": 4.2938,
      "step": 2652
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.6176308881777606,
      "learning_rate": 0.00021254932874900325,
      "loss": 4.3531,
      "step": 2653
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.4881255043094301,
      "learning_rate": 0.00021233687427524428,
      "loss": 4.2239,
      "step": 2654
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.5261693157953908,
      "learning_rate": 0.00021212446786002977,
      "loss": 4.3056,
      "step": 2655
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.7860846878830503,
      "learning_rate": 0.00021191210961980493,
      "loss": 4.15,
      "step": 2656
    },
    {
      "epoch": 3.04,
      "grad_norm": 1.4787732598354335,
      "learning_rate": 0.0002116997996709885,
      "loss": 4.0094,
      "step": 2657
    },
    {
      "epoch": 3.04,
      "grad_norm": 1.4980757634326116,
      "learning_rate": 0.0002114875381299726,
      "loss": 4.0987,
      "step": 2658
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.9545345050100846,
      "learning_rate": 0.00021127532511312297,
      "loss": 4.3616,
      "step": 2659
    },
    {
      "epoch": 3.04,
      "grad_norm": 2.329289846801104,
      "learning_rate": 0.00021106316073677875,
      "loss": 4.0631,
      "step": 2660
    },
    {
      "epoch": 3.04,
      "grad_norm": 2.065529521756144,
      "learning_rate": 0.00021085104511725243,
      "loss": 4.1527,
      "step": 2661
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.8745820558575126,
      "learning_rate": 0.00021063897837082947,
      "loss": 4.1326,
      "step": 2662
    },
    {
      "epoch": 3.04,
      "grad_norm": 1.838569962989226,
      "learning_rate": 0.00021042696061376897,
      "loss": 4.2325,
      "step": 2663
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.9027832054836259,
      "learning_rate": 0.00021021499196230292,
      "loss": 4.2452,
      "step": 2664
    },
    {
      "epoch": 3.05,
      "grad_norm": 2.004640639830266,
      "learning_rate": 0.00021000307253263656,
      "loss": 4.1976,
      "step": 2665
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.0051871063020932,
      "learning_rate": 0.00020979120244094787,
      "loss": 4.3394,
      "step": 2666
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.6446721157067745,
      "learning_rate": 0.00020957938180338803,
      "loss": 4.0812,
      "step": 2667
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.9178515402890869,
      "learning_rate": 0.000209367610736081,
      "loss": 3.9515,
      "step": 2668
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.0501235274774758,
      "learning_rate": 0.00020915588935512363,
      "loss": 4.1962,
      "step": 2669
    },
    {
      "epoch": 3.05,
      "grad_norm": 2.0750672473350504,
      "learning_rate": 0.00020894421777658544,
      "loss": 4.0438,
      "step": 2670
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.9479414065952017,
      "learning_rate": 0.00020873259611650874,
      "loss": 3.9567,
      "step": 2671
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.811164421476797,
      "learning_rate": 0.00020852102449090838,
      "loss": 4.0245,
      "step": 2672
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.5447498867380527,
      "learning_rate": 0.0002083095030157719,
      "loss": 4.3063,
      "step": 2673
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.0998796403756723,
      "learning_rate": 0.00020809803180705928,
      "loss": 4.3357,
      "step": 2674
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.1688581819045187,
      "learning_rate": 0.00020788661098070286,
      "loss": 4.2906,
      "step": 2675
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.6925154779110498,
      "learning_rate": 0.0002076752406526075,
      "loss": 4.2833,
      "step": 2676
    },
    {
      "epoch": 3.06,
      "grad_norm": 2.2082989049842885,
      "learning_rate": 0.0002074639209386503,
      "loss": 4.0648,
      "step": 2677
    },
    {
      "epoch": 3.06,
      "grad_norm": 2.1793976202979435,
      "learning_rate": 0.00020725265195468073,
      "loss": 4.1357,
      "step": 2678
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.5134876071321495,
      "learning_rate": 0.00020704143381652015,
      "loss": 3.9925,
      "step": 2679
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.3598460313990763,
      "learning_rate": 0.00020683026663996238,
      "loss": 4.2257,
      "step": 2680
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.9858963700981123,
      "learning_rate": 0.00020661915054077316,
      "loss": 4.0519,
      "step": 2681
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.178321340340512,
      "learning_rate": 0.00020640808563469022,
      "loss": 4.214,
      "step": 2682
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.7172981860105319,
      "learning_rate": 0.0002061970720374232,
      "loss": 4.3686,
      "step": 2683
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.0477243872266038,
      "learning_rate": 0.0002059861098646537,
      "loss": 4.1729,
      "step": 2684
    },
    {
      "epoch": 3.07,
      "grad_norm": 2.6117109457857097,
      "learning_rate": 0.00020577519923203498,
      "loss": 4.2265,
      "step": 2685
    },
    {
      "epoch": 3.07,
      "grad_norm": 2.698562561026729,
      "learning_rate": 0.00020556434025519242,
      "loss": 4.2613,
      "step": 2686
    },
    {
      "epoch": 3.07,
      "grad_norm": 3.1708653692198214,
      "learning_rate": 0.00020535353304972247,
      "loss": 4.2421,
      "step": 2687
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.9094911589561623,
      "learning_rate": 0.0002051427777311937,
      "loss": 4.2907,
      "step": 2688
    },
    {
      "epoch": 3.07,
      "grad_norm": 4.570019846548903,
      "learning_rate": 0.00020493207441514598,
      "loss": 4.2938,
      "step": 2689
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.3031766399729952,
      "learning_rate": 0.00020472142321709087,
      "loss": 4.2353,
      "step": 2690
    },
    {
      "epoch": 3.08,
      "grad_norm": 2.0044634562762234,
      "learning_rate": 0.00020451082425251136,
      "loss": 4.0893,
      "step": 2691
    },
    {
      "epoch": 3.08,
      "grad_norm": 4.039965494960537,
      "learning_rate": 0.00020430027763686137,
      "loss": 4.2877,
      "step": 2692
    },
    {
      "epoch": 3.08,
      "grad_norm": 2.4430676462275334,
      "learning_rate": 0.0002040897834855665,
      "loss": 4.321,
      "step": 2693
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.5555970763940636,
      "learning_rate": 0.0002038793419140238,
      "loss": 4.185,
      "step": 2694
    },
    {
      "epoch": 3.08,
      "grad_norm": 2.8700582176254485,
      "learning_rate": 0.0002036689530376011,
      "loss": 4.2749,
      "step": 2695
    },
    {
      "epoch": 3.08,
      "grad_norm": 3.247703315247495,
      "learning_rate": 0.0002034586169716373,
      "loss": 4.464,
      "step": 2696
    },
    {
      "epoch": 3.08,
      "grad_norm": 2.6545343780434347,
      "learning_rate": 0.0002032483338314427,
      "loss": 4.2278,
      "step": 2697
    },
    {
      "epoch": 3.08,
      "grad_norm": 2.3617182896214204,
      "learning_rate": 0.0002030381037322983,
      "loss": 4.2908,
      "step": 2698
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.8786949773119794,
      "learning_rate": 0.00020282792678945636,
      "loss": 4.206,
      "step": 2699
    },
    {
      "epoch": 3.09,
      "grad_norm": 2.542431933100143,
      "learning_rate": 0.00020261780311813936,
      "loss": 4.1357,
      "step": 2700
    },
    {
      "epoch": 3.09,
      "eval_blimp_filtered_avg": 0.5353731343283582,
      "eval_blimp_filtered_std": 0.0056318479878109074,
      "step": 2700
    },
    {
      "epoch": 3.09,
      "eval_blimp_supplement_avg": 0.5258620689655172,
      "eval_blimp_supplement_std": 0.022937658138571546,
      "step": 2700
    },
    {
      "epoch": 3.09,
      "eval_vqa_filtered_avg": 0.33,
      "eval_vqa_filtered_std": 0.047258156262526045,
      "step": 2700
    },
    {
      "epoch": 3.09,
      "eval_winoground_filtered_avg": 0.49,
      "eval_winoground_filtered_std": 0.05024183937956912,
      "step": 2700
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.7292275735702843,
      "learning_rate": 0.00020240773283354123,
      "loss": 4.1754,
      "step": 2701
    },
    {
      "epoch": 3.09,
      "grad_norm": 2.6074000685821312,
      "learning_rate": 0.00020219771605082634,
      "loss": 4.1111,
      "step": 2702
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.993304006344121,
      "learning_rate": 0.0002019877528851299,
      "loss": 4.4819,
      "step": 2703
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.6207996773244095,
      "learning_rate": 0.00020177784345155728,
      "loss": 4.2965,
      "step": 2704
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.5531651560433702,
      "learning_rate": 0.00020156798786518496,
      "loss": 4.265,
      "step": 2705
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.8497052989198064,
      "learning_rate": 0.00020135818624105958,
      "loss": 4.2033,
      "step": 2706
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.488630127697065,
      "learning_rate": 0.00020114843869419824,
      "loss": 4.2435,
      "step": 2707
    },
    {
      "epoch": 3.09,
      "grad_norm": 2.0388096750792433,
      "learning_rate": 0.00020093874533958853,
      "loss": 4.3619,
      "step": 2708
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.9224294577437181,
      "learning_rate": 0.00020072910629218807,
      "loss": 4.1371,
      "step": 2709
    },
    {
      "epoch": 3.1,
      "grad_norm": 3.002375925776754,
      "learning_rate": 0.00020051952166692493,
      "loss": 4.0652,
      "step": 2710
    },
    {
      "epoch": 3.1,
      "grad_norm": 2.6158857213290725,
      "learning_rate": 0.00020030999157869724,
      "loss": 4.4687,
      "step": 2711
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.7934475196892388,
      "learning_rate": 0.00020010051614237335,
      "loss": 4.2267,
      "step": 2712
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.9783242840319151,
      "learning_rate": 0.00019989109547279143,
      "loss": 4.3585,
      "step": 2713
    },
    {
      "epoch": 3.1,
      "grad_norm": 2.050480858390498,
      "learning_rate": 0.00019968172968475985,
      "loss": 4.4526,
      "step": 2714
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.9919909558947734,
      "learning_rate": 0.00019947241889305676,
      "loss": 4.2746,
      "step": 2715
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.9857840157863261,
      "learning_rate": 0.00019926316321243029,
      "loss": 4.2104,
      "step": 2716
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.5760178940515555,
      "learning_rate": 0.00019905396275759815,
      "loss": 4.2138,
      "step": 2717
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.9199717972661715,
      "learning_rate": 0.00019884481764324794,
      "loss": 4.4121,
      "step": 2718
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.316906455578091,
      "learning_rate": 0.00019863572798403683,
      "loss": 4.1376,
      "step": 2719
    },
    {
      "epoch": 3.11,
      "grad_norm": 3.6629572339659133,
      "learning_rate": 0.0001984266938945918,
      "loss": 4.069,
      "step": 2720
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.0645284802737562,
      "learning_rate": 0.00019821771548950902,
      "loss": 4.1634,
      "step": 2721
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.9629434886401738,
      "learning_rate": 0.00019800879288335433,
      "loss": 4.376,
      "step": 2722
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.360000287660664,
      "learning_rate": 0.000197799926190663,
      "loss": 3.9352,
      "step": 2723
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.8424103133351177,
      "learning_rate": 0.00019759111552593962,
      "loss": 4.1944,
      "step": 2724
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.1651973218605336,
      "learning_rate": 0.00019738236100365813,
      "loss": 4.3105,
      "step": 2725
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.1297791010771425,
      "learning_rate": 0.00019717366273826146,
      "loss": 4.2934,
      "step": 2726
    },
    {
      "epoch": 3.12,
      "grad_norm": 2.4291880833848025,
      "learning_rate": 0.0001969650208441619,
      "loss": 4.1806,
      "step": 2727
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.2374885655487293,
      "learning_rate": 0.0001967564354357408,
      "loss": 4.1317,
      "step": 2728
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.8721553688523321,
      "learning_rate": 0.00019654790662734862,
      "loss": 4.4364,
      "step": 2729
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.822995933702198,
      "learning_rate": 0.0001963394345333046,
      "loss": 4.3738,
      "step": 2730
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.8880750647582601,
      "learning_rate": 0.00019613101926789698,
      "loss": 4.2323,
      "step": 2731
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.7334582507362306,
      "learning_rate": 0.00019592266094538283,
      "loss": 4.2155,
      "step": 2732
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.3524743414472449,
      "learning_rate": 0.00019571435967998828,
      "loss": 4.2045,
      "step": 2733
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.6546135611244575,
      "learning_rate": 0.00019550611558590765,
      "loss": 4.2247,
      "step": 2734
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.8353773935104888,
      "learning_rate": 0.00019529792877730426,
      "loss": 4.2016,
      "step": 2735
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.9534438873953378,
      "learning_rate": 0.00019508979936830995,
      "loss": 4.221,
      "step": 2736
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.7253657811979985,
      "learning_rate": 0.0001948817274730254,
      "loss": 4.4386,
      "step": 2737
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.8620214751918487,
      "learning_rate": 0.00019467371320551903,
      "loss": 4.2381,
      "step": 2738
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.8938925562649608,
      "learning_rate": 0.0001944657566798283,
      "loss": 4.2226,
      "step": 2739
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.9793237651848,
      "learning_rate": 0.00019425785800995877,
      "loss": 4.163,
      "step": 2740
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.6287904392424756,
      "learning_rate": 0.0001940500173098845,
      "loss": 4.0035,
      "step": 2741
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.8996685690407565,
      "learning_rate": 0.00019384223469354742,
      "loss": 4.4299,
      "step": 2742
    },
    {
      "epoch": 3.13,
      "grad_norm": 2.6857428303972153,
      "learning_rate": 0.00019363451027485778,
      "loss": 4.3645,
      "step": 2743
    },
    {
      "epoch": 3.14,
      "grad_norm": 2.2128391034201176,
      "learning_rate": 0.00019342684416769404,
      "loss": 4.1948,
      "step": 2744
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.4568577274529326,
      "learning_rate": 0.00019321923648590262,
      "loss": 4.0352,
      "step": 2745
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.7772223415659458,
      "learning_rate": 0.00019301168734329793,
      "loss": 4.2246,
      "step": 2746
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.014222072539158,
      "learning_rate": 0.00019280419685366193,
      "loss": 4.0578,
      "step": 2747
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.3241057053060967,
      "learning_rate": 0.00019259676513074501,
      "loss": 4.2741,
      "step": 2748
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.176105843541019,
      "learning_rate": 0.0001923893922882651,
      "loss": 4.2011,
      "step": 2749
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.4545692757015989,
      "learning_rate": 0.00019218207843990778,
      "loss": 4.078,
      "step": 2750
    },
    {
      "epoch": 3.14,
      "grad_norm": 2.769358037826632,
      "learning_rate": 0.00019197482369932614,
      "loss": 4.317,
      "step": 2751
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.9663605095477744,
      "learning_rate": 0.00019176762818014125,
      "loss": 4.2099,
      "step": 2752
    },
    {
      "epoch": 3.15,
      "grad_norm": 2.8197785374562043,
      "learning_rate": 0.00019156049199594153,
      "loss": 4.345,
      "step": 2753
    },
    {
      "epoch": 3.15,
      "grad_norm": 2.205186944227695,
      "learning_rate": 0.0001913534152602828,
      "loss": 4.2379,
      "step": 2754
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.9675083150671928,
      "learning_rate": 0.00019114639808668834,
      "loss": 4.2494,
      "step": 2755
    },
    {
      "epoch": 3.15,
      "grad_norm": 2.801212887122284,
      "learning_rate": 0.00019093944058864879,
      "loss": 4.3394,
      "step": 2756
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.5755646295888077,
      "learning_rate": 0.0001907325428796221,
      "loss": 4.1915,
      "step": 2757
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.867687368817391,
      "learning_rate": 0.00019052570507303344,
      "loss": 4.1686,
      "step": 2758
    },
    {
      "epoch": 3.15,
      "grad_norm": 3.870990645689623,
      "learning_rate": 0.00019031892728227503,
      "loss": 4.0606,
      "step": 2759
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.024705655299172,
      "learning_rate": 0.00019011220962070633,
      "loss": 4.1624,
      "step": 2760
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.7441593562682862,
      "learning_rate": 0.00018990555220165377,
      "loss": 4.3822,
      "step": 2761
    },
    {
      "epoch": 3.16,
      "grad_norm": 4.759590525575753,
      "learning_rate": 0.00018969895513841074,
      "loss": 4.1238,
      "step": 2762
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.9968501678298581,
      "learning_rate": 0.00018949241854423771,
      "loss": 4.235,
      "step": 2763
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.451454375080533,
      "learning_rate": 0.00018928594253236165,
      "loss": 4.3202,
      "step": 2764
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.3257194039653943,
      "learning_rate": 0.00018907952721597662,
      "loss": 4.3438,
      "step": 2765
    },
    {
      "epoch": 3.16,
      "grad_norm": 2.1374024861455907,
      "learning_rate": 0.00018887317270824326,
      "loss": 4.2327,
      "step": 2766
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.275263829009234,
      "learning_rate": 0.00018866687912228906,
      "loss": 4.1781,
      "step": 2767
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.06338026796848,
      "learning_rate": 0.00018846064657120778,
      "loss": 4.1637,
      "step": 2768
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.9782036180312333,
      "learning_rate": 0.00018825447516806,
      "loss": 4.2968,
      "step": 2769
    },
    {
      "epoch": 3.17,
      "grad_norm": 2.7796977175907625,
      "learning_rate": 0.00018804836502587277,
      "loss": 4.2908,
      "step": 2770
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.3630411955029582,
      "learning_rate": 0.00018784231625763943,
      "loss": 4.1112,
      "step": 2771
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.934893774875886,
      "learning_rate": 0.00018763632897631963,
      "loss": 4.2957,
      "step": 2772
    },
    {
      "epoch": 3.17,
      "grad_norm": 2.4674779980346835,
      "learning_rate": 0.00018743040329483946,
      "loss": 4.4524,
      "step": 2773
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.6088177691005527,
      "learning_rate": 0.00018722453932609121,
      "loss": 4.0494,
      "step": 2774
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.2479713377836712,
      "learning_rate": 0.00018701873718293333,
      "loss": 4.1084,
      "step": 2775
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.6510558887609793,
      "learning_rate": 0.00018681299697819028,
      "loss": 4.4719,
      "step": 2776
    },
    {
      "epoch": 3.17,
      "grad_norm": 3.6068418026326805,
      "learning_rate": 0.00018660731882465267,
      "loss": 4.2634,
      "step": 2777
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.814194776720197,
      "learning_rate": 0.00018640170283507705,
      "loss": 4.2141,
      "step": 2778
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.439611143975839,
      "learning_rate": 0.00018619614912218584,
      "loss": 4.2321,
      "step": 2779
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.3044665021327235,
      "learning_rate": 0.0001859906577986677,
      "loss": 4.1679,
      "step": 2780
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.8740949296319453,
      "learning_rate": 0.00018578522897717636,
      "loss": 4.1068,
      "step": 2781
    },
    {
      "epoch": 3.18,
      "grad_norm": 4.139889055194011,
      "learning_rate": 0.0001855798627703319,
      "loss": 4.2989,
      "step": 2782
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.703480693619845,
      "learning_rate": 0.00018537455929071975,
      "loss": 4.2224,
      "step": 2783
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.9046471578387805,
      "learning_rate": 0.00018516931865089135,
      "loss": 4.0958,
      "step": 2784
    },
    {
      "epoch": 3.18,
      "grad_norm": 2.0730089257085496,
      "learning_rate": 0.00018496414096336308,
      "loss": 4.0995,
      "step": 2785
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.1309095981747728,
      "learning_rate": 0.00018475902634061724,
      "loss": 4.4566,
      "step": 2786
    },
    {
      "epoch": 3.19,
      "grad_norm": 3.5848310589617807,
      "learning_rate": 0.00018455397489510145,
      "loss": 4.1497,
      "step": 2787
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.9759554264190222,
      "learning_rate": 0.0001843489867392289,
      "loss": 4.2198,
      "step": 2788
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.174977086817324,
      "learning_rate": 0.00018414406198537762,
      "loss": 4.1557,
      "step": 2789
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.4517393729020691,
      "learning_rate": 0.00018393920074589114,
      "loss": 4.1268,
      "step": 2790
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.3096553096237091,
      "learning_rate": 0.00018373440313307834,
      "loss": 4.0351,
      "step": 2791
    },
    {
      "epoch": 3.19,
      "grad_norm": 2.714336817661011,
      "learning_rate": 0.00018352966925921314,
      "loss": 4.2838,
      "step": 2792
    },
    {
      "epoch": 3.19,
      "grad_norm": 2.124072644121655,
      "learning_rate": 0.00018332499923653416,
      "loss": 4.2268,
      "step": 2793
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.8543296956502366,
      "learning_rate": 0.00018312039317724534,
      "loss": 4.2252,
      "step": 2794
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.8846544033255543,
      "learning_rate": 0.0001829158511935156,
      "loss": 4.4064,
      "step": 2795
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.5569247935975998,
      "learning_rate": 0.00018271137339747858,
      "loss": 4.1093,
      "step": 2796
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.2665889513708246,
      "learning_rate": 0.00018250695990123297,
      "loss": 4.2194,
      "step": 2797
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.8271480060841104,
      "learning_rate": 0.00018230261081684168,
      "loss": 4.2065,
      "step": 2798
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.9493994903903273,
      "learning_rate": 0.00018209832625633285,
      "loss": 4.1319,
      "step": 2799
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.9277148937095753,
      "learning_rate": 0.00018189410633169908,
      "loss": 4.2486,
      "step": 2800
    },
    {
      "epoch": 3.2,
      "eval_blimp_filtered_avg": 0.5273134328358209,
      "eval_blimp_filtered_std": 0.005565775425191919,
      "step": 2800
    },
    {
      "epoch": 3.2,
      "eval_blimp_supplement_avg": 0.5301724137931034,
      "eval_blimp_supplement_std": 0.023129847773216545,
      "step": 2800
    },
    {
      "epoch": 3.2,
      "eval_vqa_filtered_avg": 0.38,
      "eval_vqa_filtered_std": 0.048783173121456316,
      "step": 2800
    },
    {
      "epoch": 3.2,
      "eval_winoground_filtered_avg": 0.51,
      "eval_winoground_filtered_std": 0.05024183937956912,
      "step": 2800
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.630683049379475,
      "learning_rate": 0.00018168995115489745,
      "loss": 4.3305,
      "step": 2801
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.9216934625829307,
      "learning_rate": 0.00018148586083784955,
      "loss": 4.1906,
      "step": 2802
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.173095770955462,
      "learning_rate": 0.00018128183549244155,
      "loss": 4.1595,
      "step": 2803
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.475995638245296,
      "learning_rate": 0.00018107787523052385,
      "loss": 4.4443,
      "step": 2804
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.3687966744577391,
      "learning_rate": 0.0001808739801639113,
      "loss": 4.0034,
      "step": 2805
    },
    {
      "epoch": 3.21,
      "grad_norm": 2.797063435389921,
      "learning_rate": 0.0001806701504043829,
      "loss": 4.148,
      "step": 2806
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.3105120131032841,
      "learning_rate": 0.0001804663860636818,
      "loss": 4.3193,
      "step": 2807
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.45735139256422,
      "learning_rate": 0.00018026268725351548,
      "loss": 3.9888,
      "step": 2808
    },
    {
      "epoch": 3.21,
      "grad_norm": 3.0334698428277767,
      "learning_rate": 0.0001800590540855555,
      "loss": 4.0786,
      "step": 2809
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.2674645044954693,
      "learning_rate": 0.00017985548667143707,
      "loss": 4.1715,
      "step": 2810
    },
    {
      "epoch": 3.21,
      "grad_norm": 2.042616524649647,
      "learning_rate": 0.00017965198512275975,
      "loss": 4.1544,
      "step": 2811
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.3436950328277817,
      "learning_rate": 0.00017944854955108684,
      "loss": 4.1261,
      "step": 2812
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.8908413602736949,
      "learning_rate": 0.0001792451800679454,
      "loss": 4.4054,
      "step": 2813
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.1175925899970909,
      "learning_rate": 0.00017904187678482653,
      "loss": 4.1414,
      "step": 2814
    },
    {
      "epoch": 3.22,
      "grad_norm": 3.7512680533101066,
      "learning_rate": 0.00017883863981318462,
      "loss": 4.1348,
      "step": 2815
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.003180208125442,
      "learning_rate": 0.00017863546926443803,
      "loss": 4.1727,
      "step": 2816
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.7800670935374088,
      "learning_rate": 0.0001784323652499686,
      "loss": 4.1945,
      "step": 2817
    },
    {
      "epoch": 3.22,
      "grad_norm": 2.123772761726997,
      "learning_rate": 0.0001782293278811218,
      "loss": 4.1553,
      "step": 2818
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.9025643186003013,
      "learning_rate": 0.0001780263572692063,
      "loss": 4.046,
      "step": 2819
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.767623134027193,
      "learning_rate": 0.0001778234535254944,
      "loss": 4.3584,
      "step": 2820
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.742710690728317,
      "learning_rate": 0.00017762061676122168,
      "loss": 4.2054,
      "step": 2821
    },
    {
      "epoch": 3.23,
      "grad_norm": 2.236017233257212,
      "learning_rate": 0.00017741784708758709,
      "loss": 4.1117,
      "step": 2822
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.5003693279750003,
      "learning_rate": 0.00017721514461575252,
      "loss": 3.9894,
      "step": 2823
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.9296100255091893,
      "learning_rate": 0.00017701250945684338,
      "loss": 4.1613,
      "step": 2824
    },
    {
      "epoch": 3.23,
      "grad_norm": 4.9122113548548185,
      "learning_rate": 0.00017680994172194792,
      "loss": 4.3209,
      "step": 2825
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.2134152007691492,
      "learning_rate": 0.00017660744152211757,
      "loss": 4.2426,
      "step": 2826
    },
    {
      "epoch": 3.23,
      "grad_norm": 2.178043480665856,
      "learning_rate": 0.00017640500896836663,
      "loss": 3.8948,
      "step": 2827
    },
    {
      "epoch": 3.23,
      "grad_norm": 2.3159068914153194,
      "learning_rate": 0.00017620264417167233,
      "loss": 4.1773,
      "step": 2828
    },
    {
      "epoch": 3.23,
      "grad_norm": 4.269184055570103,
      "learning_rate": 0.0001760003472429749,
      "loss": 4.4059,
      "step": 2829
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.3435879904144112,
      "learning_rate": 0.00017579811829317721,
      "loss": 4.2174,
      "step": 2830
    },
    {
      "epoch": 3.24,
      "grad_norm": 2.0081736077872283,
      "learning_rate": 0.00017559595743314483,
      "loss": 4.2244,
      "step": 2831
    },
    {
      "epoch": 3.24,
      "grad_norm": 2.735090810337554,
      "learning_rate": 0.00017539386477370618,
      "loss": 4.0842,
      "step": 2832
    },
    {
      "epoch": 3.24,
      "grad_norm": 3.5569650378515165,
      "learning_rate": 0.00017519184042565203,
      "loss": 4.2274,
      "step": 2833
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.8216625663317314,
      "learning_rate": 0.00017498988449973614,
      "loss": 4.0313,
      "step": 2834
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.271137937095734,
      "learning_rate": 0.0001747879971066742,
      "loss": 4.3292,
      "step": 2835
    },
    {
      "epoch": 3.24,
      "grad_norm": 3.214508185974161,
      "learning_rate": 0.00017458617835714477,
      "loss": 4.051,
      "step": 2836
    },
    {
      "epoch": 3.24,
      "grad_norm": 2.937214686056647,
      "learning_rate": 0.0001743844283617885,
      "loss": 4.2712,
      "step": 2837
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.5105531541861679,
      "learning_rate": 0.00017418274723120857,
      "loss": 3.9988,
      "step": 2838
    },
    {
      "epoch": 3.24,
      "grad_norm": 2.0023142527305517,
      "learning_rate": 0.00017398113507597035,
      "loss": 4.2166,
      "step": 2839
    },
    {
      "epoch": 3.25,
      "grad_norm": 2.4779953064375047,
      "learning_rate": 0.00017377959200660118,
      "loss": 3.9077,
      "step": 2840
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.317787121310388,
      "learning_rate": 0.00017357811813359074,
      "loss": 4.214,
      "step": 2841
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.9256972091848976,
      "learning_rate": 0.00017337671356739076,
      "loss": 4.1372,
      "step": 2842
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.4817489697778077,
      "learning_rate": 0.00017317537841841492,
      "loss": 4.1376,
      "step": 2843
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.2855792677187068,
      "learning_rate": 0.00017297411279703888,
      "loss": 3.9826,
      "step": 2844
    },
    {
      "epoch": 3.25,
      "grad_norm": 2.318716505291194,
      "learning_rate": 0.00017277291681360018,
      "loss": 4.0206,
      "step": 2845
    },
    {
      "epoch": 3.25,
      "grad_norm": 2.642236361428631,
      "learning_rate": 0.0001725717905783982,
      "loss": 4.2664,
      "step": 2846
    },
    {
      "epoch": 3.25,
      "grad_norm": 2.47296432818547,
      "learning_rate": 0.00017237073420169416,
      "loss": 4.1272,
      "step": 2847
    },
    {
      "epoch": 3.25,
      "grad_norm": 3.01850789171758,
      "learning_rate": 0.00017216974779371056,
      "loss": 4.2043,
      "step": 2848
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.1187419125300315,
      "learning_rate": 0.0001719688314646321,
      "loss": 4.2573,
      "step": 2849
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.0302262446464754,
      "learning_rate": 0.00017176798532460473,
      "loss": 4.1824,
      "step": 2850
    },
    {
      "epoch": 3.26,
      "grad_norm": 2.0804115302692514,
      "learning_rate": 0.00017156720948373593,
      "loss": 4.2497,
      "step": 2851
    },
    {
      "epoch": 3.26,
      "grad_norm": 2.4948952132446385,
      "learning_rate": 0.00017136650405209515,
      "loss": 4.2766,
      "step": 2852
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.6536561659707587,
      "learning_rate": 0.00017116586913971235,
      "loss": 3.9676,
      "step": 2853
    },
    {
      "epoch": 3.26,
      "grad_norm": 2.476760113222643,
      "learning_rate": 0.00017096530485657943,
      "loss": 4.1596,
      "step": 2854
    },
    {
      "epoch": 3.26,
      "grad_norm": 7.433123354072842,
      "learning_rate": 0.00017076481131264947,
      "loss": 4.0824,
      "step": 2855
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.7065925266498387,
      "learning_rate": 0.00017056438861783686,
      "loss": 4.1428,
      "step": 2856
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.6243008569365676,
      "learning_rate": 0.00017036403688201678,
      "loss": 4.093,
      "step": 2857
    },
    {
      "epoch": 3.27,
      "grad_norm": 3.1863518320588144,
      "learning_rate": 0.00017016375621502582,
      "loss": 4.2422,
      "step": 2858
    },
    {
      "epoch": 3.27,
      "grad_norm": 3.8071043921466723,
      "learning_rate": 0.00016996354672666147,
      "loss": 4.3557,
      "step": 2859
    },
    {
      "epoch": 3.27,
      "grad_norm": 2.0617686646828197,
      "learning_rate": 0.0001697634085266826,
      "loss": 4.087,
      "step": 2860
    },
    {
      "epoch": 3.27,
      "grad_norm": 9.1727244499222,
      "learning_rate": 0.00016956334172480838,
      "loss": 4.3879,
      "step": 2861
    },
    {
      "epoch": 3.27,
      "grad_norm": 5.261908166753962,
      "learning_rate": 0.0001693633464307191,
      "loss": 4.9178,
      "step": 2862
    },
    {
      "epoch": 3.27,
      "grad_norm": 6.060845270750296,
      "learning_rate": 0.00016916342275405605,
      "loss": 4.9404,
      "step": 2863
    },
    {
      "epoch": 3.27,
      "grad_norm": 2.271478110637955,
      "learning_rate": 0.00016896357080442104,
      "loss": 4.9488,
      "step": 2864
    },
    {
      "epoch": 3.27,
      "grad_norm": 2.0197099285268916,
      "learning_rate": 0.00016876379069137652,
      "loss": 4.7329,
      "step": 2865
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.872812430044207,
      "learning_rate": 0.00016856408252444548,
      "loss": 4.5932,
      "step": 2866
    },
    {
      "epoch": 3.28,
      "grad_norm": 4.116240753372435,
      "learning_rate": 0.00016836444641311194,
      "loss": 4.63,
      "step": 2867
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.9219396269261015,
      "learning_rate": 0.00016816488246682002,
      "loss": 4.5567,
      "step": 2868
    },
    {
      "epoch": 3.28,
      "grad_norm": 3.5831235722854804,
      "learning_rate": 0.0001679653907949744,
      "loss": 4.726,
      "step": 2869
    },
    {
      "epoch": 3.28,
      "grad_norm": 2.940435912801413,
      "learning_rate": 0.00016776597150693995,
      "loss": 4.4976,
      "step": 2870
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.8163356851602066,
      "learning_rate": 0.00016756662471204204,
      "loss": 4.3669,
      "step": 2871
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.40440816874465,
      "learning_rate": 0.00016736735051956633,
      "loss": 4.454,
      "step": 2872
    },
    {
      "epoch": 3.28,
      "grad_norm": 2.216907232220189,
      "learning_rate": 0.00016716814903875856,
      "loss": 4.3039,
      "step": 2873
    },
    {
      "epoch": 3.28,
      "grad_norm": 2.98365223977936,
      "learning_rate": 0.00016696902037882468,
      "loss": 4.4578,
      "step": 2874
    },
    {
      "epoch": 3.29,
      "grad_norm": 9.014507515227566,
      "learning_rate": 0.00016676996464893074,
      "loss": 4.5878,
      "step": 2875
    },
    {
      "epoch": 3.29,
      "grad_norm": 5.043157847690633,
      "learning_rate": 0.0001665709819582027,
      "loss": 4.4125,
      "step": 2876
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.5068394288891527,
      "learning_rate": 0.0001663720724157267,
      "loss": 4.4419,
      "step": 2877
    },
    {
      "epoch": 3.29,
      "grad_norm": 3.8361428654282195,
      "learning_rate": 0.0001661732361305484,
      "loss": 4.4068,
      "step": 2878
    },
    {
      "epoch": 3.29,
      "grad_norm": 3.872461133171505,
      "learning_rate": 0.00016597447321167364,
      "loss": 4.5011,
      "step": 2879
    },
    {
      "epoch": 3.29,
      "grad_norm": 2.6401463943459262,
      "learning_rate": 0.00016577578376806787,
      "loss": 4.3456,
      "step": 2880
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.640577001122683,
      "learning_rate": 0.00016557716790865636,
      "loss": 4.2418,
      "step": 2881
    },
    {
      "epoch": 3.29,
      "grad_norm": 8.599084439651879,
      "learning_rate": 0.00016537862574232397,
      "loss": 4.3751,
      "step": 2882
    },
    {
      "epoch": 3.29,
      "grad_norm": 3.5581856191368715,
      "learning_rate": 0.0001651801573779152,
      "loss": 4.4489,
      "step": 2883
    },
    {
      "epoch": 3.3,
      "grad_norm": 4.1828779171024015,
      "learning_rate": 0.00016498176292423412,
      "loss": 4.5774,
      "step": 2884
    },
    {
      "epoch": 3.3,
      "grad_norm": 3.1045328923276987,
      "learning_rate": 0.00016478344249004412,
      "loss": 4.5375,
      "step": 2885
    },
    {
      "epoch": 3.3,
      "grad_norm": 2.754031983637328,
      "learning_rate": 0.00016458519618406834,
      "loss": 4.597,
      "step": 2886
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.7748134900925425,
      "learning_rate": 0.00016438702411498878,
      "loss": 4.3639,
      "step": 2887
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.849536286646378,
      "learning_rate": 0.00016418892639144715,
      "loss": 4.5251,
      "step": 2888
    },
    {
      "epoch": 3.3,
      "grad_norm": 2.542259761762008,
      "learning_rate": 0.00016399090312204433,
      "loss": 4.609,
      "step": 2889
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.9888744187862533,
      "learning_rate": 0.0001637929544153403,
      "loss": 4.3906,
      "step": 2890
    },
    {
      "epoch": 3.3,
      "grad_norm": 3.2179599384776476,
      "learning_rate": 0.00016359508037985416,
      "loss": 4.3373,
      "step": 2891
    },
    {
      "epoch": 3.31,
      "grad_norm": 2.770506910609272,
      "learning_rate": 0.00016339728112406415,
      "loss": 4.5812,
      "step": 2892
    },
    {
      "epoch": 3.31,
      "grad_norm": 2.788267997463789,
      "learning_rate": 0.00016319955675640746,
      "loss": 4.4812,
      "step": 2893
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.7771413862165066,
      "learning_rate": 0.0001630019073852804,
      "loss": 4.2795,
      "step": 2894
    },
    {
      "epoch": 3.31,
      "grad_norm": 3.9014400560439415,
      "learning_rate": 0.00016280433311903768,
      "loss": 4.5254,
      "step": 2895
    },
    {
      "epoch": 3.31,
      "grad_norm": 2.303044382020593,
      "learning_rate": 0.00016260683406599336,
      "loss": 4.2817,
      "step": 2896
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.972416149284079,
      "learning_rate": 0.00016240941033442006,
      "loss": 4.1829,
      "step": 2897
    },
    {
      "epoch": 3.31,
      "grad_norm": 3.4124325812236163,
      "learning_rate": 0.00016221206203254909,
      "loss": 4.2986,
      "step": 2898
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.6106578382569272,
      "learning_rate": 0.00016201478926857046,
      "loss": 4.2009,
      "step": 2899
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.4507975904090147,
      "learning_rate": 0.00016181759215063273,
      "loss": 4.287,
      "step": 2900
    },
    {
      "epoch": 3.31,
      "eval_blimp_filtered_avg": 0.5256716417910448,
      "eval_blimp_filtered_std": 0.005630789097487669,
      "step": 2900
    },
    {
      "epoch": 3.31,
      "eval_blimp_supplement_avg": 0.5172413793103449,
      "eval_blimp_supplement_std": 0.02271711993467397,
      "step": 2900
    },
    {
      "epoch": 3.31,
      "eval_vqa_filtered_avg": 0.38,
      "eval_vqa_filtered_std": 0.048783173121456316,
      "step": 2900
    },
    {
      "epoch": 3.31,
      "eval_winoground_filtered_avg": 0.51,
      "eval_winoground_filtered_std": 0.05024183937956912,
      "step": 2900
    },
    {
      "epoch": 3.32,
      "grad_norm": 2.9598082300237327,
      "learning_rate": 0.00016162047078684296,
      "loss": 4.146,
      "step": 2901
    },
    {
      "epoch": 3.32,
      "grad_norm": 2.307619242709991,
      "learning_rate": 0.00016142342528526675,
      "loss": 4.4024,
      "step": 2902
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.1504692737669255,
      "learning_rate": 0.00016122645575392824,
      "loss": 4.15,
      "step": 2903
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.394519406337347,
      "learning_rate": 0.00016102956230080954,
      "loss": 3.9863,
      "step": 2904
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.554043638712319,
      "learning_rate": 0.00016083274503385128,
      "loss": 4.2885,
      "step": 2905
    },
    {
      "epoch": 3.32,
      "grad_norm": 3.105550375785096,
      "learning_rate": 0.00016063600406095234,
      "loss": 4.233,
      "step": 2906
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.6580135872056228,
      "learning_rate": 0.00016043933948997,
      "loss": 4.1895,
      "step": 2907
    },
    {
      "epoch": 3.32,
      "grad_norm": 2.1140479815888535,
      "learning_rate": 0.0001602427514287191,
      "loss": 4.3222,
      "step": 2908
    },
    {
      "epoch": 3.32,
      "grad_norm": 2.4932858413247754,
      "learning_rate": 0.000160046239984973,
      "loss": 4.237,
      "step": 2909
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.9660725508142398,
      "learning_rate": 0.00015984980526646288,
      "loss": 4.2804,
      "step": 2910
    },
    {
      "epoch": 3.33,
      "grad_norm": 2.5568933880403906,
      "learning_rate": 0.00015965344738087795,
      "loss": 4.0959,
      "step": 2911
    },
    {
      "epoch": 3.33,
      "grad_norm": 10.2083464851958,
      "learning_rate": 0.00015945716643586504,
      "loss": 4.1955,
      "step": 2912
    },
    {
      "epoch": 3.33,
      "grad_norm": 3.150228399424194,
      "learning_rate": 0.00015926096253902897,
      "loss": 4.3907,
      "step": 2913
    },
    {
      "epoch": 3.33,
      "grad_norm": 5.30523949270125,
      "learning_rate": 0.00015906483579793262,
      "loss": 4.4763,
      "step": 2914
    },
    {
      "epoch": 3.33,
      "grad_norm": 3.627813877196551,
      "learning_rate": 0.00015886878632009622,
      "loss": 4.4606,
      "step": 2915
    },
    {
      "epoch": 3.33,
      "grad_norm": 2.711163356296052,
      "learning_rate": 0.00015867281421299754,
      "loss": 4.3531,
      "step": 2916
    },
    {
      "epoch": 3.33,
      "grad_norm": 2.1567278220400987,
      "learning_rate": 0.00015847691958407223,
      "loss": 4.2971,
      "step": 2917
    },
    {
      "epoch": 3.33,
      "grad_norm": 3.3714256259490907,
      "learning_rate": 0.00015828110254071328,
      "loss": 4.1778,
      "step": 2918
    },
    {
      "epoch": 3.34,
      "grad_norm": 3.1382605079898656,
      "learning_rate": 0.00015808536319027134,
      "loss": 4.2477,
      "step": 2919
    },
    {
      "epoch": 3.34,
      "grad_norm": 2.2571459281214685,
      "learning_rate": 0.00015788970164005403,
      "loss": 4.1207,
      "step": 2920
    },
    {
      "epoch": 3.34,
      "grad_norm": 2.7798026822841564,
      "learning_rate": 0.00015769411799732697,
      "loss": 4.2812,
      "step": 2921
    },
    {
      "epoch": 3.34,
      "grad_norm": 4.758955793179381,
      "learning_rate": 0.0001574986123693126,
      "loss": 4.2582,
      "step": 2922
    },
    {
      "epoch": 3.34,
      "grad_norm": 2.6956204142506626,
      "learning_rate": 0.0001573031848631907,
      "loss": 4.3179,
      "step": 2923
    },
    {
      "epoch": 3.34,
      "grad_norm": 3.8310735404436747,
      "learning_rate": 0.00015710783558609833,
      "loss": 4.2581,
      "step": 2924
    },
    {
      "epoch": 3.34,
      "grad_norm": 3.3037076939664254,
      "learning_rate": 0.0001569125646451294,
      "loss": 4.188,
      "step": 2925
    },
    {
      "epoch": 3.34,
      "grad_norm": 8.273292743295352,
      "learning_rate": 0.0001567173721473351,
      "loss": 4.3601,
      "step": 2926
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.99025392767604,
      "learning_rate": 0.00015652225819972365,
      "loss": 4.1211,
      "step": 2927
    },
    {
      "epoch": 3.35,
      "grad_norm": 2.208580300716964,
      "learning_rate": 0.00015632722290926006,
      "loss": 4.1798,
      "step": 2928
    },
    {
      "epoch": 3.35,
      "grad_norm": 3.3852444702939843,
      "learning_rate": 0.00015613226638286625,
      "loss": 4.5403,
      "step": 2929
    },
    {
      "epoch": 3.35,
      "grad_norm": 3.3386385704084036,
      "learning_rate": 0.00015593738872742103,
      "loss": 4.4593,
      "step": 2930
    },
    {
      "epoch": 3.35,
      "grad_norm": 2.448766082647211,
      "learning_rate": 0.00015574259004975995,
      "loss": 4.2114,
      "step": 2931
    },
    {
      "epoch": 3.35,
      "grad_norm": 18.315394578160113,
      "learning_rate": 0.0001555478704566753,
      "loss": 4.2419,
      "step": 2932
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.86409821675218,
      "learning_rate": 0.0001553532300549158,
      "loss": 4.2527,
      "step": 2933
    },
    {
      "epoch": 3.35,
      "grad_norm": 8.458458282283749,
      "learning_rate": 0.00015515866895118692,
      "loss": 4.4318,
      "step": 2934
    },
    {
      "epoch": 3.35,
      "grad_norm": 4.521578618935152,
      "learning_rate": 0.00015496418725215077,
      "loss": 4.6523,
      "step": 2935
    },
    {
      "epoch": 3.36,
      "grad_norm": 3.584620714036975,
      "learning_rate": 0.00015476978506442573,
      "loss": 5.0449,
      "step": 2936
    },
    {
      "epoch": 3.36,
      "grad_norm": 5.997341126347501,
      "learning_rate": 0.0001545754624945867,
      "loss": 5.342,
      "step": 2937
    },
    {
      "epoch": 3.36,
      "grad_norm": 4.858225399009792,
      "learning_rate": 0.0001543812196491649,
      "loss": 5.2245,
      "step": 2938
    },
    {
      "epoch": 3.36,
      "grad_norm": 2.8337829197990927,
      "learning_rate": 0.00015418705663464784,
      "loss": 4.7901,
      "step": 2939
    },
    {
      "epoch": 3.36,
      "grad_norm": 3.274873438435907,
      "learning_rate": 0.00015399297355747925,
      "loss": 4.603,
      "step": 2940
    },
    {
      "epoch": 3.36,
      "grad_norm": 2.1292859099597155,
      "learning_rate": 0.00015379897052405925,
      "loss": 4.4665,
      "step": 2941
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.5918541976301666,
      "learning_rate": 0.00015360504764074358,
      "loss": 4.2902,
      "step": 2942
    },
    {
      "epoch": 3.36,
      "grad_norm": 2.8409557328666515,
      "learning_rate": 0.00015341120501384444,
      "loss": 4.6114,
      "step": 2943
    },
    {
      "epoch": 3.36,
      "grad_norm": 3.744619664073681,
      "learning_rate": 0.00015321744274963002,
      "loss": 4.4554,
      "step": 2944
    },
    {
      "epoch": 3.37,
      "grad_norm": 3.343654972012537,
      "learning_rate": 0.00015302376095432428,
      "loss": 4.6114,
      "step": 2945
    },
    {
      "epoch": 3.37,
      "grad_norm": 10.898752165364591,
      "learning_rate": 0.00015283015973410727,
      "loss": 4.736,
      "step": 2946
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.7686477370442448,
      "learning_rate": 0.00015263663919511466,
      "loss": 4.388,
      "step": 2947
    },
    {
      "epoch": 3.37,
      "grad_norm": 2.089048244529463,
      "learning_rate": 0.00015244319944343805,
      "loss": 4.5622,
      "step": 2948
    },
    {
      "epoch": 3.37,
      "grad_norm": 2.835059420065904,
      "learning_rate": 0.0001522498405851247,
      "loss": 4.3033,
      "step": 2949
    },
    {
      "epoch": 3.37,
      "grad_norm": 4.149785371751094,
      "learning_rate": 0.0001520565627261774,
      "loss": 4.6362,
      "step": 2950
    },
    {
      "epoch": 3.37,
      "grad_norm": 2.140474037529553,
      "learning_rate": 0.00015186336597255466,
      "loss": 4.576,
      "step": 2951
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.713768363907163,
      "learning_rate": 0.0001516702504301706,
      "loss": 4.5662,
      "step": 2952
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.493897242723164,
      "learning_rate": 0.00015147721620489463,
      "loss": 4.5068,
      "step": 2953
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.574675355602964,
      "learning_rate": 0.00015128426340255173,
      "loss": 4.41,
      "step": 2954
    },
    {
      "epoch": 3.38,
      "grad_norm": 2.1609263911570866,
      "learning_rate": 0.0001510913921289222,
      "loss": 4.301,
      "step": 2955
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.6909327197188215,
      "learning_rate": 0.00015089860248974156,
      "loss": 4.2753,
      "step": 2956
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.7276914340539895,
      "learning_rate": 0.00015070589459070078,
      "loss": 4.3441,
      "step": 2957
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.311909407953645,
      "learning_rate": 0.0001505132685374459,
      "loss": 4.4631,
      "step": 2958
    },
    {
      "epoch": 3.38,
      "grad_norm": 2.4276800759224995,
      "learning_rate": 0.0001503207244355779,
      "loss": 4.271,
      "step": 2959
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.8315908800579928,
      "learning_rate": 0.00015012826239065296,
      "loss": 4.3313,
      "step": 2960
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.4606940995719733,
      "learning_rate": 0.0001499358825081826,
      "loss": 4.2892,
      "step": 2961
    },
    {
      "epoch": 3.39,
      "grad_norm": 2.5579467010319803,
      "learning_rate": 0.00014974358489363302,
      "loss": 4.4045,
      "step": 2962
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.85642265753712,
      "learning_rate": 0.00014955136965242512,
      "loss": 4.2922,
      "step": 2963
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.9675815192792815,
      "learning_rate": 0.0001493592368899349,
      "loss": 4.3169,
      "step": 2964
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.6453911761982585,
      "learning_rate": 0.0001491671867114931,
      "loss": 4.3375,
      "step": 2965
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.3985374264470756,
      "learning_rate": 0.0001489752192223853,
      "loss": 4.3316,
      "step": 2966
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.8361200779531592,
      "learning_rate": 0.00014878333452785132,
      "loss": 4.4045,
      "step": 2967
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.6830535150135522,
      "learning_rate": 0.00014859153273308622,
      "loss": 4.3132,
      "step": 2968
    },
    {
      "epoch": 3.39,
      "grad_norm": 2.627165650528088,
      "learning_rate": 0.00014839981394323913,
      "loss": 4.1185,
      "step": 2969
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.7901149211557768,
      "learning_rate": 0.00014820817826341397,
      "loss": 4.0683,
      "step": 2970
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.2623048840601314,
      "learning_rate": 0.00014801662579866872,
      "loss": 4.3559,
      "step": 2971
    },
    {
      "epoch": 3.4,
      "grad_norm": 5.381616290351572,
      "learning_rate": 0.00014782515665401612,
      "loss": 4.471,
      "step": 2972
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.8693086404483586,
      "learning_rate": 0.00014763377093442306,
      "loss": 4.2935,
      "step": 2973
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.3339542615824342,
      "learning_rate": 0.00014744246874481078,
      "loss": 4.3158,
      "step": 2974
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.5561431322745403,
      "learning_rate": 0.0001472512501900546,
      "loss": 4.3639,
      "step": 2975
    },
    {
      "epoch": 3.4,
      "grad_norm": 4.125034249480485,
      "learning_rate": 0.0001470601153749841,
      "loss": 4.2365,
      "step": 2976
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.0603037867648728,
      "learning_rate": 0.00014686906440438294,
      "loss": 3.9552,
      "step": 2977
    },
    {
      "epoch": 3.4,
      "grad_norm": 2.7302665605442855,
      "learning_rate": 0.00014667809738298874,
      "loss": 4.3139,
      "step": 2978
    },
    {
      "epoch": 3.4,
      "grad_norm": 2.891402776067834,
      "learning_rate": 0.0001464872144154933,
      "loss": 4.3518,
      "step": 2979
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.9166486193528715,
      "learning_rate": 0.00014629641560654196,
      "loss": 4.2925,
      "step": 2980
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.0573575381822362,
      "learning_rate": 0.0001461057010607342,
      "loss": 4.2999,
      "step": 2981
    },
    {
      "epoch": 3.41,
      "grad_norm": 3.1213229024465416,
      "learning_rate": 0.00014591507088262335,
      "loss": 4.4475,
      "step": 2982
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.20700919827106,
      "learning_rate": 0.00014572452517671634,
      "loss": 4.0381,
      "step": 2983
    },
    {
      "epoch": 3.41,
      "grad_norm": 3.185279291648942,
      "learning_rate": 0.00014553406404747387,
      "loss": 4.1514,
      "step": 2984
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.106280758692437,
      "learning_rate": 0.0001453436875993102,
      "loss": 4.2481,
      "step": 2985
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.9784269703891317,
      "learning_rate": 0.0001451533959365933,
      "loss": 4.2724,
      "step": 2986
    },
    {
      "epoch": 3.41,
      "grad_norm": 2.3426580021491925,
      "learning_rate": 0.00014496318916364462,
      "loss": 4.2256,
      "step": 2987
    },
    {
      "epoch": 3.41,
      "grad_norm": 2.306280176719563,
      "learning_rate": 0.00014477306738473877,
      "loss": 4.2732,
      "step": 2988
    },
    {
      "epoch": 3.42,
      "grad_norm": 2.1196994893100745,
      "learning_rate": 0.00014458303070410424,
      "loss": 4.262,
      "step": 2989
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.606187711912442,
      "learning_rate": 0.00014439307922592255,
      "loss": 4.3625,
      "step": 2990
    },
    {
      "epoch": 3.42,
      "grad_norm": 2.1791598765739124,
      "learning_rate": 0.00014420321305432868,
      "loss": 4.3896,
      "step": 2991
    },
    {
      "epoch": 3.42,
      "grad_norm": 3.29815582498659,
      "learning_rate": 0.0001440134322934107,
      "loss": 4.1172,
      "step": 2992
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.8883013305514043,
      "learning_rate": 0.00014382373704721003,
      "loss": 4.2294,
      "step": 2993
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.3866533030398822,
      "learning_rate": 0.000143634127419721,
      "loss": 4.2159,
      "step": 2994
    },
    {
      "epoch": 3.42,
      "grad_norm": 2.1449402200607484,
      "learning_rate": 0.00014344460351489114,
      "loss": 4.4003,
      "step": 2995
    },
    {
      "epoch": 3.42,
      "grad_norm": 2.065395906421541,
      "learning_rate": 0.0001432551654366211,
      "loss": 4.1781,
      "step": 2996
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.5146049691480032,
      "learning_rate": 0.00014306581328876408,
      "loss": 4.0698,
      "step": 2997
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.975996406718577,
      "learning_rate": 0.00014287654717512656,
      "loss": 4.2594,
      "step": 2998
    },
    {
      "epoch": 3.43,
      "grad_norm": 3.5128526951227372,
      "learning_rate": 0.00014268736719946757,
      "loss": 4.3284,
      "step": 2999
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.0798124625667997,
      "learning_rate": 0.00014249827346549945,
      "loss": 4.3498,
      "step": 3000
    },
    {
      "epoch": 3.43,
      "eval_blimp_filtered_avg": 0.5197014925373135,
      "eval_blimp_filtered_std": 0.005688596935991008,
      "step": 3000
    },
    {
      "epoch": 3.43,
      "eval_blimp_supplement_avg": 0.4870689655172414,
      "eval_blimp_supplement_std": 0.02269928606166752,
      "step": 3000
    },
    {
      "epoch": 3.43,
      "eval_vqa_filtered_avg": 0.34,
      "eval_vqa_filtered_std": 0.047609522856952365,
      "step": 3000
    },
    {
      "epoch": 3.43,
      "eval_winoground_filtered_avg": 0.51,
      "eval_winoground_filtered_std": 0.05024183937956911,
      "step": 3000
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.6436427330455934,
      "learning_rate": 0.0001423092660768865,
      "loss": 4.231,
      "step": 3001
    },
    {
      "epoch": 3.43,
      "grad_norm": 2.3511960811735206,
      "learning_rate": 0.0001421203451372462,
      "loss": 4.2542,
      "step": 3002
    },
    {
      "epoch": 3.43,
      "grad_norm": 4.66370572454176,
      "learning_rate": 0.00014193151075014848,
      "loss": 4.3803,
      "step": 3003
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.4421822848651475,
      "learning_rate": 0.000141742763019116,
      "loss": 4.2102,
      "step": 3004
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.8080879933134022,
      "learning_rate": 0.00014155410204762342,
      "loss": 4.2323,
      "step": 3005
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.7669487860711126,
      "learning_rate": 0.0001413655279390984,
      "loss": 4.3904,
      "step": 3006
    },
    {
      "epoch": 3.44,
      "grad_norm": 3.0361315737654704,
      "learning_rate": 0.00014117704079692054,
      "loss": 4.4187,
      "step": 3007
    },
    {
      "epoch": 3.44,
      "grad_norm": 4.701207336255201,
      "learning_rate": 0.00014098864072442242,
      "loss": 4.242,
      "step": 3008
    },
    {
      "epoch": 3.44,
      "grad_norm": 2.2495166091713634,
      "learning_rate": 0.00014080032782488798,
      "loss": 4.275,
      "step": 3009
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.5044501990168426,
      "learning_rate": 0.00014061210220155402,
      "loss": 4.3108,
      "step": 3010
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.0523883606747981,
      "learning_rate": 0.00014042396395760927,
      "loss": 4.1203,
      "step": 3011
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.4149422150403703,
      "learning_rate": 0.00014023591319619458,
      "loss": 4.2414,
      "step": 3012
    },
    {
      "epoch": 3.44,
      "grad_norm": 3.024893102883082,
      "learning_rate": 0.000140047950020403,
      "loss": 4.2553,
      "step": 3013
    },
    {
      "epoch": 3.44,
      "grad_norm": 2.0032707195912827,
      "learning_rate": 0.00013986007453327908,
      "loss": 4.2786,
      "step": 3014
    },
    {
      "epoch": 3.45,
      "grad_norm": 2.138933631785984,
      "learning_rate": 0.00013967228683781994,
      "loss": 4.2051,
      "step": 3015
    },
    {
      "epoch": 3.45,
      "grad_norm": 2.1864625768375885,
      "learning_rate": 0.00013948458703697415,
      "loss": 4.2906,
      "step": 3016
    },
    {
      "epoch": 3.45,
      "grad_norm": 4.4129812990416255,
      "learning_rate": 0.00013929697523364234,
      "loss": 4.0548,
      "step": 3017
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.190366441428171,
      "learning_rate": 0.00013910945153067654,
      "loss": 4.1696,
      "step": 3018
    },
    {
      "epoch": 3.45,
      "grad_norm": 4.144525146551349,
      "learning_rate": 0.0001389220160308808,
      "loss": 4.4661,
      "step": 3019
    },
    {
      "epoch": 3.45,
      "grad_norm": 3.718542328929559,
      "learning_rate": 0.0001387346688370107,
      "loss": 4.4426,
      "step": 3020
    },
    {
      "epoch": 3.45,
      "grad_norm": 5.630238420384331,
      "learning_rate": 0.0001385474100517736,
      "loss": 4.362,
      "step": 3021
    },
    {
      "epoch": 3.45,
      "grad_norm": 3.0696245153162534,
      "learning_rate": 0.00013836023977782785,
      "loss": 4.3613,
      "step": 3022
    },
    {
      "epoch": 3.45,
      "grad_norm": 3.096382582601289,
      "learning_rate": 0.00013817315811778404,
      "loss": 4.2421,
      "step": 3023
    },
    {
      "epoch": 3.46,
      "grad_norm": 2.6621883514489473,
      "learning_rate": 0.00013798616517420362,
      "loss": 4.3431,
      "step": 3024
    },
    {
      "epoch": 3.46,
      "grad_norm": 3.5680718762937573,
      "learning_rate": 0.0001377992610495997,
      "loss": 4.2494,
      "step": 3025
    },
    {
      "epoch": 3.46,
      "grad_norm": 4.230844347163315,
      "learning_rate": 0.00013761244584643636,
      "loss": 4.3728,
      "step": 3026
    },
    {
      "epoch": 3.46,
      "grad_norm": 2.2602157284183937,
      "learning_rate": 0.00013742571966712927,
      "loss": 4.3519,
      "step": 3027
    },
    {
      "epoch": 3.46,
      "grad_norm": 15.745097039324783,
      "learning_rate": 0.00013723908261404513,
      "loss": 4.3262,
      "step": 3028
    },
    {
      "epoch": 3.46,
      "grad_norm": 2.3267379190513386,
      "learning_rate": 0.00013705253478950186,
      "loss": 4.2876,
      "step": 3029
    },
    {
      "epoch": 3.46,
      "grad_norm": 4.41852755468411,
      "learning_rate": 0.00013686607629576842,
      "loss": 4.4346,
      "step": 3030
    },
    {
      "epoch": 3.46,
      "grad_norm": 4.2123105714209625,
      "learning_rate": 0.00013667970723506474,
      "loss": 4.5282,
      "step": 3031
    },
    {
      "epoch": 3.47,
      "grad_norm": 3.904808167090843,
      "learning_rate": 0.00013649342770956187,
      "loss": 4.3823,
      "step": 3032
    },
    {
      "epoch": 3.47,
      "grad_norm": 5.936279287109261,
      "learning_rate": 0.0001363072378213816,
      "loss": 4.5298,
      "step": 3033
    },
    {
      "epoch": 3.47,
      "grad_norm": 2.9108129013374904,
      "learning_rate": 0.00013612113767259685,
      "loss": 4.3436,
      "step": 3034
    },
    {
      "epoch": 3.47,
      "grad_norm": 3.3831317926379496,
      "learning_rate": 0.0001359351273652309,
      "loss": 4.4461,
      "step": 3035
    },
    {
      "epoch": 3.47,
      "grad_norm": 4.18154453168728,
      "learning_rate": 0.00013574920700125815,
      "loss": 4.4323,
      "step": 3036
    },
    {
      "epoch": 3.47,
      "grad_norm": 16.5528760122221,
      "learning_rate": 0.00013556337668260357,
      "loss": 4.2725,
      "step": 3037
    },
    {
      "epoch": 3.47,
      "grad_norm": 5.79312054115292,
      "learning_rate": 0.0001353776365111428,
      "loss": 4.4751,
      "step": 3038
    },
    {
      "epoch": 3.47,
      "grad_norm": 6.047200595299549,
      "learning_rate": 0.00013519198658870198,
      "loss": 4.7292,
      "step": 3039
    },
    {
      "epoch": 3.47,
      "grad_norm": 2.9419929081228426,
      "learning_rate": 0.0001350064270170579,
      "loss": 4.5264,
      "step": 3040
    },
    {
      "epoch": 3.48,
      "grad_norm": 2.7878373552144584,
      "learning_rate": 0.0001348209578979377,
      "loss": 4.4431,
      "step": 3041
    },
    {
      "epoch": 3.48,
      "grad_norm": 8.223590376889106,
      "learning_rate": 0.00013463557933301909,
      "loss": 4.322,
      "step": 3042
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.519733603102312,
      "learning_rate": 0.0001344502914239298,
      "loss": 4.3892,
      "step": 3043
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.9832647442314424,
      "learning_rate": 0.00013426509427224816,
      "loss": 4.2745,
      "step": 3044
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.897762371893992,
      "learning_rate": 0.00013407998797950274,
      "loss": 4.2918,
      "step": 3045
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.8535216277567776,
      "learning_rate": 0.0001338949726471721,
      "loss": 4.5687,
      "step": 3046
    },
    {
      "epoch": 3.48,
      "grad_norm": 2.33197887578216,
      "learning_rate": 0.00013371004837668538,
      "loss": 4.361,
      "step": 3047
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.6676976878923677,
      "learning_rate": 0.00013352521526942108,
      "loss": 4.3779,
      "step": 3048
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.6722103435287188,
      "learning_rate": 0.0001333404734267084,
      "loss": 4.2572,
      "step": 3049
    },
    {
      "epoch": 3.49,
      "grad_norm": 4.393290722937419,
      "learning_rate": 0.00013315582294982605,
      "loss": 4.275,
      "step": 3050
    },
    {
      "epoch": 3.49,
      "grad_norm": 2.8532827231007953,
      "learning_rate": 0.00013297126394000306,
      "loss": 4.4644,
      "step": 3051
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.871736919305201,
      "learning_rate": 0.0001327867964984178,
      "loss": 4.4496,
      "step": 3052
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.4888363751711573,
      "learning_rate": 0.0001326024207261989,
      "loss": 4.2669,
      "step": 3053
    },
    {
      "epoch": 3.49,
      "grad_norm": 24.74357002132999,
      "learning_rate": 0.00013241813672442443,
      "loss": 4.3262,
      "step": 3054
    },
    {
      "epoch": 3.49,
      "grad_norm": 3.737967202176959,
      "learning_rate": 0.00013223394459412259,
      "loss": 4.2771,
      "step": 3055
    },
    {
      "epoch": 3.49,
      "grad_norm": 2.886326663166909,
      "learning_rate": 0.00013204984443627058,
      "loss": 4.2825,
      "step": 3056
    },
    {
      "epoch": 3.49,
      "grad_norm": 3.6332651417966773,
      "learning_rate": 0.00013186583635179565,
      "loss": 4.2213,
      "step": 3057
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.4589196300023728,
      "learning_rate": 0.00013168192044157436,
      "loss": 4.1446,
      "step": 3058
    },
    {
      "epoch": 3.5,
      "grad_norm": 2.465725797608277,
      "learning_rate": 0.000131498096806433,
      "loss": 4.1862,
      "step": 3059
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.2582594539753975,
      "learning_rate": 0.00013131436554714686,
      "loss": 4.2512,
      "step": 3060
    },
    {
      "epoch": 3.5,
      "grad_norm": 3.150010817797145,
      "learning_rate": 0.00013113072676444076,
      "loss": 4.445,
      "step": 3061
    },
    {
      "epoch": 3.5,
      "grad_norm": 3.756535486477678,
      "learning_rate": 0.00013094718055898908,
      "loss": 4.1179,
      "step": 3062
    },
    {
      "epoch": 3.5,
      "grad_norm": 2.101984529989566,
      "learning_rate": 0.0001307637270314152,
      "loss": 4.4073,
      "step": 3063
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.8832664707958415,
      "learning_rate": 0.00013058036628229178,
      "loss": 4.2018,
      "step": 3064
    },
    {
      "epoch": 3.5,
      "grad_norm": 2.969209726838927,
      "learning_rate": 0.0001303970984121404,
      "loss": 4.4063,
      "step": 3065
    },
    {
      "epoch": 3.5,
      "grad_norm": 3.1092568640442844,
      "learning_rate": 0.00013021392352143194,
      "loss": 4.1874,
      "step": 3066
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.557779816221156,
      "learning_rate": 0.00013003084171058632,
      "loss": 4.4365,
      "step": 3067
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.9065960211963298,
      "learning_rate": 0.00012984785307997232,
      "loss": 4.4596,
      "step": 3068
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.6412265781645923,
      "learning_rate": 0.00012966495772990773,
      "loss": 4.3305,
      "step": 3069
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.9166560653950269,
      "learning_rate": 0.0001294821557606591,
      "loss": 4.1937,
      "step": 3070
    },
    {
      "epoch": 3.51,
      "grad_norm": 3.823990664818364,
      "learning_rate": 0.0001292994472724419,
      "loss": 4.1603,
      "step": 3071
    },
    {
      "epoch": 3.51,
      "grad_norm": 2.176613747211886,
      "learning_rate": 0.00012911683236542043,
      "loss": 4.2939,
      "step": 3072
    },
    {
      "epoch": 3.51,
      "grad_norm": 5.705871117814191,
      "learning_rate": 0.00012893431113970724,
      "loss": 4.2088,
      "step": 3073
    },
    {
      "epoch": 3.51,
      "grad_norm": 3.225952730177573,
      "learning_rate": 0.000128751883695364,
      "loss": 4.2714,
      "step": 3074
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.8308644730945545,
      "learning_rate": 0.00012856955013240085,
      "loss": 4.2944,
      "step": 3075
    },
    {
      "epoch": 3.52,
      "grad_norm": 2.669974973757755,
      "learning_rate": 0.00012838731055077635,
      "loss": 4.3037,
      "step": 3076
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.2651499432267799,
      "learning_rate": 0.0001282051650503976,
      "loss": 4.1507,
      "step": 3077
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.8335588641497982,
      "learning_rate": 0.00012802311373112017,
      "loss": 4.1116,
      "step": 3078
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.236075470456007,
      "learning_rate": 0.00012784115669274795,
      "loss": 4.4823,
      "step": 3079
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.5548947412162477,
      "learning_rate": 0.00012765929403503313,
      "loss": 4.4347,
      "step": 3080
    },
    {
      "epoch": 3.52,
      "grad_norm": 2.994862463900862,
      "learning_rate": 0.00012747752585767628,
      "loss": 4.2839,
      "step": 3081
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.2930882751661772,
      "learning_rate": 0.0001272958522603259,
      "loss": 4.341,
      "step": 3082
    },
    {
      "epoch": 3.52,
      "grad_norm": 2.0226806945517026,
      "learning_rate": 0.00012711427334257888,
      "loss": 4.2458,
      "step": 3083
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.5543192466290554,
      "learning_rate": 0.00012693278920398015,
      "loss": 4.2223,
      "step": 3084
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.3588071336539917,
      "learning_rate": 0.00012675139994402265,
      "loss": 4.3066,
      "step": 3085
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.4047977383946104,
      "learning_rate": 0.00012657010566214733,
      "loss": 4.2968,
      "step": 3086
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.3922275213392041,
      "learning_rate": 0.00012638890645774308,
      "loss": 4.4457,
      "step": 3087
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.3825169856842991,
      "learning_rate": 0.00012620780243014661,
      "loss": 4.3136,
      "step": 3088
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.0878809705456396,
      "learning_rate": 0.00012602679367864264,
      "loss": 4.1826,
      "step": 3089
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.9086193017497569,
      "learning_rate": 0.0001258458803024633,
      "loss": 4.3718,
      "step": 3090
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.2267699920465889,
      "learning_rate": 0.00012566506240078873,
      "loss": 4.3077,
      "step": 3091
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.9264328262098767,
      "learning_rate": 0.00012548434007274664,
      "loss": 4.0587,
      "step": 3092
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.7499985178392582,
      "learning_rate": 0.0001253037134174124,
      "loss": 4.2844,
      "step": 3093
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.1674273928111123,
      "learning_rate": 0.00012512318253380875,
      "loss": 4.0244,
      "step": 3094
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.9022225282894355,
      "learning_rate": 0.00012494274752090616,
      "loss": 4.3325,
      "step": 3095
    },
    {
      "epoch": 3.54,
      "grad_norm": 2.146768678380222,
      "learning_rate": 0.00012476240847762244,
      "loss": 4.3459,
      "step": 3096
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.9810808689992113,
      "learning_rate": 0.00012458216550282277,
      "loss": 4.1667,
      "step": 3097
    },
    {
      "epoch": 3.54,
      "grad_norm": 6.0582922200120235,
      "learning_rate": 0.00012440201869531975,
      "loss": 4.3355,
      "step": 3098
    },
    {
      "epoch": 3.54,
      "grad_norm": 2.519643169606475,
      "learning_rate": 0.000124221968153873,
      "loss": 4.3129,
      "step": 3099
    },
    {
      "epoch": 3.54,
      "grad_norm": 2.3701151590782272,
      "learning_rate": 0.00012404201397718967,
      "loss": 4.2838,
      "step": 3100
    },
    {
      "epoch": 3.54,
      "eval_blimp_filtered_avg": 0.524179104477612,
      "eval_blimp_filtered_std": 0.0056781049766562505,
      "step": 3100
    },
    {
      "epoch": 3.54,
      "eval_blimp_supplement_avg": 0.49137931034482757,
      "eval_blimp_supplement_std": 0.022838162296649936,
      "step": 3100
    },
    {
      "epoch": 3.54,
      "eval_vqa_filtered_avg": 0.35,
      "eval_vqa_filtered_std": 0.0479372485441102,
      "step": 3100
    },
    {
      "epoch": 3.54,
      "eval_winoground_filtered_avg": 0.48,
      "eval_winoground_filtered_std": 0.05021167315686779,
      "step": 3100
    },
    {
      "epoch": 3.54,
      "grad_norm": 2.9375096741935365,
      "learning_rate": 0.00012386215626392384,
      "loss": 4.5588,
      "step": 3101
    },
    {
      "epoch": 3.55,
      "grad_norm": 4.378407301602557,
      "learning_rate": 0.00012368239511267715,
      "loss": 4.3454,
      "step": 3102
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.089014409595611,
      "learning_rate": 0.00012350273062199766,
      "loss": 3.9342,
      "step": 3103
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.1699366863977172,
      "learning_rate": 0.00012332316289038087,
      "loss": 4.1125,
      "step": 3104
    },
    {
      "epoch": 3.55,
      "grad_norm": 3.204872955934147,
      "learning_rate": 0.00012314369201626911,
      "loss": 4.2535,
      "step": 3105
    },
    {
      "epoch": 3.55,
      "grad_norm": 4.7205064960448455,
      "learning_rate": 0.00012296431809805182,
      "loss": 4.2734,
      "step": 3106
    },
    {
      "epoch": 3.55,
      "grad_norm": 3.202148509512982,
      "learning_rate": 0.00012278504123406478,
      "loss": 4.4724,
      "step": 3107
    },
    {
      "epoch": 3.55,
      "grad_norm": 2.7686944574745014,
      "learning_rate": 0.00012260586152259092,
      "loss": 4.5066,
      "step": 3108
    },
    {
      "epoch": 3.55,
      "grad_norm": 8.71885220518892,
      "learning_rate": 0.00012242677906186005,
      "loss": 4.243,
      "step": 3109
    },
    {
      "epoch": 3.55,
      "grad_norm": 2.7544285409264044,
      "learning_rate": 0.00012224779395004852,
      "loss": 4.2614,
      "step": 3110
    },
    {
      "epoch": 3.56,
      "grad_norm": 4.294749690811909,
      "learning_rate": 0.00012206890628527904,
      "loss": 4.436,
      "step": 3111
    },
    {
      "epoch": 3.56,
      "grad_norm": 3.7598462872058005,
      "learning_rate": 0.00012189011616562127,
      "loss": 4.5016,
      "step": 3112
    },
    {
      "epoch": 3.56,
      "grad_norm": 4.225818506964256,
      "learning_rate": 0.00012171142368909123,
      "loss": 4.2988,
      "step": 3113
    },
    {
      "epoch": 3.56,
      "grad_norm": 2.806252745533295,
      "learning_rate": 0.00012153282895365153,
      "loss": 4.405,
      "step": 3114
    },
    {
      "epoch": 3.56,
      "grad_norm": 3.067706310068835,
      "learning_rate": 0.00012135433205721077,
      "loss": 4.1811,
      "step": 3115
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.1953221608599018,
      "learning_rate": 0.00012117593309762463,
      "loss": 4.0753,
      "step": 3116
    },
    {
      "epoch": 3.56,
      "grad_norm": 2.186683538193241,
      "learning_rate": 0.00012099763217269456,
      "loss": 4.3185,
      "step": 3117
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.9057106230099483,
      "learning_rate": 0.00012081942938016842,
      "loss": 4.3192,
      "step": 3118
    },
    {
      "epoch": 3.56,
      "grad_norm": 3.359578691851843,
      "learning_rate": 0.0001206413248177404,
      "loss": 4.2813,
      "step": 3119
    },
    {
      "epoch": 3.57,
      "grad_norm": 3.6819529150124284,
      "learning_rate": 0.00012046331858305047,
      "loss": 4.5102,
      "step": 3120
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.667478221373015,
      "learning_rate": 0.00012028541077368506,
      "loss": 4.3778,
      "step": 3121
    },
    {
      "epoch": 3.57,
      "grad_norm": 2.014952398636443,
      "learning_rate": 0.00012010760148717653,
      "loss": 4.372,
      "step": 3122
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.5770755840992674,
      "learning_rate": 0.0001199298908210032,
      "loss": 4.2297,
      "step": 3123
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.266365585949548,
      "learning_rate": 0.00011975227887258933,
      "loss": 4.1576,
      "step": 3124
    },
    {
      "epoch": 3.57,
      "grad_norm": 2.9924150220920467,
      "learning_rate": 0.00011957476573930507,
      "loss": 4.1433,
      "step": 3125
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.8245270224822032,
      "learning_rate": 0.00011939735151846642,
      "loss": 4.3717,
      "step": 3126
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.694543978596632,
      "learning_rate": 0.00011922003630733522,
      "loss": 4.3395,
      "step": 3127
    },
    {
      "epoch": 3.57,
      "grad_norm": 2.447020811583496,
      "learning_rate": 0.00011904282020311871,
      "loss": 4.1883,
      "step": 3128
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.0138835020090087,
      "learning_rate": 0.00011886570330297018,
      "loss": 4.4642,
      "step": 3129
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.5319929330773685,
      "learning_rate": 0.00011868868570398838,
      "loss": 4.1568,
      "step": 3130
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.0368096271825178,
      "learning_rate": 0.00011851176750321761,
      "loss": 4.3352,
      "step": 3131
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.8418310586664867,
      "learning_rate": 0.00011833494879764769,
      "loss": 4.3116,
      "step": 3132
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.4333606281546263,
      "learning_rate": 0.00011815822968421394,
      "loss": 4.1728,
      "step": 3133
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.3765481568000348,
      "learning_rate": 0.00011798161025979698,
      "loss": 4.166,
      "step": 3134
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.409694056079145,
      "learning_rate": 0.0001178050906212229,
      "loss": 4.1542,
      "step": 3135
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.2859801934562816,
      "learning_rate": 0.00011762867086526309,
      "loss": 4.2194,
      "step": 3136
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.0927998615615322,
      "learning_rate": 0.00011745235108863393,
      "loss": 4.1262,
      "step": 3137
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.941342255091278,
      "learning_rate": 0.00011727613138799725,
      "loss": 4.3441,
      "step": 3138
    },
    {
      "epoch": 3.59,
      "grad_norm": 5.159742348676343,
      "learning_rate": 0.00011710001185996,
      "loss": 4.2176,
      "step": 3139
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.7638751716054368,
      "learning_rate": 0.0001169239926010741,
      "loss": 4.141,
      "step": 3140
    },
    {
      "epoch": 3.59,
      "grad_norm": 2.5520958370345697,
      "learning_rate": 0.00011674807370783662,
      "loss": 4.2331,
      "step": 3141
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.1683412045532688,
      "learning_rate": 0.00011657225527668947,
      "loss": 4.1658,
      "step": 3142
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.8746378592857438,
      "learning_rate": 0.00011639653740401966,
      "loss": 4.0201,
      "step": 3143
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.5440608451044426,
      "learning_rate": 0.00011622092018615895,
      "loss": 4.236,
      "step": 3144
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.8732967379437504,
      "learning_rate": 0.00011604540371938386,
      "loss": 3.9769,
      "step": 3145
    },
    {
      "epoch": 3.6,
      "grad_norm": 2.442074961300071,
      "learning_rate": 0.0001158699880999158,
      "loss": 4.1153,
      "step": 3146
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.9822758352214246,
      "learning_rate": 0.00011569467342392083,
      "loss": 4.2418,
      "step": 3147
    },
    {
      "epoch": 3.6,
      "grad_norm": 2.7592449958393845,
      "learning_rate": 0.00011551945978750975,
      "loss": 4.2548,
      "step": 3148
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.1706133308532032,
      "learning_rate": 0.00011534434728673787,
      "loss": 4.2264,
      "step": 3149
    },
    {
      "epoch": 3.6,
      "grad_norm": 2.8029338713224283,
      "learning_rate": 0.00011516933601760513,
      "loss": 4.1473,
      "step": 3150
    },
    {
      "epoch": 3.6,
      "grad_norm": 2.222635934742633,
      "learning_rate": 0.00011499442607605588,
      "loss": 4.3166,
      "step": 3151
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.885673349352131,
      "learning_rate": 0.00011481961755797905,
      "loss": 4.2234,
      "step": 3152
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.3806666382794839,
      "learning_rate": 0.00011464491055920799,
      "loss": 4.222,
      "step": 3153
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.9658888238027514,
      "learning_rate": 0.00011447030517552006,
      "loss": 4.0435,
      "step": 3154
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.3749139340789496,
      "learning_rate": 0.00011429580150263716,
      "loss": 4.2326,
      "step": 3155
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.695970526585966,
      "learning_rate": 0.00011412139963622567,
      "loss": 4.272,
      "step": 3156
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.1535700634350357,
      "learning_rate": 0.00011394709967189588,
      "loss": 4.1741,
      "step": 3157
    },
    {
      "epoch": 3.61,
      "grad_norm": 7.115065204279877,
      "learning_rate": 0.00011377290170520206,
      "loss": 4.0572,
      "step": 3158
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.4388851785556964,
      "learning_rate": 0.00011359880583164287,
      "loss": 4.2236,
      "step": 3159
    },
    {
      "epoch": 3.61,
      "grad_norm": 3.0934226093900263,
      "learning_rate": 0.00011342481214666084,
      "loss": 4.1748,
      "step": 3160
    },
    {
      "epoch": 3.61,
      "grad_norm": 2.7281681461147604,
      "learning_rate": 0.00011325092074564264,
      "loss": 4.2886,
      "step": 3161
    },
    {
      "epoch": 3.61,
      "grad_norm": 2.542614991541466,
      "learning_rate": 0.00011307713172391848,
      "loss": 4.1472,
      "step": 3162
    },
    {
      "epoch": 3.61,
      "grad_norm": 2.2095345612314774,
      "learning_rate": 0.00011290344517676295,
      "loss": 4.3034,
      "step": 3163
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.2965357461673255,
      "learning_rate": 0.00011272986119939418,
      "loss": 4.0354,
      "step": 3164
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.963141755544075,
      "learning_rate": 0.0001125563798869742,
      "loss": 4.1113,
      "step": 3165
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.789941778426034,
      "learning_rate": 0.00011238300133460845,
      "loss": 4.2851,
      "step": 3166
    },
    {
      "epoch": 3.62,
      "grad_norm": 3.002049566007964,
      "learning_rate": 0.00011220972563734632,
      "loss": 4.2174,
      "step": 3167
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.7274097260894765,
      "learning_rate": 0.00011203655289018086,
      "loss": 4.1929,
      "step": 3168
    },
    {
      "epoch": 3.62,
      "grad_norm": 2.106778316585251,
      "learning_rate": 0.00011186348318804845,
      "loss": 4.2726,
      "step": 3169
    },
    {
      "epoch": 3.62,
      "grad_norm": 2.8378010754076683,
      "learning_rate": 0.00011169051662582916,
      "loss": 4.1998,
      "step": 3170
    },
    {
      "epoch": 3.62,
      "grad_norm": 2.1867282485230852,
      "learning_rate": 0.00011151765329834646,
      "loss": 4.1752,
      "step": 3171
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.4948749714851985,
      "learning_rate": 0.00011134489330036719,
      "loss": 4.3977,
      "step": 3172
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.9128013797905492,
      "learning_rate": 0.00011117223672660158,
      "loss": 4.1496,
      "step": 3173
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.656615464499988,
      "learning_rate": 0.00011099968367170323,
      "loss": 4.2879,
      "step": 3174
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.416655072767785,
      "learning_rate": 0.00011082723423026874,
      "loss": 4.0094,
      "step": 3175
    },
    {
      "epoch": 3.63,
      "grad_norm": 2.5893064361488327,
      "learning_rate": 0.00011065488849683818,
      "loss": 4.2027,
      "step": 3176
    },
    {
      "epoch": 3.63,
      "grad_norm": 2.2537767036179863,
      "learning_rate": 0.00011048264656589465,
      "loss": 4.1892,
      "step": 3177
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.0060924162646234,
      "learning_rate": 0.00011031050853186438,
      "loss": 4.0344,
      "step": 3178
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.3366283840610342,
      "learning_rate": 0.0001101384744891166,
      "loss": 4.3908,
      "step": 3179
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.8687553511873695,
      "learning_rate": 0.00010996654453196356,
      "loss": 4.1052,
      "step": 3180
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.1887440945028214,
      "learning_rate": 0.0001097947187546604,
      "loss": 4.1931,
      "step": 3181
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.3985423700493347,
      "learning_rate": 0.00010962299725140534,
      "loss": 4.2072,
      "step": 3182
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.8968060361497714,
      "learning_rate": 0.00010945138011633903,
      "loss": 4.3608,
      "step": 3183
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.857585430734205,
      "learning_rate": 0.00010927986744354533,
      "loss": 4.1073,
      "step": 3184
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.0413404818229504,
      "learning_rate": 0.00010910845932705054,
      "loss": 4.1282,
      "step": 3185
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.7905769906002212,
      "learning_rate": 0.00010893715586082387,
      "loss": 4.1826,
      "step": 3186
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.6463764620123684,
      "learning_rate": 0.00010876595713877697,
      "loss": 4.061,
      "step": 3187
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.8462381259341787,
      "learning_rate": 0.00010859486325476415,
      "loss": 4.1246,
      "step": 3188
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.5099689144825883,
      "learning_rate": 0.00010842387430258225,
      "loss": 4.011,
      "step": 3189
    },
    {
      "epoch": 3.65,
      "grad_norm": 2.1441839036541523,
      "learning_rate": 0.00010825299037597061,
      "loss": 4.25,
      "step": 3190
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.9719093985140902,
      "learning_rate": 0.00010808221156861099,
      "loss": 4.2564,
      "step": 3191
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.68468448544817,
      "learning_rate": 0.00010791153797412731,
      "loss": 4.1785,
      "step": 3192
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.6631575954279587,
      "learning_rate": 0.00010774096968608607,
      "loss": 4.3495,
      "step": 3193
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.8665137587518054,
      "learning_rate": 0.0001075705067979959,
      "loss": 4.1427,
      "step": 3194
    },
    {
      "epoch": 3.65,
      "grad_norm": 2.7799527670244935,
      "learning_rate": 0.00010740014940330798,
      "loss": 4.1706,
      "step": 3195
    },
    {
      "epoch": 3.65,
      "grad_norm": 2.6702246367087508,
      "learning_rate": 0.00010722989759541505,
      "loss": 4.2128,
      "step": 3196
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.758468281368008,
      "learning_rate": 0.00010705975146765246,
      "loss": 3.9807,
      "step": 3197
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.255780622128858,
      "learning_rate": 0.00010688971111329741,
      "loss": 4.1501,
      "step": 3198
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.1504997281648268,
      "learning_rate": 0.00010671977662556928,
      "loss": 4.0531,
      "step": 3199
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.089672354865873,
      "learning_rate": 0.00010654994809762908,
      "loss": 4.1766,
      "step": 3200
    },
    {
      "epoch": 3.66,
      "eval_blimp_filtered_avg": 0.5319402985074627,
      "eval_blimp_filtered_std": 0.005633106409838074,
      "step": 3200
    },
    {
      "epoch": 3.66,
      "eval_blimp_supplement_avg": 0.4525862068965517,
      "eval_blimp_supplement_std": 0.022746037902290208,
      "step": 3200
    },
    {
      "epoch": 3.66,
      "eval_vqa_filtered_avg": 0.35,
      "eval_vqa_filtered_std": 0.0479372485441102,
      "step": 3200
    },
    {
      "epoch": 3.66,
      "eval_winoground_filtered_avg": 0.49,
      "eval_winoground_filtered_std": 0.05024183937956912,
      "step": 3200
    },
    {
      "epoch": 3.66,
      "grad_norm": 17.624596657582504,
      "learning_rate": 0.00010638022562258005,
      "loss": 3.9853,
      "step": 3201
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.6263463493549037,
      "learning_rate": 0.0001062106092934671,
      "loss": 4.097,
      "step": 3202
    },
    {
      "epoch": 3.66,
      "grad_norm": 3.209920700115739,
      "learning_rate": 0.0001060410992032773,
      "loss": 4.2575,
      "step": 3203
    },
    {
      "epoch": 3.66,
      "grad_norm": 2.783076178958525,
      "learning_rate": 0.00010587169544493896,
      "loss": 4.2889,
      "step": 3204
    },
    {
      "epoch": 3.66,
      "grad_norm": 2.996439327742222,
      "learning_rate": 0.00010570239811132243,
      "loss": 3.9889,
      "step": 3205
    },
    {
      "epoch": 3.66,
      "grad_norm": 2.0579089603522176,
      "learning_rate": 0.00010553320729523962,
      "loss": 4.2335,
      "step": 3206
    },
    {
      "epoch": 3.67,
      "grad_norm": 4.133424724314067,
      "learning_rate": 0.00010536412308944408,
      "loss": 4.2679,
      "step": 3207
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.1283417793065755,
      "learning_rate": 0.00010519514558663098,
      "loss": 4.1976,
      "step": 3208
    },
    {
      "epoch": 3.67,
      "grad_norm": 2.988478409147808,
      "learning_rate": 0.0001050262748794366,
      "loss": 3.9076,
      "step": 3209
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.7153340804430834,
      "learning_rate": 0.00010485751106043932,
      "loss": 4.2141,
      "step": 3210
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.3737954092140663,
      "learning_rate": 0.00010468885422215845,
      "loss": 4.2177,
      "step": 3211
    },
    {
      "epoch": 3.67,
      "grad_norm": 14.555643417005202,
      "learning_rate": 0.00010452030445705487,
      "loss": 4.0235,
      "step": 3212
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.4220353288260394,
      "learning_rate": 0.00010435186185753051,
      "loss": 4.0247,
      "step": 3213
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.3890010121995073,
      "learning_rate": 0.00010418352651592879,
      "loss": 4.2261,
      "step": 3214
    },
    {
      "epoch": 3.67,
      "grad_norm": 2.4361133666767323,
      "learning_rate": 0.00010401529852453425,
      "loss": 4.1769,
      "step": 3215
    },
    {
      "epoch": 3.68,
      "grad_norm": 2.9002874396660743,
      "learning_rate": 0.0001038471779755727,
      "loss": 4.1125,
      "step": 3216
    },
    {
      "epoch": 3.68,
      "grad_norm": 2.557519639292111,
      "learning_rate": 0.00010367916496121065,
      "loss": 4.0642,
      "step": 3217
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.3727300775540912,
      "learning_rate": 0.00010351125957355618,
      "loss": 4.1876,
      "step": 3218
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.1193755135877852,
      "learning_rate": 0.00010334346190465808,
      "loss": 4.0663,
      "step": 3219
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.1391317213768857,
      "learning_rate": 0.00010317577204650618,
      "loss": 4.13,
      "step": 3220
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.3626984297008808,
      "learning_rate": 0.00010300819009103097,
      "loss": 4.1271,
      "step": 3221
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.2064215194048413,
      "learning_rate": 0.00010284071613010411,
      "loss": 4.1464,
      "step": 3222
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.6506468988556247,
      "learning_rate": 0.0001026733502555379,
      "loss": 4.2576,
      "step": 3223
    },
    {
      "epoch": 3.68,
      "grad_norm": 2.4408219361705012,
      "learning_rate": 0.0001025060925590854,
      "loss": 4.3597,
      "step": 3224
    },
    {
      "epoch": 3.69,
      "grad_norm": 2.975816973957695,
      "learning_rate": 0.00010233894313244034,
      "loss": 4.2039,
      "step": 3225
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.6923327159694936,
      "learning_rate": 0.00010217190206723715,
      "loss": 4.1259,
      "step": 3226
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.75617780631274,
      "learning_rate": 0.00010200496945505084,
      "loss": 4.2911,
      "step": 3227
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.8381129235319367,
      "learning_rate": 0.00010183814538739694,
      "loss": 4.0165,
      "step": 3228
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.4280732725291616,
      "learning_rate": 0.00010167142995573154,
      "loss": 4.1849,
      "step": 3229
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.4982453256081794,
      "learning_rate": 0.000101504823251451,
      "loss": 4.1591,
      "step": 3230
    },
    {
      "epoch": 3.69,
      "grad_norm": 3.7469732626833023,
      "learning_rate": 0.00010133832536589224,
      "loss": 4.2803,
      "step": 3231
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.1529116702576152,
      "learning_rate": 0.00010117193639033249,
      "loss": 4.0944,
      "step": 3232
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.1672656585798726,
      "learning_rate": 0.00010100565641598926,
      "loss": 4.0324,
      "step": 3233
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.9166189043871061,
      "learning_rate": 0.0001008394855340203,
      "loss": 4.3547,
      "step": 3234
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.9693910228231015,
      "learning_rate": 0.00010067342383552357,
      "loss": 4.0933,
      "step": 3235
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.3116432040666297,
      "learning_rate": 0.0001005074714115371,
      "loss": 4.1041,
      "step": 3236
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.7211312162696133,
      "learning_rate": 0.00010034162835303927,
      "loss": 4.3542,
      "step": 3237
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.462357344516288,
      "learning_rate": 0.00010017589475094801,
      "loss": 4.1624,
      "step": 3238
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.0650186947599927,
      "learning_rate": 0.00010001027069612171,
      "loss": 4.3794,
      "step": 3239
    },
    {
      "epoch": 3.7,
      "grad_norm": 3.207204542131583,
      "learning_rate": 9.984475627935852e-05,
      "loss": 4.1557,
      "step": 3240
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.937017961251534,
      "learning_rate": 9.96793515913964e-05,
      "loss": 3.9887,
      "step": 3241
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.0408674193195766,
      "learning_rate": 9.951405672291355e-05,
      "loss": 4.1979,
      "step": 3242
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.223234740869277,
      "learning_rate": 9.934887176452736e-05,
      "loss": 4.2106,
      "step": 3243
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.3700182092203745,
      "learning_rate": 9.918379680679539e-05,
      "loss": 4.2406,
      "step": 3244
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.5267077243572735,
      "learning_rate": 9.901883194021478e-05,
      "loss": 4.288,
      "step": 3245
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.9884526486185409,
      "learning_rate": 9.88539772552224e-05,
      "loss": 4.021,
      "step": 3246
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.0788574716826151,
      "learning_rate": 9.868923284219444e-05,
      "loss": 4.2422,
      "step": 3247
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.4943213159696211,
      "learning_rate": 9.852459879144692e-05,
      "loss": 4.1849,
      "step": 3248
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.9806223802360323,
      "learning_rate": 9.836007519323522e-05,
      "loss": 4.4687,
      "step": 3249
    },
    {
      "epoch": 3.71,
      "grad_norm": 3.269250434137489,
      "learning_rate": 9.819566213775443e-05,
      "loss": 4.0501,
      "step": 3250
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.6761704985054842,
      "learning_rate": 9.80313597151386e-05,
      "loss": 4.1408,
      "step": 3251
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.5472677601860343,
      "learning_rate": 9.786716801546142e-05,
      "loss": 4.0619,
      "step": 3252
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.1358888897344144,
      "learning_rate": 9.770308712873581e-05,
      "loss": 4.3279,
      "step": 3253
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.480067135718931,
      "learning_rate": 9.753911714491404e-05,
      "loss": 4.3261,
      "step": 3254
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.0811280153870966,
      "learning_rate": 9.737525815388729e-05,
      "loss": 4.0897,
      "step": 3255
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.0434933165104079,
      "learning_rate": 9.721151024548611e-05,
      "loss": 4.0344,
      "step": 3256
    },
    {
      "epoch": 3.72,
      "grad_norm": 2.185854350221175,
      "learning_rate": 9.704787350948029e-05,
      "loss": 4.2038,
      "step": 3257
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.9744530113388489,
      "learning_rate": 9.68843480355784e-05,
      "loss": 4.1134,
      "step": 3258
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.8843149093418575,
      "learning_rate": 9.672093391342825e-05,
      "loss": 4.2737,
      "step": 3259
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.8910872823814119,
      "learning_rate": 9.655763123261631e-05,
      "loss": 3.9506,
      "step": 3260
    },
    {
      "epoch": 3.73,
      "grad_norm": 5.2844768925124335,
      "learning_rate": 9.639444008266817e-05,
      "loss": 4.1122,
      "step": 3261
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.2949832238663532,
      "learning_rate": 9.623136055304828e-05,
      "loss": 4.1638,
      "step": 3262
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.7135253741172285,
      "learning_rate": 9.606839273315978e-05,
      "loss": 4.4198,
      "step": 3263
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.2490755765761812,
      "learning_rate": 9.590553671234472e-05,
      "loss": 4.2548,
      "step": 3264
    },
    {
      "epoch": 3.73,
      "grad_norm": 2.0797117499592335,
      "learning_rate": 9.574279257988374e-05,
      "loss": 4.0479,
      "step": 3265
    },
    {
      "epoch": 3.73,
      "grad_norm": 2.2804598360879362,
      "learning_rate": 9.558016042499618e-05,
      "loss": 3.988,
      "step": 3266
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.2480021694670407,
      "learning_rate": 9.541764033684007e-05,
      "loss": 4.161,
      "step": 3267
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.2354616235579359,
      "learning_rate": 9.525523240451175e-05,
      "loss": 4.031,
      "step": 3268
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.4343356140305725,
      "learning_rate": 9.509293671704638e-05,
      "loss": 4.2028,
      "step": 3269
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.956698078951955,
      "learning_rate": 9.493075336341737e-05,
      "loss": 4.424,
      "step": 3270
    },
    {
      "epoch": 3.74,
      "grad_norm": 2.0366110761678033,
      "learning_rate": 9.476868243253671e-05,
      "loss": 4.3983,
      "step": 3271
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.785195998698921,
      "learning_rate": 9.460672401325466e-05,
      "loss": 4.2526,
      "step": 3272
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.3155356385914907,
      "learning_rate": 9.444487819435977e-05,
      "loss": 4.1495,
      "step": 3273
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.056860310659244,
      "learning_rate": 9.428314506457894e-05,
      "loss": 4.3833,
      "step": 3274
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.1653591407948571,
      "learning_rate": 9.412152471257728e-05,
      "loss": 4.2599,
      "step": 3275
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.8286454115353377,
      "learning_rate": 9.396001722695809e-05,
      "loss": 4.2253,
      "step": 3276
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.5861643368028653,
      "learning_rate": 9.379862269626262e-05,
      "loss": 4.1681,
      "step": 3277
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.091196265825571,
      "learning_rate": 9.363734120897043e-05,
      "loss": 4.3002,
      "step": 3278
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.3100802908434868,
      "learning_rate": 9.347617285349894e-05,
      "loss": 4.2181,
      "step": 3279
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.2825737921492437,
      "learning_rate": 9.331511771820368e-05,
      "loss": 4.1342,
      "step": 3280
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.616139149561849,
      "learning_rate": 9.315417589137805e-05,
      "loss": 4.2311,
      "step": 3281
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.2390516585918243,
      "learning_rate": 9.299334746125333e-05,
      "loss": 4.0458,
      "step": 3282
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.6710136589694056,
      "learning_rate": 9.283263251599861e-05,
      "loss": 4.1816,
      "step": 3283
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.5045541227550596,
      "learning_rate": 9.26720311437209e-05,
      "loss": 3.975,
      "step": 3284
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.256729370826007,
      "learning_rate": 9.251154343246463e-05,
      "loss": 4.1423,
      "step": 3285
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.1622868091854917,
      "learning_rate": 9.235116947021225e-05,
      "loss": 4.336,
      "step": 3286
    },
    {
      "epoch": 3.76,
      "grad_norm": 2.160335785602276,
      "learning_rate": 9.219090934488372e-05,
      "loss": 4.1452,
      "step": 3287
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.4205535958718132,
      "learning_rate": 9.203076314433661e-05,
      "loss": 4.208,
      "step": 3288
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.5322958624443106,
      "learning_rate": 9.1870730956366e-05,
      "loss": 4.2827,
      "step": 3289
    },
    {
      "epoch": 3.76,
      "grad_norm": 2.1817129910339945,
      "learning_rate": 9.171081286870453e-05,
      "loss": 4.1515,
      "step": 3290
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.7385081501840363,
      "learning_rate": 9.15510089690222e-05,
      "loss": 4.1317,
      "step": 3291
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.6654703054418087,
      "learning_rate": 9.13913193449266e-05,
      "loss": 4.0178,
      "step": 3292
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.2069041400896545,
      "learning_rate": 9.123174408396229e-05,
      "loss": 4.2794,
      "step": 3293
    },
    {
      "epoch": 3.76,
      "grad_norm": 2.015210008039759,
      "learning_rate": 9.10722832736115e-05,
      "loss": 4.0029,
      "step": 3294
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.9164855033969092,
      "learning_rate": 9.091293700129361e-05,
      "loss": 4.0899,
      "step": 3295
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.1380643386489189,
      "learning_rate": 9.075370535436506e-05,
      "loss": 4.1216,
      "step": 3296
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.9514335143760437,
      "learning_rate": 9.05945884201199e-05,
      "loss": 4.2012,
      "step": 3297
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.1594640738818303,
      "learning_rate": 9.043558628578865e-05,
      "loss": 4.0528,
      "step": 3298
    },
    {
      "epoch": 3.77,
      "grad_norm": 2.2595121842179187,
      "learning_rate": 9.027669903853931e-05,
      "loss": 4.2702,
      "step": 3299
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.2534810818327904,
      "learning_rate": 9.011792676547684e-05,
      "loss": 4.0686,
      "step": 3300
    },
    {
      "epoch": 3.77,
      "eval_blimp_filtered_avg": 0.5229850746268657,
      "eval_blimp_filtered_std": 0.005626151609407658,
      "step": 3300
    },
    {
      "epoch": 3.77,
      "eval_blimp_supplement_avg": 0.4504310344827586,
      "eval_blimp_supplement_std": 0.022740295965290036,
      "step": 3300
    },
    {
      "epoch": 3.77,
      "eval_vqa_filtered_avg": 0.34,
      "eval_vqa_filtered_std": 0.047609522856952365,
      "step": 3300
    },
    {
      "epoch": 3.77,
      "eval_winoground_filtered_avg": 0.5,
      "eval_winoground_filtered_std": 0.050251890762960605,
      "step": 3300
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.0518850399469923,
      "learning_rate": 8.995926955364321e-05,
      "loss": 4.2433,
      "step": 3301
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.0509857439020014,
      "learning_rate": 8.980072749001705e-05,
      "loss": 4.117,
      "step": 3302
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.8478959003575024,
      "learning_rate": 8.964230066151402e-05,
      "loss": 4.1293,
      "step": 3303
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.109300122773905,
      "learning_rate": 8.948398915498687e-05,
      "loss": 3.9388,
      "step": 3304
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.1520503153024664,
      "learning_rate": 8.932579305722483e-05,
      "loss": 4.0455,
      "step": 3305
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.895551270095727,
      "learning_rate": 8.916771245495378e-05,
      "loss": 4.0789,
      "step": 3306
    },
    {
      "epoch": 3.78,
      "grad_norm": 2.1332757130979663,
      "learning_rate": 8.900974743483653e-05,
      "loss": 4.1813,
      "step": 3307
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.1609411492499668,
      "learning_rate": 8.885189808347236e-05,
      "loss": 4.1828,
      "step": 3308
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.4489678159425914,
      "learning_rate": 8.869416448739738e-05,
      "loss": 4.12,
      "step": 3309
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.4367374373064303,
      "learning_rate": 8.853654673308369e-05,
      "loss": 4.1435,
      "step": 3310
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.2100124306485411,
      "learning_rate": 8.837904490694058e-05,
      "loss": 3.9043,
      "step": 3311
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.2966200936585779,
      "learning_rate": 8.822165909531337e-05,
      "loss": 4.0811,
      "step": 3312
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.1941675155615084,
      "learning_rate": 8.806438938448387e-05,
      "loss": 4.1644,
      "step": 3313
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.1219675197471404,
      "learning_rate": 8.790723586067035e-05,
      "loss": 4.0698,
      "step": 3314
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.5081193398970874,
      "learning_rate": 8.775019861002702e-05,
      "loss": 4.0178,
      "step": 3315
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.5457141789824853,
      "learning_rate": 8.759327771864474e-05,
      "loss": 4.1889,
      "step": 3316
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.380264371571982,
      "learning_rate": 8.743647327255044e-05,
      "loss": 3.9785,
      "step": 3317
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.5035487046486857,
      "learning_rate": 8.727978535770723e-05,
      "loss": 4.245,
      "step": 3318
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.5052412096108791,
      "learning_rate": 8.712321406001429e-05,
      "loss": 4.2113,
      "step": 3319
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.2145066584937283,
      "learning_rate": 8.69667594653069e-05,
      "loss": 3.9557,
      "step": 3320
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.3810731096633102,
      "learning_rate": 8.681042165935639e-05,
      "loss": 4.0995,
      "step": 3321
    },
    {
      "epoch": 3.8,
      "grad_norm": 2.52899171534432,
      "learning_rate": 8.665420072787006e-05,
      "loss": 4.1049,
      "step": 3322
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.2048700129434262,
      "learning_rate": 8.6498096756491e-05,
      "loss": 4.0844,
      "step": 3323
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.9201501953325104,
      "learning_rate": 8.634210983079831e-05,
      "loss": 4.0,
      "step": 3324
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.4299876860595098,
      "learning_rate": 8.618624003630697e-05,
      "loss": 4.1994,
      "step": 3325
    },
    {
      "epoch": 3.8,
      "grad_norm": 2.186811082666166,
      "learning_rate": 8.603048745846763e-05,
      "loss": 4.1379,
      "step": 3326
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.3239673438426975,
      "learning_rate": 8.587485218266678e-05,
      "loss": 4.3094,
      "step": 3327
    },
    {
      "epoch": 3.8,
      "grad_norm": 2.2924551137383666,
      "learning_rate": 8.571933429422651e-05,
      "loss": 4.4288,
      "step": 3328
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.9953091235208018,
      "learning_rate": 8.556393387840459e-05,
      "loss": 4.0379,
      "step": 3329
    },
    {
      "epoch": 3.81,
      "grad_norm": 2.789892874428047,
      "learning_rate": 8.540865102039446e-05,
      "loss": 4.1444,
      "step": 3330
    },
    {
      "epoch": 3.81,
      "grad_norm": 2.0718610060641565,
      "learning_rate": 8.525348580532505e-05,
      "loss": 4.1718,
      "step": 3331
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.3370701933465095,
      "learning_rate": 8.509843831826072e-05,
      "loss": 4.2672,
      "step": 3332
    },
    {
      "epoch": 3.81,
      "grad_norm": 2.4978867507769316,
      "learning_rate": 8.494350864420137e-05,
      "loss": 4.0762,
      "step": 3333
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.1861386484406944,
      "learning_rate": 8.478869686808234e-05,
      "loss": 4.2148,
      "step": 3334
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.7387935642305796,
      "learning_rate": 8.463400307477433e-05,
      "loss": 4.0148,
      "step": 3335
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.7661508322279906,
      "learning_rate": 8.447942734908326e-05,
      "loss": 4.0503,
      "step": 3336
    },
    {
      "epoch": 3.81,
      "grad_norm": 2.8985263223264894,
      "learning_rate": 8.43249697757505e-05,
      "loss": 4.4129,
      "step": 3337
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.745210864212912,
      "learning_rate": 8.417063043945246e-05,
      "loss": 4.1317,
      "step": 3338
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.335980176248465,
      "learning_rate": 8.401640942480096e-05,
      "loss": 3.9982,
      "step": 3339
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.4163264696432207,
      "learning_rate": 8.38623068163426e-05,
      "loss": 4.1394,
      "step": 3340
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.8563595317017516,
      "learning_rate": 8.370832269855935e-05,
      "loss": 4.1426,
      "step": 3341
    },
    {
      "epoch": 3.82,
      "grad_norm": 2.3596455334181647,
      "learning_rate": 8.355445715586821e-05,
      "loss": 4.1113,
      "step": 3342
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.3091047882021514,
      "learning_rate": 8.340071027262105e-05,
      "loss": 4.167,
      "step": 3343
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.7401030820387218,
      "learning_rate": 8.324708213310474e-05,
      "loss": 4.2679,
      "step": 3344
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.1252546817407043,
      "learning_rate": 8.309357282154112e-05,
      "loss": 3.9514,
      "step": 3345
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.4927864410503962,
      "learning_rate": 8.294018242208682e-05,
      "loss": 4.0077,
      "step": 3346
    },
    {
      "epoch": 3.83,
      "grad_norm": 3.60887229490552,
      "learning_rate": 8.278691101883328e-05,
      "loss": 4.3733,
      "step": 3347
    },
    {
      "epoch": 3.83,
      "grad_norm": 2.246449890851229,
      "learning_rate": 8.263375869580675e-05,
      "loss": 4.11,
      "step": 3348
    },
    {
      "epoch": 3.83,
      "grad_norm": 2.3579661327637096,
      "learning_rate": 8.248072553696808e-05,
      "loss": 4.2004,
      "step": 3349
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.4487181363456263,
      "learning_rate": 8.23278116262128e-05,
      "loss": 4.1187,
      "step": 3350
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.5196724204886165,
      "learning_rate": 8.217501704737135e-05,
      "loss": 4.0816,
      "step": 3351
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.0694949856804647,
      "learning_rate": 8.202234188420855e-05,
      "loss": 4.1055,
      "step": 3352
    },
    {
      "epoch": 3.83,
      "grad_norm": 2.4708918732542076,
      "learning_rate": 8.186978622042352e-05,
      "loss": 4.3279,
      "step": 3353
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.4407763788600658,
      "learning_rate": 8.17173501396502e-05,
      "loss": 4.0728,
      "step": 3354
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.3887405088736797,
      "learning_rate": 8.156503372545693e-05,
      "loss": 4.2407,
      "step": 3355
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.4420209082965683,
      "learning_rate": 8.141283706134637e-05,
      "loss": 4.1959,
      "step": 3356
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.2150378240584478,
      "learning_rate": 8.126076023075535e-05,
      "loss": 4.3334,
      "step": 3357
    },
    {
      "epoch": 3.84,
      "grad_norm": 3.073322560711939,
      "learning_rate": 8.110880331705541e-05,
      "loss": 4.1578,
      "step": 3358
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.2427332547998782,
      "learning_rate": 8.095696640355209e-05,
      "loss": 4.137,
      "step": 3359
    },
    {
      "epoch": 3.84,
      "grad_norm": 2.346414043082525,
      "learning_rate": 8.080524957348523e-05,
      "loss": 3.9984,
      "step": 3360
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.778507204037714,
      "learning_rate": 8.06536529100287e-05,
      "loss": 4.142,
      "step": 3361
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.2149825008595843,
      "learning_rate": 8.050217649629059e-05,
      "loss": 4.2951,
      "step": 3362
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.538892449511593,
      "learning_rate": 8.035082041531312e-05,
      "loss": 4.029,
      "step": 3363
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.9662110692401356,
      "learning_rate": 8.019958475007251e-05,
      "loss": 4.2029,
      "step": 3364
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.6275817571829883,
      "learning_rate": 8.004846958347889e-05,
      "loss": 4.0334,
      "step": 3365
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.6186152654939423,
      "learning_rate": 7.98974749983764e-05,
      "loss": 4.0779,
      "step": 3366
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.5549236912061863,
      "learning_rate": 7.974660107754308e-05,
      "loss": 4.144,
      "step": 3367
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.8173061154013437,
      "learning_rate": 7.959584790369078e-05,
      "loss": 3.9799,
      "step": 3368
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.7511657868524069,
      "learning_rate": 7.944521555946525e-05,
      "loss": 4.0315,
      "step": 3369
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.4861240778857125,
      "learning_rate": 7.929470412744579e-05,
      "loss": 4.1782,
      "step": 3370
    },
    {
      "epoch": 3.85,
      "grad_norm": 2.254123818750728,
      "learning_rate": 7.914431369014562e-05,
      "loss": 4.0544,
      "step": 3371
    },
    {
      "epoch": 3.85,
      "grad_norm": 2.414319803813102,
      "learning_rate": 7.899404433001154e-05,
      "loss": 3.8913,
      "step": 3372
    },
    {
      "epoch": 3.85,
      "grad_norm": 3.4859679339262124,
      "learning_rate": 7.884389612942399e-05,
      "loss": 4.0949,
      "step": 3373
    },
    {
      "epoch": 3.86,
      "grad_norm": 2.2079682893138184,
      "learning_rate": 7.869386917069699e-05,
      "loss": 4.2035,
      "step": 3374
    },
    {
      "epoch": 3.86,
      "grad_norm": 2.9409598737506397,
      "learning_rate": 7.854396353607811e-05,
      "loss": 4.0075,
      "step": 3375
    },
    {
      "epoch": 3.86,
      "grad_norm": 3.1761597778366055,
      "learning_rate": 7.839417930774836e-05,
      "loss": 4.019,
      "step": 3376
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.6833246507199882,
      "learning_rate": 7.824451656782231e-05,
      "loss": 4.0529,
      "step": 3377
    },
    {
      "epoch": 3.86,
      "grad_norm": 5.64332792586859,
      "learning_rate": 7.809497539834766e-05,
      "loss": 4.2422,
      "step": 3378
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.4092665455236553,
      "learning_rate": 7.794555588130575e-05,
      "loss": 4.0162,
      "step": 3379
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.9097182089230074,
      "learning_rate": 7.779625809861114e-05,
      "loss": 3.9794,
      "step": 3380
    },
    {
      "epoch": 3.86,
      "grad_norm": 2.384410809231374,
      "learning_rate": 7.764708213211156e-05,
      "loss": 4.1199,
      "step": 3381
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.7043668772202547,
      "learning_rate": 7.749802806358806e-05,
      "loss": 4.338,
      "step": 3382
    },
    {
      "epoch": 3.87,
      "grad_norm": 2.3790667133190353,
      "learning_rate": 7.734909597475487e-05,
      "loss": 4.1018,
      "step": 3383
    },
    {
      "epoch": 3.87,
      "grad_norm": 2.3448952554154694,
      "learning_rate": 7.720028594725929e-05,
      "loss": 4.2448,
      "step": 3384
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.3004885138203928,
      "learning_rate": 7.705159806268169e-05,
      "loss": 4.1088,
      "step": 3385
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.844620391872864,
      "learning_rate": 7.690303240253563e-05,
      "loss": 4.1142,
      "step": 3386
    },
    {
      "epoch": 3.87,
      "grad_norm": 2.068357525581232,
      "learning_rate": 7.675458904826739e-05,
      "loss": 4.041,
      "step": 3387
    },
    {
      "epoch": 3.87,
      "grad_norm": 3.7256970080300156,
      "learning_rate": 7.660626808125642e-05,
      "loss": 3.8905,
      "step": 3388
    },
    {
      "epoch": 3.87,
      "grad_norm": 2.548347864090204,
      "learning_rate": 7.645806958281497e-05,
      "loss": 4.2503,
      "step": 3389
    },
    {
      "epoch": 3.87,
      "grad_norm": 2.211094582740078,
      "learning_rate": 7.630999363418837e-05,
      "loss": 4.3927,
      "step": 3390
    },
    {
      "epoch": 3.88,
      "grad_norm": 3.0899305078334107,
      "learning_rate": 7.616204031655441e-05,
      "loss": 4.1496,
      "step": 3391
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.6230703505809985,
      "learning_rate": 7.601420971102389e-05,
      "loss": 4.0591,
      "step": 3392
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.494468773094374,
      "learning_rate": 7.586650189864029e-05,
      "loss": 4.2755,
      "step": 3393
    },
    {
      "epoch": 3.88,
      "grad_norm": 2.5228537110169174,
      "learning_rate": 7.571891696037979e-05,
      "loss": 3.9725,
      "step": 3394
    },
    {
      "epoch": 3.88,
      "grad_norm": 3.8015271475298165,
      "learning_rate": 7.557145497715109e-05,
      "loss": 4.2322,
      "step": 3395
    },
    {
      "epoch": 3.88,
      "grad_norm": 2.957810301426153,
      "learning_rate": 7.542411602979558e-05,
      "loss": 3.9961,
      "step": 3396
    },
    {
      "epoch": 3.88,
      "grad_norm": 2.5038145612528013,
      "learning_rate": 7.527690019908718e-05,
      "loss": 4.1258,
      "step": 3397
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.729948682922274,
      "learning_rate": 7.512980756573253e-05,
      "loss": 3.9408,
      "step": 3398
    },
    {
      "epoch": 3.88,
      "grad_norm": 3.319429538393257,
      "learning_rate": 7.498283821037032e-05,
      "loss": 4.2339,
      "step": 3399
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.847789669201742,
      "learning_rate": 7.483599221357188e-05,
      "loss": 4.0012,
      "step": 3400
    },
    {
      "epoch": 3.89,
      "eval_blimp_filtered_avg": 0.5219402985074627,
      "eval_blimp_filtered_std": 0.005628011061357745,
      "step": 3400
    },
    {
      "epoch": 3.89,
      "eval_blimp_supplement_avg": 0.4482758620689655,
      "eval_blimp_supplement_std": 0.022533914102181193,
      "step": 3400
    },
    {
      "epoch": 3.89,
      "eval_vqa_filtered_avg": 0.36,
      "eval_vqa_filtered_std": 0.04824181513244218,
      "step": 3400
    },
    {
      "epoch": 3.89,
      "eval_winoground_filtered_avg": 0.5,
      "eval_winoground_filtered_std": 0.050251890762960605,
      "step": 3400
    },
    {
      "epoch": 3.89,
      "grad_norm": 3.470018721832849,
      "learning_rate": 7.468926965584098e-05,
      "loss": 4.1482,
      "step": 3401
    },
    {
      "epoch": 3.89,
      "grad_norm": 2.2397420095128404,
      "learning_rate": 7.454267061761357e-05,
      "loss": 4.2162,
      "step": 3402
    },
    {
      "epoch": 3.89,
      "grad_norm": 2.089243034152301,
      "learning_rate": 7.439619517925806e-05,
      "loss": 4.2122,
      "step": 3403
    },
    {
      "epoch": 3.89,
      "grad_norm": 2.504611289602091,
      "learning_rate": 7.424984342107472e-05,
      "loss": 4.0375,
      "step": 3404
    },
    {
      "epoch": 3.89,
      "grad_norm": 2.235982658376364,
      "learning_rate": 7.410361542329657e-05,
      "loss": 4.0465,
      "step": 3405
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.7538312831215086,
      "learning_rate": 7.395751126608842e-05,
      "loss": 4.1694,
      "step": 3406
    },
    {
      "epoch": 3.89,
      "grad_norm": 4.792053702622619,
      "learning_rate": 7.38115310295473e-05,
      "loss": 4.0679,
      "step": 3407
    },
    {
      "epoch": 3.89,
      "grad_norm": 2.739059841655377,
      "learning_rate": 7.366567479370215e-05,
      "loss": 4.0756,
      "step": 3408
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.5481163498455444,
      "learning_rate": 7.351994263851413e-05,
      "loss": 4.2122,
      "step": 3409
    },
    {
      "epoch": 3.9,
      "grad_norm": 2.3789177623479203,
      "learning_rate": 7.337433464387631e-05,
      "loss": 4.1076,
      "step": 3410
    },
    {
      "epoch": 3.9,
      "grad_norm": 2.469553958482088,
      "learning_rate": 7.322885088961372e-05,
      "loss": 4.187,
      "step": 3411
    },
    {
      "epoch": 3.9,
      "grad_norm": 3.314921111075954,
      "learning_rate": 7.3083491455483e-05,
      "loss": 4.4027,
      "step": 3412
    },
    {
      "epoch": 3.9,
      "grad_norm": 2.4711899022192307,
      "learning_rate": 7.293825642117317e-05,
      "loss": 4.0521,
      "step": 3413
    },
    {
      "epoch": 3.9,
      "grad_norm": 2.349120990676839,
      "learning_rate": 7.279314586630462e-05,
      "loss": 4.1017,
      "step": 3414
    },
    {
      "epoch": 3.9,
      "grad_norm": 2.1873148832148375,
      "learning_rate": 7.26481598704298e-05,
      "loss": 4.1716,
      "step": 3415
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.481339653012873,
      "learning_rate": 7.250329851303244e-05,
      "loss": 4.2414,
      "step": 3416
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.3175360464850165,
      "learning_rate": 7.235856187352838e-05,
      "loss": 4.1332,
      "step": 3417
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.9848752307300288,
      "learning_rate": 7.221395003126487e-05,
      "loss": 4.1389,
      "step": 3418
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.4332955355371615,
      "learning_rate": 7.20694630655208e-05,
      "loss": 4.0235,
      "step": 3419
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.4860826971494037,
      "learning_rate": 7.192510105550663e-05,
      "loss": 4.1448,
      "step": 3420
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.5502279076412384,
      "learning_rate": 7.17808640803642e-05,
      "loss": 3.9278,
      "step": 3421
    },
    {
      "epoch": 3.91,
      "grad_norm": 2.2184124746945,
      "learning_rate": 7.163675221916699e-05,
      "loss": 4.0612,
      "step": 3422
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.2521161196858128,
      "learning_rate": 7.149276555091972e-05,
      "loss": 3.9441,
      "step": 3423
    },
    {
      "epoch": 3.91,
      "grad_norm": 2.6998207519284954,
      "learning_rate": 7.13489041545586e-05,
      "loss": 4.0083,
      "step": 3424
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.6948036794712216,
      "learning_rate": 7.1205168108951e-05,
      "loss": 4.0673,
      "step": 3425
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.9083136054388012,
      "learning_rate": 7.106155749289571e-05,
      "loss": 4.1857,
      "step": 3426
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.9336598926323545,
      "learning_rate": 7.091807238512276e-05,
      "loss": 4.1656,
      "step": 3427
    },
    {
      "epoch": 3.92,
      "grad_norm": 2.1693045260384753,
      "learning_rate": 7.077471286429328e-05,
      "loss": 4.2116,
      "step": 3428
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.7180440452043282,
      "learning_rate": 7.063147900899968e-05,
      "loss": 4.3377,
      "step": 3429
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.9922263375765297,
      "learning_rate": 7.048837089776535e-05,
      "loss": 4.0492,
      "step": 3430
    },
    {
      "epoch": 3.92,
      "grad_norm": 5.8110245405092265,
      "learning_rate": 7.034538860904482e-05,
      "loss": 4.0022,
      "step": 3431
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.3078704921577244,
      "learning_rate": 7.020253222122369e-05,
      "loss": 4.16,
      "step": 3432
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.348654826934665,
      "learning_rate": 7.005980181261836e-05,
      "loss": 4.1607,
      "step": 3433
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.5297713544558276,
      "learning_rate": 6.991719746147628e-05,
      "loss": 4.1094,
      "step": 3434
    },
    {
      "epoch": 3.93,
      "grad_norm": 2.5183613811567174,
      "learning_rate": 6.977471924597585e-05,
      "loss": 4.2665,
      "step": 3435
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.380223629748389,
      "learning_rate": 6.963236724422609e-05,
      "loss": 3.9095,
      "step": 3436
    },
    {
      "epoch": 3.93,
      "grad_norm": 4.325580821237642,
      "learning_rate": 6.949014153426738e-05,
      "loss": 4.17,
      "step": 3437
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.5847543043490915,
      "learning_rate": 6.934804219407011e-05,
      "loss": 4.3351,
      "step": 3438
    },
    {
      "epoch": 3.93,
      "grad_norm": 2.8867950722683484,
      "learning_rate": 6.920606930153592e-05,
      "loss": 4.0453,
      "step": 3439
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.757867175262845,
      "learning_rate": 6.90642229344969e-05,
      "loss": 4.1525,
      "step": 3440
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.8266086516935927,
      "learning_rate": 6.892250317071603e-05,
      "loss": 4.3103,
      "step": 3441
    },
    {
      "epoch": 3.93,
      "grad_norm": 3.118193753256394,
      "learning_rate": 6.878091008788647e-05,
      "loss": 3.9629,
      "step": 3442
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.692339092807929,
      "learning_rate": 6.863944376363222e-05,
      "loss": 4.0449,
      "step": 3443
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.151465531504573,
      "learning_rate": 6.849810427550772e-05,
      "loss": 4.2318,
      "step": 3444
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.6242586378911261,
      "learning_rate": 6.835689170099811e-05,
      "loss": 4.2257,
      "step": 3445
    },
    {
      "epoch": 3.94,
      "grad_norm": 3.24438797367478,
      "learning_rate": 6.821580611751848e-05,
      "loss": 4.1249,
      "step": 3446
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.8605871689855538,
      "learning_rate": 6.80748476024146e-05,
      "loss": 4.0569,
      "step": 3447
    },
    {
      "epoch": 3.94,
      "grad_norm": 3.101605645042912,
      "learning_rate": 6.793401623296257e-05,
      "loss": 4.1084,
      "step": 3448
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.3958115241882771,
      "learning_rate": 6.779331208636883e-05,
      "loss": 3.9488,
      "step": 3449
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.739262636989102,
      "learning_rate": 6.765273523976982e-05,
      "loss": 4.043,
      "step": 3450
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.9290841423785303,
      "learning_rate": 6.751228577023237e-05,
      "loss": 4.1535,
      "step": 3451
    },
    {
      "epoch": 3.95,
      "grad_norm": 2.2970291115809727,
      "learning_rate": 6.737196375475362e-05,
      "loss": 4.2475,
      "step": 3452
    },
    {
      "epoch": 3.95,
      "grad_norm": 2.6318991757245245,
      "learning_rate": 6.723176927026062e-05,
      "loss": 4.2696,
      "step": 3453
    },
    {
      "epoch": 3.95,
      "grad_norm": 2.8133433344600167,
      "learning_rate": 6.709170239361065e-05,
      "loss": 4.1249,
      "step": 3454
    },
    {
      "epoch": 3.95,
      "grad_norm": 2.8220270673672188,
      "learning_rate": 6.695176320159081e-05,
      "loss": 4.0947,
      "step": 3455
    },
    {
      "epoch": 3.95,
      "grad_norm": 1.599630642936928,
      "learning_rate": 6.681195177091842e-05,
      "loss": 4.1151,
      "step": 3456
    },
    {
      "epoch": 3.95,
      "grad_norm": 3.1824872160472784,
      "learning_rate": 6.667226817824067e-05,
      "loss": 4.0812,
      "step": 3457
    },
    {
      "epoch": 3.95,
      "grad_norm": 3.848714515958676,
      "learning_rate": 6.653271250013473e-05,
      "loss": 4.1854,
      "step": 3458
    },
    {
      "epoch": 3.95,
      "grad_norm": 2.559022294045578,
      "learning_rate": 6.63932848131076e-05,
      "loss": 3.9643,
      "step": 3459
    },
    {
      "epoch": 3.95,
      "grad_norm": 1.8088261615900627,
      "learning_rate": 6.625398519359605e-05,
      "loss": 3.9422,
      "step": 3460
    },
    {
      "epoch": 3.96,
      "grad_norm": 1.8519859070854041,
      "learning_rate": 6.611481371796674e-05,
      "loss": 4.214,
      "step": 3461
    },
    {
      "epoch": 3.96,
      "grad_norm": 2.0255100265750485,
      "learning_rate": 6.597577046251618e-05,
      "loss": 3.9239,
      "step": 3462
    },
    {
      "epoch": 3.96,
      "grad_norm": 3.4315138335817554,
      "learning_rate": 6.583685550347017e-05,
      "loss": 4.1546,
      "step": 3463
    },
    {
      "epoch": 3.96,
      "grad_norm": 2.0871922543234698,
      "learning_rate": 6.569806891698466e-05,
      "loss": 4.0996,
      "step": 3464
    },
    {
      "epoch": 3.96,
      "grad_norm": 1.6781911138929437,
      "learning_rate": 6.555941077914493e-05,
      "loss": 4.0037,
      "step": 3465
    },
    {
      "epoch": 3.96,
      "grad_norm": 1.9052747457383474,
      "learning_rate": 6.542088116596598e-05,
      "loss": 4.0219,
      "step": 3466
    },
    {
      "epoch": 3.96,
      "grad_norm": 2.45016693315942,
      "learning_rate": 6.528248015339225e-05,
      "loss": 4.2443,
      "step": 3467
    },
    {
      "epoch": 3.96,
      "grad_norm": 2.292821470969717,
      "learning_rate": 6.51442078172978e-05,
      "loss": 4.1301,
      "step": 3468
    },
    {
      "epoch": 3.96,
      "grad_norm": 2.488655856975399,
      "learning_rate": 6.500606423348604e-05,
      "loss": 4.1392,
      "step": 3469
    },
    {
      "epoch": 3.97,
      "grad_norm": 2.010321010272779,
      "learning_rate": 6.486804947768984e-05,
      "loss": 4.2417,
      "step": 3470
    },
    {
      "epoch": 3.97,
      "grad_norm": 2.393672572345543,
      "learning_rate": 6.473016362557155e-05,
      "loss": 3.9808,
      "step": 3471
    },
    {
      "epoch": 3.97,
      "grad_norm": 2.1518823646805685,
      "learning_rate": 6.459240675272252e-05,
      "loss": 4.3066,
      "step": 3472
    },
    {
      "epoch": 3.97,
      "grad_norm": 2.1963810693386523,
      "learning_rate": 6.445477893466372e-05,
      "loss": 4.0091,
      "step": 3473
    },
    {
      "epoch": 3.97,
      "grad_norm": 1.936092188681743,
      "learning_rate": 6.431728024684525e-05,
      "loss": 4.1202,
      "step": 3474
    },
    {
      "epoch": 3.97,
      "grad_norm": 3.073665176499354,
      "learning_rate": 6.417991076464648e-05,
      "loss": 4.207,
      "step": 3475
    },
    {
      "epoch": 3.97,
      "grad_norm": 2.905600596477906,
      "learning_rate": 6.404267056337584e-05,
      "loss": 4.1479,
      "step": 3476
    },
    {
      "epoch": 3.97,
      "grad_norm": 3.7100267208133517,
      "learning_rate": 6.390555971827095e-05,
      "loss": 4.103,
      "step": 3477
    },
    {
      "epoch": 3.97,
      "grad_norm": 2.2875590363680427,
      "learning_rate": 6.376857830449854e-05,
      "loss": 4.2382,
      "step": 3478
    },
    {
      "epoch": 3.98,
      "grad_norm": 1.9457960919051724,
      "learning_rate": 6.363172639715441e-05,
      "loss": 4.2627,
      "step": 3479
    },
    {
      "epoch": 3.98,
      "grad_norm": 3.1182339816371907,
      "learning_rate": 6.349500407126312e-05,
      "loss": 4.1978,
      "step": 3480
    },
    {
      "epoch": 3.98,
      "grad_norm": 4.991612699971542,
      "learning_rate": 6.335841140177851e-05,
      "loss": 4.0617,
      "step": 3481
    },
    {
      "epoch": 3.98,
      "grad_norm": 2.1690204897626857,
      "learning_rate": 6.322194846358317e-05,
      "loss": 3.9048,
      "step": 3482
    },
    {
      "epoch": 3.98,
      "grad_norm": 1.5472513968828265,
      "learning_rate": 6.30856153314886e-05,
      "loss": 3.9185,
      "step": 3483
    },
    {
      "epoch": 3.98,
      "grad_norm": 1.898410892349771,
      "learning_rate": 6.294941208023515e-05,
      "loss": 4.272,
      "step": 3484
    },
    {
      "epoch": 3.98,
      "grad_norm": 4.954931111265682,
      "learning_rate": 6.281333878449191e-05,
      "loss": 4.2094,
      "step": 3485
    },
    {
      "epoch": 3.98,
      "grad_norm": 2.3556245215414036,
      "learning_rate": 6.267739551885687e-05,
      "loss": 4.2815,
      "step": 3486
    },
    {
      "epoch": 3.99,
      "grad_norm": 3.0246034456165964,
      "learning_rate": 6.254158235785668e-05,
      "loss": 4.0454,
      "step": 3487
    },
    {
      "epoch": 3.99,
      "grad_norm": 2.1724604598500123,
      "learning_rate": 6.24058993759464e-05,
      "loss": 4.1188,
      "step": 3488
    },
    {
      "epoch": 3.99,
      "grad_norm": 1.3642957031518947,
      "learning_rate": 6.227034664751009e-05,
      "loss": 4.1455,
      "step": 3489
    },
    {
      "epoch": 3.99,
      "grad_norm": 2.2041174720177215,
      "learning_rate": 6.213492424686029e-05,
      "loss": 4.1365,
      "step": 3490
    },
    {
      "epoch": 3.99,
      "grad_norm": 1.807827488810114,
      "learning_rate": 6.199963224823788e-05,
      "loss": 4.0472,
      "step": 3491
    },
    {
      "epoch": 3.99,
      "grad_norm": 2.335851456663086,
      "learning_rate": 6.186447072581277e-05,
      "loss": 3.9763,
      "step": 3492
    },
    {
      "epoch": 3.99,
      "grad_norm": 2.883482320528189,
      "learning_rate": 6.172943975368269e-05,
      "loss": 4.1042,
      "step": 3493
    },
    {
      "epoch": 3.99,
      "grad_norm": 2.2436354701525016,
      "learning_rate": 6.159453940587424e-05,
      "loss": 4.1076,
      "step": 3494
    },
    {
      "epoch": 3.99,
      "grad_norm": 1.5966055793397647,
      "learning_rate": 6.145976975634228e-05,
      "loss": 3.9576,
      "step": 3495
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.6825985345290422,
      "learning_rate": 6.132513087897007e-05,
      "loss": 4.108,
      "step": 3496
    },
    {
      "epoch": 4.0,
      "grad_norm": 3.9566097888547604,
      "learning_rate": 6.119062284756897e-05,
      "loss": 4.0415,
      "step": 3497
    },
    {
      "epoch": 4.0,
      "grad_norm": 2.706239460238766,
      "learning_rate": 6.105624573587882e-05,
      "loss": 4.2283,
      "step": 3498
    },
    {
      "epoch": 4.0,
      "grad_norm": 2.3765801577541357,
      "learning_rate": 6.092199961756774e-05,
      "loss": 4.0625,
      "step": 3499
    },
    {
      "epoch": 4.0,
      "grad_norm": 2.739307432103461,
      "learning_rate": 6.078788456623197e-05,
      "loss": 4.0601,
      "step": 3500
    },
    {
      "epoch": 4.0,
      "eval_blimp_filtered_avg": 0.5240298507462686,
      "eval_blimp_filtered_std": 0.005630169654235808,
      "step": 3500
    },
    {
      "epoch": 4.0,
      "eval_blimp_supplement_avg": 0.4504310344827586,
      "eval_blimp_supplement_std": 0.02258237285959225,
      "step": 3500
    },
    {
      "epoch": 4.0,
      "eval_vqa_filtered_avg": 0.37,
      "eval_vqa_filtered_std": 0.048523658709391,
      "step": 3500
    },
    {
      "epoch": 4.0,
      "eval_winoground_filtered_avg": 0.51,
      "eval_winoground_filtered_std": 0.05024183937956912,
      "step": 3500
    },
    {
      "epoch": 4.0,
      "eval_blimp_filtered_avg": 0.5240298507462686,
      "eval_blimp_filtered_std": 0.005630169654235808,
      "step": 3500
    },
    {
      "epoch": 4.0,
      "eval_blimp_supplement_avg": 0.4504310344827586,
      "eval_blimp_supplement_std": 0.02258237285959225,
      "step": 3500
    },
    {
      "epoch": 4.0,
      "eval_vqa_filtered_avg": 0.37,
      "eval_vqa_filtered_std": 0.048523658709391,
      "step": 3500
    },
    {
      "epoch": 4.0,
      "eval_winoground_filtered_avg": 0.51,
      "eval_winoground_filtered_std": 0.05024183937956912,
      "step": 3500
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.7075372915311746,
      "learning_rate": 6.065390065539566e-05,
      "loss": 3.9221,
      "step": 3501
    },
    {
      "epoch": 4.0,
      "grad_norm": 2.598589349431351,
      "learning_rate": 6.052004795851135e-05,
      "loss": 4.113,
      "step": 3502
    },
    {
      "epoch": 4.0,
      "grad_norm": 2.820915136060971,
      "learning_rate": 6.0386326548959586e-05,
      "loss": 4.2593,
      "step": 3503
    },
    {
      "epoch": 4.0,
      "grad_norm": 3.191741811289664,
      "learning_rate": 6.025273650004898e-05,
      "loss": 4.0402,
      "step": 3504
    },
    {
      "epoch": 4.01,
      "grad_norm": 1.4727093427177709,
      "learning_rate": 6.0119277885015825e-05,
      "loss": 4.1636,
      "step": 3505
    },
    {
      "epoch": 4.01,
      "grad_norm": 2.2456756520920482,
      "learning_rate": 5.998595077702483e-05,
      "loss": 4.2336,
      "step": 3506
    },
    {
      "epoch": 4.01,
      "grad_norm": 2.2095242615007376,
      "learning_rate": 5.9852755249168315e-05,
      "loss": 3.8525,
      "step": 3507
    },
    {
      "epoch": 4.01,
      "grad_norm": 2.468290611430893,
      "learning_rate": 5.9719691374466516e-05,
      "loss": 4.1854,
      "step": 3508
    },
    {
      "epoch": 4.01,
      "grad_norm": 3.040043344784734,
      "learning_rate": 5.958675922586761e-05,
      "loss": 4.2067,
      "step": 3509
    },
    {
      "epoch": 4.01,
      "grad_norm": 2.6184799625210005,
      "learning_rate": 5.945395887624723e-05,
      "loss": 3.9965,
      "step": 3510
    },
    {
      "epoch": 4.01,
      "grad_norm": 2.80929383019444,
      "learning_rate": 5.9321290398409147e-05,
      "loss": 4.144,
      "step": 3511
    },
    {
      "epoch": 4.01,
      "grad_norm": 2.3100933172494087,
      "learning_rate": 5.918875386508462e-05,
      "loss": 4.1339,
      "step": 3512
    },
    {
      "epoch": 4.01,
      "grad_norm": 2.3247771387771095,
      "learning_rate": 5.905634934893264e-05,
      "loss": 4.2282,
      "step": 3513
    },
    {
      "epoch": 4.02,
      "grad_norm": 2.1392157259734397,
      "learning_rate": 5.89240769225398e-05,
      "loss": 4.3983,
      "step": 3514
    },
    {
      "epoch": 4.02,
      "grad_norm": 1.5940489516236678,
      "learning_rate": 5.879193665842028e-05,
      "loss": 4.0937,
      "step": 3515
    },
    {
      "epoch": 4.02,
      "grad_norm": 2.4128766487299433,
      "learning_rate": 5.865992862901581e-05,
      "loss": 4.0056,
      "step": 3516
    },
    {
      "epoch": 4.02,
      "grad_norm": 2.378453104810622,
      "learning_rate": 5.8528052906695754e-05,
      "loss": 4.0268,
      "step": 3517
    },
    {
      "epoch": 4.02,
      "grad_norm": 1.734870061248331,
      "learning_rate": 5.8396309563756595e-05,
      "loss": 4.0503,
      "step": 3518
    },
    {
      "epoch": 4.02,
      "grad_norm": 2.22939070117633,
      "learning_rate": 5.8264698672422613e-05,
      "loss": 4.0084,
      "step": 3519
    },
    {
      "epoch": 4.02,
      "grad_norm": 2.6419235359217965,
      "learning_rate": 5.8133220304845276e-05,
      "loss": 4.0112,
      "step": 3520
    },
    {
      "epoch": 4.02,
      "grad_norm": 2.420327913540527,
      "learning_rate": 5.8001874533103475e-05,
      "loss": 4.2076,
      "step": 3521
    },
    {
      "epoch": 4.03,
      "grad_norm": 4.4122761669434185,
      "learning_rate": 5.787066142920339e-05,
      "loss": 4.1702,
      "step": 3522
    },
    {
      "epoch": 4.03,
      "grad_norm": 3.6482888068528774,
      "learning_rate": 5.773958106507844e-05,
      "loss": 4.0976,
      "step": 3523
    },
    {
      "epoch": 4.03,
      "grad_norm": 2.4409284668029243,
      "learning_rate": 5.7608633512589353e-05,
      "loss": 4.3172,
      "step": 3524
    },
    {
      "epoch": 4.03,
      "grad_norm": 1.9424071671925356,
      "learning_rate": 5.747781884352396e-05,
      "loss": 4.0566,
      "step": 3525
    },
    {
      "epoch": 4.03,
      "grad_norm": 1.98026865969247,
      "learning_rate": 5.734713712959736e-05,
      "loss": 4.2095,
      "step": 3526
    },
    {
      "epoch": 4.03,
      "grad_norm": 2.878657245771692,
      "learning_rate": 5.721658844245153e-05,
      "loss": 4.1011,
      "step": 3527
    },
    {
      "epoch": 4.03,
      "grad_norm": 3.230259443936672,
      "learning_rate": 5.7086172853655734e-05,
      "loss": 4.1763,
      "step": 3528
    },
    {
      "epoch": 4.03,
      "grad_norm": 2.7201923801798156,
      "learning_rate": 5.6955890434706254e-05,
      "loss": 3.9993,
      "step": 3529
    },
    {
      "epoch": 4.03,
      "grad_norm": 3.067605526787616,
      "learning_rate": 5.682574125702626e-05,
      "loss": 4.0871,
      "step": 3530
    },
    {
      "epoch": 4.04,
      "grad_norm": 12.791950912037768,
      "learning_rate": 5.6695725391965975e-05,
      "loss": 4.1648,
      "step": 3531
    },
    {
      "epoch": 4.04,
      "grad_norm": 2.5082280405687865,
      "learning_rate": 5.656584291080248e-05,
      "loss": 4.0296,
      "step": 3532
    },
    {
      "epoch": 4.04,
      "grad_norm": 2.1402752935860803,
      "learning_rate": 5.643609388473973e-05,
      "loss": 4.1179,
      "step": 3533
    },
    {
      "epoch": 4.04,
      "grad_norm": 2.522998470014395,
      "learning_rate": 5.6306478384908685e-05,
      "loss": 4.1138,
      "step": 3534
    },
    {
      "epoch": 4.04,
      "grad_norm": 2.5947858296007853,
      "learning_rate": 5.61769964823667e-05,
      "loss": 4.0997,
      "step": 3535
    },
    {
      "epoch": 4.04,
      "grad_norm": 5.321972315426501,
      "learning_rate": 5.6047648248098295e-05,
      "loss": 4.0776,
      "step": 3536
    },
    {
      "epoch": 4.04,
      "grad_norm": 5.344490044865733,
      "learning_rate": 5.591843375301456e-05,
      "loss": 4.1193,
      "step": 3537
    },
    {
      "epoch": 4.04,
      "grad_norm": 3.444675384158124,
      "learning_rate": 5.578935306795325e-05,
      "loss": 4.275,
      "step": 3538
    },
    {
      "epoch": 4.04,
      "grad_norm": 2.848810277902967,
      "learning_rate": 5.566040626367876e-05,
      "loss": 4.0741,
      "step": 3539
    },
    {
      "epoch": 4.05,
      "grad_norm": 4.010707911723712,
      "learning_rate": 5.553159341088211e-05,
      "loss": 4.0755,
      "step": 3540
    },
    {
      "epoch": 4.05,
      "grad_norm": 5.028743821661657,
      "learning_rate": 5.540291458018096e-05,
      "loss": 4.1993,
      "step": 3541
    },
    {
      "epoch": 4.05,
      "grad_norm": 5.068107858066767,
      "learning_rate": 5.5274369842119325e-05,
      "loss": 3.9684,
      "step": 3542
    },
    {
      "epoch": 4.05,
      "grad_norm": 2.3737456782388175,
      "learning_rate": 5.514595926716795e-05,
      "loss": 4.0375,
      "step": 3543
    },
    {
      "epoch": 4.05,
      "grad_norm": 3.317066957801331,
      "learning_rate": 5.501768292572368e-05,
      "loss": 4.0848,
      "step": 3544
    },
    {
      "epoch": 4.05,
      "grad_norm": 4.001136557476432,
      "learning_rate": 5.4889540888110006e-05,
      "loss": 4.2648,
      "step": 3545
    },
    {
      "epoch": 4.05,
      "grad_norm": 5.015089213842579,
      "learning_rate": 5.476153322457689e-05,
      "loss": 4.3071,
      "step": 3546
    },
    {
      "epoch": 4.05,
      "grad_norm": 4.332737110312137,
      "learning_rate": 5.463366000530054e-05,
      "loss": 4.1417,
      "step": 3547
    },
    {
      "epoch": 4.05,
      "grad_norm": 2.142962660200713,
      "learning_rate": 5.4505921300383206e-05,
      "loss": 4.0741,
      "step": 3548
    },
    {
      "epoch": 4.06,
      "grad_norm": 4.579399725138693,
      "learning_rate": 5.437831717985374e-05,
      "loss": 4.1862,
      "step": 3549
    },
    {
      "epoch": 4.06,
      "grad_norm": 3.606865827222683,
      "learning_rate": 5.4250847713667014e-05,
      "loss": 4.0912,
      "step": 3550
    },
    {
      "epoch": 4.06,
      "grad_norm": 2.3965773942002677,
      "learning_rate": 5.412351297170426e-05,
      "loss": 4.0709,
      "step": 3551
    },
    {
      "epoch": 4.06,
      "grad_norm": 5.31315330392113,
      "learning_rate": 5.399631302377252e-05,
      "loss": 4.3931,
      "step": 3552
    },
    {
      "epoch": 4.06,
      "grad_norm": 9.79507585167842,
      "learning_rate": 5.386924793960533e-05,
      "loss": 4.1638,
      "step": 3553
    },
    {
      "epoch": 4.06,
      "grad_norm": 1.9000302223317878,
      "learning_rate": 5.374231778886211e-05,
      "loss": 4.1287,
      "step": 3554
    },
    {
      "epoch": 4.06,
      "grad_norm": 3.090100560824642,
      "learning_rate": 5.361552264112833e-05,
      "loss": 4.1297,
      "step": 3555
    },
    {
      "epoch": 4.06,
      "grad_norm": 3.462344971719606,
      "learning_rate": 5.3488862565915314e-05,
      "loss": 4.0514,
      "step": 3556
    },
    {
      "epoch": 4.07,
      "grad_norm": 2.529639624401169,
      "learning_rate": 5.3362337632660526e-05,
      "loss": 4.2577,
      "step": 3557
    },
    {
      "epoch": 4.07,
      "grad_norm": 2.268078829755966,
      "learning_rate": 5.323594791072724e-05,
      "loss": 4.175,
      "step": 3558
    },
    {
      "epoch": 4.07,
      "grad_norm": 4.416049110535458,
      "learning_rate": 5.310969346940467e-05,
      "loss": 4.1469,
      "step": 3559
    },
    {
      "epoch": 4.07,
      "grad_norm": 1.6177420254825572,
      "learning_rate": 5.298357437790781e-05,
      "loss": 4.1866,
      "step": 3560
    },
    {
      "epoch": 4.07,
      "grad_norm": 3.3316744867948227,
      "learning_rate": 5.2857590705377495e-05,
      "loss": 4.2168,
      "step": 3561
    },
    {
      "epoch": 4.07,
      "grad_norm": 14.004467923721895,
      "learning_rate": 5.2731742520880306e-05,
      "loss": 4.061,
      "step": 3562
    },
    {
      "epoch": 4.07,
      "grad_norm": 2.905782346746633,
      "learning_rate": 5.2606029893408486e-05,
      "loss": 4.1482,
      "step": 3563
    },
    {
      "epoch": 4.07,
      "grad_norm": 3.4514697186083922,
      "learning_rate": 5.248045289188017e-05,
      "loss": 4.4161,
      "step": 3564
    },
    {
      "epoch": 4.07,
      "grad_norm": 5.935484268848983,
      "learning_rate": 5.235501158513881e-05,
      "loss": 4.4268,
      "step": 3565
    },
    {
      "epoch": 4.08,
      "grad_norm": 4.275284440554483,
      "learning_rate": 5.22297060419537e-05,
      "loss": 4.3941,
      "step": 3566
    },
    {
      "epoch": 4.08,
      "grad_norm": 9.082787689708622,
      "learning_rate": 5.210453633101968e-05,
      "loss": 4.1896,
      "step": 3567
    },
    {
      "epoch": 4.08,
      "grad_norm": 5.867315258788226,
      "learning_rate": 5.197950252095711e-05,
      "loss": 4.3694,
      "step": 3568
    },
    {
      "epoch": 4.08,
      "grad_norm": 4.671059362352427,
      "learning_rate": 5.1854604680311765e-05,
      "loss": 3.989,
      "step": 3569
    },
    {
      "epoch": 4.08,
      "grad_norm": 3.247118131856371,
      "learning_rate": 5.1729842877555036e-05,
      "loss": 3.9313,
      "step": 3570
    },
    {
      "epoch": 4.08,
      "grad_norm": 2.9378153984924147,
      "learning_rate": 5.1605217181083594e-05,
      "loss": 4.0907,
      "step": 3571
    },
    {
      "epoch": 4.08,
      "grad_norm": 3.081252416293258,
      "learning_rate": 5.1480727659219604e-05,
      "loss": 4.1651,
      "step": 3572
    },
    {
      "epoch": 4.08,
      "grad_norm": 3.4445595433517697,
      "learning_rate": 5.135637438021044e-05,
      "loss": 4.0493,
      "step": 3573
    },
    {
      "epoch": 4.08,
      "grad_norm": 3.0666528845879157,
      "learning_rate": 5.123215741222888e-05,
      "loss": 4.2087,
      "step": 3574
    },
    {
      "epoch": 4.09,
      "grad_norm": 3.5851993840612653,
      "learning_rate": 5.1108076823373014e-05,
      "loss": 4.2625,
      "step": 3575
    },
    {
      "epoch": 4.09,
      "grad_norm": 4.3938222369688305,
      "learning_rate": 5.0984132681666054e-05,
      "loss": 4.1969,
      "step": 3576
    },
    {
      "epoch": 4.09,
      "grad_norm": 3.0662066044007963,
      "learning_rate": 5.08603250550565e-05,
      "loss": 4.0751,
      "step": 3577
    },
    {
      "epoch": 4.09,
      "grad_norm": 3.3923288014385773,
      "learning_rate": 5.073665401141796e-05,
      "loss": 4.0649,
      "step": 3578
    },
    {
      "epoch": 4.09,
      "grad_norm": 2.3796999722476486,
      "learning_rate": 5.0613119618549235e-05,
      "loss": 4.1409,
      "step": 3579
    },
    {
      "epoch": 4.09,
      "grad_norm": 2.490782452264804,
      "learning_rate": 5.04897219441741e-05,
      "loss": 4.2024,
      "step": 3580
    },
    {
      "epoch": 4.09,
      "grad_norm": 3.7546327372179626,
      "learning_rate": 5.0366461055941545e-05,
      "loss": 4.2641,
      "step": 3581
    },
    {
      "epoch": 4.09,
      "grad_norm": 4.282670696850256,
      "learning_rate": 5.0243337021425314e-05,
      "loss": 4.1234,
      "step": 3582
    },
    {
      "epoch": 4.09,
      "grad_norm": 5.554601141617358,
      "learning_rate": 5.0120349908124314e-05,
      "loss": 3.9754,
      "step": 3583
    },
    {
      "epoch": 4.1,
      "grad_norm": 4.407669682737058,
      "learning_rate": 4.9997499783462346e-05,
      "loss": 4.2378,
      "step": 3584
    },
    {
      "epoch": 4.1,
      "grad_norm": 27.888943282567677,
      "learning_rate": 4.987478671478831e-05,
      "loss": 4.0176,
      "step": 3585
    },
    {
      "epoch": 4.1,
      "grad_norm": 2.396506910788156,
      "learning_rate": 4.975221076937551e-05,
      "loss": 4.2301,
      "step": 3586
    },
    {
      "epoch": 4.1,
      "grad_norm": 2.8792619358918006,
      "learning_rate": 4.962977201442248e-05,
      "loss": 4.1278,
      "step": 3587
    },
    {
      "epoch": 4.1,
      "grad_norm": 5.883881528022609,
      "learning_rate": 4.9507470517052375e-05,
      "loss": 4.2444,
      "step": 3588
    },
    {
      "epoch": 4.1,
      "grad_norm": 4.2078105581336,
      "learning_rate": 4.93853063443132e-05,
      "loss": 4.2882,
      "step": 3589
    },
    {
      "epoch": 4.1,
      "grad_norm": 3.3397891771736252,
      "learning_rate": 4.926327956317749e-05,
      "loss": 4.2494,
      "step": 3590
    },
    {
      "epoch": 4.1,
      "grad_norm": 3.3646523372931747,
      "learning_rate": 4.914139024054267e-05,
      "loss": 4.4453,
      "step": 3591
    },
    {
      "epoch": 4.11,
      "grad_norm": 3.1083894987660416,
      "learning_rate": 4.901963844323057e-05,
      "loss": 4.4633,
      "step": 3592
    },
    {
      "epoch": 4.11,
      "grad_norm": 2.1172080040575127,
      "learning_rate": 4.889802423798807e-05,
      "loss": 4.1917,
      "step": 3593
    },
    {
      "epoch": 4.11,
      "grad_norm": 2.6615660917725013,
      "learning_rate": 4.877654769148606e-05,
      "loss": 4.3003,
      "step": 3594
    },
    {
      "epoch": 4.11,
      "grad_norm": 2.5523401723525385,
      "learning_rate": 4.865520887032034e-05,
      "loss": 4.2777,
      "step": 3595
    },
    {
      "epoch": 4.11,
      "grad_norm": 12.185781555777368,
      "learning_rate": 4.8534007841011043e-05,
      "loss": 4.2261,
      "step": 3596
    },
    {
      "epoch": 4.11,
      "grad_norm": 3.22281287487506,
      "learning_rate": 4.841294467000283e-05,
      "loss": 4.1856,
      "step": 3597
    },
    {
      "epoch": 4.11,
      "grad_norm": 3.4459971142781867,
      "learning_rate": 4.829201942366483e-05,
      "loss": 4.108,
      "step": 3598
    },
    {
      "epoch": 4.11,
      "grad_norm": 2.720613178267878,
      "learning_rate": 4.817123216829032e-05,
      "loss": 4.2964,
      "step": 3599
    },
    {
      "epoch": 4.11,
      "grad_norm": 2.846198699750679,
      "learning_rate": 4.8050582970097295e-05,
      "loss": 4.2912,
      "step": 3600
    },
    {
      "epoch": 4.11,
      "eval_blimp_filtered_avg": 0.524179104477612,
      "eval_blimp_filtered_std": 0.005679808762812842,
      "step": 3600
    },
    {
      "epoch": 4.11,
      "eval_blimp_supplement_avg": 0.45905172413793105,
      "eval_blimp_supplement_std": 0.022908982054118343,
      "step": 3600
    },
    {
      "epoch": 4.11,
      "eval_vqa_filtered_avg": 0.37,
      "eval_vqa_filtered_std": 0.048523658709391,
      "step": 3600
    },
    {
      "epoch": 4.11,
      "eval_winoground_filtered_avg": 0.48,
      "eval_winoground_filtered_std": 0.05021167315686779,
      "step": 3600
    },
    {
      "epoch": 4.12,
      "grad_norm": 5.131782034283068,
      "learning_rate": 4.7930071895227785e-05,
      "loss": 4.3504,
      "step": 3601
    },
    {
      "epoch": 4.12,
      "grad_norm": 5.581362767895976,
      "learning_rate": 4.78096990097483e-05,
      "loss": 4.174,
      "step": 3602
    },
    {
      "epoch": 4.12,
      "grad_norm": 6.110857285708107,
      "learning_rate": 4.76894643796493e-05,
      "loss": 3.9011,
      "step": 3603
    },
    {
      "epoch": 4.12,
      "grad_norm": 2.5832484369995896,
      "learning_rate": 4.7569368070845685e-05,
      "loss": 4.1102,
      "step": 3604
    },
    {
      "epoch": 4.12,
      "grad_norm": 8.0145152780317,
      "learning_rate": 4.7449410149176516e-05,
      "loss": 4.0851,
      "step": 3605
    },
    {
      "epoch": 4.12,
      "grad_norm": 3.8413681816037544,
      "learning_rate": 4.732959068040499e-05,
      "loss": 4.132,
      "step": 3606
    },
    {
      "epoch": 4.12,
      "grad_norm": 3.3359890694387717,
      "learning_rate": 4.720990973021814e-05,
      "loss": 4.1289,
      "step": 3607
    },
    {
      "epoch": 4.12,
      "grad_norm": 3.6610589894518455,
      "learning_rate": 4.709036736422751e-05,
      "loss": 4.1985,
      "step": 3608
    },
    {
      "epoch": 4.12,
      "grad_norm": 3.572399539195982,
      "learning_rate": 4.6970963647968306e-05,
      "loss": 4.3221,
      "step": 3609
    },
    {
      "epoch": 4.13,
      "grad_norm": 6.618275238979667,
      "learning_rate": 4.685169864689997e-05,
      "loss": 4.0706,
      "step": 3610
    },
    {
      "epoch": 4.13,
      "grad_norm": 2.397455503599506,
      "learning_rate": 4.673257242640563e-05,
      "loss": 4.0416,
      "step": 3611
    },
    {
      "epoch": 4.13,
      "grad_norm": 2.8466327292153877,
      "learning_rate": 4.6613585051792523e-05,
      "loss": 4.0198,
      "step": 3612
    },
    {
      "epoch": 4.13,
      "grad_norm": 2.5325309387337795,
      "learning_rate": 4.649473658829176e-05,
      "loss": 4.0148,
      "step": 3613
    },
    {
      "epoch": 4.13,
      "grad_norm": 2.949252771556712,
      "learning_rate": 4.637602710105825e-05,
      "loss": 4.1447,
      "step": 3614
    },
    {
      "epoch": 4.13,
      "grad_norm": 3.881652909076067,
      "learning_rate": 4.6257456655170723e-05,
      "loss": 4.0515,
      "step": 3615
    },
    {
      "epoch": 4.13,
      "grad_norm": 2.984135481559177,
      "learning_rate": 4.6139025315631664e-05,
      "loss": 4.1306,
      "step": 3616
    },
    {
      "epoch": 4.13,
      "grad_norm": 3.8769972117637828,
      "learning_rate": 4.602073314736737e-05,
      "loss": 4.3417,
      "step": 3617
    },
    {
      "epoch": 4.13,
      "grad_norm": 3.2123672206147575,
      "learning_rate": 4.590258021522778e-05,
      "loss": 4.0543,
      "step": 3618
    },
    {
      "epoch": 4.14,
      "grad_norm": 3.2556661805071196,
      "learning_rate": 4.5784566583986536e-05,
      "loss": 4.025,
      "step": 3619
    },
    {
      "epoch": 4.14,
      "grad_norm": 2.8773568905651734,
      "learning_rate": 4.5666692318340825e-05,
      "loss": 4.1821,
      "step": 3620
    },
    {
      "epoch": 4.14,
      "grad_norm": 3.3582550092351373,
      "learning_rate": 4.5548957482911484e-05,
      "loss": 4.2361,
      "step": 3621
    },
    {
      "epoch": 4.14,
      "grad_norm": 3.2113267285914255,
      "learning_rate": 4.543136214224299e-05,
      "loss": 3.9839,
      "step": 3622
    },
    {
      "epoch": 4.14,
      "grad_norm": 4.594280805041271,
      "learning_rate": 4.531390636080326e-05,
      "loss": 3.9496,
      "step": 3623
    },
    {
      "epoch": 4.14,
      "grad_norm": 4.298898080151559,
      "learning_rate": 4.519659020298371e-05,
      "loss": 4.1084,
      "step": 3624
    },
    {
      "epoch": 4.14,
      "grad_norm": 5.881376028009944,
      "learning_rate": 4.507941373309927e-05,
      "loss": 4.1397,
      "step": 3625
    },
    {
      "epoch": 4.14,
      "grad_norm": 4.368712011424131,
      "learning_rate": 4.496237701538816e-05,
      "loss": 4.1113,
      "step": 3626
    },
    {
      "epoch": 4.15,
      "grad_norm": 7.438650643518618,
      "learning_rate": 4.48454801140122e-05,
      "loss": 4.1301,
      "step": 3627
    },
    {
      "epoch": 4.15,
      "grad_norm": 3.721389973320724,
      "learning_rate": 4.472872309305632e-05,
      "loss": 4.1596,
      "step": 3628
    },
    {
      "epoch": 4.15,
      "grad_norm": 4.77930486982069,
      "learning_rate": 4.461210601652885e-05,
      "loss": 4.1862,
      "step": 3629
    },
    {
      "epoch": 4.15,
      "grad_norm": 7.783998153597915,
      "learning_rate": 4.449562894836152e-05,
      "loss": 4.0013,
      "step": 3630
    },
    {
      "epoch": 4.15,
      "grad_norm": 4.382224339566421,
      "learning_rate": 4.4379291952409076e-05,
      "loss": 4.1968,
      "step": 3631
    },
    {
      "epoch": 4.15,
      "grad_norm": 3.709805678169295,
      "learning_rate": 4.426309509244985e-05,
      "loss": 4.3482,
      "step": 3632
    },
    {
      "epoch": 4.15,
      "grad_norm": 5.232961325614194,
      "learning_rate": 4.4147038432184914e-05,
      "loss": 4.2531,
      "step": 3633
    },
    {
      "epoch": 4.15,
      "grad_norm": 7.864860839854472,
      "learning_rate": 4.4031122035238664e-05,
      "loss": 4.2552,
      "step": 3634
    },
    {
      "epoch": 4.15,
      "grad_norm": 4.732335112526321,
      "learning_rate": 4.391534596515871e-05,
      "loss": 4.125,
      "step": 3635
    },
    {
      "epoch": 4.16,
      "grad_norm": 3.686163598572516,
      "learning_rate": 4.379971028541565e-05,
      "loss": 4.1646,
      "step": 3636
    },
    {
      "epoch": 4.16,
      "grad_norm": 2.7408830581682406,
      "learning_rate": 4.368421505940293e-05,
      "loss": 4.1597,
      "step": 3637
    },
    {
      "epoch": 4.16,
      "grad_norm": 8.371716413049333,
      "learning_rate": 4.3568860350437276e-05,
      "loss": 4.1837,
      "step": 3638
    },
    {
      "epoch": 4.16,
      "grad_norm": 5.308445809471748,
      "learning_rate": 4.345364622175823e-05,
      "loss": 4.0191,
      "step": 3639
    },
    {
      "epoch": 4.16,
      "grad_norm": 6.182580288963607,
      "learning_rate": 4.333857273652844e-05,
      "loss": 4.2567,
      "step": 3640
    },
    {
      "epoch": 4.16,
      "grad_norm": 8.454844929772934,
      "learning_rate": 4.3223639957833175e-05,
      "loss": 4.1818,
      "step": 3641
    },
    {
      "epoch": 4.16,
      "grad_norm": 3.2326631131096333,
      "learning_rate": 4.31088479486807e-05,
      "loss": 4.4526,
      "step": 3642
    },
    {
      "epoch": 4.16,
      "grad_norm": 5.6622054249693035,
      "learning_rate": 4.299419677200221e-05,
      "loss": 4.1159,
      "step": 3643
    },
    {
      "epoch": 4.16,
      "grad_norm": 6.582481844127717,
      "learning_rate": 4.287968649065159e-05,
      "loss": 3.8922,
      "step": 3644
    },
    {
      "epoch": 4.17,
      "grad_norm": 6.120517525425776,
      "learning_rate": 4.276531716740539e-05,
      "loss": 4.1076,
      "step": 3645
    },
    {
      "epoch": 4.17,
      "grad_norm": 3.354250745850937,
      "learning_rate": 4.265108886496299e-05,
      "loss": 3.9887,
      "step": 3646
    },
    {
      "epoch": 4.17,
      "grad_norm": 6.343302935636311,
      "learning_rate": 4.2537001645946566e-05,
      "loss": 4.2847,
      "step": 3647
    },
    {
      "epoch": 4.17,
      "grad_norm": 5.704364224084587,
      "learning_rate": 4.242305557290076e-05,
      "loss": 4.0646,
      "step": 3648
    },
    {
      "epoch": 4.17,
      "grad_norm": 6.401201939345437,
      "learning_rate": 4.230925070829302e-05,
      "loss": 4.1448,
      "step": 3649
    },
    {
      "epoch": 4.17,
      "grad_norm": 3.1569911886799606,
      "learning_rate": 4.219558711451312e-05,
      "loss": 4.0598,
      "step": 3650
    },
    {
      "epoch": 4.17,
      "grad_norm": 5.37262696430018,
      "learning_rate": 4.208206485387364e-05,
      "loss": 3.9892,
      "step": 3651
    },
    {
      "epoch": 4.17,
      "grad_norm": 8.92468002418296,
      "learning_rate": 4.196868398860951e-05,
      "loss": 4.2384,
      "step": 3652
    },
    {
      "epoch": 4.17,
      "grad_norm": 9.768349531865256,
      "learning_rate": 4.1855444580878295e-05,
      "loss": 3.9139,
      "step": 3653
    },
    {
      "epoch": 4.18,
      "grad_norm": 2.4203135655101966,
      "learning_rate": 4.174234669275984e-05,
      "loss": 4.064,
      "step": 3654
    },
    {
      "epoch": 4.18,
      "grad_norm": 5.488266848769192,
      "learning_rate": 4.162939038625659e-05,
      "loss": 3.9972,
      "step": 3655
    },
    {
      "epoch": 4.18,
      "grad_norm": 4.78295241978529,
      "learning_rate": 4.151657572329323e-05,
      "loss": 4.0129,
      "step": 3656
    },
    {
      "epoch": 4.18,
      "grad_norm": 5.263118229560215,
      "learning_rate": 4.14039027657169e-05,
      "loss": 4.2694,
      "step": 3657
    },
    {
      "epoch": 4.18,
      "grad_norm": 2.205895123491167,
      "learning_rate": 4.1291371575296887e-05,
      "loss": 4.2937,
      "step": 3658
    },
    {
      "epoch": 4.18,
      "grad_norm": 2.63810083641438,
      "learning_rate": 4.1178982213724896e-05,
      "loss": 4.4265,
      "step": 3659
    },
    {
      "epoch": 4.18,
      "grad_norm": 3.2736196137722655,
      "learning_rate": 4.106673474261489e-05,
      "loss": 4.3539,
      "step": 3660
    },
    {
      "epoch": 4.18,
      "grad_norm": 3.1748114105906446,
      "learning_rate": 4.095462922350301e-05,
      "loss": 4.1853,
      "step": 3661
    },
    {
      "epoch": 4.19,
      "grad_norm": 2.8014063567239433,
      "learning_rate": 4.084266571784755e-05,
      "loss": 3.9337,
      "step": 3662
    },
    {
      "epoch": 4.19,
      "grad_norm": 3.732077700571058,
      "learning_rate": 4.073084428702899e-05,
      "loss": 4.1371,
      "step": 3663
    },
    {
      "epoch": 4.19,
      "grad_norm": 3.8921418110020287,
      "learning_rate": 4.061916499234986e-05,
      "loss": 4.1039,
      "step": 3664
    },
    {
      "epoch": 4.19,
      "grad_norm": 4.319331091223462,
      "learning_rate": 4.050762789503488e-05,
      "loss": 4.1437,
      "step": 3665
    },
    {
      "epoch": 4.19,
      "grad_norm": 2.4677873087719195,
      "learning_rate": 4.0396233056230765e-05,
      "loss": 4.1445,
      "step": 3666
    },
    {
      "epoch": 4.19,
      "grad_norm": 3.6364411472694735,
      "learning_rate": 4.028498053700607e-05,
      "loss": 4.3646,
      "step": 3667
    },
    {
      "epoch": 4.19,
      "grad_norm": 2.4230227854747644,
      "learning_rate": 4.0173870398351616e-05,
      "loss": 4.1127,
      "step": 3668
    },
    {
      "epoch": 4.19,
      "grad_norm": 4.083442276206653,
      "learning_rate": 4.006290270117996e-05,
      "loss": 4.2346,
      "step": 3669
    },
    {
      "epoch": 4.19,
      "grad_norm": 3.76395978151018,
      "learning_rate": 3.99520775063257e-05,
      "loss": 4.1627,
      "step": 3670
    },
    {
      "epoch": 4.2,
      "grad_norm": 6.133603414028325,
      "learning_rate": 3.984139487454524e-05,
      "loss": 4.3092,
      "step": 3671
    },
    {
      "epoch": 4.2,
      "grad_norm": 2.7708459921013695,
      "learning_rate": 3.973085486651682e-05,
      "loss": 4.1053,
      "step": 3672
    },
    {
      "epoch": 4.2,
      "grad_norm": 2.0740382253985166,
      "learning_rate": 3.9620457542840546e-05,
      "loss": 4.0934,
      "step": 3673
    },
    {
      "epoch": 4.2,
      "grad_norm": 4.445493466660575,
      "learning_rate": 3.9510202964038316e-05,
      "loss": 4.0642,
      "step": 3674
    },
    {
      "epoch": 4.2,
      "grad_norm": 6.326323298806858,
      "learning_rate": 3.94000911905536e-05,
      "loss": 4.0063,
      "step": 3675
    },
    {
      "epoch": 4.2,
      "grad_norm": 5.6882323317049055,
      "learning_rate": 3.929012228275176e-05,
      "loss": 4.355,
      "step": 3676
    },
    {
      "epoch": 4.2,
      "grad_norm": 2.943847960274427,
      "learning_rate": 3.9180296300919775e-05,
      "loss": 4.0447,
      "step": 3677
    },
    {
      "epoch": 4.2,
      "grad_norm": 4.275925193578393,
      "learning_rate": 3.907061330526633e-05,
      "loss": 4.1654,
      "step": 3678
    },
    {
      "epoch": 4.2,
      "grad_norm": 5.155874484271656,
      "learning_rate": 3.896107335592161e-05,
      "loss": 4.1328,
      "step": 3679
    },
    {
      "epoch": 4.21,
      "grad_norm": 5.628279075093127,
      "learning_rate": 3.885167651293746e-05,
      "loss": 4.1944,
      "step": 3680
    },
    {
      "epoch": 4.21,
      "grad_norm": 5.205842245851854,
      "learning_rate": 3.8742422836287254e-05,
      "loss": 3.9919,
      "step": 3681
    },
    {
      "epoch": 4.21,
      "grad_norm": 3.568618505820273,
      "learning_rate": 3.8633312385865935e-05,
      "loss": 3.9594,
      "step": 3682
    },
    {
      "epoch": 4.21,
      "grad_norm": 3.730447177801977,
      "learning_rate": 3.852434522148974e-05,
      "loss": 4.0065,
      "step": 3683
    },
    {
      "epoch": 4.21,
      "grad_norm": 5.287833821358675,
      "learning_rate": 3.841552140289656e-05,
      "loss": 4.1788,
      "step": 3684
    },
    {
      "epoch": 4.21,
      "grad_norm": 2.863253284778557,
      "learning_rate": 3.830684098974556e-05,
      "loss": 4.0146,
      "step": 3685
    },
    {
      "epoch": 4.21,
      "grad_norm": 3.540012932846599,
      "learning_rate": 3.8198304041617386e-05,
      "loss": 4.0053,
      "step": 3686
    },
    {
      "epoch": 4.21,
      "grad_norm": 5.922100358543876,
      "learning_rate": 3.808991061801412e-05,
      "loss": 4.0111,
      "step": 3687
    },
    {
      "epoch": 4.21,
      "grad_norm": 3.7399586712416495,
      "learning_rate": 3.798166077835885e-05,
      "loss": 4.0448,
      "step": 3688
    },
    {
      "epoch": 4.22,
      "grad_norm": 3.604636442640428,
      "learning_rate": 3.78735545819962e-05,
      "loss": 3.9785,
      "step": 3689
    },
    {
      "epoch": 4.22,
      "grad_norm": 3.8329153142536105,
      "learning_rate": 3.776559208819204e-05,
      "loss": 4.1237,
      "step": 3690
    },
    {
      "epoch": 4.22,
      "grad_norm": 3.82260847405284,
      "learning_rate": 3.765777335613339e-05,
      "loss": 3.923,
      "step": 3691
    },
    {
      "epoch": 4.22,
      "grad_norm": 4.560116597937605,
      "learning_rate": 3.75500984449284e-05,
      "loss": 4.1643,
      "step": 3692
    },
    {
      "epoch": 4.22,
      "grad_norm": 5.250877080088921,
      "learning_rate": 3.744256741360638e-05,
      "loss": 3.9902,
      "step": 3693
    },
    {
      "epoch": 4.22,
      "grad_norm": 5.443282896168814,
      "learning_rate": 3.733518032111805e-05,
      "loss": 4.0529,
      "step": 3694
    },
    {
      "epoch": 4.22,
      "grad_norm": 3.1403446913079986,
      "learning_rate": 3.722793722633491e-05,
      "loss": 4.3579,
      "step": 3695
    },
    {
      "epoch": 4.22,
      "grad_norm": 3.9121579359221776,
      "learning_rate": 3.712083818804951e-05,
      "loss": 4.1671,
      "step": 3696
    },
    {
      "epoch": 4.23,
      "grad_norm": 5.068376304786552,
      "learning_rate": 3.70138832649756e-05,
      "loss": 3.9384,
      "step": 3697
    },
    {
      "epoch": 4.23,
      "grad_norm": 5.419705714112912,
      "learning_rate": 3.690707251574781e-05,
      "loss": 3.9901,
      "step": 3698
    },
    {
      "epoch": 4.23,
      "grad_norm": 2.984841634948577,
      "learning_rate": 3.680040599892189e-05,
      "loss": 4.0834,
      "step": 3699
    },
    {
      "epoch": 4.23,
      "grad_norm": 3.2984953665908674,
      "learning_rate": 3.6693883772974144e-05,
      "loss": 4.1972,
      "step": 3700
    },
    {
      "epoch": 4.23,
      "eval_blimp_filtered_avg": 0.5216417910447761,
      "eval_blimp_filtered_std": 0.005742395525976493,
      "step": 3700
    },
    {
      "epoch": 4.23,
      "eval_blimp_supplement_avg": 0.4676724137931034,
      "eval_blimp_supplement_std": 0.02280067506813048,
      "step": 3700
    },
    {
      "epoch": 4.23,
      "eval_vqa_filtered_avg": 0.36,
      "eval_vqa_filtered_std": 0.04824181513244218,
      "step": 3700
    },
    {
      "epoch": 4.23,
      "eval_winoground_filtered_avg": 0.48,
      "eval_winoground_filtered_std": 0.05021167315686779,
      "step": 3700
    },
    {
      "epoch": 4.23,
      "grad_norm": 2.7526079330224382,
      "learning_rate": 3.6587505896302227e-05,
      "loss": 4.008,
      "step": 3701
    },
    {
      "epoch": 4.23,
      "grad_norm": 4.090880567337932,
      "learning_rate": 3.6481272427224393e-05,
      "loss": 4.0863,
      "step": 3702
    },
    {
      "epoch": 4.23,
      "grad_norm": 4.035055688477091,
      "learning_rate": 3.637518342397979e-05,
      "loss": 4.0816,
      "step": 3703
    },
    {
      "epoch": 4.23,
      "grad_norm": 4.21178928416299,
      "learning_rate": 3.6269238944728466e-05,
      "loss": 4.1727,
      "step": 3704
    },
    {
      "epoch": 4.23,
      "grad_norm": 6.540815260444656,
      "learning_rate": 3.6163439047550946e-05,
      "loss": 3.9811,
      "step": 3705
    },
    {
      "epoch": 4.24,
      "grad_norm": 2.628634926556877,
      "learning_rate": 3.6057783790448814e-05,
      "loss": 4.1279,
      "step": 3706
    },
    {
      "epoch": 4.24,
      "grad_norm": 2.4956472283533713,
      "learning_rate": 3.595227323134423e-05,
      "loss": 3.9846,
      "step": 3707
    },
    {
      "epoch": 4.24,
      "grad_norm": 4.577987377296766,
      "learning_rate": 3.584690742808e-05,
      "loss": 4.0898,
      "step": 3708
    },
    {
      "epoch": 4.24,
      "grad_norm": 3.8111908989259886,
      "learning_rate": 3.5741686438419604e-05,
      "loss": 4.1382,
      "step": 3709
    },
    {
      "epoch": 4.24,
      "grad_norm": 5.582132003604005,
      "learning_rate": 3.563661032004714e-05,
      "loss": 3.978,
      "step": 3710
    },
    {
      "epoch": 4.24,
      "grad_norm": 3.4143706995306538,
      "learning_rate": 3.55316791305673e-05,
      "loss": 4.0962,
      "step": 3711
    },
    {
      "epoch": 4.24,
      "grad_norm": 3.9935429504166398,
      "learning_rate": 3.5426892927505335e-05,
      "loss": 4.068,
      "step": 3712
    },
    {
      "epoch": 4.24,
      "grad_norm": 3.0170125351571455,
      "learning_rate": 3.532225176830685e-05,
      "loss": 4.1724,
      "step": 3713
    },
    {
      "epoch": 4.24,
      "grad_norm": 4.967612913607452,
      "learning_rate": 3.521775571033817e-05,
      "loss": 4.1945,
      "step": 3714
    },
    {
      "epoch": 4.25,
      "grad_norm": 4.635174250852855,
      "learning_rate": 3.511340481088589e-05,
      "loss": 4.136,
      "step": 3715
    },
    {
      "epoch": 4.25,
      "grad_norm": 2.6502685813190148,
      "learning_rate": 3.500919912715716e-05,
      "loss": 4.1644,
      "step": 3716
    },
    {
      "epoch": 4.25,
      "grad_norm": 3.3716035684295087,
      "learning_rate": 3.4905138716279415e-05,
      "loss": 4.2265,
      "step": 3717
    },
    {
      "epoch": 4.25,
      "grad_norm": 3.2420457368299807,
      "learning_rate": 3.4801223635300546e-05,
      "loss": 4.3306,
      "step": 3718
    },
    {
      "epoch": 4.25,
      "grad_norm": 2.907021711673663,
      "learning_rate": 3.469745394118866e-05,
      "loss": 4.0928,
      "step": 3719
    },
    {
      "epoch": 4.25,
      "grad_norm": 5.26213464272231,
      "learning_rate": 3.459382969083227e-05,
      "loss": 3.9577,
      "step": 3720
    },
    {
      "epoch": 4.25,
      "grad_norm": 4.495568705430152,
      "learning_rate": 3.449035094104012e-05,
      "loss": 4.1986,
      "step": 3721
    },
    {
      "epoch": 4.25,
      "grad_norm": 4.58724297607965,
      "learning_rate": 3.438701774854109e-05,
      "loss": 4.0034,
      "step": 3722
    },
    {
      "epoch": 4.25,
      "grad_norm": 3.989978937260802,
      "learning_rate": 3.4283830169984395e-05,
      "loss": 4.1339,
      "step": 3723
    },
    {
      "epoch": 4.26,
      "grad_norm": 3.2465861828309355,
      "learning_rate": 3.4180788261939386e-05,
      "loss": 4.0685,
      "step": 3724
    },
    {
      "epoch": 4.26,
      "grad_norm": 4.35157579539799,
      "learning_rate": 3.4077892080895474e-05,
      "loss": 3.9266,
      "step": 3725
    },
    {
      "epoch": 4.26,
      "grad_norm": 4.0691757092400636,
      "learning_rate": 3.397514168326231e-05,
      "loss": 4.2285,
      "step": 3726
    },
    {
      "epoch": 4.26,
      "grad_norm": 4.8986787404401495,
      "learning_rate": 3.3872537125369556e-05,
      "loss": 4.0275,
      "step": 3727
    },
    {
      "epoch": 4.26,
      "grad_norm": 3.7319715601089967,
      "learning_rate": 3.37700784634669e-05,
      "loss": 4.0674,
      "step": 3728
    },
    {
      "epoch": 4.26,
      "grad_norm": 3.2988017560727783,
      "learning_rate": 3.3667765753724164e-05,
      "loss": 4.138,
      "step": 3729
    },
    {
      "epoch": 4.26,
      "grad_norm": 3.827153042688276,
      "learning_rate": 3.356559905223092e-05,
      "loss": 4.1378,
      "step": 3730
    },
    {
      "epoch": 4.26,
      "grad_norm": 3.939090366109162,
      "learning_rate": 3.346357841499686e-05,
      "loss": 4.2655,
      "step": 3731
    },
    {
      "epoch": 4.27,
      "grad_norm": 4.017774049601149,
      "learning_rate": 3.336170389795169e-05,
      "loss": 4.1322,
      "step": 3732
    },
    {
      "epoch": 4.27,
      "grad_norm": 3.1147577234720685,
      "learning_rate": 3.3259975556944785e-05,
      "loss": 3.9091,
      "step": 3733
    },
    {
      "epoch": 4.27,
      "grad_norm": 3.5275982207146868,
      "learning_rate": 3.315839344774558e-05,
      "loss": 4.0453,
      "step": 3734
    },
    {
      "epoch": 4.27,
      "grad_norm": 4.952638702099205,
      "learning_rate": 3.305695762604326e-05,
      "loss": 4.1446,
      "step": 3735
    },
    {
      "epoch": 4.27,
      "grad_norm": 4.884078842353125,
      "learning_rate": 3.2955668147446765e-05,
      "loss": 4.0492,
      "step": 3736
    },
    {
      "epoch": 4.27,
      "grad_norm": 4.091707525966017,
      "learning_rate": 3.2854525067484894e-05,
      "loss": 4.2128,
      "step": 3737
    },
    {
      "epoch": 4.27,
      "grad_norm": 2.515506749746765,
      "learning_rate": 3.275352844160621e-05,
      "loss": 4.0877,
      "step": 3738
    },
    {
      "epoch": 4.27,
      "grad_norm": 7.483745604250074,
      "learning_rate": 3.2652678325178774e-05,
      "loss": 4.0434,
      "step": 3739
    },
    {
      "epoch": 4.27,
      "grad_norm": 4.007635169698222,
      "learning_rate": 3.255197477349054e-05,
      "loss": 4.2162,
      "step": 3740
    },
    {
      "epoch": 4.28,
      "grad_norm": 4.798756583783412,
      "learning_rate": 3.245141784174913e-05,
      "loss": 4.1417,
      "step": 3741
    },
    {
      "epoch": 4.28,
      "grad_norm": 3.580542706622596,
      "learning_rate": 3.2351007585081735e-05,
      "loss": 4.0517,
      "step": 3742
    },
    {
      "epoch": 4.28,
      "grad_norm": 4.361008295125797,
      "learning_rate": 3.225074405853499e-05,
      "loss": 3.9871,
      "step": 3743
    },
    {
      "epoch": 4.28,
      "grad_norm": 3.1100676214393204,
      "learning_rate": 3.215062731707524e-05,
      "loss": 4.2718,
      "step": 3744
    },
    {
      "epoch": 4.28,
      "grad_norm": 5.751638130430259,
      "learning_rate": 3.20506574155884e-05,
      "loss": 3.9137,
      "step": 3745
    },
    {
      "epoch": 4.28,
      "grad_norm": 2.3515136844064055,
      "learning_rate": 3.19508344088798e-05,
      "loss": 4.0942,
      "step": 3746
    },
    {
      "epoch": 4.28,
      "grad_norm": 5.397947829425868,
      "learning_rate": 3.18511583516741e-05,
      "loss": 4.2324,
      "step": 3747
    },
    {
      "epoch": 4.28,
      "grad_norm": 3.8921474048531586,
      "learning_rate": 3.175162929861579e-05,
      "loss": 4.0793,
      "step": 3748
    },
    {
      "epoch": 4.28,
      "grad_norm": 4.447867557357014,
      "learning_rate": 3.165224730426839e-05,
      "loss": 4.0637,
      "step": 3749
    },
    {
      "epoch": 4.29,
      "grad_norm": 5.527495024213583,
      "learning_rate": 3.155301242311504e-05,
      "loss": 4.0633,
      "step": 3750
    },
    {
      "epoch": 4.29,
      "grad_norm": 3.1689117324916527,
      "learning_rate": 3.145392470955804e-05,
      "loss": 4.0836,
      "step": 3751
    },
    {
      "epoch": 4.29,
      "grad_norm": 4.497776433030314,
      "learning_rate": 3.1354984217919165e-05,
      "loss": 3.9597,
      "step": 3752
    },
    {
      "epoch": 4.29,
      "grad_norm": 4.116794843385568,
      "learning_rate": 3.1256191002439356e-05,
      "loss": 3.94,
      "step": 3753
    },
    {
      "epoch": 4.29,
      "grad_norm": 4.893382217233808,
      "learning_rate": 3.115754511727896e-05,
      "loss": 3.9929,
      "step": 3754
    },
    {
      "epoch": 4.29,
      "grad_norm": 4.537792445403418,
      "learning_rate": 3.105904661651741e-05,
      "loss": 4.2817,
      "step": 3755
    },
    {
      "epoch": 4.29,
      "grad_norm": 3.891264404204084,
      "learning_rate": 3.0960695554153426e-05,
      "loss": 4.1314,
      "step": 3756
    },
    {
      "epoch": 4.29,
      "grad_norm": 3.302319126658895,
      "learning_rate": 3.086249198410487e-05,
      "loss": 3.9794,
      "step": 3757
    },
    {
      "epoch": 4.29,
      "grad_norm": 6.0110122041397185,
      "learning_rate": 3.076443596020877e-05,
      "loss": 4.1296,
      "step": 3758
    },
    {
      "epoch": 4.3,
      "grad_norm": 5.411576387974142,
      "learning_rate": 3.06665275362213e-05,
      "loss": 4.1065,
      "step": 3759
    },
    {
      "epoch": 4.3,
      "grad_norm": 2.6826169263789414,
      "learning_rate": 3.0568766765817556e-05,
      "loss": 4.1219,
      "step": 3760
    },
    {
      "epoch": 4.3,
      "grad_norm": 10.468453446906986,
      "learning_rate": 3.0471153702591833e-05,
      "loss": 3.9644,
      "step": 3761
    },
    {
      "epoch": 4.3,
      "grad_norm": 2.965988932966465,
      "learning_rate": 3.037368840005746e-05,
      "loss": 3.9549,
      "step": 3762
    },
    {
      "epoch": 4.3,
      "grad_norm": 4.041988346680554,
      "learning_rate": 3.0276370911646665e-05,
      "loss": 4.216,
      "step": 3763
    },
    {
      "epoch": 4.3,
      "grad_norm": 6.7040440028784865,
      "learning_rate": 3.0179201290710696e-05,
      "loss": 4.0198,
      "step": 3764
    },
    {
      "epoch": 4.3,
      "grad_norm": 2.806691721193337,
      "learning_rate": 3.0082179590519784e-05,
      "loss": 4.0215,
      "step": 3765
    },
    {
      "epoch": 4.3,
      "grad_norm": 2.831086579574943,
      "learning_rate": 2.9985305864262942e-05,
      "loss": 4.1985,
      "step": 3766
    },
    {
      "epoch": 4.31,
      "grad_norm": 3.8080458885178774,
      "learning_rate": 2.988858016504826e-05,
      "loss": 3.9085,
      "step": 3767
    },
    {
      "epoch": 4.31,
      "grad_norm": 4.363826542631269,
      "learning_rate": 2.9792002545902415e-05,
      "loss": 4.0654,
      "step": 3768
    },
    {
      "epoch": 4.31,
      "grad_norm": 3.2087842588090716,
      "learning_rate": 2.9695573059771127e-05,
      "loss": 4.1361,
      "step": 3769
    },
    {
      "epoch": 4.31,
      "grad_norm": 5.534174294269441,
      "learning_rate": 2.9599291759518738e-05,
      "loss": 4.0574,
      "step": 3770
    },
    {
      "epoch": 4.31,
      "grad_norm": 3.947671130679422,
      "learning_rate": 2.9503158697928532e-05,
      "loss": 4.1496,
      "step": 3771
    },
    {
      "epoch": 4.31,
      "grad_norm": 6.107559652326083,
      "learning_rate": 2.9407173927702344e-05,
      "loss": 4.1242,
      "step": 3772
    },
    {
      "epoch": 4.31,
      "grad_norm": 4.533618936030546,
      "learning_rate": 2.931133750146085e-05,
      "loss": 3.9442,
      "step": 3773
    },
    {
      "epoch": 4.31,
      "grad_norm": 7.433255515494958,
      "learning_rate": 2.9215649471743364e-05,
      "loss": 4.1676,
      "step": 3774
    },
    {
      "epoch": 4.31,
      "grad_norm": 3.3608388532613396,
      "learning_rate": 2.912010989100778e-05,
      "loss": 4.032,
      "step": 3775
    },
    {
      "epoch": 4.32,
      "grad_norm": 4.493878294886374,
      "learning_rate": 2.9024718811630744e-05,
      "loss": 4.1605,
      "step": 3776
    },
    {
      "epoch": 4.32,
      "grad_norm": 5.934285836371989,
      "learning_rate": 2.8929476285907317e-05,
      "loss": 4.1568,
      "step": 3777
    },
    {
      "epoch": 4.32,
      "grad_norm": 5.215548634543467,
      "learning_rate": 2.88343823660512e-05,
      "loss": 3.8765,
      "step": 3778
    },
    {
      "epoch": 4.32,
      "grad_norm": 13.316160606592792,
      "learning_rate": 2.8739437104194664e-05,
      "loss": 4.0842,
      "step": 3779
    },
    {
      "epoch": 4.32,
      "grad_norm": 3.254142631962817,
      "learning_rate": 2.8644640552388565e-05,
      "loss": 4.2111,
      "step": 3780
    },
    {
      "epoch": 4.32,
      "grad_norm": 3.489192298091436,
      "learning_rate": 2.854999276260197e-05,
      "loss": 4.2083,
      "step": 3781
    },
    {
      "epoch": 4.32,
      "grad_norm": 6.319543110610224,
      "learning_rate": 2.845549378672257e-05,
      "loss": 4.0847,
      "step": 3782
    },
    {
      "epoch": 4.32,
      "grad_norm": 6.885325918411502,
      "learning_rate": 2.836114367655653e-05,
      "loss": 4.3508,
      "step": 3783
    },
    {
      "epoch": 4.32,
      "grad_norm": 5.5876465695815,
      "learning_rate": 2.8266942483828303e-05,
      "loss": 4.1069,
      "step": 3784
    },
    {
      "epoch": 4.33,
      "grad_norm": 4.7869598042959955,
      "learning_rate": 2.817289026018068e-05,
      "loss": 4.0376,
      "step": 3785
    },
    {
      "epoch": 4.33,
      "grad_norm": 3.988014212724107,
      "learning_rate": 2.8078987057174783e-05,
      "loss": 3.9891,
      "step": 3786
    },
    {
      "epoch": 4.33,
      "grad_norm": 4.069915436113912,
      "learning_rate": 2.7985232926290134e-05,
      "loss": 4.265,
      "step": 3787
    },
    {
      "epoch": 4.33,
      "grad_norm": 3.690971499610017,
      "learning_rate": 2.7891627918924587e-05,
      "loss": 4.1365,
      "step": 3788
    },
    {
      "epoch": 4.33,
      "grad_norm": 7.3708266141108965,
      "learning_rate": 2.7798172086393957e-05,
      "loss": 3.9809,
      "step": 3789
    },
    {
      "epoch": 4.33,
      "grad_norm": 2.165809087849634,
      "learning_rate": 2.7704865479932558e-05,
      "loss": 4.1794,
      "step": 3790
    },
    {
      "epoch": 4.33,
      "grad_norm": 3.3979500301935404,
      "learning_rate": 2.7611708150692745e-05,
      "loss": 4.0439,
      "step": 3791
    },
    {
      "epoch": 4.33,
      "grad_norm": 4.24502569473556,
      "learning_rate": 2.7518700149745133e-05,
      "loss": 4.0118,
      "step": 3792
    },
    {
      "epoch": 4.33,
      "grad_norm": 6.597760823323774,
      "learning_rate": 2.7425841528078462e-05,
      "loss": 4.0947,
      "step": 3793
    },
    {
      "epoch": 4.34,
      "grad_norm": 2.730048463899977,
      "learning_rate": 2.733313233659935e-05,
      "loss": 4.013,
      "step": 3794
    },
    {
      "epoch": 4.34,
      "grad_norm": 5.367709487329057,
      "learning_rate": 2.7240572626132906e-05,
      "loss": 4.2193,
      "step": 3795
    },
    {
      "epoch": 4.34,
      "grad_norm": 4.331914344538612,
      "learning_rate": 2.7148162447422005e-05,
      "loss": 4.2947,
      "step": 3796
    },
    {
      "epoch": 4.34,
      "grad_norm": 3.5072764017284928,
      "learning_rate": 2.7055901851127625e-05,
      "loss": 4.1933,
      "step": 3797
    },
    {
      "epoch": 4.34,
      "grad_norm": 5.724973039135592,
      "learning_rate": 2.6963790887828673e-05,
      "loss": 4.1064,
      "step": 3798
    },
    {
      "epoch": 4.34,
      "grad_norm": 5.148735889942877,
      "learning_rate": 2.6871829608022088e-05,
      "loss": 3.942,
      "step": 3799
    },
    {
      "epoch": 4.34,
      "grad_norm": 4.0269166696466305,
      "learning_rate": 2.678001806212281e-05,
      "loss": 4.1367,
      "step": 3800
    },
    {
      "epoch": 4.34,
      "eval_blimp_filtered_avg": 0.5238805970149254,
      "eval_blimp_filtered_std": 0.00569646309909148,
      "step": 3800
    },
    {
      "epoch": 4.34,
      "eval_blimp_supplement_avg": 0.49137931034482757,
      "eval_blimp_supplement_std": 0.02270775243605513,
      "step": 3800
    },
    {
      "epoch": 4.34,
      "eval_vqa_filtered_avg": 0.37,
      "eval_vqa_filtered_std": 0.048523658709391,
      "step": 3800
    },
    {
      "epoch": 4.34,
      "eval_winoground_filtered_avg": 0.51,
      "eval_winoground_filtered_std": 0.05024183937956912,
      "step": 3800
    },
    {
      "epoch": 4.34,
      "grad_norm": 4.7550151190419125,
      "learning_rate": 2.6688356300463576e-05,
      "loss": 4.07,
      "step": 3801
    },
    {
      "epoch": 4.35,
      "grad_norm": 5.8967227015819,
      "learning_rate": 2.659684437329499e-05,
      "loss": 4.1135,
      "step": 3802
    },
    {
      "epoch": 4.35,
      "grad_norm": 20.765623988872903,
      "learning_rate": 2.6505482330785688e-05,
      "loss": 4.0943,
      "step": 3803
    },
    {
      "epoch": 4.35,
      "grad_norm": 5.008706615866194,
      "learning_rate": 2.6414270223022007e-05,
      "loss": 4.1475,
      "step": 3804
    },
    {
      "epoch": 4.35,
      "grad_norm": 3.8910604432650433,
      "learning_rate": 2.632320810000814e-05,
      "loss": 4.256,
      "step": 3805
    },
    {
      "epoch": 4.35,
      "grad_norm": 6.890204550357026,
      "learning_rate": 2.6232296011665933e-05,
      "loss": 4.2697,
      "step": 3806
    },
    {
      "epoch": 4.35,
      "grad_norm": 3.2895377547913154,
      "learning_rate": 2.614153400783511e-05,
      "loss": 4.0574,
      "step": 3807
    },
    {
      "epoch": 4.35,
      "grad_norm": 24.293456052943537,
      "learning_rate": 2.605092213827307e-05,
      "loss": 3.9068,
      "step": 3808
    },
    {
      "epoch": 4.35,
      "grad_norm": 6.945292892163397,
      "learning_rate": 2.596046045265494e-05,
      "loss": 4.2262,
      "step": 3809
    },
    {
      "epoch": 4.35,
      "grad_norm": 3.9698742558846565,
      "learning_rate": 2.587014900057349e-05,
      "loss": 4.0786,
      "step": 3810
    },
    {
      "epoch": 4.36,
      "grad_norm": 3.2519604057316114,
      "learning_rate": 2.5779987831539083e-05,
      "loss": 4.0985,
      "step": 3811
    },
    {
      "epoch": 4.36,
      "grad_norm": 4.938974606795237,
      "learning_rate": 2.5689976994979812e-05,
      "loss": 4.1996,
      "step": 3812
    },
    {
      "epoch": 4.36,
      "grad_norm": 4.4893931064334565,
      "learning_rate": 2.56001165402412e-05,
      "loss": 4.0054,
      "step": 3813
    },
    {
      "epoch": 4.36,
      "grad_norm": 3.354378933825482,
      "learning_rate": 2.551040651658648e-05,
      "loss": 3.9906,
      "step": 3814
    },
    {
      "epoch": 4.36,
      "grad_norm": 4.123552870622024,
      "learning_rate": 2.5420846973196273e-05,
      "loss": 4.057,
      "step": 3815
    },
    {
      "epoch": 4.36,
      "grad_norm": 5.37460518893227,
      "learning_rate": 2.533143795916881e-05,
      "loss": 4.3089,
      "step": 3816
    },
    {
      "epoch": 4.36,
      "grad_norm": 4.293479156755431,
      "learning_rate": 2.5242179523519744e-05,
      "loss": 4.0919,
      "step": 3817
    },
    {
      "epoch": 4.36,
      "grad_norm": 3.8544130527148943,
      "learning_rate": 2.5153071715182205e-05,
      "loss": 4.1266,
      "step": 3818
    },
    {
      "epoch": 4.36,
      "grad_norm": 6.56426917808491,
      "learning_rate": 2.506411458300678e-05,
      "loss": 4.085,
      "step": 3819
    },
    {
      "epoch": 4.37,
      "grad_norm": 4.1210491802769,
      "learning_rate": 2.4975308175761356e-05,
      "loss": 4.0192,
      "step": 3820
    },
    {
      "epoch": 4.37,
      "grad_norm": 2.9384723610382792,
      "learning_rate": 2.488665254213129e-05,
      "loss": 4.3127,
      "step": 3821
    },
    {
      "epoch": 4.37,
      "grad_norm": 3.7590516793194664,
      "learning_rate": 2.479814773071924e-05,
      "loss": 4.0799,
      "step": 3822
    },
    {
      "epoch": 4.37,
      "grad_norm": 2.420579156358614,
      "learning_rate": 2.470979379004513e-05,
      "loss": 3.9447,
      "step": 3823
    },
    {
      "epoch": 4.37,
      "grad_norm": 8.877477216620878,
      "learning_rate": 2.4621590768546238e-05,
      "loss": 4.1021,
      "step": 3824
    },
    {
      "epoch": 4.37,
      "grad_norm": 4.164044314940858,
      "learning_rate": 2.4533538714577095e-05,
      "loss": 4.2391,
      "step": 3825
    },
    {
      "epoch": 4.37,
      "grad_norm": 6.4109933695554595,
      "learning_rate": 2.4445637676409412e-05,
      "loss": 4.1356,
      "step": 3826
    },
    {
      "epoch": 4.37,
      "grad_norm": 4.381753132626693,
      "learning_rate": 2.4357887702232347e-05,
      "loss": 4.2066,
      "step": 3827
    },
    {
      "epoch": 4.37,
      "grad_norm": 2.1834975332567375,
      "learning_rate": 2.4270288840151852e-05,
      "loss": 4.3794,
      "step": 3828
    },
    {
      "epoch": 4.38,
      "grad_norm": 3.2518974572145205,
      "learning_rate": 2.418284113819131e-05,
      "loss": 4.2172,
      "step": 3829
    },
    {
      "epoch": 4.38,
      "grad_norm": 3.5774363515226115,
      "learning_rate": 2.4095544644291177e-05,
      "loss": 4.2209,
      "step": 3830
    },
    {
      "epoch": 4.38,
      "grad_norm": 3.1999596120992013,
      "learning_rate": 2.4008399406309063e-05,
      "loss": 4.1799,
      "step": 3831
    },
    {
      "epoch": 4.38,
      "grad_norm": 3.124570791711505,
      "learning_rate": 2.3921405472019462e-05,
      "loss": 4.3332,
      "step": 3832
    },
    {
      "epoch": 4.38,
      "grad_norm": 3.1565157956334753,
      "learning_rate": 2.383456288911415e-05,
      "loss": 4.215,
      "step": 3833
    },
    {
      "epoch": 4.38,
      "grad_norm": 3.4734008546167416,
      "learning_rate": 2.3747871705201716e-05,
      "loss": 4.1438,
      "step": 3834
    },
    {
      "epoch": 4.38,
      "grad_norm": 7.566821586523008,
      "learning_rate": 2.3661331967808107e-05,
      "loss": 4.1506,
      "step": 3835
    },
    {
      "epoch": 4.38,
      "grad_norm": 3.269024291708266,
      "learning_rate": 2.357494372437577e-05,
      "loss": 4.3316,
      "step": 3836
    },
    {
      "epoch": 4.39,
      "grad_norm": 2.6711250873853905,
      "learning_rate": 2.3488707022264464e-05,
      "loss": 4.0621,
      "step": 3837
    },
    {
      "epoch": 4.39,
      "grad_norm": 4.40859373247557,
      "learning_rate": 2.3402621908750694e-05,
      "loss": 4.1106,
      "step": 3838
    },
    {
      "epoch": 4.39,
      "grad_norm": 4.170365817322545,
      "learning_rate": 2.3316688431027952e-05,
      "loss": 3.9865,
      "step": 3839
    },
    {
      "epoch": 4.39,
      "grad_norm": 6.189474034725737,
      "learning_rate": 2.3230906636206482e-05,
      "loss": 4.081,
      "step": 3840
    },
    {
      "epoch": 4.39,
      "grad_norm": 3.812676991876167,
      "learning_rate": 2.314527657131343e-05,
      "loss": 4.0213,
      "step": 3841
    },
    {
      "epoch": 4.39,
      "grad_norm": 5.425686111449401,
      "learning_rate": 2.3059798283292887e-05,
      "loss": 4.023,
      "step": 3842
    },
    {
      "epoch": 4.39,
      "grad_norm": 6.398138794209846,
      "learning_rate": 2.297447181900559e-05,
      "loss": 4.071,
      "step": 3843
    },
    {
      "epoch": 4.39,
      "grad_norm": 3.1383567278285667,
      "learning_rate": 2.288929722522913e-05,
      "loss": 4.1399,
      "step": 3844
    },
    {
      "epoch": 4.39,
      "grad_norm": 6.609205842692176,
      "learning_rate": 2.2804274548657687e-05,
      "loss": 3.9689,
      "step": 3845
    },
    {
      "epoch": 4.4,
      "grad_norm": 3.390935025418311,
      "learning_rate": 2.2719403835902307e-05,
      "loss": 4.0398,
      "step": 3846
    },
    {
      "epoch": 4.4,
      "grad_norm": 3.5678875269113712,
      "learning_rate": 2.2634685133490672e-05,
      "loss": 4.0065,
      "step": 3847
    },
    {
      "epoch": 4.4,
      "grad_norm": 4.000732662092567,
      "learning_rate": 2.25501184878672e-05,
      "loss": 4.1565,
      "step": 3848
    },
    {
      "epoch": 4.4,
      "grad_norm": 3.3728431842826514,
      "learning_rate": 2.2465703945392797e-05,
      "loss": 4.2647,
      "step": 3849
    },
    {
      "epoch": 4.4,
      "grad_norm": 4.700006963145578,
      "learning_rate": 2.2381441552345124e-05,
      "loss": 4.2847,
      "step": 3850
    },
    {
      "epoch": 4.4,
      "grad_norm": 3.7698096040659426,
      "learning_rate": 2.229733135491839e-05,
      "loss": 4.1292,
      "step": 3851
    },
    {
      "epoch": 4.4,
      "grad_norm": 3.815575865875677,
      "learning_rate": 2.2213373399223378e-05,
      "loss": 4.0985,
      "step": 3852
    },
    {
      "epoch": 4.4,
      "grad_norm": 4.642830124100103,
      "learning_rate": 2.2129567731287334e-05,
      "loss": 4.1428,
      "step": 3853
    },
    {
      "epoch": 4.4,
      "grad_norm": 3.211811374695382,
      "learning_rate": 2.2045914397054077e-05,
      "loss": 4.2421,
      "step": 3854
    },
    {
      "epoch": 4.41,
      "grad_norm": 3.3386146964600076,
      "learning_rate": 2.196241344238392e-05,
      "loss": 4.0981,
      "step": 3855
    },
    {
      "epoch": 4.41,
      "grad_norm": 5.274851722822476,
      "learning_rate": 2.1879064913053646e-05,
      "loss": 4.2274,
      "step": 3856
    },
    {
      "epoch": 4.41,
      "grad_norm": 3.404555699431231,
      "learning_rate": 2.179586885475647e-05,
      "loss": 4.0496,
      "step": 3857
    },
    {
      "epoch": 4.41,
      "grad_norm": 7.233665538698727,
      "learning_rate": 2.1712825313101946e-05,
      "loss": 4.0601,
      "step": 3858
    },
    {
      "epoch": 4.41,
      "grad_norm": 3.9364645522227626,
      "learning_rate": 2.162993433361615e-05,
      "loss": 4.0828,
      "step": 3859
    },
    {
      "epoch": 4.41,
      "grad_norm": 4.970550785926615,
      "learning_rate": 2.154719596174147e-05,
      "loss": 4.125,
      "step": 3860
    },
    {
      "epoch": 4.41,
      "grad_norm": 4.30931270122184,
      "learning_rate": 2.1464610242836487e-05,
      "loss": 4.3251,
      "step": 3861
    },
    {
      "epoch": 4.41,
      "grad_norm": 11.205974375714858,
      "learning_rate": 2.138217722217629e-05,
      "loss": 4.2038,
      "step": 3862
    },
    {
      "epoch": 4.41,
      "grad_norm": 3.812922715254097,
      "learning_rate": 2.1299896944952156e-05,
      "loss": 4.2154,
      "step": 3863
    },
    {
      "epoch": 4.42,
      "grad_norm": 4.20413541849482,
      "learning_rate": 2.1217769456271706e-05,
      "loss": 4.149,
      "step": 3864
    },
    {
      "epoch": 4.42,
      "grad_norm": 5.742868949595403,
      "learning_rate": 2.1135794801158722e-05,
      "loss": 4.0124,
      "step": 3865
    },
    {
      "epoch": 4.42,
      "grad_norm": 3.3863844613197833,
      "learning_rate": 2.1053973024553193e-05,
      "loss": 4.1052,
      "step": 3866
    },
    {
      "epoch": 4.42,
      "grad_norm": 4.9377932511227876,
      "learning_rate": 2.097230417131137e-05,
      "loss": 4.0703,
      "step": 3867
    },
    {
      "epoch": 4.42,
      "grad_norm": 4.121486202712329,
      "learning_rate": 2.0890788286205606e-05,
      "loss": 4.145,
      "step": 3868
    },
    {
      "epoch": 4.42,
      "grad_norm": 4.412624094734753,
      "learning_rate": 2.0809425413924452e-05,
      "loss": 4.0887,
      "step": 3869
    },
    {
      "epoch": 4.42,
      "grad_norm": 4.472232765225054,
      "learning_rate": 2.072821559907246e-05,
      "loss": 3.9357,
      "step": 3870
    },
    {
      "epoch": 4.42,
      "grad_norm": 4.50211320367663,
      "learning_rate": 2.064715888617039e-05,
      "loss": 4.1617,
      "step": 3871
    },
    {
      "epoch": 4.43,
      "grad_norm": 4.035795294862242,
      "learning_rate": 2.0566255319655056e-05,
      "loss": 4.1874,
      "step": 3872
    },
    {
      "epoch": 4.43,
      "grad_norm": 2.7987871680955223,
      "learning_rate": 2.048550494387925e-05,
      "loss": 4.0861,
      "step": 3873
    },
    {
      "epoch": 4.43,
      "grad_norm": 5.454777599491666,
      "learning_rate": 2.0404907803111827e-05,
      "loss": 4.0772,
      "step": 3874
    },
    {
      "epoch": 4.43,
      "grad_norm": 7.163448286190371,
      "learning_rate": 2.032446394153764e-05,
      "loss": 3.9419,
      "step": 3875
    },
    {
      "epoch": 4.43,
      "grad_norm": 3.669604924607669,
      "learning_rate": 2.024417340325748e-05,
      "loss": 4.1684,
      "step": 3876
    },
    {
      "epoch": 4.43,
      "grad_norm": 3.4064725605343105,
      "learning_rate": 2.0164036232288205e-05,
      "loss": 4.1339,
      "step": 3877
    },
    {
      "epoch": 4.43,
      "grad_norm": 4.257421647330552,
      "learning_rate": 2.0084052472562305e-05,
      "loss": 4.2213,
      "step": 3878
    },
    {
      "epoch": 4.43,
      "grad_norm": 2.9029162854211825,
      "learning_rate": 2.0004222167928474e-05,
      "loss": 3.9255,
      "step": 3879
    },
    {
      "epoch": 4.43,
      "grad_norm": 3.7620326421586947,
      "learning_rate": 1.9924545362151134e-05,
      "loss": 4.1038,
      "step": 3880
    },
    {
      "epoch": 4.44,
      "grad_norm": 5.058812665617539,
      "learning_rate": 1.9845022098910545e-05,
      "loss": 4.0792,
      "step": 3881
    },
    {
      "epoch": 4.44,
      "grad_norm": 3.644749911471363,
      "learning_rate": 1.9765652421802968e-05,
      "loss": 4.0537,
      "step": 3882
    },
    {
      "epoch": 4.44,
      "grad_norm": 3.4658443829354626,
      "learning_rate": 1.9686436374340197e-05,
      "loss": 4.0949,
      "step": 3883
    },
    {
      "epoch": 4.44,
      "grad_norm": 3.5054732597974407,
      "learning_rate": 1.9607373999949927e-05,
      "loss": 4.0198,
      "step": 3884
    },
    {
      "epoch": 4.44,
      "grad_norm": 4.177892876419979,
      "learning_rate": 1.9528465341975653e-05,
      "loss": 4.0142,
      "step": 3885
    },
    {
      "epoch": 4.44,
      "grad_norm": 3.2291717176165857,
      "learning_rate": 1.94497104436766e-05,
      "loss": 3.8081,
      "step": 3886
    },
    {
      "epoch": 4.44,
      "grad_norm": 3.8489401845469153,
      "learning_rate": 1.9371109348227577e-05,
      "loss": 4.048,
      "step": 3887
    },
    {
      "epoch": 4.44,
      "grad_norm": 3.9313916773883264,
      "learning_rate": 1.9292662098719137e-05,
      "loss": 4.1949,
      "step": 3888
    },
    {
      "epoch": 4.44,
      "grad_norm": 4.03253098206425,
      "learning_rate": 1.9214368738157615e-05,
      "loss": 3.9993,
      "step": 3889
    },
    {
      "epoch": 4.45,
      "grad_norm": 3.3824406785044854,
      "learning_rate": 1.9136229309464877e-05,
      "loss": 4.0129,
      "step": 3890
    },
    {
      "epoch": 4.45,
      "grad_norm": 3.853547916210705,
      "learning_rate": 1.9058243855478294e-05,
      "loss": 3.8151,
      "step": 3891
    },
    {
      "epoch": 4.45,
      "grad_norm": 3.758952968753591,
      "learning_rate": 1.8980412418950998e-05,
      "loss": 4.2905,
      "step": 3892
    },
    {
      "epoch": 4.45,
      "grad_norm": 4.11413985775139,
      "learning_rate": 1.8902735042551588e-05,
      "loss": 3.9722,
      "step": 3893
    },
    {
      "epoch": 4.45,
      "grad_norm": 1.7573348727455205,
      "learning_rate": 1.8825211768864333e-05,
      "loss": 4.0775,
      "step": 3894
    },
    {
      "epoch": 4.45,
      "grad_norm": 2.991314772878765,
      "learning_rate": 1.8747842640388745e-05,
      "loss": 4.0089,
      "step": 3895
    },
    {
      "epoch": 4.45,
      "grad_norm": 5.895143857480241,
      "learning_rate": 1.867062769954015e-05,
      "loss": 4.1536,
      "step": 3896
    },
    {
      "epoch": 4.45,
      "grad_norm": 3.891802227163627,
      "learning_rate": 1.8593566988649166e-05,
      "loss": 4.026,
      "step": 3897
    },
    {
      "epoch": 4.45,
      "grad_norm": 3.0807485937177264,
      "learning_rate": 1.8516660549961893e-05,
      "loss": 4.128,
      "step": 3898
    },
    {
      "epoch": 4.46,
      "grad_norm": 3.54402614713139,
      "learning_rate": 1.843990842563994e-05,
      "loss": 3.9487,
      "step": 3899
    },
    {
      "epoch": 4.46,
      "grad_norm": 1.8864535885239648,
      "learning_rate": 1.836331065776011e-05,
      "loss": 4.232,
      "step": 3900
    },
    {
      "epoch": 4.46,
      "eval_blimp_filtered_avg": 0.5240298507462686,
      "eval_blimp_filtered_std": 0.005720055690947627,
      "step": 3900
    },
    {
      "epoch": 4.46,
      "eval_blimp_supplement_avg": 0.47844827586206895,
      "eval_blimp_supplement_std": 0.022917691086984965,
      "step": 3900
    },
    {
      "epoch": 4.46,
      "eval_vqa_filtered_avg": 0.37,
      "eval_vqa_filtered_std": 0.048523658709391,
      "step": 3900
    },
    {
      "epoch": 4.46,
      "eval_winoground_filtered_avg": 0.52,
      "eval_winoground_filtered_std": 0.05021167315686779,
      "step": 3900
    },
    {
      "epoch": 4.46,
      "grad_norm": 4.036710324362638,
      "learning_rate": 1.8286867288314754e-05,
      "loss": 4.2407,
      "step": 3901
    },
    {
      "epoch": 4.46,
      "grad_norm": 3.8138743009811846,
      "learning_rate": 1.821057835921158e-05,
      "loss": 3.9976,
      "step": 3902
    },
    {
      "epoch": 4.46,
      "grad_norm": 3.3309643983346153,
      "learning_rate": 1.8134443912273534e-05,
      "loss": 4.14,
      "step": 3903
    },
    {
      "epoch": 4.46,
      "grad_norm": 4.305424559034067,
      "learning_rate": 1.805846398923896e-05,
      "loss": 4.1639,
      "step": 3904
    },
    {
      "epoch": 4.46,
      "grad_norm": 3.012289768678228,
      "learning_rate": 1.798263863176147e-05,
      "loss": 4.1773,
      "step": 3905
    },
    {
      "epoch": 4.46,
      "grad_norm": 2.7629929835496863,
      "learning_rate": 1.7906967881409918e-05,
      "loss": 4.196,
      "step": 3906
    },
    {
      "epoch": 4.47,
      "grad_norm": 6.07052158876355,
      "learning_rate": 1.7831451779668417e-05,
      "loss": 3.9719,
      "step": 3907
    },
    {
      "epoch": 4.47,
      "grad_norm": 3.420958220761249,
      "learning_rate": 1.7756090367936215e-05,
      "loss": 4.3165,
      "step": 3908
    },
    {
      "epoch": 4.47,
      "grad_norm": 3.823735914002226,
      "learning_rate": 1.768088368752787e-05,
      "loss": 4.097,
      "step": 3909
    },
    {
      "epoch": 4.47,
      "grad_norm": 3.482087814711469,
      "learning_rate": 1.7605831779673098e-05,
      "loss": 4.2187,
      "step": 3910
    },
    {
      "epoch": 4.47,
      "grad_norm": 3.6068470399106167,
      "learning_rate": 1.7530934685516752e-05,
      "loss": 3.9933,
      "step": 3911
    },
    {
      "epoch": 4.47,
      "grad_norm": 3.899476326589671,
      "learning_rate": 1.7456192446118754e-05,
      "loss": 4.0122,
      "step": 3912
    },
    {
      "epoch": 4.47,
      "grad_norm": 3.677258195689179,
      "learning_rate": 1.73816051024542e-05,
      "loss": 4.0117,
      "step": 3913
    },
    {
      "epoch": 4.47,
      "grad_norm": 4.467881427083908,
      "learning_rate": 1.730717269541324e-05,
      "loss": 3.9551,
      "step": 3914
    },
    {
      "epoch": 4.47,
      "grad_norm": 3.8131732636552607,
      "learning_rate": 1.723289526580114e-05,
      "loss": 4.0231,
      "step": 3915
    },
    {
      "epoch": 4.48,
      "grad_norm": 3.273003399644787,
      "learning_rate": 1.7158772854338155e-05,
      "loss": 4.2193,
      "step": 3916
    },
    {
      "epoch": 4.48,
      "grad_norm": 2.7661766774048364,
      "learning_rate": 1.708480550165948e-05,
      "loss": 4.0163,
      "step": 3917
    },
    {
      "epoch": 4.48,
      "grad_norm": 3.811222476293808,
      "learning_rate": 1.701099324831545e-05,
      "loss": 4.1843,
      "step": 3918
    },
    {
      "epoch": 4.48,
      "grad_norm": 5.703627188338064,
      "learning_rate": 1.6937336134771263e-05,
      "loss": 4.2539,
      "step": 3919
    },
    {
      "epoch": 4.48,
      "grad_norm": 6.876104372623802,
      "learning_rate": 1.6863834201407144e-05,
      "loss": 3.9532,
      "step": 3920
    },
    {
      "epoch": 4.48,
      "grad_norm": 4.001749009849482,
      "learning_rate": 1.67904874885182e-05,
      "loss": 4.0959,
      "step": 3921
    },
    {
      "epoch": 4.48,
      "grad_norm": 4.656915046948665,
      "learning_rate": 1.6717296036314463e-05,
      "loss": 4.0539,
      "step": 3922
    },
    {
      "epoch": 4.48,
      "grad_norm": 2.795198613585797,
      "learning_rate": 1.664425988492084e-05,
      "loss": 3.9267,
      "step": 3923
    },
    {
      "epoch": 4.48,
      "grad_norm": 2.683432220071694,
      "learning_rate": 1.6571379074377134e-05,
      "loss": 4.1989,
      "step": 3924
    },
    {
      "epoch": 4.49,
      "grad_norm": 4.945996840061124,
      "learning_rate": 1.649865364463785e-05,
      "loss": 4.0086,
      "step": 3925
    },
    {
      "epoch": 4.49,
      "grad_norm": 3.926512945913584,
      "learning_rate": 1.642608363557246e-05,
      "loss": 4.1477,
      "step": 3926
    },
    {
      "epoch": 4.49,
      "grad_norm": 4.725766723767193,
      "learning_rate": 1.6353669086965216e-05,
      "loss": 4.1185,
      "step": 3927
    },
    {
      "epoch": 4.49,
      "grad_norm": 3.931582848248367,
      "learning_rate": 1.6281410038515053e-05,
      "loss": 4.2814,
      "step": 3928
    },
    {
      "epoch": 4.49,
      "grad_norm": 4.394203660682757,
      "learning_rate": 1.6209306529835785e-05,
      "loss": 4.1393,
      "step": 3929
    },
    {
      "epoch": 4.49,
      "grad_norm": 3.5199227968388143,
      "learning_rate": 1.613735860045583e-05,
      "loss": 4.3944,
      "step": 3930
    },
    {
      "epoch": 4.49,
      "grad_norm": 3.1359824199952717,
      "learning_rate": 1.60655662898184e-05,
      "loss": 4.0104,
      "step": 3931
    },
    {
      "epoch": 4.49,
      "grad_norm": 3.49308502029687,
      "learning_rate": 1.5993929637281344e-05,
      "loss": 4.1539,
      "step": 3932
    },
    {
      "epoch": 4.49,
      "grad_norm": 7.451362666064646,
      "learning_rate": 1.5922448682117282e-05,
      "loss": 4.0051,
      "step": 3933
    },
    {
      "epoch": 4.5,
      "grad_norm": 6.211810619419057,
      "learning_rate": 1.5851123463513272e-05,
      "loss": 4.1456,
      "step": 3934
    },
    {
      "epoch": 4.5,
      "grad_norm": 4.614487481744874,
      "learning_rate": 1.577995402057113e-05,
      "loss": 4.0574,
      "step": 3935
    },
    {
      "epoch": 4.5,
      "grad_norm": 3.0930080769480437,
      "learning_rate": 1.570894039230731e-05,
      "loss": 4.0438,
      "step": 3936
    },
    {
      "epoch": 4.5,
      "grad_norm": 2.6053256819118706,
      "learning_rate": 1.5638082617652847e-05,
      "loss": 4.2941,
      "step": 3937
    },
    {
      "epoch": 4.5,
      "grad_norm": 5.958967936147854,
      "learning_rate": 1.5567380735453193e-05,
      "loss": 4.0967,
      "step": 3938
    },
    {
      "epoch": 4.5,
      "grad_norm": 4.512421496299159,
      "learning_rate": 1.5496834784468414e-05,
      "loss": 4.126,
      "step": 3939
    },
    {
      "epoch": 4.5,
      "grad_norm": 6.807071535368659,
      "learning_rate": 1.542644480337317e-05,
      "loss": 4.0106,
      "step": 3940
    },
    {
      "epoch": 4.5,
      "grad_norm": 3.0776721785249777,
      "learning_rate": 1.5356210830756533e-05,
      "loss": 4.0628,
      "step": 3941
    },
    {
      "epoch": 4.51,
      "grad_norm": 3.9227529645873216,
      "learning_rate": 1.5286132905121962e-05,
      "loss": 4.1682,
      "step": 3942
    },
    {
      "epoch": 4.51,
      "grad_norm": 4.6294698762562705,
      "learning_rate": 1.5216211064887628e-05,
      "loss": 4.032,
      "step": 3943
    },
    {
      "epoch": 4.51,
      "grad_norm": 5.290366610125381,
      "learning_rate": 1.5146445348385872e-05,
      "loss": 4.1652,
      "step": 3944
    },
    {
      "epoch": 4.51,
      "grad_norm": 2.374226090818631,
      "learning_rate": 1.5076835793863629e-05,
      "loss": 4.2616,
      "step": 3945
    },
    {
      "epoch": 4.51,
      "grad_norm": 6.486212150480482,
      "learning_rate": 1.5007382439482074e-05,
      "loss": 3.8761,
      "step": 3946
    },
    {
      "epoch": 4.51,
      "grad_norm": 6.1765575786850935,
      "learning_rate": 1.4938085323316817e-05,
      "loss": 4.1172,
      "step": 3947
    },
    {
      "epoch": 4.51,
      "grad_norm": 3.488875532024985,
      "learning_rate": 1.4868944483357836e-05,
      "loss": 4.0738,
      "step": 3948
    },
    {
      "epoch": 4.51,
      "grad_norm": 5.962398051882274,
      "learning_rate": 1.479995995750941e-05,
      "loss": 3.9247,
      "step": 3949
    },
    {
      "epoch": 4.51,
      "grad_norm": 4.513844327520555,
      "learning_rate": 1.4731131783590156e-05,
      "loss": 4.201,
      "step": 3950
    },
    {
      "epoch": 4.52,
      "grad_norm": 3.936683447187226,
      "learning_rate": 1.466245999933292e-05,
      "loss": 4.0537,
      "step": 3951
    },
    {
      "epoch": 4.52,
      "grad_norm": 3.680512408427049,
      "learning_rate": 1.4593944642384859e-05,
      "loss": 4.1327,
      "step": 3952
    },
    {
      "epoch": 4.52,
      "grad_norm": 3.9738319548586585,
      "learning_rate": 1.4525585750307356e-05,
      "loss": 4.0406,
      "step": 3953
    },
    {
      "epoch": 4.52,
      "grad_norm": 5.738394397694501,
      "learning_rate": 1.445738336057607e-05,
      "loss": 4.1436,
      "step": 3954
    },
    {
      "epoch": 4.52,
      "grad_norm": 4.224894510638059,
      "learning_rate": 1.4389337510580689e-05,
      "loss": 4.0665,
      "step": 3955
    },
    {
      "epoch": 4.52,
      "grad_norm": 3.620272063865558,
      "learning_rate": 1.4321448237625277e-05,
      "loss": 4.1797,
      "step": 3956
    },
    {
      "epoch": 4.52,
      "grad_norm": 2.7039392136664415,
      "learning_rate": 1.4253715578927993e-05,
      "loss": 4.1106,
      "step": 3957
    },
    {
      "epoch": 4.52,
      "grad_norm": 2.851070299104164,
      "learning_rate": 1.4186139571621102e-05,
      "loss": 3.9647,
      "step": 3958
    },
    {
      "epoch": 4.52,
      "grad_norm": 4.378771558868778,
      "learning_rate": 1.4118720252751037e-05,
      "loss": 4.2093,
      "step": 3959
    },
    {
      "epoch": 4.53,
      "grad_norm": 2.6907952261477015,
      "learning_rate": 1.4051457659278332e-05,
      "loss": 4.1396,
      "step": 3960
    },
    {
      "epoch": 4.53,
      "grad_norm": 4.774372302687286,
      "learning_rate": 1.3984351828077557e-05,
      "loss": 4.0438,
      "step": 3961
    },
    {
      "epoch": 4.53,
      "grad_norm": 4.552835756715325,
      "learning_rate": 1.3917402795937383e-05,
      "loss": 4.0981,
      "step": 3962
    },
    {
      "epoch": 4.53,
      "grad_norm": 2.539530089940984,
      "learning_rate": 1.3850610599560486e-05,
      "loss": 4.2572,
      "step": 3963
    },
    {
      "epoch": 4.53,
      "grad_norm": 3.3820212224055775,
      "learning_rate": 1.3783975275563608e-05,
      "loss": 4.0154,
      "step": 3964
    },
    {
      "epoch": 4.53,
      "grad_norm": 2.8523872159918247,
      "learning_rate": 1.3717496860477428e-05,
      "loss": 3.8822,
      "step": 3965
    },
    {
      "epoch": 4.53,
      "grad_norm": 20.36807453832221,
      "learning_rate": 1.3651175390746694e-05,
      "loss": 4.1767,
      "step": 3966
    },
    {
      "epoch": 4.53,
      "grad_norm": 3.0787810487871905,
      "learning_rate": 1.3585010902729987e-05,
      "loss": 4.0287,
      "step": 3967
    },
    {
      "epoch": 4.53,
      "grad_norm": 4.44348091336858,
      "learning_rate": 1.351900343269996e-05,
      "loss": 4.1939,
      "step": 3968
    },
    {
      "epoch": 4.54,
      "grad_norm": 4.778921311712089,
      "learning_rate": 1.3453153016843132e-05,
      "loss": 4.2358,
      "step": 3969
    },
    {
      "epoch": 4.54,
      "grad_norm": 4.7137092158355545,
      "learning_rate": 1.3387459691259861e-05,
      "loss": 4.1348,
      "step": 3970
    },
    {
      "epoch": 4.54,
      "grad_norm": 3.291489950315583,
      "learning_rate": 1.3321923491964537e-05,
      "loss": 4.0196,
      "step": 3971
    },
    {
      "epoch": 4.54,
      "grad_norm": 6.781358490648424,
      "learning_rate": 1.325654445488522e-05,
      "loss": 4.0413,
      "step": 3972
    },
    {
      "epoch": 4.54,
      "grad_norm": 7.388259637001667,
      "learning_rate": 1.3191322615863909e-05,
      "loss": 4.2396,
      "step": 3973
    },
    {
      "epoch": 4.54,
      "grad_norm": 3.548430117398899,
      "learning_rate": 1.3126258010656466e-05,
      "loss": 4.0966,
      "step": 3974
    },
    {
      "epoch": 4.54,
      "grad_norm": 3.7574643280789672,
      "learning_rate": 1.3061350674932525e-05,
      "loss": 3.886,
      "step": 3975
    },
    {
      "epoch": 4.54,
      "grad_norm": 5.26509357266034,
      "learning_rate": 1.2996600644275423e-05,
      "loss": 3.9666,
      "step": 3976
    },
    {
      "epoch": 4.55,
      "grad_norm": 3.1844830808337496,
      "learning_rate": 1.2932007954182366e-05,
      "loss": 4.2156,
      "step": 3977
    },
    {
      "epoch": 4.55,
      "grad_norm": 4.49461094628924,
      "learning_rate": 1.286757264006426e-05,
      "loss": 4.2779,
      "step": 3978
    },
    {
      "epoch": 4.55,
      "grad_norm": 3.88705236971674,
      "learning_rate": 1.2803294737245751e-05,
      "loss": 4.089,
      "step": 3979
    },
    {
      "epoch": 4.55,
      "grad_norm": 2.7741634579103582,
      "learning_rate": 1.2739174280965115e-05,
      "loss": 4.1153,
      "step": 3980
    },
    {
      "epoch": 4.55,
      "grad_norm": 4.393215941798447,
      "learning_rate": 1.2675211306374367e-05,
      "loss": 4.1399,
      "step": 3981
    },
    {
      "epoch": 4.55,
      "grad_norm": 3.415989022053479,
      "learning_rate": 1.2611405848539224e-05,
      "loss": 4.0104,
      "step": 3982
    },
    {
      "epoch": 4.55,
      "grad_norm": 2.5251455459752945,
      "learning_rate": 1.2547757942439074e-05,
      "loss": 3.967,
      "step": 3983
    },
    {
      "epoch": 4.55,
      "grad_norm": 4.329278282412647,
      "learning_rate": 1.2484267622966771e-05,
      "loss": 4.0165,
      "step": 3984
    },
    {
      "epoch": 4.55,
      "grad_norm": 2.810106587340108,
      "learning_rate": 1.2420934924928938e-05,
      "loss": 4.1595,
      "step": 3985
    },
    {
      "epoch": 4.56,
      "grad_norm": 4.4992396329999815,
      "learning_rate": 1.2357759883045737e-05,
      "loss": 4.1164,
      "step": 3986
    },
    {
      "epoch": 4.56,
      "grad_norm": 4.331106530135752,
      "learning_rate": 1.2294742531950863e-05,
      "loss": 4.1665,
      "step": 3987
    },
    {
      "epoch": 4.56,
      "grad_norm": 6.651653581189197,
      "learning_rate": 1.2231882906191615e-05,
      "loss": 4.1875,
      "step": 3988
    },
    {
      "epoch": 4.56,
      "grad_norm": 2.74141841354632,
      "learning_rate": 1.216918104022876e-05,
      "loss": 3.9784,
      "step": 3989
    },
    {
      "epoch": 4.56,
      "grad_norm": 2.806923778733947,
      "learning_rate": 1.2106636968436633e-05,
      "loss": 3.8698,
      "step": 3990
    },
    {
      "epoch": 4.56,
      "grad_norm": 2.706510973596713,
      "learning_rate": 1.2044250725103111e-05,
      "loss": 3.9779,
      "step": 3991
    },
    {
      "epoch": 4.56,
      "grad_norm": 5.949757706081025,
      "learning_rate": 1.198202234442943e-05,
      "loss": 4.0682,
      "step": 3992
    },
    {
      "epoch": 4.56,
      "grad_norm": 5.9484985151257455,
      "learning_rate": 1.1919951860530341e-05,
      "loss": 3.9659,
      "step": 3993
    },
    {
      "epoch": 4.56,
      "grad_norm": 3.77239696220753,
      "learning_rate": 1.1858039307434019e-05,
      "loss": 4.022,
      "step": 3994
    },
    {
      "epoch": 4.57,
      "grad_norm": 3.5662279255718463,
      "learning_rate": 1.1796284719082084e-05,
      "loss": 4.2784,
      "step": 3995
    },
    {
      "epoch": 4.57,
      "grad_norm": 3.6062102579474313,
      "learning_rate": 1.1734688129329583e-05,
      "loss": 3.9701,
      "step": 3996
    },
    {
      "epoch": 4.57,
      "grad_norm": 4.376554116708603,
      "learning_rate": 1.167324957194481e-05,
      "loss": 3.9548,
      "step": 3997
    },
    {
      "epoch": 4.57,
      "grad_norm": 4.617633363802044,
      "learning_rate": 1.1611969080609585e-05,
      "loss": 4.1449,
      "step": 3998
    },
    {
      "epoch": 4.57,
      "grad_norm": 1.878363997098849,
      "learning_rate": 1.1550846688919001e-05,
      "loss": 4.1431,
      "step": 3999
    },
    {
      "epoch": 4.57,
      "grad_norm": 4.915068865934237,
      "learning_rate": 1.1489882430381492e-05,
      "loss": 4.161,
      "step": 4000
    },
    {
      "epoch": 4.57,
      "eval_blimp_filtered_avg": 0.5243283582089552,
      "eval_blimp_filtered_std": 0.0056953767134424104,
      "step": 4000
    },
    {
      "epoch": 4.57,
      "eval_blimp_supplement_avg": 0.47198275862068967,
      "eval_blimp_supplement_std": 0.022638377289993485,
      "step": 4000
    },
    {
      "epoch": 4.57,
      "eval_vqa_filtered_avg": 0.37,
      "eval_vqa_filtered_std": 0.048523658709391,
      "step": 4000
    },
    {
      "epoch": 4.57,
      "eval_winoground_filtered_avg": 0.49,
      "eval_winoground_filtered_std": 0.05024183937956912,
      "step": 4000
    },
    {
      "epoch": 4.57,
      "grad_norm": 4.083826968389988,
      "learning_rate": 1.1429076338418719e-05,
      "loss": 3.8051,
      "step": 4001
    },
    {
      "epoch": 4.57,
      "grad_norm": 3.7118379079289876,
      "learning_rate": 1.1368428446365785e-05,
      "loss": 4.244,
      "step": 4002
    },
    {
      "epoch": 4.57,
      "grad_norm": 2.9136873029362773,
      "learning_rate": 1.1307938787470927e-05,
      "loss": 4.1234,
      "step": 4003
    },
    {
      "epoch": 4.58,
      "grad_norm": 2.562723323907262,
      "learning_rate": 1.1247607394895752e-05,
      "loss": 3.9621,
      "step": 4004
    },
    {
      "epoch": 4.58,
      "grad_norm": 6.998071279862643,
      "learning_rate": 1.1187434301715003e-05,
      "loss": 4.2605,
      "step": 4005
    },
    {
      "epoch": 4.58,
      "grad_norm": 3.916202270413966,
      "learning_rate": 1.1127419540916693e-05,
      "loss": 4.3383,
      "step": 4006
    },
    {
      "epoch": 4.58,
      "grad_norm": 3.412082759718138,
      "learning_rate": 1.1067563145402004e-05,
      "loss": 3.9657,
      "step": 4007
    },
    {
      "epoch": 4.58,
      "grad_norm": 27.996960225640553,
      "learning_rate": 1.1007865147985317e-05,
      "loss": 3.9547,
      "step": 4008
    },
    {
      "epoch": 4.58,
      "grad_norm": 5.356330623285319,
      "learning_rate": 1.0948325581394257e-05,
      "loss": 4.1502,
      "step": 4009
    },
    {
      "epoch": 4.58,
      "grad_norm": 3.056214928207012,
      "learning_rate": 1.0888944478269412e-05,
      "loss": 4.2495,
      "step": 4010
    },
    {
      "epoch": 4.58,
      "grad_norm": 2.5108412335351638,
      "learning_rate": 1.0829721871164643e-05,
      "loss": 4.1524,
      "step": 4011
    },
    {
      "epoch": 4.59,
      "grad_norm": 2.807973791115616,
      "learning_rate": 1.0770657792546844e-05,
      "loss": 4.1149,
      "step": 4012
    },
    {
      "epoch": 4.59,
      "grad_norm": 2.263668308563047,
      "learning_rate": 1.0711752274796082e-05,
      "loss": 4.0738,
      "step": 4013
    },
    {
      "epoch": 4.59,
      "grad_norm": 4.333137465929719,
      "learning_rate": 1.0653005350205423e-05,
      "loss": 4.1973,
      "step": 4014
    },
    {
      "epoch": 4.59,
      "grad_norm": 5.613112714212941,
      "learning_rate": 1.0594417050981009e-05,
      "loss": 4.1275,
      "step": 4015
    },
    {
      "epoch": 4.59,
      "grad_norm": 5.0690285261432075,
      "learning_rate": 1.0535987409242042e-05,
      "loss": 4.1298,
      "step": 4016
    },
    {
      "epoch": 4.59,
      "grad_norm": 3.123488549908844,
      "learning_rate": 1.0477716457020801e-05,
      "loss": 4.0367,
      "step": 4017
    },
    {
      "epoch": 4.59,
      "grad_norm": 3.019879265796207,
      "learning_rate": 1.0419604226262335e-05,
      "loss": 4.1774,
      "step": 4018
    },
    {
      "epoch": 4.59,
      "grad_norm": 3.6594758114212547,
      "learning_rate": 1.036165074882499e-05,
      "loss": 4.1817,
      "step": 4019
    },
    {
      "epoch": 4.59,
      "grad_norm": 2.3794368414126112,
      "learning_rate": 1.0303856056479821e-05,
      "loss": 4.135,
      "step": 4020
    },
    {
      "epoch": 4.6,
      "grad_norm": 5.127374048209246,
      "learning_rate": 1.0246220180911013e-05,
      "loss": 4.1546,
      "step": 4021
    },
    {
      "epoch": 4.6,
      "grad_norm": 3.1718755881347525,
      "learning_rate": 1.0188743153715628e-05,
      "loss": 4.0469,
      "step": 4022
    },
    {
      "epoch": 4.6,
      "grad_norm": 6.618431231927645,
      "learning_rate": 1.0131425006403626e-05,
      "loss": 3.8927,
      "step": 4023
    },
    {
      "epoch": 4.6,
      "grad_norm": 4.382420453857184,
      "learning_rate": 1.0074265770397838e-05,
      "loss": 4.2084,
      "step": 4024
    },
    {
      "epoch": 4.6,
      "grad_norm": 4.592646055659577,
      "learning_rate": 1.0017265477033998e-05,
      "loss": 3.9096,
      "step": 4025
    },
    {
      "epoch": 4.6,
      "grad_norm": 2.711023067891441,
      "learning_rate": 9.960424157560842e-06,
      "loss": 4.0323,
      "step": 4026
    },
    {
      "epoch": 4.6,
      "grad_norm": 2.041185437705355,
      "learning_rate": 9.903741843139679e-06,
      "loss": 4.1031,
      "step": 4027
    },
    {
      "epoch": 4.6,
      "grad_norm": 3.450932742159155,
      "learning_rate": 9.847218564844884e-06,
      "loss": 4.0284,
      "step": 4028
    },
    {
      "epoch": 4.6,
      "grad_norm": 5.255149283288173,
      "learning_rate": 9.790854353663502e-06,
      "loss": 4.3427,
      "step": 4029
    },
    {
      "epoch": 4.61,
      "grad_norm": 3.3120025119127794,
      "learning_rate": 9.734649240495517e-06,
      "loss": 3.9292,
      "step": 4030
    },
    {
      "epoch": 4.61,
      "grad_norm": 2.9272064803868387,
      "learning_rate": 9.678603256153584e-06,
      "loss": 4.0939,
      "step": 4031
    },
    {
      "epoch": 4.61,
      "grad_norm": 2.900200339971581,
      "learning_rate": 9.622716431363086e-06,
      "loss": 4.0414,
      "step": 4032
    },
    {
      "epoch": 4.61,
      "grad_norm": 2.5699663865176037,
      "learning_rate": 9.566988796762288e-06,
      "loss": 4.1209,
      "step": 4033
    },
    {
      "epoch": 4.61,
      "grad_norm": 2.81355824986362,
      "learning_rate": 9.511420382902113e-06,
      "loss": 4.2825,
      "step": 4034
    },
    {
      "epoch": 4.61,
      "grad_norm": 3.1218294492556358,
      "learning_rate": 9.45601122024613e-06,
      "loss": 3.9088,
      "step": 4035
    },
    {
      "epoch": 4.61,
      "grad_norm": 3.816996579772586,
      "learning_rate": 9.400761339170637e-06,
      "loss": 3.9956,
      "step": 4036
    },
    {
      "epoch": 4.61,
      "grad_norm": 3.636511063204475,
      "learning_rate": 9.345670769964742e-06,
      "loss": 4.0131,
      "step": 4037
    },
    {
      "epoch": 4.61,
      "grad_norm": 5.7478907603635685,
      "learning_rate": 9.29073954283005e-06,
      "loss": 3.9302,
      "step": 4038
    },
    {
      "epoch": 4.62,
      "grad_norm": 5.3301843211156905,
      "learning_rate": 9.235967687880941e-06,
      "loss": 4.0518,
      "step": 4039
    },
    {
      "epoch": 4.62,
      "grad_norm": 2.1229866602177982,
      "learning_rate": 9.181355235144261e-06,
      "loss": 4.1877,
      "step": 4040
    },
    {
      "epoch": 4.62,
      "grad_norm": 3.4213828393386927,
      "learning_rate": 9.126902214559628e-06,
      "loss": 3.9532,
      "step": 4041
    },
    {
      "epoch": 4.62,
      "grad_norm": 4.325072401805332,
      "learning_rate": 9.072608655979164e-06,
      "loss": 4.2427,
      "step": 4042
    },
    {
      "epoch": 4.62,
      "grad_norm": 3.6586162614661917,
      "learning_rate": 9.01847458916759e-06,
      "loss": 4.0063,
      "step": 4043
    },
    {
      "epoch": 4.62,
      "grad_norm": 5.446208301834507,
      "learning_rate": 8.96450004380227e-06,
      "loss": 4.2332,
      "step": 4044
    },
    {
      "epoch": 4.62,
      "grad_norm": 5.394530438747241,
      "learning_rate": 8.910685049472965e-06,
      "loss": 3.9655,
      "step": 4045
    },
    {
      "epoch": 4.62,
      "grad_norm": 2.751997561818143,
      "learning_rate": 8.85702963568211e-06,
      "loss": 4.0227,
      "step": 4046
    },
    {
      "epoch": 4.63,
      "grad_norm": 2.4020260851078032,
      "learning_rate": 8.803533831844577e-06,
      "loss": 4.2375,
      "step": 4047
    },
    {
      "epoch": 4.63,
      "grad_norm": 3.445989928682379,
      "learning_rate": 8.750197667287706e-06,
      "loss": 4.1277,
      "step": 4048
    },
    {
      "epoch": 4.63,
      "grad_norm": 3.0769752399109507,
      "learning_rate": 8.69702117125144e-06,
      "loss": 4.0396,
      "step": 4049
    },
    {
      "epoch": 4.63,
      "grad_norm": 4.893744754028155,
      "learning_rate": 8.644004372888059e-06,
      "loss": 4.1491,
      "step": 4050
    },
    {
      "epoch": 4.63,
      "grad_norm": 3.148103616379697,
      "learning_rate": 8.59114730126238e-06,
      "loss": 4.2078,
      "step": 4051
    },
    {
      "epoch": 4.63,
      "grad_norm": 2.6100897352198538,
      "learning_rate": 8.538449985351625e-06,
      "loss": 4.0135,
      "step": 4052
    },
    {
      "epoch": 4.63,
      "grad_norm": 4.324480730484191,
      "learning_rate": 8.485912454045419e-06,
      "loss": 3.8184,
      "step": 4053
    },
    {
      "epoch": 4.63,
      "grad_norm": 2.8950820782747355,
      "learning_rate": 8.433534736145819e-06,
      "loss": 4.0826,
      "step": 4054
    },
    {
      "epoch": 4.63,
      "grad_norm": 3.5512675793392017,
      "learning_rate": 8.381316860367293e-06,
      "loss": 4.3048,
      "step": 4055
    },
    {
      "epoch": 4.64,
      "grad_norm": 2.3506913970544767,
      "learning_rate": 8.329258855336574e-06,
      "loss": 4.1,
      "step": 4056
    },
    {
      "epoch": 4.64,
      "grad_norm": 4.752256431195318,
      "learning_rate": 8.277360749592866e-06,
      "loss": 4.2809,
      "step": 4057
    },
    {
      "epoch": 4.64,
      "grad_norm": 3.815785242295265,
      "learning_rate": 8.225622571587643e-06,
      "loss": 3.9356,
      "step": 4058
    },
    {
      "epoch": 4.64,
      "grad_norm": 4.990748870661661,
      "learning_rate": 8.174044349684783e-06,
      "loss": 4.013,
      "step": 4059
    },
    {
      "epoch": 4.64,
      "grad_norm": 39.63887652495673,
      "learning_rate": 8.122626112160336e-06,
      "loss": 4.0079,
      "step": 4060
    },
    {
      "epoch": 4.64,
      "grad_norm": 2.6873759274096294,
      "learning_rate": 8.071367887202817e-06,
      "loss": 3.9895,
      "step": 4061
    },
    {
      "epoch": 4.64,
      "grad_norm": 6.538519490937643,
      "learning_rate": 8.020269702912885e-06,
      "loss": 4.1062,
      "step": 4062
    },
    {
      "epoch": 4.64,
      "grad_norm": 4.871153169002898,
      "learning_rate": 7.969331587303496e-06,
      "loss": 4.0802,
      "step": 4063
    },
    {
      "epoch": 4.64,
      "grad_norm": 2.650666804190377,
      "learning_rate": 7.918553568299912e-06,
      "loss": 4.1916,
      "step": 4064
    },
    {
      "epoch": 4.65,
      "grad_norm": 3.952448132545441,
      "learning_rate": 7.867935673739534e-06,
      "loss": 4.1941,
      "step": 4065
    },
    {
      "epoch": 4.65,
      "grad_norm": 3.543736276089882,
      "learning_rate": 7.817477931372063e-06,
      "loss": 4.0595,
      "step": 4066
    },
    {
      "epoch": 4.65,
      "grad_norm": 5.520349085421009,
      "learning_rate": 7.767180368859305e-06,
      "loss": 3.9056,
      "step": 4067
    },
    {
      "epoch": 4.65,
      "grad_norm": 4.191234944668208,
      "learning_rate": 7.717043013775404e-06,
      "loss": 4.1407,
      "step": 4068
    },
    {
      "epoch": 4.65,
      "grad_norm": 3.5499970724822005,
      "learning_rate": 7.667065893606506e-06,
      "loss": 3.8692,
      "step": 4069
    },
    {
      "epoch": 4.65,
      "grad_norm": 2.342515503237021,
      "learning_rate": 7.6172490357510275e-06,
      "loss": 4.1101,
      "step": 4070
    },
    {
      "epoch": 4.65,
      "grad_norm": 3.292414967483802,
      "learning_rate": 7.56759246751949e-06,
      "loss": 4.0174,
      "step": 4071
    },
    {
      "epoch": 4.65,
      "grad_norm": 3.6403307810700034,
      "learning_rate": 7.51809621613455e-06,
      "loss": 3.9366,
      "step": 4072
    },
    {
      "epoch": 4.65,
      "grad_norm": 3.391342845461786,
      "learning_rate": 7.468760308730937e-06,
      "loss": 4.2533,
      "step": 4073
    },
    {
      "epoch": 4.66,
      "grad_norm": 3.2566946755404818,
      "learning_rate": 7.419584772355514e-06,
      "loss": 3.9729,
      "step": 4074
    },
    {
      "epoch": 4.66,
      "grad_norm": 2.6961725848341884,
      "learning_rate": 7.370569633967216e-06,
      "loss": 4.0817,
      "step": 4075
    },
    {
      "epoch": 4.66,
      "grad_norm": 5.928782245637853,
      "learning_rate": 7.321714920437049e-06,
      "loss": 3.987,
      "step": 4076
    },
    {
      "epoch": 4.66,
      "grad_norm": 2.1358984546770157,
      "learning_rate": 7.273020658548123e-06,
      "loss": 4.1768,
      "step": 4077
    },
    {
      "epoch": 4.66,
      "grad_norm": 4.3711733486853035,
      "learning_rate": 7.22448687499545e-06,
      "loss": 4.0668,
      "step": 4078
    },
    {
      "epoch": 4.66,
      "grad_norm": 3.351720278093051,
      "learning_rate": 7.176113596386213e-06,
      "loss": 4.1144,
      "step": 4079
    },
    {
      "epoch": 4.66,
      "grad_norm": 2.927710686504573,
      "learning_rate": 7.1279008492394655e-06,
      "loss": 4.0912,
      "step": 4080
    },
    {
      "epoch": 4.66,
      "grad_norm": 3.995003234978026,
      "learning_rate": 7.079848659986432e-06,
      "loss": 3.979,
      "step": 4081
    },
    {
      "epoch": 4.67,
      "grad_norm": 4.808852220992887,
      "learning_rate": 7.031957054970072e-06,
      "loss": 4.0423,
      "step": 4082
    },
    {
      "epoch": 4.67,
      "grad_norm": 4.57601012863105,
      "learning_rate": 6.984226060445519e-06,
      "loss": 3.9875,
      "step": 4083
    },
    {
      "epoch": 4.67,
      "grad_norm": 3.057954630628553,
      "learning_rate": 6.936655702579807e-06,
      "loss": 4.1043,
      "step": 4084
    },
    {
      "epoch": 4.67,
      "grad_norm": 4.833889604894008,
      "learning_rate": 6.88924600745191e-06,
      "loss": 4.2445,
      "step": 4085
    },
    {
      "epoch": 4.67,
      "grad_norm": 4.050909005029455,
      "learning_rate": 6.841997001052602e-06,
      "loss": 3.9401,
      "step": 4086
    },
    {
      "epoch": 4.67,
      "grad_norm": 3.273187578006022,
      "learning_rate": 6.79490870928473e-06,
      "loss": 4.1666,
      "step": 4087
    },
    {
      "epoch": 4.67,
      "grad_norm": 2.2644079227607077,
      "learning_rate": 6.747981157962912e-06,
      "loss": 4.0261,
      "step": 4088
    },
    {
      "epoch": 4.67,
      "grad_norm": 3.610146476168009,
      "learning_rate": 6.701214372813802e-06,
      "loss": 3.9117,
      "step": 4089
    },
    {
      "epoch": 4.67,
      "grad_norm": 2.371341100941479,
      "learning_rate": 6.654608379475656e-06,
      "loss": 4.0189,
      "step": 4090
    },
    {
      "epoch": 4.68,
      "grad_norm": 2.7828718714905083,
      "learning_rate": 6.608163203498906e-06,
      "loss": 3.9445,
      "step": 4091
    },
    {
      "epoch": 4.68,
      "grad_norm": 3.3083006592788493,
      "learning_rate": 6.561878870345549e-06,
      "loss": 4.3049,
      "step": 4092
    },
    {
      "epoch": 4.68,
      "grad_norm": 1.9106554604435018,
      "learning_rate": 6.515755405389522e-06,
      "loss": 4.0196,
      "step": 4093
    },
    {
      "epoch": 4.68,
      "grad_norm": 3.35902876998861,
      "learning_rate": 6.469792833916632e-06,
      "loss": 4.1701,
      "step": 4094
    },
    {
      "epoch": 4.68,
      "grad_norm": 4.164324827294454,
      "learning_rate": 6.423991181124355e-06,
      "loss": 4.1308,
      "step": 4095
    },
    {
      "epoch": 4.68,
      "grad_norm": 3.2068679085442238,
      "learning_rate": 6.378350472121973e-06,
      "loss": 4.321,
      "step": 4096
    },
    {
      "epoch": 4.68,
      "grad_norm": 4.10691182137336,
      "learning_rate": 6.332870731930606e-06,
      "loss": 4.1513,
      "step": 4097
    },
    {
      "epoch": 4.68,
      "grad_norm": 5.137208000432623,
      "learning_rate": 6.287551985483075e-06,
      "loss": 4.0206,
      "step": 4098
    },
    {
      "epoch": 4.68,
      "grad_norm": 3.960813120268537,
      "learning_rate": 6.242394257623973e-06,
      "loss": 3.9969,
      "step": 4099
    },
    {
      "epoch": 4.69,
      "grad_norm": 5.24228697276774,
      "learning_rate": 6.1973975731095634e-06,
      "loss": 4.2676,
      "step": 4100
    },
    {
      "epoch": 4.69,
      "eval_blimp_filtered_avg": 0.5204477611940298,
      "eval_blimp_filtered_std": 0.005696443348474412,
      "step": 4100
    },
    {
      "epoch": 4.69,
      "eval_blimp_supplement_avg": 0.47629310344827586,
      "eval_blimp_supplement_std": 0.022712727381912646,
      "step": 4100
    },
    {
      "epoch": 4.69,
      "eval_vqa_filtered_avg": 0.36,
      "eval_vqa_filtered_std": 0.04824181513244218,
      "step": 4100
    },
    {
      "epoch": 4.69,
      "eval_winoground_filtered_avg": 0.5,
      "eval_winoground_filtered_std": 0.050251890762960605,
      "step": 4100
    },
    {
      "epoch": 4.69,
      "grad_norm": 2.6149711046487107,
      "learning_rate": 6.152561956607915e-06,
      "loss": 4.1195,
      "step": 4101
    },
    {
      "epoch": 4.69,
      "grad_norm": 2.277802517595267,
      "learning_rate": 6.107887432698699e-06,
      "loss": 4.1484,
      "step": 4102
    },
    {
      "epoch": 4.69,
      "grad_norm": 4.514651931816453,
      "learning_rate": 6.063374025873324e-06,
      "loss": 4.313,
      "step": 4103
    },
    {
      "epoch": 4.69,
      "grad_norm": 2.9516369775750326,
      "learning_rate": 6.0190217605348704e-06,
      "loss": 4.2599,
      "step": 4104
    },
    {
      "epoch": 4.69,
      "grad_norm": 2.81229472610806,
      "learning_rate": 5.97483066099802e-06,
      "loss": 3.9711,
      "step": 4105
    },
    {
      "epoch": 4.69,
      "grad_norm": 3.212383496486994,
      "learning_rate": 5.930800751489228e-06,
      "loss": 4.1372,
      "step": 4106
    },
    {
      "epoch": 4.69,
      "grad_norm": 5.8711344141211805,
      "learning_rate": 5.886932056146454e-06,
      "loss": 3.8939,
      "step": 4107
    },
    {
      "epoch": 4.69,
      "grad_norm": 2.817584409686935,
      "learning_rate": 5.843224599019325e-06,
      "loss": 4.171,
      "step": 4108
    },
    {
      "epoch": 4.7,
      "grad_norm": 3.627385570816045,
      "learning_rate": 5.799678404069108e-06,
      "loss": 4.0284,
      "step": 4109
    },
    {
      "epoch": 4.7,
      "grad_norm": 5.903950099636898,
      "learning_rate": 5.756293495168607e-06,
      "loss": 4.1023,
      "step": 4110
    },
    {
      "epoch": 4.7,
      "grad_norm": 4.004538078157361,
      "learning_rate": 5.713069896102263e-06,
      "loss": 4.166,
      "step": 4111
    },
    {
      "epoch": 4.7,
      "grad_norm": 3.7016423947538026,
      "learning_rate": 5.67000763056602e-06,
      "loss": 4.0772,
      "step": 4112
    },
    {
      "epoch": 4.7,
      "grad_norm": 6.482951501421826,
      "learning_rate": 5.627106722167396e-06,
      "loss": 3.9197,
      "step": 4113
    },
    {
      "epoch": 4.7,
      "grad_norm": 2.2694869640960103,
      "learning_rate": 5.584367194425476e-06,
      "loss": 4.1161,
      "step": 4114
    },
    {
      "epoch": 4.7,
      "grad_norm": 2.831325612127297,
      "learning_rate": 5.541789070770852e-06,
      "loss": 4.0728,
      "step": 4115
    },
    {
      "epoch": 4.7,
      "grad_norm": 2.5351871512577775,
      "learning_rate": 5.499372374545652e-06,
      "loss": 4.0256,
      "step": 4116
    },
    {
      "epoch": 4.71,
      "grad_norm": 2.4748892820694204,
      "learning_rate": 5.457117129003474e-06,
      "loss": 4.1725,
      "step": 4117
    },
    {
      "epoch": 4.71,
      "grad_norm": 3.236820638804631,
      "learning_rate": 5.415023357309456e-06,
      "loss": 4.1681,
      "step": 4118
    },
    {
      "epoch": 4.71,
      "grad_norm": 4.835061284108866,
      "learning_rate": 5.373091082540171e-06,
      "loss": 4.2828,
      "step": 4119
    },
    {
      "epoch": 4.71,
      "grad_norm": 2.9945199114469996,
      "learning_rate": 5.331320327683664e-06,
      "loss": 4.1166,
      "step": 4120
    },
    {
      "epoch": 4.71,
      "grad_norm": 2.368580364427367,
      "learning_rate": 5.289711115639384e-06,
      "loss": 3.9402,
      "step": 4121
    },
    {
      "epoch": 4.71,
      "grad_norm": 2.988604763486571,
      "learning_rate": 5.2482634692183504e-06,
      "loss": 4.1609,
      "step": 4122
    },
    {
      "epoch": 4.71,
      "grad_norm": 3.1501703987649634,
      "learning_rate": 5.206977411142854e-06,
      "loss": 4.0472,
      "step": 4123
    },
    {
      "epoch": 4.71,
      "grad_norm": 2.720187649860745,
      "learning_rate": 5.165852964046724e-06,
      "loss": 4.0094,
      "step": 4124
    },
    {
      "epoch": 4.71,
      "grad_norm": 3.275650154581212,
      "learning_rate": 5.124890150475158e-06,
      "loss": 4.0372,
      "step": 4125
    },
    {
      "epoch": 4.72,
      "grad_norm": 3.5771965814390256,
      "learning_rate": 5.084088992884661e-06,
      "loss": 4.1532,
      "step": 4126
    },
    {
      "epoch": 4.72,
      "grad_norm": 3.4139027295565674,
      "learning_rate": 5.043449513643239e-06,
      "loss": 4.1858,
      "step": 4127
    },
    {
      "epoch": 4.72,
      "grad_norm": 3.6394965180709673,
      "learning_rate": 5.00297173503017e-06,
      "loss": 3.91,
      "step": 4128
    },
    {
      "epoch": 4.72,
      "grad_norm": 4.833828983780835,
      "learning_rate": 4.962655679236138e-06,
      "loss": 4.0769,
      "step": 4129
    },
    {
      "epoch": 4.72,
      "grad_norm": 2.8037701755015374,
      "learning_rate": 4.9225013683631296e-06,
      "loss": 3.9953,
      "step": 4130
    },
    {
      "epoch": 4.72,
      "grad_norm": 3.318653125697597,
      "learning_rate": 4.8825088244245025e-06,
      "loss": 4.1853,
      "step": 4131
    },
    {
      "epoch": 4.72,
      "grad_norm": 3.845439781261716,
      "learning_rate": 4.842678069344952e-06,
      "loss": 4.0148,
      "step": 4132
    },
    {
      "epoch": 4.72,
      "grad_norm": 3.2586936237347817,
      "learning_rate": 4.803009124960344e-06,
      "loss": 4.0513,
      "step": 4133
    },
    {
      "epoch": 4.72,
      "grad_norm": 2.6450403752270324,
      "learning_rate": 4.76350201301795e-06,
      "loss": 4.1923,
      "step": 4134
    },
    {
      "epoch": 4.73,
      "grad_norm": 3.7118694120438356,
      "learning_rate": 4.724156755176311e-06,
      "loss": 4.1852,
      "step": 4135
    },
    {
      "epoch": 4.73,
      "grad_norm": 4.102684565829334,
      "learning_rate": 4.684973373005274e-06,
      "loss": 4.1537,
      "step": 4136
    },
    {
      "epoch": 4.73,
      "grad_norm": 3.735714042891191,
      "learning_rate": 4.6459518879857885e-06,
      "loss": 3.93,
      "step": 4137
    },
    {
      "epoch": 4.73,
      "grad_norm": 4.78599579963367,
      "learning_rate": 4.6070923215102085e-06,
      "loss": 3.9531,
      "step": 4138
    },
    {
      "epoch": 4.73,
      "grad_norm": 4.676409226759489,
      "learning_rate": 4.568394694882094e-06,
      "loss": 4.0574,
      "step": 4139
    },
    {
      "epoch": 4.73,
      "grad_norm": 6.173571341935304,
      "learning_rate": 4.529859029316174e-06,
      "loss": 4.1035,
      "step": 4140
    },
    {
      "epoch": 4.73,
      "grad_norm": 4.590076334845145,
      "learning_rate": 4.491485345938417e-06,
      "loss": 4.0306,
      "step": 4141
    },
    {
      "epoch": 4.73,
      "grad_norm": 3.4769994206731942,
      "learning_rate": 4.453273665785928e-06,
      "loss": 3.8919,
      "step": 4142
    },
    {
      "epoch": 4.73,
      "grad_norm": 2.429257278084257,
      "learning_rate": 4.41522400980715e-06,
      "loss": 3.8789,
      "step": 4143
    },
    {
      "epoch": 4.74,
      "grad_norm": 3.084300406476649,
      "learning_rate": 4.377336398861564e-06,
      "loss": 4.136,
      "step": 4144
    },
    {
      "epoch": 4.74,
      "grad_norm": 5.009343215140627,
      "learning_rate": 4.3396108537198215e-06,
      "loss": 4.0704,
      "step": 4145
    },
    {
      "epoch": 4.74,
      "grad_norm": 2.6738966447902937,
      "learning_rate": 4.302047395063813e-06,
      "loss": 4.0965,
      "step": 4146
    },
    {
      "epoch": 4.74,
      "grad_norm": 3.197313815231188,
      "learning_rate": 4.264646043486531e-06,
      "loss": 4.0851,
      "step": 4147
    },
    {
      "epoch": 4.74,
      "grad_norm": 3.318072290847935,
      "learning_rate": 4.22740681949204e-06,
      "loss": 3.9411,
      "step": 4148
    },
    {
      "epoch": 4.74,
      "grad_norm": 3.322493926119286,
      "learning_rate": 4.190329743495613e-06,
      "loss": 4.0534,
      "step": 4149
    },
    {
      "epoch": 4.74,
      "grad_norm": 3.443883311067105,
      "learning_rate": 4.153414835823521e-06,
      "loss": 3.9464,
      "step": 4150
    },
    {
      "epoch": 4.74,
      "grad_norm": 6.15298458701985,
      "learning_rate": 4.116662116713276e-06,
      "loss": 4.2068,
      "step": 4151
    },
    {
      "epoch": 4.75,
      "grad_norm": 3.180087174152525,
      "learning_rate": 4.080071606313362e-06,
      "loss": 3.9826,
      "step": 4152
    },
    {
      "epoch": 4.75,
      "grad_norm": 2.7083159766677936,
      "learning_rate": 4.043643324683366e-06,
      "loss": 4.0189,
      "step": 4153
    },
    {
      "epoch": 4.75,
      "grad_norm": 5.406095752350838,
      "learning_rate": 4.007377291793945e-06,
      "loss": 4.121,
      "step": 4154
    },
    {
      "epoch": 4.75,
      "grad_norm": 6.365366498001963,
      "learning_rate": 3.9712735275268306e-06,
      "loss": 4.0305,
      "step": 4155
    },
    {
      "epoch": 4.75,
      "grad_norm": 6.4125106864791865,
      "learning_rate": 3.9353320516747555e-06,
      "loss": 4.0175,
      "step": 4156
    },
    {
      "epoch": 4.75,
      "grad_norm": 3.5904505361391523,
      "learning_rate": 3.8995528839415595e-06,
      "loss": 3.9473,
      "step": 4157
    },
    {
      "epoch": 4.75,
      "grad_norm": 4.227464933977833,
      "learning_rate": 3.8639360439419175e-06,
      "loss": 3.9673,
      "step": 4158
    },
    {
      "epoch": 4.75,
      "grad_norm": 5.1514159128125385,
      "learning_rate": 3.828481551201745e-06,
      "loss": 4.2055,
      "step": 4159
    },
    {
      "epoch": 4.75,
      "grad_norm": 5.956088992937854,
      "learning_rate": 3.7931894251577942e-06,
      "loss": 4.0795,
      "step": 4160
    },
    {
      "epoch": 4.76,
      "grad_norm": 3.2632840821980027,
      "learning_rate": 3.758059685157888e-06,
      "loss": 3.867,
      "step": 4161
    },
    {
      "epoch": 4.76,
      "grad_norm": 4.00905065418712,
      "learning_rate": 3.723092350460788e-06,
      "loss": 4.0146,
      "step": 4162
    },
    {
      "epoch": 4.76,
      "grad_norm": 4.147756156578193,
      "learning_rate": 3.6882874402362596e-06,
      "loss": 3.9144,
      "step": 4163
    },
    {
      "epoch": 4.76,
      "grad_norm": 2.4859674266689074,
      "learning_rate": 3.6536449735649397e-06,
      "loss": 4.2,
      "step": 4164
    },
    {
      "epoch": 4.76,
      "grad_norm": 6.487107352157862,
      "learning_rate": 3.6191649694385038e-06,
      "loss": 4.0864,
      "step": 4165
    },
    {
      "epoch": 4.76,
      "grad_norm": 3.9734564660950773,
      "learning_rate": 3.5848474467595644e-06,
      "loss": 4.0906,
      "step": 4166
    },
    {
      "epoch": 4.76,
      "grad_norm": 4.854503198710445,
      "learning_rate": 3.5506924243415726e-06,
      "loss": 4.1284,
      "step": 4167
    },
    {
      "epoch": 4.76,
      "grad_norm": 3.5918630458342444,
      "learning_rate": 3.5166999209089496e-06,
      "loss": 3.9251,
      "step": 4168
    },
    {
      "epoch": 4.76,
      "grad_norm": 4.556655383855815,
      "learning_rate": 3.482869955096956e-06,
      "loss": 4.0307,
      "step": 4169
    },
    {
      "epoch": 4.77,
      "grad_norm": 4.412540688597168,
      "learning_rate": 3.4492025454519566e-06,
      "loss": 4.0176,
      "step": 4170
    },
    {
      "epoch": 4.77,
      "grad_norm": 11.507594320247751,
      "learning_rate": 3.41569771043092e-06,
      "loss": 4.0502,
      "step": 4171
    },
    {
      "epoch": 4.77,
      "grad_norm": 6.203366523882198,
      "learning_rate": 3.3823554684018204e-06,
      "loss": 4.1232,
      "step": 4172
    },
    {
      "epoch": 4.77,
      "grad_norm": 2.4346679836100935,
      "learning_rate": 3.3491758376435697e-06,
      "loss": 4.1323,
      "step": 4173
    },
    {
      "epoch": 4.77,
      "grad_norm": 2.689411964647328,
      "learning_rate": 3.3161588363457837e-06,
      "loss": 4.2082,
      "step": 4174
    },
    {
      "epoch": 4.77,
      "grad_norm": 4.598436617239837,
      "learning_rate": 3.283304482609017e-06,
      "loss": 4.0,
      "step": 4175
    },
    {
      "epoch": 4.77,
      "grad_norm": 2.1490909066566743,
      "learning_rate": 3.250612794444629e-06,
      "loss": 4.2136,
      "step": 4176
    },
    {
      "epoch": 4.77,
      "grad_norm": 2.7089804031300773,
      "learning_rate": 3.2180837897747837e-06,
      "loss": 4.1596,
      "step": 4177
    },
    {
      "epoch": 4.77,
      "grad_norm": 5.468794429537608,
      "learning_rate": 3.1857174864325483e-06,
      "loss": 4.104,
      "step": 4178
    },
    {
      "epoch": 4.78,
      "grad_norm": 2.2817485068869248,
      "learning_rate": 3.153513902161664e-06,
      "loss": 4.069,
      "step": 4179
    },
    {
      "epoch": 4.78,
      "grad_norm": 4.155132628270886,
      "learning_rate": 3.121473054616741e-06,
      "loss": 4.1985,
      "step": 4180
    },
    {
      "epoch": 4.78,
      "grad_norm": 4.115890093220942,
      "learning_rate": 3.0895949613631953e-06,
      "loss": 4.1008,
      "step": 4181
    },
    {
      "epoch": 4.78,
      "grad_norm": 3.511188157229788,
      "learning_rate": 3.057879639877148e-06,
      "loss": 4.184,
      "step": 4182
    },
    {
      "epoch": 4.78,
      "grad_norm": 4.294932916599916,
      "learning_rate": 3.0263271075455565e-06,
      "loss": 3.864,
      "step": 4183
    },
    {
      "epoch": 4.78,
      "grad_norm": 2.4660159509038078,
      "learning_rate": 2.994937381666085e-06,
      "loss": 4.1631,
      "step": 4184
    },
    {
      "epoch": 4.78,
      "grad_norm": 5.705390488624052,
      "learning_rate": 2.963710479447168e-06,
      "loss": 4.2812,
      "step": 4185
    },
    {
      "epoch": 4.78,
      "grad_norm": 5.203760861932563,
      "learning_rate": 2.9326464180079444e-06,
      "loss": 3.8155,
      "step": 4186
    },
    {
      "epoch": 4.79,
      "grad_norm": 4.339879197030374,
      "learning_rate": 2.9017452143783595e-06,
      "loss": 3.9618,
      "step": 4187
    },
    {
      "epoch": 4.79,
      "grad_norm": 2.8428504898493556,
      "learning_rate": 2.8710068854989943e-06,
      "loss": 3.8488,
      "step": 4188
    },
    {
      "epoch": 4.79,
      "grad_norm": 5.223396823981209,
      "learning_rate": 2.8404314482211367e-06,
      "loss": 3.993,
      "step": 4189
    },
    {
      "epoch": 4.79,
      "grad_norm": 2.985357396010778,
      "learning_rate": 2.8100189193068444e-06,
      "loss": 4.3529,
      "step": 4190
    },
    {
      "epoch": 4.79,
      "grad_norm": 4.508292793179489,
      "learning_rate": 2.779769315428848e-06,
      "loss": 4.1144,
      "step": 4191
    },
    {
      "epoch": 4.79,
      "grad_norm": 3.5488049162484185,
      "learning_rate": 2.7496826531704817e-06,
      "loss": 4.1042,
      "step": 4192
    },
    {
      "epoch": 4.79,
      "grad_norm": 3.538211888417464,
      "learning_rate": 2.7197589490258187e-06,
      "loss": 4.1615,
      "step": 4193
    },
    {
      "epoch": 4.79,
      "grad_norm": 4.560958382284151,
      "learning_rate": 2.689998219399636e-06,
      "loss": 4.1783,
      "step": 4194
    },
    {
      "epoch": 4.79,
      "grad_norm": 3.4849233319067023,
      "learning_rate": 2.6604004806072833e-06,
      "loss": 4.2348,
      "step": 4195
    },
    {
      "epoch": 4.8,
      "grad_norm": 4.248879689872377,
      "learning_rate": 2.63096574887478e-06,
      "loss": 3.9831,
      "step": 4196
    },
    {
      "epoch": 4.8,
      "grad_norm": 2.7331674647925515,
      "learning_rate": 2.60169404033882e-06,
      "loss": 4.0615,
      "step": 4197
    },
    {
      "epoch": 4.8,
      "grad_norm": 4.865499172092791,
      "learning_rate": 2.5725853710466314e-06,
      "loss": 4.0719,
      "step": 4198
    },
    {
      "epoch": 4.8,
      "grad_norm": 3.0325299751515638,
      "learning_rate": 2.543639756956184e-06,
      "loss": 4.1211,
      "step": 4199
    },
    {
      "epoch": 4.8,
      "grad_norm": 2.330675982887287,
      "learning_rate": 2.5148572139359836e-06,
      "loss": 3.9945,
      "step": 4200
    },
    {
      "epoch": 4.8,
      "eval_blimp_filtered_avg": 0.5231343283582089,
      "eval_blimp_filtered_std": 0.005716553515323113,
      "step": 4200
    },
    {
      "epoch": 4.8,
      "eval_blimp_supplement_avg": 0.46120689655172414,
      "eval_blimp_supplement_std": 0.022616394405843454,
      "step": 4200
    },
    {
      "epoch": 4.8,
      "eval_vqa_filtered_avg": 0.36,
      "eval_vqa_filtered_std": 0.04824181513244218,
      "step": 4200
    },
    {
      "epoch": 4.8,
      "eval_winoground_filtered_avg": 0.51,
      "eval_winoground_filtered_std": 0.05024183937956912,
      "step": 4200
    },
    {
      "epoch": 4.8,
      "grad_norm": 3.7547579273425833,
      "learning_rate": 2.4862377577651415e-06,
      "loss": 4.0467,
      "step": 4201
    },
    {
      "epoch": 4.8,
      "grad_norm": 4.07934525725909,
      "learning_rate": 2.457781404133374e-06,
      "loss": 4.0441,
      "step": 4202
    },
    {
      "epoch": 4.8,
      "grad_norm": 4.051522547069704,
      "learning_rate": 2.429488168641003e-06,
      "loss": 3.9748,
      "step": 4203
    },
    {
      "epoch": 4.8,
      "grad_norm": 7.472598926045134,
      "learning_rate": 2.4013580667988865e-06,
      "loss": 4.0727,
      "step": 4204
    },
    {
      "epoch": 4.81,
      "grad_norm": 4.010529503372069,
      "learning_rate": 2.3733911140284554e-06,
      "loss": 4.1224,
      "step": 4205
    },
    {
      "epoch": 4.81,
      "grad_norm": 4.005729006674253,
      "learning_rate": 2.345587325661713e-06,
      "loss": 4.0498,
      "step": 4206
    },
    {
      "epoch": 4.81,
      "grad_norm": 2.8121572135451705,
      "learning_rate": 2.3179467169411992e-06,
      "loss": 4.2631,
      "step": 4207
    },
    {
      "epoch": 4.81,
      "grad_norm": 5.087392132720708,
      "learning_rate": 2.290469303020026e-06,
      "loss": 3.9391,
      "step": 4208
    },
    {
      "epoch": 4.81,
      "grad_norm": 2.91428516227794,
      "learning_rate": 2.263155098961844e-06,
      "loss": 4.1292,
      "step": 4209
    },
    {
      "epoch": 4.81,
      "grad_norm": 5.581014378424899,
      "learning_rate": 2.236004119740742e-06,
      "loss": 4.0435,
      "step": 4210
    },
    {
      "epoch": 4.81,
      "grad_norm": 3.348495432086857,
      "learning_rate": 2.209016380241413e-06,
      "loss": 4.1711,
      "step": 4211
    },
    {
      "epoch": 4.81,
      "grad_norm": 3.611437043206666,
      "learning_rate": 2.1821918952590887e-06,
      "loss": 3.9138,
      "step": 4212
    },
    {
      "epoch": 4.81,
      "grad_norm": 5.9296352876818235,
      "learning_rate": 2.1555306794993733e-06,
      "loss": 4.2207,
      "step": 4213
    },
    {
      "epoch": 4.82,
      "grad_norm": 3.348326475461506,
      "learning_rate": 2.129032747578474e-06,
      "loss": 4.2114,
      "step": 4214
    },
    {
      "epoch": 4.82,
      "grad_norm": 3.4671733132709837,
      "learning_rate": 2.102698114023005e-06,
      "loss": 4.1773,
      "step": 4215
    },
    {
      "epoch": 4.82,
      "grad_norm": 4.400638138211643,
      "learning_rate": 2.07652679327015e-06,
      "loss": 4.0814,
      "step": 4216
    },
    {
      "epoch": 4.82,
      "grad_norm": 5.513436027559663,
      "learning_rate": 2.0505187996675e-06,
      "loss": 4.0981,
      "step": 4217
    },
    {
      "epoch": 4.82,
      "grad_norm": 6.272452406993846,
      "learning_rate": 2.0246741474731155e-06,
      "loss": 4.1431,
      "step": 4218
    },
    {
      "epoch": 4.82,
      "grad_norm": 4.465939717062978,
      "learning_rate": 1.9989928508554964e-06,
      "loss": 4.1365,
      "step": 4219
    },
    {
      "epoch": 4.82,
      "grad_norm": 2.4813228046095595,
      "learning_rate": 1.9734749238936143e-06,
      "loss": 4.0744,
      "step": 4220
    },
    {
      "epoch": 4.82,
      "grad_norm": 4.240061597998089,
      "learning_rate": 1.9481203805768784e-06,
      "loss": 3.9837,
      "step": 4221
    },
    {
      "epoch": 4.83,
      "grad_norm": 5.433178681054289,
      "learning_rate": 1.9229292348050707e-06,
      "loss": 4.257,
      "step": 4222
    },
    {
      "epoch": 4.83,
      "grad_norm": 4.488418544816571,
      "learning_rate": 1.89790150038851e-06,
      "loss": 3.9483,
      "step": 4223
    },
    {
      "epoch": 4.83,
      "grad_norm": 3.222550174555159,
      "learning_rate": 1.8730371910477881e-06,
      "loss": 4.1021,
      "step": 4224
    },
    {
      "epoch": 4.83,
      "grad_norm": 3.0709484894187664,
      "learning_rate": 1.8483363204140678e-06,
      "loss": 4.0688,
      "step": 4225
    },
    {
      "epoch": 4.83,
      "grad_norm": 3.6248980556689094,
      "learning_rate": 1.8237989020287835e-06,
      "loss": 3.8968,
      "step": 4226
    },
    {
      "epoch": 4.83,
      "grad_norm": 5.8921432998410355,
      "learning_rate": 1.799424949343742e-06,
      "loss": 3.9676,
      "step": 4227
    },
    {
      "epoch": 4.83,
      "grad_norm": 3.4399832279655778,
      "learning_rate": 1.7752144757212872e-06,
      "loss": 4.1346,
      "step": 4228
    },
    {
      "epoch": 4.83,
      "grad_norm": 3.4524870727730548,
      "learning_rate": 1.7511674944340026e-06,
      "loss": 4.0019,
      "step": 4229
    },
    {
      "epoch": 4.83,
      "grad_norm": 3.320787145718268,
      "learning_rate": 1.7272840186648762e-06,
      "loss": 4.1001,
      "step": 4230
    },
    {
      "epoch": 4.84,
      "grad_norm": 7.611934753941828,
      "learning_rate": 1.7035640615072675e-06,
      "loss": 3.9197,
      "step": 4231
    },
    {
      "epoch": 4.84,
      "grad_norm": 3.5839858280563117,
      "learning_rate": 1.6800076359649417e-06,
      "loss": 3.9388,
      "step": 4232
    },
    {
      "epoch": 4.84,
      "grad_norm": 2.8072818012250047,
      "learning_rate": 1.656614754951935e-06,
      "loss": 4.1738,
      "step": 4233
    },
    {
      "epoch": 4.84,
      "grad_norm": 3.5634375543039423,
      "learning_rate": 1.6333854312926553e-06,
      "loss": 3.9566,
      "step": 4234
    },
    {
      "epoch": 4.84,
      "grad_norm": 3.525067469446047,
      "learning_rate": 1.6103196777218497e-06,
      "loss": 4.239,
      "step": 4235
    },
    {
      "epoch": 4.84,
      "grad_norm": 4.225297022597572,
      "learning_rate": 1.587417506884603e-06,
      "loss": 4.169,
      "step": 4236
    },
    {
      "epoch": 4.84,
      "grad_norm": 2.338187060397291,
      "learning_rate": 1.5646789313362718e-06,
      "loss": 3.9228,
      "step": 4237
    },
    {
      "epoch": 4.84,
      "grad_norm": 4.444444305731669,
      "learning_rate": 1.542103963542618e-06,
      "loss": 4.0758,
      "step": 4238
    },
    {
      "epoch": 4.84,
      "grad_norm": 4.219639453849816,
      "learning_rate": 1.5196926158796086e-06,
      "loss": 4.1944,
      "step": 4239
    },
    {
      "epoch": 4.85,
      "grad_norm": 3.8808928755468837,
      "learning_rate": 1.4974449006335488e-06,
      "loss": 3.9017,
      "step": 4240
    },
    {
      "epoch": 4.85,
      "grad_norm": 3.6848037384655767,
      "learning_rate": 1.4753608300011154e-06,
      "loss": 4.0909,
      "step": 4241
    },
    {
      "epoch": 4.85,
      "grad_norm": 3.6720856821322267,
      "learning_rate": 1.4534404160891578e-06,
      "loss": 4.1376,
      "step": 4242
    },
    {
      "epoch": 4.85,
      "grad_norm": 3.2732453471440524,
      "learning_rate": 1.4316836709148293e-06,
      "loss": 4.1344,
      "step": 4243
    },
    {
      "epoch": 4.85,
      "grad_norm": 2.9077824328001842,
      "learning_rate": 1.4100906064056228e-06,
      "loss": 4.0221,
      "step": 4244
    },
    {
      "epoch": 4.85,
      "grad_norm": 3.5855252111728406,
      "learning_rate": 1.3886612343992353e-06,
      "loss": 4.1997,
      "step": 4245
    },
    {
      "epoch": 4.85,
      "grad_norm": 7.88933377398823,
      "learning_rate": 1.3673955666436697e-06,
      "loss": 3.9867,
      "step": 4246
    },
    {
      "epoch": 4.85,
      "grad_norm": 3.3340897741680715,
      "learning_rate": 1.3462936147971004e-06,
      "loss": 4.0624,
      "step": 4247
    },
    {
      "epoch": 4.85,
      "grad_norm": 3.9989394213696885,
      "learning_rate": 1.3253553904280733e-06,
      "loss": 4.2275,
      "step": 4248
    },
    {
      "epoch": 4.86,
      "grad_norm": 2.9809093475882076,
      "learning_rate": 1.3045809050152733e-06,
      "loss": 4.0985,
      "step": 4249
    },
    {
      "epoch": 4.86,
      "grad_norm": 2.8112113409700425,
      "learning_rate": 1.2839701699476569e-06,
      "loss": 3.9454,
      "step": 4250
    },
    {
      "epoch": 4.86,
      "grad_norm": 7.13902909286275,
      "learning_rate": 1.2635231965244186e-06,
      "loss": 3.9962,
      "step": 4251
    },
    {
      "epoch": 4.86,
      "grad_norm": 2.2712482703131482,
      "learning_rate": 1.2432399959549589e-06,
      "loss": 3.9627,
      "step": 4252
    },
    {
      "epoch": 4.86,
      "grad_norm": 2.2415012708509496,
      "learning_rate": 1.2231205793588828e-06,
      "loss": 4.2479,
      "step": 4253
    },
    {
      "epoch": 4.86,
      "grad_norm": 2.3796691939702925,
      "learning_rate": 1.2031649577660674e-06,
      "loss": 4.2083,
      "step": 4254
    },
    {
      "epoch": 4.86,
      "grad_norm": 4.100473524706443,
      "learning_rate": 1.1833731421165616e-06,
      "loss": 4.0673,
      "step": 4255
    },
    {
      "epoch": 4.86,
      "grad_norm": 6.9469106066208965,
      "learning_rate": 1.1637451432605528e-06,
      "loss": 4.1317,
      "step": 4256
    },
    {
      "epoch": 4.87,
      "grad_norm": 3.5245754122375295,
      "learning_rate": 1.1442809719585e-06,
      "loss": 3.918,
      "step": 4257
    },
    {
      "epoch": 4.87,
      "grad_norm": 8.20987723325389,
      "learning_rate": 1.1249806388810344e-06,
      "loss": 4.1029,
      "step": 4258
    },
    {
      "epoch": 4.87,
      "grad_norm": 2.704097506477064,
      "learning_rate": 1.1058441546089258e-06,
      "loss": 4.253,
      "step": 4259
    },
    {
      "epoch": 4.87,
      "grad_norm": 3.0459610104815202,
      "learning_rate": 1.086871529633182e-06,
      "loss": 3.9569,
      "step": 4260
    },
    {
      "epoch": 4.87,
      "grad_norm": 2.4722202042835764,
      "learning_rate": 1.0680627743549164e-06,
      "loss": 3.9733,
      "step": 4261
    },
    {
      "epoch": 4.87,
      "grad_norm": 5.232631801695797,
      "learning_rate": 1.0494178990854473e-06,
      "loss": 3.8992,
      "step": 4262
    },
    {
      "epoch": 4.87,
      "grad_norm": 3.644816440535391,
      "learning_rate": 1.030936914046232e-06,
      "loss": 4.1171,
      "step": 4263
    },
    {
      "epoch": 4.87,
      "grad_norm": 2.6119471961733955,
      "learning_rate": 1.0126198293688992e-06,
      "loss": 4.1262,
      "step": 4264
    },
    {
      "epoch": 4.87,
      "grad_norm": 5.622430384165323,
      "learning_rate": 9.944666550952163e-07,
      "loss": 4.0512,
      "step": 4265
    },
    {
      "epoch": 4.88,
      "grad_norm": 7.430417663698003,
      "learning_rate": 9.764774011770559e-07,
      "loss": 4.0526,
      "step": 4266
    },
    {
      "epoch": 4.88,
      "grad_norm": 5.018548846294173,
      "learning_rate": 9.586520774765294e-07,
      "loss": 3.9743,
      "step": 4267
    },
    {
      "epoch": 4.88,
      "grad_norm": 5.763522486441058,
      "learning_rate": 9.409906937657197e-07,
      "loss": 3.939,
      "step": 4268
    },
    {
      "epoch": 4.88,
      "grad_norm": 2.8102831014498286,
      "learning_rate": 9.234932597270151e-07,
      "loss": 4.1408,
      "step": 4269
    },
    {
      "epoch": 4.88,
      "grad_norm": 3.847536595128641,
      "learning_rate": 9.061597849527425e-07,
      "loss": 4.0972,
      "step": 4270
    },
    {
      "epoch": 4.88,
      "grad_norm": 3.2299951960972715,
      "learning_rate": 8.889902789454673e-07,
      "loss": 4.0997,
      "step": 4271
    },
    {
      "epoch": 4.88,
      "grad_norm": 3.011594287761476,
      "learning_rate": 8.719847511178935e-07,
      "loss": 3.9894,
      "step": 4272
    },
    {
      "epoch": 4.88,
      "grad_norm": 4.473138311084413,
      "learning_rate": 8.551432107926637e-07,
      "loss": 3.9967,
      "step": 4273
    },
    {
      "epoch": 4.88,
      "grad_norm": 4.026569643627682,
      "learning_rate": 8.384656672026924e-07,
      "loss": 4.0304,
      "step": 4274
    },
    {
      "epoch": 4.89,
      "grad_norm": 3.0303858704766107,
      "learning_rate": 8.21952129490866e-07,
      "loss": 4.2153,
      "step": 4275
    },
    {
      "epoch": 4.89,
      "grad_norm": 3.301139705225723,
      "learning_rate": 8.056026067102761e-07,
      "loss": 4.2036,
      "step": 4276
    },
    {
      "epoch": 4.89,
      "grad_norm": 10.711978451032289,
      "learning_rate": 7.89417107823953e-07,
      "loss": 4.1081,
      "step": 4277
    },
    {
      "epoch": 4.89,
      "grad_norm": 6.137029322210865,
      "learning_rate": 7.733956417051323e-07,
      "loss": 4.0727,
      "step": 4278
    },
    {
      "epoch": 4.89,
      "grad_norm": 4.837632274165734,
      "learning_rate": 7.57538217137088e-07,
      "loss": 4.1043,
      "step": 4279
    },
    {
      "epoch": 4.89,
      "grad_norm": 5.114966585297645,
      "learning_rate": 7.41844842813133e-07,
      "loss": 4.1703,
      "step": 4280
    },
    {
      "epoch": 4.89,
      "grad_norm": 3.290231766103547,
      "learning_rate": 7.263155273366849e-07,
      "loss": 4.1182,
      "step": 4281
    },
    {
      "epoch": 4.89,
      "grad_norm": 3.6300041077518688,
      "learning_rate": 7.109502792211674e-07,
      "loss": 4.0192,
      "step": 4282
    },
    {
      "epoch": 4.89,
      "grad_norm": 3.7960047694475705,
      "learning_rate": 6.957491068901089e-07,
      "loss": 4.1291,
      "step": 4283
    },
    {
      "epoch": 4.9,
      "grad_norm": 4.578634840949222,
      "learning_rate": 6.807120186770765e-07,
      "loss": 4.1268,
      "step": 4284
    },
    {
      "epoch": 4.9,
      "grad_norm": 2.604940314489177,
      "learning_rate": 6.65839022825676e-07,
      "loss": 3.9997,
      "step": 4285
    },
    {
      "epoch": 4.9,
      "grad_norm": 6.39501249214444,
      "learning_rate": 6.511301274895853e-07,
      "loss": 3.993,
      "step": 4286
    },
    {
      "epoch": 4.9,
      "grad_norm": 2.7532436713612354,
      "learning_rate": 6.365853407324206e-07,
      "loss": 4.1458,
      "step": 4287
    },
    {
      "epoch": 4.9,
      "grad_norm": 3.516494498617141,
      "learning_rate": 6.222046705280038e-07,
      "loss": 4.2123,
      "step": 4288
    },
    {
      "epoch": 4.9,
      "grad_norm": 4.365080384876002,
      "learning_rate": 6.07988124759995e-07,
      "loss": 4.0238,
      "step": 4289
    },
    {
      "epoch": 4.9,
      "grad_norm": 3.445366122923398,
      "learning_rate": 5.939357112222265e-07,
      "loss": 4.1211,
      "step": 4290
    },
    {
      "epoch": 4.9,
      "grad_norm": 4.123339418003066,
      "learning_rate": 5.800474376184361e-07,
      "loss": 4.031,
      "step": 4291
    },
    {
      "epoch": 4.91,
      "grad_norm": 4.498494920250651,
      "learning_rate": 5.663233115624999e-07,
      "loss": 4.1058,
      "step": 4292
    },
    {
      "epoch": 4.91,
      "grad_norm": 4.470794214364937,
      "learning_rate": 5.527633405781662e-07,
      "loss": 4.001,
      "step": 4293
    },
    {
      "epoch": 4.91,
      "grad_norm": 3.458606025466353,
      "learning_rate": 5.393675320993219e-07,
      "loss": 4.1887,
      "step": 4294
    },
    {
      "epoch": 4.91,
      "grad_norm": 4.931505254193795,
      "learning_rate": 5.261358934697591e-07,
      "loss": 4.1386,
      "step": 4295
    },
    {
      "epoch": 4.91,
      "grad_norm": 5.090000354377346,
      "learning_rate": 5.130684319433087e-07,
      "loss": 4.3127,
      "step": 4296
    },
    {
      "epoch": 4.91,
      "grad_norm": 4.323501093793252,
      "learning_rate": 5.001651546838071e-07,
      "loss": 4.145,
      "step": 4297
    },
    {
      "epoch": 4.91,
      "grad_norm": 4.76244980723696,
      "learning_rate": 4.874260687650622e-07,
      "loss": 4.1077,
      "step": 4298
    },
    {
      "epoch": 4.91,
      "grad_norm": 3.529134593421125,
      "learning_rate": 4.748511811708877e-07,
      "loss": 4.0039,
      "step": 4299
    },
    {
      "epoch": 4.91,
      "grad_norm": 3.4278601172633723,
      "learning_rate": 4.624404987950692e-07,
      "loss": 4.1369,
      "step": 4300
    },
    {
      "epoch": 4.91,
      "eval_blimp_filtered_avg": 0.5222388059701493,
      "eval_blimp_filtered_std": 0.0057290963420511135,
      "step": 4300
    },
    {
      "epoch": 4.91,
      "eval_blimp_supplement_avg": 0.46120689655172414,
      "eval_blimp_supplement_std": 0.022554741357676152,
      "step": 4300
    },
    {
      "epoch": 4.91,
      "eval_vqa_filtered_avg": 0.36,
      "eval_vqa_filtered_std": 0.04824181513244218,
      "step": 4300
    },
    {
      "epoch": 4.91,
      "eval_winoground_filtered_avg": 0.53,
      "eval_winoground_filtered_std": 0.0501613558046592,
      "step": 4300
    },
    {
      "epoch": 4.92,
      "grad_norm": 3.9656988221212077,
      "learning_rate": 4.501940284413308e-07,
      "loss": 3.9099,
      "step": 4301
    },
    {
      "epoch": 4.92,
      "grad_norm": 6.409908685271376,
      "learning_rate": 4.381117768234688e-07,
      "loss": 3.8438,
      "step": 4302
    },
    {
      "epoch": 4.92,
      "grad_norm": 11.677065277470783,
      "learning_rate": 4.2619375056518467e-07,
      "loss": 4.0962,
      "step": 4303
    },
    {
      "epoch": 4.92,
      "grad_norm": 2.8510011030762916,
      "learning_rate": 4.1443995620015214e-07,
      "loss": 4.0841,
      "step": 4304
    },
    {
      "epoch": 4.92,
      "grad_norm": 3.9867019766809775,
      "learning_rate": 4.0285040017198343e-07,
      "loss": 4.0376,
      "step": 4305
    },
    {
      "epoch": 4.92,
      "grad_norm": 6.310020011240259,
      "learning_rate": 3.9142508883436284e-07,
      "loss": 4.121,
      "step": 4306
    },
    {
      "epoch": 4.92,
      "grad_norm": 3.111450967658074,
      "learning_rate": 3.8016402845078007e-07,
      "loss": 4.1149,
      "step": 4307
    },
    {
      "epoch": 4.92,
      "grad_norm": 3.7600243324669895,
      "learning_rate": 3.690672251947968e-07,
      "loss": 4.0199,
      "step": 4308
    },
    {
      "epoch": 4.92,
      "grad_norm": 2.573730214908648,
      "learning_rate": 3.581346851498801e-07,
      "loss": 4.1373,
      "step": 4309
    },
    {
      "epoch": 4.93,
      "grad_norm": 2.977436646327557,
      "learning_rate": 3.47366414309469e-07,
      "loss": 4.1751,
      "step": 4310
    },
    {
      "epoch": 4.93,
      "grad_norm": 2.270514061523291,
      "learning_rate": 3.367624185768747e-07,
      "loss": 3.8962,
      "step": 4311
    },
    {
      "epoch": 4.93,
      "grad_norm": 4.262224850159704,
      "learning_rate": 3.2632270376548033e-07,
      "loss": 4.0127,
      "step": 4312
    },
    {
      "epoch": 4.93,
      "grad_norm": 3.186231074858047,
      "learning_rate": 3.160472755984411e-07,
      "loss": 4.0488,
      "step": 4313
    },
    {
      "epoch": 4.93,
      "grad_norm": 6.113654909726644,
      "learning_rate": 3.0593613970901763e-07,
      "loss": 3.9435,
      "step": 4314
    },
    {
      "epoch": 4.93,
      "grad_norm": 3.973467150168338,
      "learning_rate": 2.959893016402759e-07,
      "loss": 4.1979,
      "step": 4315
    },
    {
      "epoch": 4.93,
      "grad_norm": 3.01520381173676,
      "learning_rate": 2.862067668452872e-07,
      "loss": 4.2707,
      "step": 4316
    },
    {
      "epoch": 4.93,
      "grad_norm": 3.7118286004966703,
      "learning_rate": 2.765885406869617e-07,
      "loss": 4.0569,
      "step": 4317
    },
    {
      "epoch": 4.93,
      "grad_norm": 4.307557593584464,
      "learning_rate": 2.6713462843824806e-07,
      "loss": 4.0975,
      "step": 4318
    },
    {
      "epoch": 4.94,
      "grad_norm": 5.424494644087411,
      "learning_rate": 2.578450352819339e-07,
      "loss": 4.3158,
      "step": 4319
    },
    {
      "epoch": 4.94,
      "grad_norm": 4.8734211009708694,
      "learning_rate": 2.487197663107121e-07,
      "loss": 4.0962,
      "step": 4320
    },
    {
      "epoch": 4.94,
      "grad_norm": 2.7765695327363455,
      "learning_rate": 2.39758826527281e-07,
      "loss": 4.3301,
      "step": 4321
    },
    {
      "epoch": 4.94,
      "grad_norm": 3.8955758623928287,
      "learning_rate": 2.3096222084417747e-07,
      "loss": 4.1206,
      "step": 4322
    },
    {
      "epoch": 4.94,
      "grad_norm": 5.4124974380313695,
      "learning_rate": 2.2232995408384414e-07,
      "loss": 3.8779,
      "step": 4323
    },
    {
      "epoch": 4.94,
      "grad_norm": 2.110442629804503,
      "learning_rate": 2.1386203097866207e-07,
      "loss": 3.9958,
      "step": 4324
    },
    {
      "epoch": 4.94,
      "grad_norm": 3.6227525322338123,
      "learning_rate": 2.055584561708845e-07,
      "loss": 4.1369,
      "step": 4325
    },
    {
      "epoch": 4.94,
      "grad_norm": 3.1139243980873474,
      "learning_rate": 1.9741923421270345e-07,
      "loss": 4.0377,
      "step": 4326
    },
    {
      "epoch": 4.95,
      "grad_norm": 6.272368813218213,
      "learning_rate": 1.8944436956621622e-07,
      "loss": 4.0219,
      "step": 4327
    },
    {
      "epoch": 4.95,
      "grad_norm": 3.1277456396298144,
      "learning_rate": 1.8163386660332567e-07,
      "loss": 3.955,
      "step": 4328
    },
    {
      "epoch": 4.95,
      "grad_norm": 3.081851706869889,
      "learning_rate": 1.7398772960593998e-07,
      "loss": 4.0061,
      "step": 4329
    },
    {
      "epoch": 4.95,
      "grad_norm": 3.64514640094717,
      "learning_rate": 1.6650596276580608e-07,
      "loss": 4.0648,
      "step": 4330
    },
    {
      "epoch": 4.95,
      "grad_norm": 4.525689201332529,
      "learning_rate": 1.5918857018457633e-07,
      "loss": 3.9569,
      "step": 4331
    },
    {
      "epoch": 4.95,
      "grad_norm": 6.700132818395476,
      "learning_rate": 1.5203555587374183e-07,
      "loss": 3.9206,
      "step": 4332
    },
    {
      "epoch": 4.95,
      "grad_norm": 3.2099007417713357,
      "learning_rate": 1.4504692375476578e-07,
      "loss": 4.2668,
      "step": 4333
    },
    {
      "epoch": 4.95,
      "grad_norm": 4.540505640331635,
      "learning_rate": 1.382226776589168e-07,
      "loss": 3.9683,
      "step": 4334
    },
    {
      "epoch": 4.95,
      "grad_norm": 7.941807372233224,
      "learning_rate": 1.3156282132736895e-07,
      "loss": 4.1645,
      "step": 4335
    },
    {
      "epoch": 4.96,
      "grad_norm": 5.7960715694419935,
      "learning_rate": 1.2506735841120164e-07,
      "loss": 4.1258,
      "step": 4336
    },
    {
      "epoch": 4.96,
      "grad_norm": 3.002632555038053,
      "learning_rate": 1.1873629247129978e-07,
      "loss": 4.0374,
      "step": 4337
    },
    {
      "epoch": 4.96,
      "grad_norm": 2.48403033346022,
      "learning_rate": 1.1256962697852035e-07,
      "loss": 4.0181,
      "step": 4338
    },
    {
      "epoch": 4.96,
      "grad_norm": 3.9689927917512273,
      "learning_rate": 1.0656736531355903e-07,
      "loss": 4.0765,
      "step": 4339
    },
    {
      "epoch": 4.96,
      "grad_norm": 7.784373544895908,
      "learning_rate": 1.0072951076688374e-07,
      "loss": 3.912,
      "step": 4340
    },
    {
      "epoch": 4.96,
      "grad_norm": 2.7297961773294372,
      "learning_rate": 9.5056066539001e-08,
      "loss": 4.0069,
      "step": 4341
    },
    {
      "epoch": 4.96,
      "grad_norm": 3.6192125267970545,
      "learning_rate": 8.95470357401562e-08,
      "loss": 4.0035,
      "step": 4342
    },
    {
      "epoch": 4.96,
      "grad_norm": 6.553771956595043,
      "learning_rate": 8.420242139050016e-08,
      "loss": 4.1604,
      "step": 4343
    },
    {
      "epoch": 4.96,
      "grad_norm": 2.3115108875947934,
      "learning_rate": 7.902222642005574e-08,
      "loss": 4.0727,
      "step": 4344
    },
    {
      "epoch": 4.97,
      "grad_norm": 3.4405287209059794,
      "learning_rate": 7.400645366871794e-08,
      "loss": 4.082,
      "step": 4345
    },
    {
      "epoch": 4.97,
      "grad_norm": 3.097730092068558,
      "learning_rate": 6.915510588618723e-08,
      "loss": 4.3861,
      "step": 4346
    },
    {
      "epoch": 4.97,
      "grad_norm": 3.382189133781147,
      "learning_rate": 6.446818573210277e-08,
      "loss": 4.1442,
      "step": 4347
    },
    {
      "epoch": 4.97,
      "grad_norm": 3.5959033960081532,
      "learning_rate": 5.994569577590924e-08,
      "loss": 4.0968,
      "step": 4348
    },
    {
      "epoch": 4.97,
      "grad_norm": 4.6422708275639115,
      "learning_rate": 5.558763849689007e-08,
      "loss": 3.939,
      "step": 4349
    },
    {
      "epoch": 4.97,
      "grad_norm": 3.21636597272728,
      "learning_rate": 5.1394016284234117e-08,
      "loss": 4.0155,
      "step": 4350
    },
    {
      "epoch": 4.97,
      "grad_norm": 4.011094645916989,
      "learning_rate": 4.736483143696901e-08,
      "loss": 4.1768,
      "step": 4351
    },
    {
      "epoch": 4.97,
      "grad_norm": 3.5146615393447362,
      "learning_rate": 4.350008616399447e-08,
      "loss": 4.1232,
      "step": 4352
    },
    {
      "epoch": 4.97,
      "grad_norm": 3.586965568572597,
      "learning_rate": 3.979978258398242e-08,
      "loss": 3.8865,
      "step": 4353
    },
    {
      "epoch": 4.98,
      "grad_norm": 3.066656773249175,
      "learning_rate": 3.626392272551015e-08,
      "loss": 4.0584,
      "step": 4354
    },
    {
      "epoch": 4.98,
      "grad_norm": 3.0692073805272875,
      "learning_rate": 3.2892508527027074e-08,
      "loss": 3.92,
      "step": 4355
    },
    {
      "epoch": 4.98,
      "grad_norm": 2.781943776469792,
      "learning_rate": 2.968554183682137e-08,
      "loss": 4.1927,
      "step": 4356
    },
    {
      "epoch": 4.98,
      "grad_norm": 2.9928594043431183,
      "learning_rate": 2.6643024412986713e-08,
      "loss": 3.9699,
      "step": 4357
    },
    {
      "epoch": 4.98,
      "grad_norm": 3.2123473659005617,
      "learning_rate": 2.3764957923455563e-08,
      "loss": 4.1257,
      "step": 4358
    },
    {
      "epoch": 4.98,
      "grad_norm": 2.361301331532532,
      "learning_rate": 2.1051343946099086e-08,
      "loss": 4.247,
      "step": 4359
    },
    {
      "epoch": 4.98,
      "grad_norm": 4.160195816468606,
      "learning_rate": 1.8502183968527318e-08,
      "loss": 4.1823,
      "step": 4360
    },
    {
      "epoch": 4.98,
      "grad_norm": 3.2862173563304613,
      "learning_rate": 1.6117479388255695e-08,
      "loss": 4.1217,
      "step": 4361
    },
    {
      "epoch": 4.99,
      "grad_norm": 8.051234267083998,
      "learning_rate": 1.3897231512638441e-08,
      "loss": 4.198,
      "step": 4362
    },
    {
      "epoch": 4.99,
      "grad_norm": 4.445218180405192,
      "learning_rate": 1.1841441558801957e-08,
      "loss": 4.1696,
      "step": 4363
    },
    {
      "epoch": 4.99,
      "grad_norm": 4.397289277421306,
      "learning_rate": 9.95011065381135e-09,
      "loss": 4.1526,
      "step": 4364
    },
    {
      "epoch": 4.99,
      "grad_norm": 2.766834149990971,
      "learning_rate": 8.22323983453721e-09,
      "loss": 4.1963,
      "step": 4365
    },
    {
      "epoch": 4.99,
      "grad_norm": 3.7251947638100575,
      "learning_rate": 6.660830047655608e-09,
      "loss": 4.0511,
      "step": 4366
    },
    {
      "epoch": 4.99,
      "grad_norm": 3.21397471458884,
      "learning_rate": 5.262882149748015e-09,
      "loss": 4.1718,
      "step": 4367
    },
    {
      "epoch": 4.99,
      "grad_norm": 3.1957737714726817,
      "learning_rate": 4.0293969071347744e-09,
      "loss": 3.9336,
      "step": 4368
    },
    {
      "epoch": 4.99,
      "grad_norm": 2.7131171639178047,
      "learning_rate": 2.960374996074932e-09,
      "loss": 4.1371,
      "step": 4369
    },
    {
      "epoch": 4.99,
      "grad_norm": 4.961320187940303,
      "learning_rate": 2.055817002599713e-09,
      "loss": 4.2498,
      "step": 4370
    },
    {
      "epoch": 5.0,
      "grad_norm": 3.3099955852103706,
      "learning_rate": 1.3157234226457426e-09,
      "loss": 4.2319,
      "step": 4371
    },
    {
      "epoch": 5.0,
      "grad_norm": 3.74023747011329,
      "learning_rate": 7.400946619218196e-10,
      "loss": 4.2399,
      "step": 4372
    },
    {
      "epoch": 5.0,
      "grad_norm": 2.528757091762162,
      "learning_rate": 3.289310359755326e-10,
      "loss": 4.0392,
      "step": 4373
    },
    {
      "epoch": 5.0,
      "grad_norm": 3.6277130145875223,
      "learning_rate": 8.22327702598713e-11,
      "loss": 4.0927,
      "step": 4374
    },
    {
      "epoch": 5.0,
      "grad_norm": 2.5685903470353098,
      "learning_rate": 0.0,
      "loss": 3.9198,
      "step": 4375
    },
    {
      "epoch": 5.0,
      "step": 4375,
      "total_flos": 75070389288960.0,
      "train_loss": 4.425508696310861,
      "train_runtime": 40327.6362,
      "train_samples_per_second": 13.872,
      "train_steps_per_second": 0.108
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 4375,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "total_flos": 75070389288960.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}