{
  "best_metric": 1.1458154916763306,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.4550625711035267,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0022753128555176336,
      "grad_norm": 1.4814585447311401,
      "learning_rate": 1e-05,
      "loss": 2.3828,
      "step": 1
    },
    {
      "epoch": 0.0022753128555176336,
      "eval_loss": 1.5660679340362549,
      "eval_runtime": 34.0892,
      "eval_samples_per_second": 21.708,
      "eval_steps_per_second": 5.427,
      "step": 1
    },
    {
      "epoch": 0.004550625711035267,
      "grad_norm": 1.809823989868164,
      "learning_rate": 2e-05,
      "loss": 3.0001,
      "step": 2
    },
    {
      "epoch": 0.006825938566552901,
      "grad_norm": 1.7375787496566772,
      "learning_rate": 3e-05,
      "loss": 2.6469,
      "step": 3
    },
    {
      "epoch": 0.009101251422070534,
      "grad_norm": 1.6638267040252686,
      "learning_rate": 4e-05,
      "loss": 3.1323,
      "step": 4
    },
    {
      "epoch": 0.011376564277588168,
      "grad_norm": 1.637603521347046,
      "learning_rate": 5e-05,
      "loss": 3.1495,
      "step": 5
    },
    {
      "epoch": 0.013651877133105802,
      "grad_norm": 1.2763593196868896,
      "learning_rate": 6e-05,
      "loss": 3.3158,
      "step": 6
    },
    {
      "epoch": 0.015927189988623434,
      "grad_norm": 1.588097333908081,
      "learning_rate": 7e-05,
      "loss": 3.4439,
      "step": 7
    },
    {
      "epoch": 0.01820250284414107,
      "grad_norm": 1.7552661895751953,
      "learning_rate": 8e-05,
      "loss": 3.5118,
      "step": 8
    },
    {
      "epoch": 0.020477815699658702,
      "grad_norm": 1.539528489112854,
      "learning_rate": 9e-05,
      "loss": 3.2236,
      "step": 9
    },
    {
      "epoch": 0.022753128555176336,
      "grad_norm": 1.5751824378967285,
      "learning_rate": 0.0001,
      "loss": 3.8917,
      "step": 10
    },
    {
      "epoch": 0.02502844141069397,
      "grad_norm": 1.8290205001831055,
      "learning_rate": 9.999316524962345e-05,
      "loss": 5.0552,
      "step": 11
    },
    {
      "epoch": 0.027303754266211604,
      "grad_norm": 1.559188961982727,
      "learning_rate": 9.997266286704631e-05,
      "loss": 4.7131,
      "step": 12
    },
    {
      "epoch": 0.02957906712172924,
      "grad_norm": 1.130842924118042,
      "learning_rate": 9.993849845741524e-05,
      "loss": 3.8074,
      "step": 13
    },
    {
      "epoch": 0.03185437997724687,
      "grad_norm": 0.9377134442329407,
      "learning_rate": 9.989068136093873e-05,
      "loss": 3.4978,
      "step": 14
    },
    {
      "epoch": 0.034129692832764506,
      "grad_norm": 1.0720447301864624,
      "learning_rate": 9.98292246503335e-05,
      "loss": 4.536,
      "step": 15
    },
    {
      "epoch": 0.03640500568828214,
      "grad_norm": 1.1191229820251465,
      "learning_rate": 9.975414512725057e-05,
      "loss": 4.8975,
      "step": 16
    },
    {
      "epoch": 0.038680318543799774,
      "grad_norm": 1.1141483783721924,
      "learning_rate": 9.966546331768191e-05,
      "loss": 4.0189,
      "step": 17
    },
    {
      "epoch": 0.040955631399317405,
      "grad_norm": 1.2463666200637817,
      "learning_rate": 9.956320346634876e-05,
      "loss": 5.0045,
      "step": 18
    },
    {
      "epoch": 0.04323094425483504,
      "grad_norm": 1.1816623210906982,
      "learning_rate": 9.944739353007344e-05,
      "loss": 4.2725,
      "step": 19
    },
    {
      "epoch": 0.04550625711035267,
      "grad_norm": 1.16055428981781,
      "learning_rate": 9.931806517013612e-05,
      "loss": 4.4338,
      "step": 20
    },
    {
      "epoch": 0.04778156996587031,
      "grad_norm": 1.2772587537765503,
      "learning_rate": 9.917525374361912e-05,
      "loss": 5.4265,
      "step": 21
    },
    {
      "epoch": 0.05005688282138794,
      "grad_norm": 1.400126338005066,
      "learning_rate": 9.901899829374047e-05,
      "loss": 4.9977,
      "step": 22
    },
    {
      "epoch": 0.05233219567690557,
      "grad_norm": 1.5492485761642456,
      "learning_rate": 9.884934153917997e-05,
      "loss": 5.8469,
      "step": 23
    },
    {
      "epoch": 0.05460750853242321,
      "grad_norm": 1.3513226509094238,
      "learning_rate": 9.86663298624003e-05,
      "loss": 6.2075,
      "step": 24
    },
    {
      "epoch": 0.05688282138794084,
      "grad_norm": 1.2896218299865723,
      "learning_rate": 9.847001329696653e-05,
      "loss": 5.7344,
      "step": 25
    },
    {
      "epoch": 0.05915813424345848,
      "grad_norm": 1.4297120571136475,
      "learning_rate": 9.826044551386744e-05,
      "loss": 6.5411,
      "step": 26
    },
    {
      "epoch": 0.06143344709897611,
      "grad_norm": 1.1831682920455933,
      "learning_rate": 9.803768380684242e-05,
      "loss": 4.8473,
      "step": 27
    },
    {
      "epoch": 0.06370875995449374,
      "grad_norm": 1.25262451171875,
      "learning_rate": 9.780178907671789e-05,
      "loss": 5.1935,
      "step": 28
    },
    {
      "epoch": 0.06598407281001138,
      "grad_norm": 1.2811355590820312,
      "learning_rate": 9.755282581475769e-05,
      "loss": 6.2861,
      "step": 29
    },
    {
      "epoch": 0.06825938566552901,
      "grad_norm": 1.473062515258789,
      "learning_rate": 9.729086208503174e-05,
      "loss": 6.5451,
      "step": 30
    },
    {
      "epoch": 0.07053469852104664,
      "grad_norm": 1.5842453241348267,
      "learning_rate": 9.701596950580806e-05,
      "loss": 7.2188,
      "step": 31
    },
    {
      "epoch": 0.07281001137656427,
      "grad_norm": 1.5610548257827759,
      "learning_rate": 9.672822322997305e-05,
      "loss": 7.254,
      "step": 32
    },
    {
      "epoch": 0.07508532423208192,
      "grad_norm": 1.6412367820739746,
      "learning_rate": 9.642770192448536e-05,
      "loss": 7.2967,
      "step": 33
    },
    {
      "epoch": 0.07736063708759955,
      "grad_norm": 1.4503934383392334,
      "learning_rate": 9.611448774886924e-05,
      "loss": 6.6187,
      "step": 34
    },
    {
      "epoch": 0.07963594994311718,
      "grad_norm": 1.8282753229141235,
      "learning_rate": 9.578866633275288e-05,
      "loss": 6.3759,
      "step": 35
    },
    {
      "epoch": 0.08191126279863481,
      "grad_norm": 1.420009970664978,
      "learning_rate": 9.545032675245813e-05,
      "loss": 5.6363,
      "step": 36
    },
    {
      "epoch": 0.08418657565415244,
      "grad_norm": 1.688437581062317,
      "learning_rate": 9.509956150664796e-05,
      "loss": 7.1337,
      "step": 37
    },
    {
      "epoch": 0.08646188850967008,
      "grad_norm": 1.7707394361495972,
      "learning_rate": 9.473646649103818e-05,
      "loss": 6.7484,
      "step": 38
    },
    {
      "epoch": 0.08873720136518772,
      "grad_norm": 1.6407301425933838,
      "learning_rate": 9.43611409721806e-05,
      "loss": 6.9463,
      "step": 39
    },
    {
      "epoch": 0.09101251422070535,
      "grad_norm": 1.7492953538894653,
      "learning_rate": 9.397368756032445e-05,
      "loss": 7.4364,
      "step": 40
    },
    {
      "epoch": 0.09328782707622298,
      "grad_norm": 1.7869080305099487,
      "learning_rate": 9.357421218136386e-05,
      "loss": 7.4718,
      "step": 41
    },
    {
      "epoch": 0.09556313993174062,
      "grad_norm": 1.9442745447158813,
      "learning_rate": 9.316282404787871e-05,
      "loss": 7.2835,
      "step": 42
    },
    {
      "epoch": 0.09783845278725825,
      "grad_norm": 2.0581507682800293,
      "learning_rate": 9.273963562927695e-05,
      "loss": 8.3224,
      "step": 43
    },
    {
      "epoch": 0.10011376564277588,
      "grad_norm": 1.7309566736221313,
      "learning_rate": 9.230476262104677e-05,
      "loss": 7.1928,
      "step": 44
    },
    {
      "epoch": 0.10238907849829351,
      "grad_norm": 1.7606233358383179,
      "learning_rate": 9.185832391312644e-05,
      "loss": 6.7532,
      "step": 45
    },
    {
      "epoch": 0.10466439135381114,
      "grad_norm": 2.086742401123047,
      "learning_rate": 9.140044155740101e-05,
      "loss": 7.9023,
      "step": 46
    },
    {
      "epoch": 0.10693970420932879,
      "grad_norm": 2.0646538734436035,
      "learning_rate": 9.093124073433463e-05,
      "loss": 7.9228,
      "step": 47
    },
    {
      "epoch": 0.10921501706484642,
      "grad_norm": 2.0563600063323975,
      "learning_rate": 9.045084971874738e-05,
      "loss": 6.8809,
      "step": 48
    },
    {
      "epoch": 0.11149032992036405,
      "grad_norm": 2.717761754989624,
      "learning_rate": 8.995939984474624e-05,
      "loss": 8.6021,
      "step": 49
    },
    {
      "epoch": 0.11376564277588168,
      "grad_norm": 4.531638145446777,
      "learning_rate": 8.945702546981969e-05,
      "loss": 9.3771,
      "step": 50
    },
    {
      "epoch": 0.11376564277588168,
      "eval_loss": 1.2955468893051147,
      "eval_runtime": 34.8798,
      "eval_samples_per_second": 21.216,
      "eval_steps_per_second": 5.304,
      "step": 50
    },
    {
      "epoch": 0.11604095563139932,
      "grad_norm": 1.4643594026565552,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.5717,
      "step": 51
    },
    {
      "epoch": 0.11831626848691695,
      "grad_norm": 1.0985218286514282,
      "learning_rate": 8.842005554284296e-05,
      "loss": 3.0408,
      "step": 52
    },
    {
      "epoch": 0.12059158134243458,
      "grad_norm": 1.0725528001785278,
      "learning_rate": 8.788574348801675e-05,
      "loss": 2.6001,
      "step": 53
    },
    {
      "epoch": 0.12286689419795221,
      "grad_norm": 0.838362991809845,
      "learning_rate": 8.73410738492077e-05,
      "loss": 2.448,
      "step": 54
    },
    {
      "epoch": 0.12514220705346984,
      "grad_norm": 0.8049578666687012,
      "learning_rate": 8.678619553365659e-05,
      "loss": 2.32,
      "step": 55
    },
    {
      "epoch": 0.12741751990898748,
      "grad_norm": 0.8058600425720215,
      "learning_rate": 8.622126023955446e-05,
      "loss": 2.2641,
      "step": 56
    },
    {
      "epoch": 0.1296928327645051,
      "grad_norm": 0.942395031452179,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.9597,
      "step": 57
    },
    {
      "epoch": 0.13196814562002276,
      "grad_norm": 1.0960770845413208,
      "learning_rate": 8.506183921362443e-05,
      "loss": 3.1577,
      "step": 58
    },
    {
      "epoch": 0.1342434584755404,
      "grad_norm": 1.0346654653549194,
      "learning_rate": 8.44676704559283e-05,
      "loss": 3.1914,
      "step": 59
    },
    {
      "epoch": 0.13651877133105803,
      "grad_norm": 1.1430764198303223,
      "learning_rate": 8.386407858128706e-05,
      "loss": 3.5692,
      "step": 60
    },
    {
      "epoch": 0.13879408418657566,
      "grad_norm": 0.9637904763221741,
      "learning_rate": 8.32512286056924e-05,
      "loss": 3.2595,
      "step": 61
    },
    {
      "epoch": 0.1410693970420933,
      "grad_norm": 1.0756419897079468,
      "learning_rate": 8.262928807620843e-05,
      "loss": 4.0026,
      "step": 62
    },
    {
      "epoch": 0.14334470989761092,
      "grad_norm": 1.0559083223342896,
      "learning_rate": 8.199842702516583e-05,
      "loss": 3.6662,
      "step": 63
    },
    {
      "epoch": 0.14562002275312855,
      "grad_norm": 1.2225507497787476,
      "learning_rate": 8.135881792367686e-05,
      "loss": 4.4012,
      "step": 64
    },
    {
      "epoch": 0.14789533560864618,
      "grad_norm": 1.1780744791030884,
      "learning_rate": 8.07106356344834e-05,
      "loss": 4.6059,
      "step": 65
    },
    {
      "epoch": 0.15017064846416384,
      "grad_norm": 1.085599422454834,
      "learning_rate": 8.005405736415126e-05,
      "loss": 3.6639,
      "step": 66
    },
    {
      "epoch": 0.15244596131968147,
      "grad_norm": 1.4645206928253174,
      "learning_rate": 7.938926261462366e-05,
      "loss": 5.5207,
      "step": 67
    },
    {
      "epoch": 0.1547212741751991,
      "grad_norm": 1.072421669960022,
      "learning_rate": 7.871643313414718e-05,
      "loss": 3.5846,
      "step": 68
    },
    {
      "epoch": 0.15699658703071673,
      "grad_norm": 1.3423982858657837,
      "learning_rate": 7.803575286758364e-05,
      "loss": 4.7186,
      "step": 69
    },
    {
      "epoch": 0.15927189988623436,
      "grad_norm": 1.454968810081482,
      "learning_rate": 7.734740790612136e-05,
      "loss": 5.1052,
      "step": 70
    },
    {
      "epoch": 0.161547212741752,
      "grad_norm": 1.1506472826004028,
      "learning_rate": 7.66515864363997e-05,
      "loss": 3.8436,
      "step": 71
    },
    {
      "epoch": 0.16382252559726962,
      "grad_norm": 1.4477007389068604,
      "learning_rate": 7.594847868906076e-05,
      "loss": 5.1956,
      "step": 72
    },
    {
      "epoch": 0.16609783845278725,
      "grad_norm": 1.6539088487625122,
      "learning_rate": 7.52382768867422e-05,
      "loss": 6.4884,
      "step": 73
    },
    {
      "epoch": 0.16837315130830488,
      "grad_norm": 1.5916926860809326,
      "learning_rate": 7.452117519152542e-05,
      "loss": 6.248,
      "step": 74
    },
    {
      "epoch": 0.17064846416382254,
      "grad_norm": 1.4304360151290894,
      "learning_rate": 7.379736965185368e-05,
      "loss": 5.4603,
      "step": 75
    },
    {
      "epoch": 0.17292377701934017,
      "grad_norm": 1.5564725399017334,
      "learning_rate": 7.30670581489344e-05,
      "loss": 5.9565,
      "step": 76
    },
    {
      "epoch": 0.1751990898748578,
      "grad_norm": 1.5194623470306396,
      "learning_rate": 7.233044034264034e-05,
      "loss": 4.3343,
      "step": 77
    },
    {
      "epoch": 0.17747440273037543,
      "grad_norm": 1.5293631553649902,
      "learning_rate": 7.158771761692464e-05,
      "loss": 5.1852,
      "step": 78
    },
    {
      "epoch": 0.17974971558589306,
      "grad_norm": 1.5281931161880493,
      "learning_rate": 7.083909302476453e-05,
      "loss": 5.276,
      "step": 79
    },
    {
      "epoch": 0.1820250284414107,
      "grad_norm": 1.7697945833206177,
      "learning_rate": 7.008477123264848e-05,
      "loss": 5.9039,
      "step": 80
    },
    {
      "epoch": 0.18430034129692832,
      "grad_norm": 1.3789178133010864,
      "learning_rate": 6.932495846462261e-05,
      "loss": 4.2209,
      "step": 81
    },
    {
      "epoch": 0.18657565415244595,
      "grad_norm": 1.9531902074813843,
      "learning_rate": 6.855986244591104e-05,
      "loss": 6.706,
      "step": 82
    },
    {
      "epoch": 0.18885096700796358,
      "grad_norm": 1.6132508516311646,
      "learning_rate": 6.778969234612584e-05,
      "loss": 5.145,
      "step": 83
    },
    {
      "epoch": 0.19112627986348124,
      "grad_norm": 1.9818165302276611,
      "learning_rate": 6.701465872208216e-05,
      "loss": 6.3047,
      "step": 84
    },
    {
      "epoch": 0.19340159271899887,
      "grad_norm": 1.673631191253662,
      "learning_rate": 6.623497346023418e-05,
      "loss": 5.9104,
      "step": 85
    },
    {
      "epoch": 0.1956769055745165,
      "grad_norm": 1.704563856124878,
      "learning_rate": 6.545084971874738e-05,
      "loss": 5.916,
      "step": 86
    },
    {
      "epoch": 0.19795221843003413,
      "grad_norm": 1.7582212686538696,
      "learning_rate": 6.466250186922325e-05,
      "loss": 5.5365,
      "step": 87
    },
    {
      "epoch": 0.20022753128555176,
      "grad_norm": 2.1068577766418457,
      "learning_rate": 6.387014543809223e-05,
      "loss": 6.0633,
      "step": 88
    },
    {
      "epoch": 0.2025028441410694,
      "grad_norm": 2.204310655593872,
      "learning_rate": 6.307399704769099e-05,
      "loss": 6.8995,
      "step": 89
    },
    {
      "epoch": 0.20477815699658702,
      "grad_norm": 1.9469975233078003,
      "learning_rate": 6.227427435703997e-05,
      "loss": 6.4982,
      "step": 90
    },
    {
      "epoch": 0.20705346985210465,
      "grad_norm": 2.011195421218872,
      "learning_rate": 6.147119600233758e-05,
      "loss": 7.0875,
      "step": 91
    },
    {
      "epoch": 0.20932878270762229,
      "grad_norm": 2.0855579376220703,
      "learning_rate": 6.066498153718735e-05,
      "loss": 6.4728,
      "step": 92
    },
    {
      "epoch": 0.21160409556313994,
      "grad_norm": 1.970643401145935,
      "learning_rate": 5.985585137257401e-05,
      "loss": 6.4405,
      "step": 93
    },
    {
      "epoch": 0.21387940841865757,
      "grad_norm": 2.05020809173584,
      "learning_rate": 5.90440267166055e-05,
      "loss": 6.6298,
      "step": 94
    },
    {
      "epoch": 0.2161547212741752,
      "grad_norm": 2.370215654373169,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 7.0923,
      "step": 95
    },
    {
      "epoch": 0.21843003412969283,
      "grad_norm": 2.514678955078125,
      "learning_rate": 5.74131823855921e-05,
      "loss": 7.519,
      "step": 96
    },
    {
      "epoch": 0.22070534698521047,
      "grad_norm": 2.4687697887420654,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 7.1018,
      "step": 97
    },
    {
      "epoch": 0.2229806598407281,
      "grad_norm": 2.858738422393799,
      "learning_rate": 5.577423184847932e-05,
      "loss": 7.9406,
      "step": 98
    },
    {
      "epoch": 0.22525597269624573,
      "grad_norm": 2.983034372329712,
      "learning_rate": 5.495227651252315e-05,
      "loss": 7.4763,
      "step": 99
    },
    {
      "epoch": 0.22753128555176336,
      "grad_norm": 4.467612266540527,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 7.841,
      "step": 100
    },
    {
      "epoch": 0.22753128555176336,
      "eval_loss": 1.2031514644622803,
      "eval_runtime": 34.9022,
      "eval_samples_per_second": 21.202,
      "eval_steps_per_second": 5.301,
      "step": 100
    },
    {
      "epoch": 0.229806598407281,
      "grad_norm": 1.0204566717147827,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.4443,
      "step": 101
    },
    {
      "epoch": 0.23208191126279865,
      "grad_norm": 1.1552355289459229,
      "learning_rate": 5.247918773366112e-05,
      "loss": 2.279,
      "step": 102
    },
    {
      "epoch": 0.23435722411831628,
      "grad_norm": 0.9076858758926392,
      "learning_rate": 5.165316846586541e-05,
      "loss": 2.23,
      "step": 103
    },
    {
      "epoch": 0.2366325369738339,
      "grad_norm": 1.1261475086212158,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 3.0607,
      "step": 104
    },
    {
      "epoch": 0.23890784982935154,
      "grad_norm": 0.8971647620201111,
      "learning_rate": 5e-05,
      "loss": 1.9419,
      "step": 105
    },
    {
      "epoch": 0.24118316268486917,
      "grad_norm": 0.9300563931465149,
      "learning_rate": 4.917330276168208e-05,
      "loss": 2.0819,
      "step": 106
    },
    {
      "epoch": 0.2434584755403868,
      "grad_norm": 0.9358783960342407,
      "learning_rate": 4.834683153413459e-05,
      "loss": 2.2509,
      "step": 107
    },
    {
      "epoch": 0.24573378839590443,
      "grad_norm": 1.107191801071167,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 3.0746,
      "step": 108
    },
    {
      "epoch": 0.24800910125142206,
      "grad_norm": 1.4491487741470337,
      "learning_rate": 4.669547078371504e-05,
      "loss": 3.9955,
      "step": 109
    },
    {
      "epoch": 0.2502844141069397,
      "grad_norm": 1.512543797492981,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 4.0009,
      "step": 110
    },
    {
      "epoch": 0.2525597269624573,
      "grad_norm": 1.2420899868011475,
      "learning_rate": 4.504772348747687e-05,
      "loss": 3.3004,
      "step": 111
    },
    {
      "epoch": 0.25483503981797495,
      "grad_norm": 1.3755626678466797,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 4.2423,
      "step": 112
    },
    {
      "epoch": 0.2571103526734926,
      "grad_norm": 1.0412871837615967,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 2.9379,
      "step": 113
    },
    {
      "epoch": 0.2593856655290102,
      "grad_norm": 1.3524209260940552,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 3.5945,
      "step": 114
    },
    {
      "epoch": 0.2616609783845279,
      "grad_norm": 1.275431752204895,
      "learning_rate": 4.17702704859633e-05,
      "loss": 3.5224,
      "step": 115
    },
    {
      "epoch": 0.26393629124004553,
      "grad_norm": 1.0613654851913452,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.7019,
      "step": 116
    },
    {
      "epoch": 0.26621160409556316,
      "grad_norm": 1.5998655557632446,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 4.7014,
      "step": 117
    },
    {
      "epoch": 0.2684869169510808,
      "grad_norm": 1.2930930852890015,
      "learning_rate": 3.933501846281267e-05,
      "loss": 3.9263,
      "step": 118
    },
    {
      "epoch": 0.2707622298065984,
      "grad_norm": 1.5960543155670166,
      "learning_rate": 3.852880399766243e-05,
      "loss": 5.0185,
      "step": 119
    },
    {
      "epoch": 0.27303754266211605,
      "grad_norm": 1.5104215145111084,
      "learning_rate": 3.772572564296005e-05,
      "loss": 4.7537,
      "step": 120
    },
    {
      "epoch": 0.2753128555176337,
      "grad_norm": 1.6494826078414917,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 4.8911,
      "step": 121
    },
    {
      "epoch": 0.2775881683731513,
      "grad_norm": 1.2384024858474731,
      "learning_rate": 3.612985456190778e-05,
      "loss": 3.3423,
      "step": 122
    },
    {
      "epoch": 0.27986348122866894,
      "grad_norm": 1.6823331117630005,
      "learning_rate": 3.533749813077677e-05,
      "loss": 5.7709,
      "step": 123
    },
    {
      "epoch": 0.2821387940841866,
      "grad_norm": 1.4126425981521606,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 3.988,
      "step": 124
    },
    {
      "epoch": 0.2844141069397042,
      "grad_norm": 1.9775968790054321,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 5.1021,
      "step": 125
    },
    {
      "epoch": 0.28668941979522183,
      "grad_norm": 1.6785099506378174,
      "learning_rate": 3.298534127791785e-05,
      "loss": 4.8588,
      "step": 126
    },
    {
      "epoch": 0.28896473265073946,
      "grad_norm": 1.749727487564087,
      "learning_rate": 3.221030765387417e-05,
      "loss": 4.4106,
      "step": 127
    },
    {
      "epoch": 0.2912400455062571,
      "grad_norm": 1.9710289239883423,
      "learning_rate": 3.144013755408895e-05,
      "loss": 6.2356,
      "step": 128
    },
    {
      "epoch": 0.2935153583617747,
      "grad_norm": 1.5526156425476074,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 4.153,
      "step": 129
    },
    {
      "epoch": 0.29579067121729236,
      "grad_norm": 1.6731418371200562,
      "learning_rate": 2.991522876735154e-05,
      "loss": 5.0893,
      "step": 130
    },
    {
      "epoch": 0.29806598407281,
      "grad_norm": 1.8265318870544434,
      "learning_rate": 2.916090697523549e-05,
      "loss": 5.7695,
      "step": 131
    },
    {
      "epoch": 0.3003412969283277,
      "grad_norm": 1.7096079587936401,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 5.8865,
      "step": 132
    },
    {
      "epoch": 0.3026166097838453,
      "grad_norm": 1.7367585897445679,
      "learning_rate": 2.766955965735968e-05,
      "loss": 5.6732,
      "step": 133
    },
    {
      "epoch": 0.30489192263936293,
      "grad_norm": 1.739740252494812,
      "learning_rate": 2.693294185106562e-05,
      "loss": 5.1278,
      "step": 134
    },
    {
      "epoch": 0.30716723549488056,
      "grad_norm": 1.9743983745574951,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 6.8044,
      "step": 135
    },
    {
      "epoch": 0.3094425483503982,
      "grad_norm": 1.8566263914108276,
      "learning_rate": 2.547882480847461e-05,
      "loss": 5.5522,
      "step": 136
    },
    {
      "epoch": 0.3117178612059158,
      "grad_norm": 2.170522689819336,
      "learning_rate": 2.476172311325783e-05,
      "loss": 6.4498,
      "step": 137
    },
    {
      "epoch": 0.31399317406143346,
      "grad_norm": 2.273266553878784,
      "learning_rate": 2.405152131093926e-05,
      "loss": 7.1787,
      "step": 138
    },
    {
      "epoch": 0.3162684869169511,
      "grad_norm": 1.8308862447738647,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 4.9864,
      "step": 139
    },
    {
      "epoch": 0.3185437997724687,
      "grad_norm": 2.220378875732422,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 6.7028,
      "step": 140
    },
    {
      "epoch": 0.32081911262798635,
      "grad_norm": 2.027402639389038,
      "learning_rate": 2.196424713241637e-05,
      "loss": 6.0877,
      "step": 141
    },
    {
      "epoch": 0.323094425483504,
      "grad_norm": 2.177976369857788,
      "learning_rate": 2.128356686585282e-05,
      "loss": 6.7621,
      "step": 142
    },
    {
      "epoch": 0.3253697383390216,
      "grad_norm": 3.0128822326660156,
      "learning_rate": 2.061073738537635e-05,
      "loss": 7.3477,
      "step": 143
    },
    {
      "epoch": 0.32764505119453924,
      "grad_norm": 2.305850028991699,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 7.4856,
      "step": 144
    },
    {
      "epoch": 0.32992036405005687,
      "grad_norm": 2.4234507083892822,
      "learning_rate": 1.928936436551661e-05,
      "loss": 7.195,
      "step": 145
    },
    {
      "epoch": 0.3321956769055745,
      "grad_norm": 2.5720205307006836,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 6.313,
      "step": 146
    },
    {
      "epoch": 0.33447098976109213,
      "grad_norm": 2.7900567054748535,
      "learning_rate": 1.800157297483417e-05,
      "loss": 7.0602,
      "step": 147
    },
    {
      "epoch": 0.33674630261660976,
      "grad_norm": 3.1076736450195312,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 7.7984,
      "step": 148
    },
    {
      "epoch": 0.3390216154721274,
      "grad_norm": 2.941718578338623,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 7.5484,
      "step": 149
    },
    {
      "epoch": 0.3412969283276451,
      "grad_norm": 5.1757049560546875,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 8.0111,
      "step": 150
    },
    {
      "epoch": 0.3412969283276451,
      "eval_loss": 1.154452919960022,
      "eval_runtime": 34.9102,
      "eval_samples_per_second": 21.197,
      "eval_steps_per_second": 5.299,
      "step": 150
    },
    {
      "epoch": 0.3435722411831627,
      "grad_norm": 1.098740816116333,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.488,
      "step": 151
    },
    {
      "epoch": 0.34584755403868034,
      "grad_norm": 1.1226850748062134,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.9497,
      "step": 152
    },
    {
      "epoch": 0.34812286689419797,
      "grad_norm": 1.0201623439788818,
      "learning_rate": 1.435357758543015e-05,
      "loss": 2.1387,
      "step": 153
    },
    {
      "epoch": 0.3503981797497156,
      "grad_norm": 1.1433717012405396,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 2.4159,
      "step": 154
    },
    {
      "epoch": 0.35267349260523323,
      "grad_norm": 1.2182257175445557,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 2.7417,
      "step": 155
    },
    {
      "epoch": 0.35494880546075086,
      "grad_norm": 1.0635778903961182,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 2.6288,
      "step": 156
    },
    {
      "epoch": 0.3572241183162685,
      "grad_norm": 1.165103554725647,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 2.7699,
      "step": 157
    },
    {
      "epoch": 0.3594994311717861,
      "grad_norm": 0.958754301071167,
      "learning_rate": 1.157994445715706e-05,
      "loss": 2.1388,
      "step": 158
    },
    {
      "epoch": 0.36177474402730375,
      "grad_norm": 1.1642366647720337,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 2.6063,
      "step": 159
    },
    {
      "epoch": 0.3640500568828214,
      "grad_norm": 1.2607970237731934,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 3.1178,
      "step": 160
    },
    {
      "epoch": 0.366325369738339,
      "grad_norm": 1.2615807056427002,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.8609,
      "step": 161
    },
    {
      "epoch": 0.36860068259385664,
      "grad_norm": 1.007006049156189,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.246,
      "step": 162
    },
    {
      "epoch": 0.3708759954493743,
      "grad_norm": 1.4469585418701172,
      "learning_rate": 9.068759265665384e-06,
      "loss": 3.4412,
      "step": 163
    },
    {
      "epoch": 0.3731513083048919,
      "grad_norm": 1.4021576642990112,
      "learning_rate": 8.599558442598998e-06,
      "loss": 3.3755,
      "step": 164
    },
    {
      "epoch": 0.37542662116040953,
      "grad_norm": 1.4966576099395752,
      "learning_rate": 8.141676086873572e-06,
      "loss": 3.2754,
      "step": 165
    },
    {
      "epoch": 0.37770193401592717,
      "grad_norm": 1.7988269329071045,
      "learning_rate": 7.695237378953223e-06,
      "loss": 3.816,
      "step": 166
    },
    {
      "epoch": 0.3799772468714448,
      "grad_norm": 1.5620849132537842,
      "learning_rate": 7.260364370723044e-06,
      "loss": 3.7642,
      "step": 167
    },
    {
      "epoch": 0.3822525597269625,
      "grad_norm": 1.3957778215408325,
      "learning_rate": 6.837175952121306e-06,
      "loss": 3.4017,
      "step": 168
    },
    {
      "epoch": 0.3845278725824801,
      "grad_norm": 1.6015712022781372,
      "learning_rate": 6.425787818636131e-06,
      "loss": 3.8844,
      "step": 169
    },
    {
      "epoch": 0.38680318543799774,
      "grad_norm": 1.5334484577178955,
      "learning_rate": 6.026312439675552e-06,
      "loss": 3.9066,
      "step": 170
    },
    {
      "epoch": 0.3890784982935154,
      "grad_norm": 1.8198069334030151,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 5.1438,
      "step": 171
    },
    {
      "epoch": 0.391353811149033,
      "grad_norm": 1.6796146631240845,
      "learning_rate": 5.263533508961827e-06,
      "loss": 4.3619,
      "step": 172
    },
    {
      "epoch": 0.39362912400455063,
      "grad_norm": 1.8552777767181396,
      "learning_rate": 4.900438493352055e-06,
      "loss": 5.3145,
      "step": 173
    },
    {
      "epoch": 0.39590443686006827,
      "grad_norm": 1.5486481189727783,
      "learning_rate": 4.549673247541875e-06,
      "loss": 4.0269,
      "step": 174
    },
    {
      "epoch": 0.3981797497155859,
      "grad_norm": 1.653618335723877,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 5.0562,
      "step": 175
    },
    {
      "epoch": 0.4004550625711035,
      "grad_norm": 1.7251213788986206,
      "learning_rate": 3.885512251130763e-06,
      "loss": 4.9044,
      "step": 176
    },
    {
      "epoch": 0.40273037542662116,
      "grad_norm": 1.5859131813049316,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 4.4779,
      "step": 177
    },
    {
      "epoch": 0.4050056882821388,
      "grad_norm": 1.9267104864120483,
      "learning_rate": 3.271776770026963e-06,
      "loss": 5.747,
      "step": 178
    },
    {
      "epoch": 0.4072810011376564,
      "grad_norm": 2.2389144897460938,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 6.019,
      "step": 179
    },
    {
      "epoch": 0.40955631399317405,
      "grad_norm": 1.8806343078613281,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 5.4059,
      "step": 180
    },
    {
      "epoch": 0.4118316268486917,
      "grad_norm": 1.830378770828247,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 4.8672,
      "step": 181
    },
    {
      "epoch": 0.4141069397042093,
      "grad_norm": 1.9920967817306519,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 5.8265,
      "step": 182
    },
    {
      "epoch": 0.41638225255972694,
      "grad_norm": 2.4261438846588135,
      "learning_rate": 1.962316193157593e-06,
      "loss": 5.9084,
      "step": 183
    },
    {
      "epoch": 0.41865756541524457,
      "grad_norm": 2.101851224899292,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 6.2117,
      "step": 184
    },
    {
      "epoch": 0.42093287827076226,
      "grad_norm": 2.309554100036621,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 6.2675,
      "step": 185
    },
    {
      "epoch": 0.4232081911262799,
      "grad_norm": 1.8390836715698242,
      "learning_rate": 1.333670137599713e-06,
      "loss": 5.395,
      "step": 186
    },
    {
      "epoch": 0.4254835039817975,
      "grad_norm": 2.0720298290252686,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 5.9934,
      "step": 187
    },
    {
      "epoch": 0.42775881683731515,
      "grad_norm": 2.215231418609619,
      "learning_rate": 9.810017062595322e-07,
      "loss": 6.0897,
      "step": 188
    },
    {
      "epoch": 0.4300341296928328,
      "grad_norm": 1.9252026081085205,
      "learning_rate": 8.247462563808817e-07,
      "loss": 4.6792,
      "step": 189
    },
    {
      "epoch": 0.4323094425483504,
      "grad_norm": 2.3703060150146484,
      "learning_rate": 6.819348298638839e-07,
      "loss": 6.0202,
      "step": 190
    },
    {
      "epoch": 0.43458475540386804,
      "grad_norm": 2.369107961654663,
      "learning_rate": 5.526064699265753e-07,
      "loss": 6.5793,
      "step": 191
    },
    {
      "epoch": 0.43686006825938567,
      "grad_norm": 2.236239433288574,
      "learning_rate": 4.367965336512403e-07,
      "loss": 6.4385,
      "step": 192
    },
    {
      "epoch": 0.4391353811149033,
      "grad_norm": 2.367990732192993,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 5.754,
      "step": 193
    },
    {
      "epoch": 0.44141069397042093,
      "grad_norm": 2.521074056625366,
      "learning_rate": 2.458548727494292e-07,
      "loss": 6.9947,
      "step": 194
    },
    {
      "epoch": 0.44368600682593856,
      "grad_norm": 2.6892969608306885,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 7.5039,
      "step": 195
    },
    {
      "epoch": 0.4459613196814562,
      "grad_norm": 2.6571784019470215,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 6.2621,
      "step": 196
    },
    {
      "epoch": 0.4482366325369738,
      "grad_norm": 2.8904409408569336,
      "learning_rate": 6.150154258476315e-08,
      "loss": 7.8708,
      "step": 197
    },
    {
      "epoch": 0.45051194539249145,
      "grad_norm": 3.4330527782440186,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 7.4755,
      "step": 198
    },
    {
      "epoch": 0.4527872582480091,
      "grad_norm": 3.3603594303131104,
      "learning_rate": 6.834750376549792e-09,
      "loss": 7.2884,
      "step": 199
    },
    {
      "epoch": 0.4550625711035267,
      "grad_norm": 4.667001724243164,
      "learning_rate": 0.0,
      "loss": 8.1919,
      "step": 200
    },
    {
      "epoch": 0.4550625711035267,
      "eval_loss": 1.1458154916763306,
      "eval_runtime": 34.903,
      "eval_samples_per_second": 21.202,
      "eval_steps_per_second": 5.3,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.50336002064384e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}