{
  "best_metric": 0.4938221871852875,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.033716862645930795,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.000168584313229654,
      "grad_norm": 0.8356221318244934,
      "learning_rate": 1e-05,
      "loss": 2.4589,
      "step": 1
    },
    {
      "epoch": 0.000168584313229654,
      "eval_loss": 0.733188807964325,
      "eval_runtime": 501.5907,
      "eval_samples_per_second": 19.917,
      "eval_steps_per_second": 4.98,
      "step": 1
    },
    {
      "epoch": 0.000337168626459308,
      "grad_norm": 1.1123285293579102,
      "learning_rate": 2e-05,
      "loss": 2.4424,
      "step": 2
    },
    {
      "epoch": 0.0005057529396889619,
      "grad_norm": 1.0420997142791748,
      "learning_rate": 3e-05,
      "loss": 2.4165,
      "step": 3
    },
    {
      "epoch": 0.000674337252918616,
      "grad_norm": 0.9809317588806152,
      "learning_rate": 4e-05,
      "loss": 2.5515,
      "step": 4
    },
    {
      "epoch": 0.0008429215661482699,
      "grad_norm": 0.8284363150596619,
      "learning_rate": 5e-05,
      "loss": 2.3991,
      "step": 5
    },
    {
      "epoch": 0.0010115058793779238,
      "grad_norm": 0.6797701716423035,
      "learning_rate": 6e-05,
      "loss": 2.3095,
      "step": 6
    },
    {
      "epoch": 0.0011800901926075778,
      "grad_norm": 0.6245788335800171,
      "learning_rate": 7e-05,
      "loss": 2.5452,
      "step": 7
    },
    {
      "epoch": 0.001348674505837232,
      "grad_norm": 0.8151345252990723,
      "learning_rate": 8e-05,
      "loss": 2.2631,
      "step": 8
    },
    {
      "epoch": 0.0015172588190668858,
      "grad_norm": 0.8989050388336182,
      "learning_rate": 9e-05,
      "loss": 2.3298,
      "step": 9
    },
    {
      "epoch": 0.0016858431322965398,
      "grad_norm": 0.7531624436378479,
      "learning_rate": 0.0001,
      "loss": 2.363,
      "step": 10
    },
    {
      "epoch": 0.0018544274455261937,
      "grad_norm": 2.898843765258789,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.4567,
      "step": 11
    },
    {
      "epoch": 0.0020230117587558476,
      "grad_norm": 0.8626514673233032,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.5136,
      "step": 12
    },
    {
      "epoch": 0.002191596071985502,
      "grad_norm": 0.8011757135391235,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.1879,
      "step": 13
    },
    {
      "epoch": 0.0023601803852151557,
      "grad_norm": 0.687286376953125,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.3944,
      "step": 14
    },
    {
      "epoch": 0.0025287646984448096,
      "grad_norm": 0.6025044322013855,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.3256,
      "step": 15
    },
    {
      "epoch": 0.002697349011674464,
      "grad_norm": 0.5510916113853455,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.202,
      "step": 16
    },
    {
      "epoch": 0.0028659333249041177,
      "grad_norm": 0.5466865301132202,
      "learning_rate": 9.966546331768191e-05,
      "loss": 2.2705,
      "step": 17
    },
    {
      "epoch": 0.0030345176381337715,
      "grad_norm": 0.5651683807373047,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.2199,
      "step": 18
    },
    {
      "epoch": 0.003203101951363426,
      "grad_norm": 0.5666970610618591,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.2768,
      "step": 19
    },
    {
      "epoch": 0.0033716862645930797,
      "grad_norm": 0.5634180903434753,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.2632,
      "step": 20
    },
    {
      "epoch": 0.0035402705778227335,
      "grad_norm": 0.5488885045051575,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.3418,
      "step": 21
    },
    {
      "epoch": 0.0037088548910523874,
      "grad_norm": 0.574619710445404,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.4511,
      "step": 22
    },
    {
      "epoch": 0.0038774392042820417,
      "grad_norm": 0.5341477394104004,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.2371,
      "step": 23
    },
    {
      "epoch": 0.004046023517511695,
      "grad_norm": 0.5474244356155396,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.2004,
      "step": 24
    },
    {
      "epoch": 0.004214607830741349,
      "grad_norm": 0.6370226144790649,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.1892,
      "step": 25
    },
    {
      "epoch": 0.004383192143971004,
      "grad_norm": 0.6040836572647095,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.2479,
      "step": 26
    },
    {
      "epoch": 0.004551776457200657,
      "grad_norm": 0.5802085995674133,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.0094,
      "step": 27
    },
    {
      "epoch": 0.004720360770430311,
      "grad_norm": 0.5974027514457703,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.2492,
      "step": 28
    },
    {
      "epoch": 0.004888945083659966,
      "grad_norm": 0.5877422094345093,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.1791,
      "step": 29
    },
    {
      "epoch": 0.005057529396889619,
      "grad_norm": 0.5708343982696533,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.0455,
      "step": 30
    },
    {
      "epoch": 0.005226113710119273,
      "grad_norm": 0.6857141852378845,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.4454,
      "step": 31
    },
    {
      "epoch": 0.005394698023348928,
      "grad_norm": 0.6374479532241821,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.2322,
      "step": 32
    },
    {
      "epoch": 0.005563282336578581,
      "grad_norm": 0.6205618381500244,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.0517,
      "step": 33
    },
    {
      "epoch": 0.005731866649808235,
      "grad_norm": 0.679690420627594,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.3632,
      "step": 34
    },
    {
      "epoch": 0.00590045096303789,
      "grad_norm": 0.6905668377876282,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.3155,
      "step": 35
    },
    {
      "epoch": 0.006069035276267543,
      "grad_norm": 0.691063404083252,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.4142,
      "step": 36
    },
    {
      "epoch": 0.006237619589497197,
      "grad_norm": 0.7013322114944458,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.271,
      "step": 37
    },
    {
      "epoch": 0.006406203902726852,
      "grad_norm": 0.7218900918960571,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.5897,
      "step": 38
    },
    {
      "epoch": 0.006574788215956505,
      "grad_norm": 0.7723600268363953,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.3272,
      "step": 39
    },
    {
      "epoch": 0.006743372529186159,
      "grad_norm": 0.8070730566978455,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.3632,
      "step": 40
    },
    {
      "epoch": 0.006911956842415813,
      "grad_norm": 0.8654877543449402,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.3817,
      "step": 41
    },
    {
      "epoch": 0.007080541155645467,
      "grad_norm": 1.0104258060455322,
      "learning_rate": 9.316282404787871e-05,
      "loss": 2.4231,
      "step": 42
    },
    {
      "epoch": 0.007249125468875121,
      "grad_norm": 0.9260497093200684,
      "learning_rate": 9.273963562927695e-05,
      "loss": 2.2641,
      "step": 43
    },
    {
      "epoch": 0.007417709782104775,
      "grad_norm": 1.331114649772644,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.4468,
      "step": 44
    },
    {
      "epoch": 0.007586294095334429,
      "grad_norm": 0.9541406631469727,
      "learning_rate": 9.185832391312644e-05,
      "loss": 2.0806,
      "step": 45
    },
    {
      "epoch": 0.007754878408564083,
      "grad_norm": 1.1738696098327637,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.0837,
      "step": 46
    },
    {
      "epoch": 0.007923462721793737,
      "grad_norm": 1.2855007648468018,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.5559,
      "step": 47
    },
    {
      "epoch": 0.00809204703502339,
      "grad_norm": 1.311841368675232,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.0784,
      "step": 48
    },
    {
      "epoch": 0.008260631348253045,
      "grad_norm": 1.547163724899292,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.3095,
      "step": 49
    },
    {
      "epoch": 0.008429215661482699,
      "grad_norm": 1.7676929235458374,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.185,
      "step": 50
    },
    {
      "epoch": 0.008429215661482699,
      "eval_loss": 0.5834899544715881,
      "eval_runtime": 503.8159,
      "eval_samples_per_second": 19.829,
      "eval_steps_per_second": 4.958,
      "step": 50
    },
    {
      "epoch": 0.008597799974712352,
      "grad_norm": 1.5893906354904175,
      "learning_rate": 8.894386393810563e-05,
      "loss": 2.225,
      "step": 51
    },
    {
      "epoch": 0.008766384287942007,
      "grad_norm": 1.3943727016448975,
      "learning_rate": 8.842005554284296e-05,
      "loss": 2.1335,
      "step": 52
    },
    {
      "epoch": 0.00893496860117166,
      "grad_norm": 1.213746190071106,
      "learning_rate": 8.788574348801675e-05,
      "loss": 2.1279,
      "step": 53
    },
    {
      "epoch": 0.009103552914401314,
      "grad_norm": 0.8661502003669739,
      "learning_rate": 8.73410738492077e-05,
      "loss": 2.0464,
      "step": 54
    },
    {
      "epoch": 0.00927213722763097,
      "grad_norm": 0.627927839756012,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.9024,
      "step": 55
    },
    {
      "epoch": 0.009440721540860623,
      "grad_norm": 0.5861731767654419,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.9038,
      "step": 56
    },
    {
      "epoch": 0.009609305854090276,
      "grad_norm": 0.6599506735801697,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.0148,
      "step": 57
    },
    {
      "epoch": 0.009777890167319931,
      "grad_norm": 0.6939547061920166,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.0994,
      "step": 58
    },
    {
      "epoch": 0.009946474480549585,
      "grad_norm": 0.7301190495491028,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.9956,
      "step": 59
    },
    {
      "epoch": 0.010115058793779238,
      "grad_norm": 0.7114450335502625,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.9967,
      "step": 60
    },
    {
      "epoch": 0.010283643107008893,
      "grad_norm": 0.6240396499633789,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.8873,
      "step": 61
    },
    {
      "epoch": 0.010452227420238547,
      "grad_norm": 0.6104307770729065,
      "learning_rate": 8.262928807620843e-05,
      "loss": 2.0076,
      "step": 62
    },
    {
      "epoch": 0.0106208117334682,
      "grad_norm": 0.5649477243423462,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.9833,
      "step": 63
    },
    {
      "epoch": 0.010789396046697855,
      "grad_norm": 0.5459277629852295,
      "learning_rate": 8.135881792367686e-05,
      "loss": 2.1804,
      "step": 64
    },
    {
      "epoch": 0.010957980359927509,
      "grad_norm": 0.5765476822853088,
      "learning_rate": 8.07106356344834e-05,
      "loss": 2.037,
      "step": 65
    },
    {
      "epoch": 0.011126564673157162,
      "grad_norm": 0.5954334139823914,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.1889,
      "step": 66
    },
    {
      "epoch": 0.011295148986386817,
      "grad_norm": 0.5802199840545654,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.1004,
      "step": 67
    },
    {
      "epoch": 0.01146373329961647,
      "grad_norm": 0.5583539605140686,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.9828,
      "step": 68
    },
    {
      "epoch": 0.011632317612846124,
      "grad_norm": 0.5685673356056213,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.001,
      "step": 69
    },
    {
      "epoch": 0.01180090192607578,
      "grad_norm": 0.5391646027565002,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.99,
      "step": 70
    },
    {
      "epoch": 0.011969486239305433,
      "grad_norm": 0.5730313062667847,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.9903,
      "step": 71
    },
    {
      "epoch": 0.012138070552535086,
      "grad_norm": 0.5506834387779236,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.9428,
      "step": 72
    },
    {
      "epoch": 0.012306654865764741,
      "grad_norm": 0.5586628317832947,
      "learning_rate": 7.52382768867422e-05,
      "loss": 2.087,
      "step": 73
    },
    {
      "epoch": 0.012475239178994395,
      "grad_norm": 0.5688629150390625,
      "learning_rate": 7.452117519152542e-05,
      "loss": 2.044,
      "step": 74
    },
    {
      "epoch": 0.012643823492224048,
      "grad_norm": 0.5850514769554138,
      "learning_rate": 7.379736965185368e-05,
      "loss": 2.1281,
      "step": 75
    },
    {
      "epoch": 0.012812407805453703,
      "grad_norm": 0.9722267389297485,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.9546,
      "step": 76
    },
    {
      "epoch": 0.012980992118683357,
      "grad_norm": 0.5938723087310791,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.9647,
      "step": 77
    },
    {
      "epoch": 0.01314957643191301,
      "grad_norm": 0.5510969161987305,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.9415,
      "step": 78
    },
    {
      "epoch": 0.013318160745142664,
      "grad_norm": 0.6177936792373657,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.057,
      "step": 79
    },
    {
      "epoch": 0.013486745058372319,
      "grad_norm": 0.5951735973358154,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.1511,
      "step": 80
    },
    {
      "epoch": 0.013655329371601972,
      "grad_norm": 0.5899169445037842,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.1427,
      "step": 81
    },
    {
      "epoch": 0.013823913684831626,
      "grad_norm": 0.6052743196487427,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.0984,
      "step": 82
    },
    {
      "epoch": 0.01399249799806128,
      "grad_norm": 0.5954638719558716,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.8308,
      "step": 83
    },
    {
      "epoch": 0.014161082311290934,
      "grad_norm": 0.6279288530349731,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.0456,
      "step": 84
    },
    {
      "epoch": 0.014329666624520588,
      "grad_norm": 0.6560235619544983,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.1687,
      "step": 85
    },
    {
      "epoch": 0.014498250937750243,
      "grad_norm": 0.6726033687591553,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.0138,
      "step": 86
    },
    {
      "epoch": 0.014666835250979896,
      "grad_norm": 0.6857135891914368,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.112,
      "step": 87
    },
    {
      "epoch": 0.01483541956420955,
      "grad_norm": 0.6859216094017029,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.9114,
      "step": 88
    },
    {
      "epoch": 0.015004003877439205,
      "grad_norm": 0.707784116268158,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.1385,
      "step": 89
    },
    {
      "epoch": 0.015172588190668858,
      "grad_norm": 0.7206366062164307,
      "learning_rate": 6.227427435703997e-05,
      "loss": 2.0489,
      "step": 90
    },
    {
      "epoch": 0.015341172503898512,
      "grad_norm": 0.7995542287826538,
      "learning_rate": 6.147119600233758e-05,
      "loss": 2.1788,
      "step": 91
    },
    {
      "epoch": 0.015509756817128167,
      "grad_norm": 0.789454996585846,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.0025,
      "step": 92
    },
    {
      "epoch": 0.01567834113035782,
      "grad_norm": 0.9014174342155457,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.2109,
      "step": 93
    },
    {
      "epoch": 0.015846925443587474,
      "grad_norm": 0.890470027923584,
      "learning_rate": 5.90440267166055e-05,
      "loss": 2.189,
      "step": 94
    },
    {
      "epoch": 0.016015509756817127,
      "grad_norm": 0.9573277831077576,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 2.0214,
      "step": 95
    },
    {
      "epoch": 0.01618409407004678,
      "grad_norm": 1.0161410570144653,
      "learning_rate": 5.74131823855921e-05,
      "loss": 2.0902,
      "step": 96
    },
    {
      "epoch": 0.016352678383276437,
      "grad_norm": 1.1503360271453857,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 2.0676,
      "step": 97
    },
    {
      "epoch": 0.01652126269650609,
      "grad_norm": 1.0322849750518799,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.8626,
      "step": 98
    },
    {
      "epoch": 0.016689847009735744,
      "grad_norm": 1.092878818511963,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.6393,
      "step": 99
    },
    {
      "epoch": 0.016858431322965398,
      "grad_norm": 1.5688586235046387,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.1295,
      "step": 100
    },
    {
      "epoch": 0.016858431322965398,
      "eval_loss": 0.5253283381462097,
      "eval_runtime": 504.2844,
      "eval_samples_per_second": 19.81,
      "eval_steps_per_second": 4.954,
      "step": 100
    },
    {
      "epoch": 0.01702701563619505,
      "grad_norm": 0.8478464484214783,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.7976,
      "step": 101
    },
    {
      "epoch": 0.017195599949424704,
      "grad_norm": 0.7851842641830444,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.8304,
      "step": 102
    },
    {
      "epoch": 0.01736418426265436,
      "grad_norm": 0.8179306387901306,
      "learning_rate": 5.165316846586541e-05,
      "loss": 2.0022,
      "step": 103
    },
    {
      "epoch": 0.017532768575884015,
      "grad_norm": 0.7113230228424072,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.7451,
      "step": 104
    },
    {
      "epoch": 0.017701352889113668,
      "grad_norm": 0.6080374717712402,
      "learning_rate": 5e-05,
      "loss": 1.9349,
      "step": 105
    },
    {
      "epoch": 0.01786993720234332,
      "grad_norm": 0.5461326241493225,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.8954,
      "step": 106
    },
    {
      "epoch": 0.018038521515572975,
      "grad_norm": 0.5062869787216187,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.9877,
      "step": 107
    },
    {
      "epoch": 0.01820710582880263,
      "grad_norm": 0.5204727649688721,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 2.0229,
      "step": 108
    },
    {
      "epoch": 0.018375690142032285,
      "grad_norm": 0.5268082618713379,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.9778,
      "step": 109
    },
    {
      "epoch": 0.01854427445526194,
      "grad_norm": 0.5715117454528809,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.9703,
      "step": 110
    },
    {
      "epoch": 0.018712858768491592,
      "grad_norm": 0.5608370304107666,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.8129,
      "step": 111
    },
    {
      "epoch": 0.018881443081721246,
      "grad_norm": 0.6107746958732605,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.9648,
      "step": 112
    },
    {
      "epoch": 0.0190500273949509,
      "grad_norm": 0.5928745269775391,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 2.0033,
      "step": 113
    },
    {
      "epoch": 0.019218611708180552,
      "grad_norm": 0.6055542826652527,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.9539,
      "step": 114
    },
    {
      "epoch": 0.01938719602141021,
      "grad_norm": 0.5719534158706665,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.1993,
      "step": 115
    },
    {
      "epoch": 0.019555780334639863,
      "grad_norm": 0.5812926888465881,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.0225,
      "step": 116
    },
    {
      "epoch": 0.019724364647869516,
      "grad_norm": 0.5546519160270691,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.9672,
      "step": 117
    },
    {
      "epoch": 0.01989294896109917,
      "grad_norm": 0.5410682559013367,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.9459,
      "step": 118
    },
    {
      "epoch": 0.020061533274328823,
      "grad_norm": 0.5316163301467896,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.9116,
      "step": 119
    },
    {
      "epoch": 0.020230117587558476,
      "grad_norm": 0.5533401966094971,
      "learning_rate": 3.772572564296005e-05,
      "loss": 2.0063,
      "step": 120
    },
    {
      "epoch": 0.020398701900788133,
      "grad_norm": 0.5578827261924744,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 2.1339,
      "step": 121
    },
    {
      "epoch": 0.020567286214017787,
      "grad_norm": 0.548750102519989,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.9597,
      "step": 122
    },
    {
      "epoch": 0.02073587052724744,
      "grad_norm": 0.5724698901176453,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.9919,
      "step": 123
    },
    {
      "epoch": 0.020904454840477094,
      "grad_norm": 0.5434842109680176,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.9448,
      "step": 124
    },
    {
      "epoch": 0.021073039153706747,
      "grad_norm": 0.6229725480079651,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 2.0906,
      "step": 125
    },
    {
      "epoch": 0.0212416234669364,
      "grad_norm": 0.5655553340911865,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.8697,
      "step": 126
    },
    {
      "epoch": 0.021410207780166054,
      "grad_norm": 0.6026455163955688,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.1017,
      "step": 127
    },
    {
      "epoch": 0.02157879209339571,
      "grad_norm": 0.5940404534339905,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.8951,
      "step": 128
    },
    {
      "epoch": 0.021747376406625364,
      "grad_norm": 0.5814304947853088,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.989,
      "step": 129
    },
    {
      "epoch": 0.021915960719855018,
      "grad_norm": 0.582674503326416,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.9003,
      "step": 130
    },
    {
      "epoch": 0.02208454503308467,
      "grad_norm": 0.652070164680481,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.2016,
      "step": 131
    },
    {
      "epoch": 0.022253129346314324,
      "grad_norm": 0.651836097240448,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.0757,
      "step": 132
    },
    {
      "epoch": 0.022421713659543978,
      "grad_norm": 0.6593762636184692,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.2828,
      "step": 133
    },
    {
      "epoch": 0.022590297972773635,
      "grad_norm": 0.6495382189750671,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.1775,
      "step": 134
    },
    {
      "epoch": 0.022758882286003288,
      "grad_norm": 0.6451095342636108,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.9447,
      "step": 135
    },
    {
      "epoch": 0.02292746659923294,
      "grad_norm": 0.696794331073761,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.1392,
      "step": 136
    },
    {
      "epoch": 0.023096050912462595,
      "grad_norm": 0.6984504461288452,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.1578,
      "step": 137
    },
    {
      "epoch": 0.02326463522569225,
      "grad_norm": 0.7092112302780151,
      "learning_rate": 2.405152131093926e-05,
      "loss": 2.1024,
      "step": 138
    },
    {
      "epoch": 0.023433219538921902,
      "grad_norm": 0.7876008152961731,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.1975,
      "step": 139
    },
    {
      "epoch": 0.02360180385215156,
      "grad_norm": 0.8380593657493591,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 2.1563,
      "step": 140
    },
    {
      "epoch": 0.023770388165381212,
      "grad_norm": 0.7053534388542175,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.7217,
      "step": 141
    },
    {
      "epoch": 0.023938972478610866,
      "grad_norm": 0.8504711389541626,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.1424,
      "step": 142
    },
    {
      "epoch": 0.02410755679184052,
      "grad_norm": 0.9147765636444092,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.0807,
      "step": 143
    },
    {
      "epoch": 0.024276141105070172,
      "grad_norm": 0.9966739416122437,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.2053,
      "step": 144
    },
    {
      "epoch": 0.024444725418299826,
      "grad_norm": 1.010679841041565,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.1144,
      "step": 145
    },
    {
      "epoch": 0.024613309731529483,
      "grad_norm": 1.0044876337051392,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.745,
      "step": 146
    },
    {
      "epoch": 0.024781894044759136,
      "grad_norm": 1.1902530193328857,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.8855,
      "step": 147
    },
    {
      "epoch": 0.02495047835798879,
      "grad_norm": 1.1966170072555542,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.909,
      "step": 148
    },
    {
      "epoch": 0.025119062671218443,
      "grad_norm": 1.2299884557724,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.6729,
      "step": 149
    },
    {
      "epoch": 0.025287646984448096,
      "grad_norm": 1.890581488609314,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.3292,
      "step": 150
    },
    {
      "epoch": 0.025287646984448096,
      "eval_loss": 0.5001963973045349,
      "eval_runtime": 504.889,
      "eval_samples_per_second": 19.787,
      "eval_steps_per_second": 4.948,
      "step": 150
    },
    {
      "epoch": 0.02545623129767775,
      "grad_norm": 0.501922607421875,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.6838,
      "step": 151
    },
    {
      "epoch": 0.025624815610907407,
      "grad_norm": 0.5581042170524597,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 2.0035,
      "step": 152
    },
    {
      "epoch": 0.02579339992413706,
      "grad_norm": 0.5630311369895935,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.7692,
      "step": 153
    },
    {
      "epoch": 0.025961984237366714,
      "grad_norm": 0.5622650384902954,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.9031,
      "step": 154
    },
    {
      "epoch": 0.026130568550596367,
      "grad_norm": 0.5657137036323547,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.9748,
      "step": 155
    },
    {
      "epoch": 0.02629915286382602,
      "grad_norm": 0.591448187828064,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.8937,
      "step": 156
    },
    {
      "epoch": 0.026467737177055674,
      "grad_norm": 0.5595288872718811,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.8565,
      "step": 157
    },
    {
      "epoch": 0.026636321490285327,
      "grad_norm": 0.5535275936126709,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.9763,
      "step": 158
    },
    {
      "epoch": 0.026804905803514984,
      "grad_norm": 0.564119279384613,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.9217,
      "step": 159
    },
    {
      "epoch": 0.026973490116744638,
      "grad_norm": 0.545243501663208,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.8693,
      "step": 160
    },
    {
      "epoch": 0.02714207442997429,
      "grad_norm": 0.5433475971221924,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.9114,
      "step": 161
    },
    {
      "epoch": 0.027310658743203944,
      "grad_norm": 0.5309762358665466,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.8492,
      "step": 162
    },
    {
      "epoch": 0.027479243056433598,
      "grad_norm": 0.5673748850822449,
      "learning_rate": 9.068759265665384e-06,
      "loss": 2.0088,
      "step": 163
    },
    {
      "epoch": 0.02764782736966325,
      "grad_norm": 0.5460406541824341,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.8557,
      "step": 164
    },
    {
      "epoch": 0.027816411682892908,
      "grad_norm": 0.5581811666488647,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.9468,
      "step": 165
    },
    {
      "epoch": 0.02798499599612256,
      "grad_norm": 0.5672568678855896,
      "learning_rate": 7.695237378953223e-06,
      "loss": 2.0391,
      "step": 166
    },
    {
      "epoch": 0.028153580309352215,
      "grad_norm": 0.5526508688926697,
      "learning_rate": 7.260364370723044e-06,
      "loss": 2.0418,
      "step": 167
    },
    {
      "epoch": 0.02832216462258187,
      "grad_norm": 0.5540817379951477,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.9121,
      "step": 168
    },
    {
      "epoch": 0.028490748935811522,
      "grad_norm": 0.5510901212692261,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.0265,
      "step": 169
    },
    {
      "epoch": 0.028659333249041175,
      "grad_norm": 0.5410937070846558,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.7815,
      "step": 170
    },
    {
      "epoch": 0.028827917562270832,
      "grad_norm": 0.5430680513381958,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 2.0071,
      "step": 171
    },
    {
      "epoch": 0.028996501875500486,
      "grad_norm": 0.5580127835273743,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.9387,
      "step": 172
    },
    {
      "epoch": 0.02916508618873014,
      "grad_norm": 0.5640109181404114,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.8506,
      "step": 173
    },
    {
      "epoch": 0.029333670501959792,
      "grad_norm": 0.6047261953353882,
      "learning_rate": 4.549673247541875e-06,
      "loss": 2.0019,
      "step": 174
    },
    {
      "epoch": 0.029502254815189446,
      "grad_norm": 0.5717324018478394,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.9162,
      "step": 175
    },
    {
      "epoch": 0.0296708391284191,
      "grad_norm": 0.5931180715560913,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.958,
      "step": 176
    },
    {
      "epoch": 0.029839423441648756,
      "grad_norm": 0.5776300430297852,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.7402,
      "step": 177
    },
    {
      "epoch": 0.03000800775487841,
      "grad_norm": 0.6053551435470581,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.9773,
      "step": 178
    },
    {
      "epoch": 0.030176592068108063,
      "grad_norm": 0.5899092555046082,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.8716,
      "step": 179
    },
    {
      "epoch": 0.030345176381337716,
      "grad_norm": 0.6354209184646606,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.0506,
      "step": 180
    },
    {
      "epoch": 0.03051376069456737,
      "grad_norm": 0.6621575951576233,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.9665,
      "step": 181
    },
    {
      "epoch": 0.030682345007797023,
      "grad_norm": 0.6359539031982422,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.8477,
      "step": 182
    },
    {
      "epoch": 0.03085092932102668,
      "grad_norm": 0.6529463529586792,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.8756,
      "step": 183
    },
    {
      "epoch": 0.031019513634256334,
      "grad_norm": 0.643500804901123,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.9785,
      "step": 184
    },
    {
      "epoch": 0.031188097947485987,
      "grad_norm": 0.6969662308692932,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.04,
      "step": 185
    },
    {
      "epoch": 0.03135668226071564,
      "grad_norm": 0.7233607172966003,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.0622,
      "step": 186
    },
    {
      "epoch": 0.0315252665739453,
      "grad_norm": 0.7131527662277222,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.0282,
      "step": 187
    },
    {
      "epoch": 0.03169385088717495,
      "grad_norm": 0.7319460511207581,
      "learning_rate": 9.810017062595322e-07,
      "loss": 2.033,
      "step": 188
    },
    {
      "epoch": 0.031862435200404604,
      "grad_norm": 0.8378353714942932,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.1209,
      "step": 189
    },
    {
      "epoch": 0.032031019513634254,
      "grad_norm": 0.7956808805465698,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.9675,
      "step": 190
    },
    {
      "epoch": 0.03219960382686391,
      "grad_norm": 0.8247827887535095,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.9993,
      "step": 191
    },
    {
      "epoch": 0.03236818814009356,
      "grad_norm": 0.9170274138450623,
      "learning_rate": 4.367965336512403e-07,
      "loss": 2.2301,
      "step": 192
    },
    {
      "epoch": 0.03253677245332322,
      "grad_norm": 0.9447985887527466,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 2.2666,
      "step": 193
    },
    {
      "epoch": 0.032705356766552875,
      "grad_norm": 0.9828495383262634,
      "learning_rate": 2.458548727494292e-07,
      "loss": 2.2058,
      "step": 194
    },
    {
      "epoch": 0.032873941079782525,
      "grad_norm": 0.9916980266571045,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 2.0004,
      "step": 195
    },
    {
      "epoch": 0.03304252539301218,
      "grad_norm": 1.0816479921340942,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.938,
      "step": 196
    },
    {
      "epoch": 0.03321110970624183,
      "grad_norm": 1.0090898275375366,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.8004,
      "step": 197
    },
    {
      "epoch": 0.03337969401947149,
      "grad_norm": 1.2615922689437866,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.6736,
      "step": 198
    },
    {
      "epoch": 0.033548278332701145,
      "grad_norm": 1.234153389930725,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.6951,
      "step": 199
    },
    {
      "epoch": 0.033716862645930795,
      "grad_norm": 1.5582042932510376,
      "learning_rate": 0.0,
      "loss": 2.0536,
      "step": 200
    },
    {
      "epoch": 0.033716862645930795,
      "eval_loss": 0.4938221871852875,
      "eval_runtime": 504.2973,
      "eval_samples_per_second": 19.81,
      "eval_steps_per_second": 4.953,
      "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": 1.535331927148462e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}