0x1202's picture
Training in progress, step 200, checkpoint
17c5726 verified
{
"best_metric": 1.746632695198059,
"best_model_checkpoint": "miner_id_24/checkpoint-200",
"epoch": 1.3722126929674099,
"eval_steps": 50,
"global_step": 200,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.00686106346483705,
"grad_norm": 3.590763568878174,
"learning_rate": 1e-05,
"loss": 8.2815,
"step": 1
},
{
"epoch": 0.00686106346483705,
"eval_loss": 2.6706011295318604,
"eval_runtime": 3.356,
"eval_samples_per_second": 73.302,
"eval_steps_per_second": 18.475,
"step": 1
},
{
"epoch": 0.0137221269296741,
"grad_norm": 4.286794185638428,
"learning_rate": 2e-05,
"loss": 8.6471,
"step": 2
},
{
"epoch": 0.02058319039451115,
"grad_norm": 4.678545951843262,
"learning_rate": 3e-05,
"loss": 9.0696,
"step": 3
},
{
"epoch": 0.0274442538593482,
"grad_norm": 4.255434036254883,
"learning_rate": 4e-05,
"loss": 8.7098,
"step": 4
},
{
"epoch": 0.03430531732418525,
"grad_norm": 4.1823811531066895,
"learning_rate": 5e-05,
"loss": 8.7977,
"step": 5
},
{
"epoch": 0.0411663807890223,
"grad_norm": 3.9178574085235596,
"learning_rate": 6e-05,
"loss": 8.0478,
"step": 6
},
{
"epoch": 0.048027444253859346,
"grad_norm": 4.1361403465271,
"learning_rate": 7e-05,
"loss": 9.3618,
"step": 7
},
{
"epoch": 0.0548885077186964,
"grad_norm": 4.153346061706543,
"learning_rate": 8e-05,
"loss": 8.2667,
"step": 8
},
{
"epoch": 0.06174957118353345,
"grad_norm": 4.951449394226074,
"learning_rate": 9e-05,
"loss": 8.1769,
"step": 9
},
{
"epoch": 0.0686106346483705,
"grad_norm": 5.21913480758667,
"learning_rate": 0.0001,
"loss": 8.7103,
"step": 10
},
{
"epoch": 0.07547169811320754,
"grad_norm": 4.755789756774902,
"learning_rate": 9.999316524962345e-05,
"loss": 8.3865,
"step": 11
},
{
"epoch": 0.0823327615780446,
"grad_norm": 4.612483024597168,
"learning_rate": 9.997266286704631e-05,
"loss": 8.1553,
"step": 12
},
{
"epoch": 0.08919382504288165,
"grad_norm": 5.011524677276611,
"learning_rate": 9.993849845741524e-05,
"loss": 7.9329,
"step": 13
},
{
"epoch": 0.09605488850771869,
"grad_norm": 5.112983226776123,
"learning_rate": 9.989068136093873e-05,
"loss": 9.1607,
"step": 14
},
{
"epoch": 0.10291595197255575,
"grad_norm": 5.1411871910095215,
"learning_rate": 9.98292246503335e-05,
"loss": 8.6013,
"step": 15
},
{
"epoch": 0.1097770154373928,
"grad_norm": 5.042740821838379,
"learning_rate": 9.975414512725057e-05,
"loss": 7.9965,
"step": 16
},
{
"epoch": 0.11663807890222985,
"grad_norm": 4.665650367736816,
"learning_rate": 9.966546331768191e-05,
"loss": 7.5944,
"step": 17
},
{
"epoch": 0.1234991423670669,
"grad_norm": 4.292209148406982,
"learning_rate": 9.956320346634876e-05,
"loss": 6.9841,
"step": 18
},
{
"epoch": 0.13036020583190394,
"grad_norm": 4.671782493591309,
"learning_rate": 9.944739353007344e-05,
"loss": 7.9687,
"step": 19
},
{
"epoch": 0.137221269296741,
"grad_norm": 5.246168613433838,
"learning_rate": 9.931806517013612e-05,
"loss": 8.1801,
"step": 20
},
{
"epoch": 0.14408233276157806,
"grad_norm": 5.219156742095947,
"learning_rate": 9.917525374361912e-05,
"loss": 7.1466,
"step": 21
},
{
"epoch": 0.1509433962264151,
"grad_norm": 4.827649116516113,
"learning_rate": 9.901899829374047e-05,
"loss": 7.4352,
"step": 22
},
{
"epoch": 0.15780445969125215,
"grad_norm": 4.873559951782227,
"learning_rate": 9.884934153917997e-05,
"loss": 7.2922,
"step": 23
},
{
"epoch": 0.1646655231560892,
"grad_norm": 5.314990520477295,
"learning_rate": 9.86663298624003e-05,
"loss": 7.871,
"step": 24
},
{
"epoch": 0.17152658662092624,
"grad_norm": 5.435187339782715,
"learning_rate": 9.847001329696653e-05,
"loss": 7.3211,
"step": 25
},
{
"epoch": 0.1783876500857633,
"grad_norm": 5.212953567504883,
"learning_rate": 9.826044551386744e-05,
"loss": 7.3988,
"step": 26
},
{
"epoch": 0.18524871355060035,
"grad_norm": 5.423251152038574,
"learning_rate": 9.803768380684242e-05,
"loss": 7.5834,
"step": 27
},
{
"epoch": 0.19210977701543738,
"grad_norm": 5.233191013336182,
"learning_rate": 9.780178907671789e-05,
"loss": 7.3138,
"step": 28
},
{
"epoch": 0.19897084048027444,
"grad_norm": 6.06533145904541,
"learning_rate": 9.755282581475769e-05,
"loss": 7.7481,
"step": 29
},
{
"epoch": 0.2058319039451115,
"grad_norm": 6.182301044464111,
"learning_rate": 9.729086208503174e-05,
"loss": 7.9384,
"step": 30
},
{
"epoch": 0.21269296740994853,
"grad_norm": 6.526544570922852,
"learning_rate": 9.701596950580806e-05,
"loss": 7.8768,
"step": 31
},
{
"epoch": 0.2195540308747856,
"grad_norm": 7.317831039428711,
"learning_rate": 9.672822322997305e-05,
"loss": 8.1528,
"step": 32
},
{
"epoch": 0.22641509433962265,
"grad_norm": 7.71596622467041,
"learning_rate": 9.642770192448536e-05,
"loss": 8.2846,
"step": 33
},
{
"epoch": 0.2332761578044597,
"grad_norm": 9.334805488586426,
"learning_rate": 9.611448774886924e-05,
"loss": 9.105,
"step": 34
},
{
"epoch": 0.24013722126929674,
"grad_norm": 12.165909767150879,
"learning_rate": 9.578866633275288e-05,
"loss": 9.4108,
"step": 35
},
{
"epoch": 0.2469982847341338,
"grad_norm": 43.22640609741211,
"learning_rate": 9.545032675245813e-05,
"loss": 14.8537,
"step": 36
},
{
"epoch": 0.2538593481989708,
"grad_norm": 8.505460739135742,
"learning_rate": 9.509956150664796e-05,
"loss": 8.2678,
"step": 37
},
{
"epoch": 0.2607204116638079,
"grad_norm": 7.500619888305664,
"learning_rate": 9.473646649103818e-05,
"loss": 8.0485,
"step": 38
},
{
"epoch": 0.26758147512864494,
"grad_norm": 6.143131256103516,
"learning_rate": 9.43611409721806e-05,
"loss": 8.2348,
"step": 39
},
{
"epoch": 0.274442538593482,
"grad_norm": 4.7102508544921875,
"learning_rate": 9.397368756032445e-05,
"loss": 7.7434,
"step": 40
},
{
"epoch": 0.28130360205831906,
"grad_norm": 4.04035758972168,
"learning_rate": 9.357421218136386e-05,
"loss": 7.6768,
"step": 41
},
{
"epoch": 0.2881646655231561,
"grad_norm": 3.8485052585601807,
"learning_rate": 9.316282404787871e-05,
"loss": 7.0839,
"step": 42
},
{
"epoch": 0.2950257289879931,
"grad_norm": 3.8183319568634033,
"learning_rate": 9.273963562927695e-05,
"loss": 7.4041,
"step": 43
},
{
"epoch": 0.3018867924528302,
"grad_norm": 3.9263436794281006,
"learning_rate": 9.230476262104677e-05,
"loss": 6.8789,
"step": 44
},
{
"epoch": 0.30874785591766724,
"grad_norm": 4.271997451782227,
"learning_rate": 9.185832391312644e-05,
"loss": 7.567,
"step": 45
},
{
"epoch": 0.3156089193825043,
"grad_norm": 4.242434024810791,
"learning_rate": 9.140044155740101e-05,
"loss": 6.8902,
"step": 46
},
{
"epoch": 0.32246998284734135,
"grad_norm": 4.1314778327941895,
"learning_rate": 9.093124073433463e-05,
"loss": 6.9011,
"step": 47
},
{
"epoch": 0.3293310463121784,
"grad_norm": 4.239864826202393,
"learning_rate": 9.045084971874738e-05,
"loss": 7.6861,
"step": 48
},
{
"epoch": 0.3361921097770154,
"grad_norm": 4.179339408874512,
"learning_rate": 8.995939984474624e-05,
"loss": 7.2055,
"step": 49
},
{
"epoch": 0.34305317324185247,
"grad_norm": 4.8224663734436035,
"learning_rate": 8.945702546981969e-05,
"loss": 7.8787,
"step": 50
},
{
"epoch": 0.34305317324185247,
"eval_loss": 1.9609556198120117,
"eval_runtime": 3.3691,
"eval_samples_per_second": 73.017,
"eval_steps_per_second": 18.403,
"step": 50
},
{
"epoch": 0.34991423670668953,
"grad_norm": 4.2829484939575195,
"learning_rate": 8.894386393810563e-05,
"loss": 7.0131,
"step": 51
},
{
"epoch": 0.3567753001715266,
"grad_norm": 4.773698806762695,
"learning_rate": 8.842005554284296e-05,
"loss": 7.0616,
"step": 52
},
{
"epoch": 0.36363636363636365,
"grad_norm": 4.639217376708984,
"learning_rate": 8.788574348801675e-05,
"loss": 7.2588,
"step": 53
},
{
"epoch": 0.3704974271012007,
"grad_norm": 4.57711124420166,
"learning_rate": 8.73410738492077e-05,
"loss": 6.2915,
"step": 54
},
{
"epoch": 0.37735849056603776,
"grad_norm": 4.3175950050354,
"learning_rate": 8.678619553365659e-05,
"loss": 6.62,
"step": 55
},
{
"epoch": 0.38421955403087477,
"grad_norm": 4.927550792694092,
"learning_rate": 8.622126023955446e-05,
"loss": 6.8197,
"step": 56
},
{
"epoch": 0.3910806174957118,
"grad_norm": 5.108540058135986,
"learning_rate": 8.564642241456986e-05,
"loss": 6.652,
"step": 57
},
{
"epoch": 0.3979416809605489,
"grad_norm": 5.311190128326416,
"learning_rate": 8.506183921362443e-05,
"loss": 6.654,
"step": 58
},
{
"epoch": 0.40480274442538594,
"grad_norm": 5.356717109680176,
"learning_rate": 8.44676704559283e-05,
"loss": 6.7666,
"step": 59
},
{
"epoch": 0.411663807890223,
"grad_norm": 5.327913761138916,
"learning_rate": 8.386407858128706e-05,
"loss": 7.324,
"step": 60
},
{
"epoch": 0.41852487135506006,
"grad_norm": 5.4165191650390625,
"learning_rate": 8.32512286056924e-05,
"loss": 6.7351,
"step": 61
},
{
"epoch": 0.42538593481989706,
"grad_norm": 5.452204704284668,
"learning_rate": 8.262928807620843e-05,
"loss": 5.4856,
"step": 62
},
{
"epoch": 0.4322469982847341,
"grad_norm": 5.71651029586792,
"learning_rate": 8.199842702516583e-05,
"loss": 6.9173,
"step": 63
},
{
"epoch": 0.4391080617495712,
"grad_norm": 6.328149318695068,
"learning_rate": 8.135881792367686e-05,
"loss": 7.5935,
"step": 64
},
{
"epoch": 0.44596912521440824,
"grad_norm": 6.928874969482422,
"learning_rate": 8.07106356344834e-05,
"loss": 7.5651,
"step": 65
},
{
"epoch": 0.4528301886792453,
"grad_norm": 6.131435394287109,
"learning_rate": 8.005405736415126e-05,
"loss": 6.1422,
"step": 66
},
{
"epoch": 0.45969125214408235,
"grad_norm": 7.036497592926025,
"learning_rate": 7.938926261462366e-05,
"loss": 8.1299,
"step": 67
},
{
"epoch": 0.4665523156089194,
"grad_norm": 7.607172966003418,
"learning_rate": 7.871643313414718e-05,
"loss": 7.5372,
"step": 68
},
{
"epoch": 0.4734133790737564,
"grad_norm": 7.075242042541504,
"learning_rate": 7.803575286758364e-05,
"loss": 7.735,
"step": 69
},
{
"epoch": 0.48027444253859347,
"grad_norm": 8.967554092407227,
"learning_rate": 7.734740790612136e-05,
"loss": 8.3584,
"step": 70
},
{
"epoch": 0.48713550600343053,
"grad_norm": 10.635770797729492,
"learning_rate": 7.66515864363997e-05,
"loss": 7.8883,
"step": 71
},
{
"epoch": 0.4939965694682676,
"grad_norm": 36.83614730834961,
"learning_rate": 7.594847868906076e-05,
"loss": 13.816,
"step": 72
},
{
"epoch": 0.5008576329331046,
"grad_norm": 5.224258899688721,
"learning_rate": 7.52382768867422e-05,
"loss": 7.5155,
"step": 73
},
{
"epoch": 0.5077186963979416,
"grad_norm": 5.5043864250183105,
"learning_rate": 7.452117519152542e-05,
"loss": 7.4326,
"step": 74
},
{
"epoch": 0.5145797598627787,
"grad_norm": 4.99073600769043,
"learning_rate": 7.379736965185368e-05,
"loss": 7.7991,
"step": 75
},
{
"epoch": 0.5214408233276158,
"grad_norm": 4.494393825531006,
"learning_rate": 7.30670581489344e-05,
"loss": 6.8002,
"step": 76
},
{
"epoch": 0.5283018867924528,
"grad_norm": 4.096364974975586,
"learning_rate": 7.233044034264034e-05,
"loss": 6.5461,
"step": 77
},
{
"epoch": 0.5351629502572899,
"grad_norm": 3.8724262714385986,
"learning_rate": 7.158771761692464e-05,
"loss": 7.0417,
"step": 78
},
{
"epoch": 0.5420240137221269,
"grad_norm": 3.8217287063598633,
"learning_rate": 7.083909302476453e-05,
"loss": 6.084,
"step": 79
},
{
"epoch": 0.548885077186964,
"grad_norm": 3.9765660762786865,
"learning_rate": 7.008477123264848e-05,
"loss": 6.8385,
"step": 80
},
{
"epoch": 0.5557461406518011,
"grad_norm": 3.7557239532470703,
"learning_rate": 6.932495846462261e-05,
"loss": 5.9077,
"step": 81
},
{
"epoch": 0.5626072041166381,
"grad_norm": 4.217108249664307,
"learning_rate": 6.855986244591104e-05,
"loss": 7.4603,
"step": 82
},
{
"epoch": 0.5694682675814752,
"grad_norm": 3.938063144683838,
"learning_rate": 6.778969234612584e-05,
"loss": 6.4279,
"step": 83
},
{
"epoch": 0.5763293310463122,
"grad_norm": 4.099449634552002,
"learning_rate": 6.701465872208216e-05,
"loss": 6.6643,
"step": 84
},
{
"epoch": 0.5831903945111492,
"grad_norm": 4.268118381500244,
"learning_rate": 6.623497346023418e-05,
"loss": 7.2316,
"step": 85
},
{
"epoch": 0.5900514579759862,
"grad_norm": 4.3306565284729,
"learning_rate": 6.545084971874738e-05,
"loss": 6.6102,
"step": 86
},
{
"epoch": 0.5969125214408233,
"grad_norm": 4.440483570098877,
"learning_rate": 6.466250186922325e-05,
"loss": 6.7896,
"step": 87
},
{
"epoch": 0.6037735849056604,
"grad_norm": 4.6268110275268555,
"learning_rate": 6.387014543809223e-05,
"loss": 6.8771,
"step": 88
},
{
"epoch": 0.6106346483704974,
"grad_norm": 4.460641384124756,
"learning_rate": 6.307399704769099e-05,
"loss": 6.5277,
"step": 89
},
{
"epoch": 0.6174957118353345,
"grad_norm": 4.795734405517578,
"learning_rate": 6.227427435703997e-05,
"loss": 7.0335,
"step": 90
},
{
"epoch": 0.6243567753001715,
"grad_norm": 4.73524808883667,
"learning_rate": 6.147119600233758e-05,
"loss": 7.2792,
"step": 91
},
{
"epoch": 0.6312178387650086,
"grad_norm": 4.746474742889404,
"learning_rate": 6.066498153718735e-05,
"loss": 6.5431,
"step": 92
},
{
"epoch": 0.6380789022298456,
"grad_norm": 4.961014747619629,
"learning_rate": 5.985585137257401e-05,
"loss": 6.9168,
"step": 93
},
{
"epoch": 0.6449399656946827,
"grad_norm": 5.02461051940918,
"learning_rate": 5.90440267166055e-05,
"loss": 6.2909,
"step": 94
},
{
"epoch": 0.6518010291595198,
"grad_norm": 5.03878927230835,
"learning_rate": 5.8229729514036705e-05,
"loss": 7.0238,
"step": 95
},
{
"epoch": 0.6586620926243568,
"grad_norm": 5.197082996368408,
"learning_rate": 5.74131823855921e-05,
"loss": 6.0466,
"step": 96
},
{
"epoch": 0.6655231560891939,
"grad_norm": 5.540366172790527,
"learning_rate": 5.6594608567103456e-05,
"loss": 6.8652,
"step": 97
},
{
"epoch": 0.6723842195540308,
"grad_norm": 5.839454650878906,
"learning_rate": 5.577423184847932e-05,
"loss": 6.4574,
"step": 98
},
{
"epoch": 0.6792452830188679,
"grad_norm": 5.818624973297119,
"learning_rate": 5.495227651252315e-05,
"loss": 6.6828,
"step": 99
},
{
"epoch": 0.6861063464837049,
"grad_norm": 5.735467433929443,
"learning_rate": 5.4128967273616625e-05,
"loss": 7.0153,
"step": 100
},
{
"epoch": 0.6861063464837049,
"eval_loss": 1.8372527360916138,
"eval_runtime": 3.3857,
"eval_samples_per_second": 72.659,
"eval_steps_per_second": 18.313,
"step": 100
},
{
"epoch": 0.692967409948542,
"grad_norm": 6.105105876922607,
"learning_rate": 5.330452921628497e-05,
"loss": 6.0957,
"step": 101
},
{
"epoch": 0.6998284734133791,
"grad_norm": 6.07030725479126,
"learning_rate": 5.247918773366112e-05,
"loss": 6.9366,
"step": 102
},
{
"epoch": 0.7066895368782161,
"grad_norm": 7.4269866943359375,
"learning_rate": 5.165316846586541e-05,
"loss": 6.7843,
"step": 103
},
{
"epoch": 0.7135506003430532,
"grad_norm": 6.641677379608154,
"learning_rate": 5.0826697238317935e-05,
"loss": 7.4297,
"step": 104
},
{
"epoch": 0.7204116638078902,
"grad_norm": 7.776499271392822,
"learning_rate": 5e-05,
"loss": 6.7605,
"step": 105
},
{
"epoch": 0.7272727272727273,
"grad_norm": 8.558423042297363,
"learning_rate": 4.917330276168208e-05,
"loss": 7.5774,
"step": 106
},
{
"epoch": 0.7341337907375644,
"grad_norm": 11.58539867401123,
"learning_rate": 4.834683153413459e-05,
"loss": 8.866,
"step": 107
},
{
"epoch": 0.7409948542024014,
"grad_norm": 26.738967895507812,
"learning_rate": 4.7520812266338885e-05,
"loss": 12.3726,
"step": 108
},
{
"epoch": 0.7478559176672385,
"grad_norm": 3.9378786087036133,
"learning_rate": 4.669547078371504e-05,
"loss": 6.9369,
"step": 109
},
{
"epoch": 0.7547169811320755,
"grad_norm": 4.1709885597229,
"learning_rate": 4.5871032726383386e-05,
"loss": 7.1027,
"step": 110
},
{
"epoch": 0.7615780445969125,
"grad_norm": 4.352792263031006,
"learning_rate": 4.504772348747687e-05,
"loss": 7.2637,
"step": 111
},
{
"epoch": 0.7684391080617495,
"grad_norm": 4.159816265106201,
"learning_rate": 4.4225768151520694e-05,
"loss": 7.4184,
"step": 112
},
{
"epoch": 0.7753001715265866,
"grad_norm": 4.296802043914795,
"learning_rate": 4.3405391432896555e-05,
"loss": 6.9394,
"step": 113
},
{
"epoch": 0.7821612349914236,
"grad_norm": 4.155566692352295,
"learning_rate": 4.2586817614407895e-05,
"loss": 7.0328,
"step": 114
},
{
"epoch": 0.7890222984562607,
"grad_norm": 4.032870769500732,
"learning_rate": 4.17702704859633e-05,
"loss": 6.0346,
"step": 115
},
{
"epoch": 0.7958833619210978,
"grad_norm": 4.005494117736816,
"learning_rate": 4.095597328339452e-05,
"loss": 7.1284,
"step": 116
},
{
"epoch": 0.8027444253859348,
"grad_norm": 4.260548114776611,
"learning_rate": 4.0144148627425993e-05,
"loss": 6.6892,
"step": 117
},
{
"epoch": 0.8096054888507719,
"grad_norm": 4.667361736297607,
"learning_rate": 3.933501846281267e-05,
"loss": 7.5277,
"step": 118
},
{
"epoch": 0.8164665523156089,
"grad_norm": 4.318197250366211,
"learning_rate": 3.852880399766243e-05,
"loss": 6.2792,
"step": 119
},
{
"epoch": 0.823327615780446,
"grad_norm": 4.119472026824951,
"learning_rate": 3.772572564296005e-05,
"loss": 5.9804,
"step": 120
},
{
"epoch": 0.8301886792452831,
"grad_norm": 4.485971450805664,
"learning_rate": 3.6926002952309016e-05,
"loss": 6.4601,
"step": 121
},
{
"epoch": 0.8370497427101201,
"grad_norm": 4.47144079208374,
"learning_rate": 3.612985456190778e-05,
"loss": 6.0583,
"step": 122
},
{
"epoch": 0.8439108061749572,
"grad_norm": 4.6258063316345215,
"learning_rate": 3.533749813077677e-05,
"loss": 7.0987,
"step": 123
},
{
"epoch": 0.8507718696397941,
"grad_norm": 4.7973246574401855,
"learning_rate": 3.4549150281252636e-05,
"loss": 6.9344,
"step": 124
},
{
"epoch": 0.8576329331046312,
"grad_norm": 4.9836249351501465,
"learning_rate": 3.3765026539765834e-05,
"loss": 6.1316,
"step": 125
},
{
"epoch": 0.8644939965694682,
"grad_norm": 5.039267063140869,
"learning_rate": 3.298534127791785e-05,
"loss": 6.878,
"step": 126
},
{
"epoch": 0.8713550600343053,
"grad_norm": 5.1987199783325195,
"learning_rate": 3.221030765387417e-05,
"loss": 6.6305,
"step": 127
},
{
"epoch": 0.8782161234991424,
"grad_norm": 4.993228912353516,
"learning_rate": 3.144013755408895e-05,
"loss": 6.1317,
"step": 128
},
{
"epoch": 0.8850771869639794,
"grad_norm": 5.320596694946289,
"learning_rate": 3.0675041535377405e-05,
"loss": 5.8034,
"step": 129
},
{
"epoch": 0.8919382504288165,
"grad_norm": 5.601073741912842,
"learning_rate": 2.991522876735154e-05,
"loss": 6.2977,
"step": 130
},
{
"epoch": 0.8987993138936535,
"grad_norm": 5.388200283050537,
"learning_rate": 2.916090697523549e-05,
"loss": 6.2854,
"step": 131
},
{
"epoch": 0.9056603773584906,
"grad_norm": 5.7619829177856445,
"learning_rate": 2.8412282383075363e-05,
"loss": 6.0629,
"step": 132
},
{
"epoch": 0.9125214408233276,
"grad_norm": 5.515921115875244,
"learning_rate": 2.766955965735968e-05,
"loss": 5.6834,
"step": 133
},
{
"epoch": 0.9193825042881647,
"grad_norm": 5.793899059295654,
"learning_rate": 2.693294185106562e-05,
"loss": 6.5223,
"step": 134
},
{
"epoch": 0.9262435677530018,
"grad_norm": 5.80432653427124,
"learning_rate": 2.6202630348146324e-05,
"loss": 6.1701,
"step": 135
},
{
"epoch": 0.9331046312178388,
"grad_norm": 6.891260623931885,
"learning_rate": 2.547882480847461e-05,
"loss": 7.1704,
"step": 136
},
{
"epoch": 0.9399656946826758,
"grad_norm": 6.534001350402832,
"learning_rate": 2.476172311325783e-05,
"loss": 6.2087,
"step": 137
},
{
"epoch": 0.9468267581475128,
"grad_norm": 7.076842308044434,
"learning_rate": 2.405152131093926e-05,
"loss": 6.7816,
"step": 138
},
{
"epoch": 0.9536878216123499,
"grad_norm": 7.151756763458252,
"learning_rate": 2.3348413563600325e-05,
"loss": 6.5743,
"step": 139
},
{
"epoch": 0.9605488850771869,
"grad_norm": 8.139583587646484,
"learning_rate": 2.2652592093878666e-05,
"loss": 6.646,
"step": 140
},
{
"epoch": 0.967409948542024,
"grad_norm": 7.985987663269043,
"learning_rate": 2.196424713241637e-05,
"loss": 7.0655,
"step": 141
},
{
"epoch": 0.9742710120068611,
"grad_norm": 9.346457481384277,
"learning_rate": 2.128356686585282e-05,
"loss": 7.9942,
"step": 142
},
{
"epoch": 0.9811320754716981,
"grad_norm": 12.354272842407227,
"learning_rate": 2.061073738537635e-05,
"loss": 8.2754,
"step": 143
},
{
"epoch": 0.9879931389365352,
"grad_norm": 26.231876373291016,
"learning_rate": 1.9945942635848748e-05,
"loss": 12.5108,
"step": 144
},
{
"epoch": 0.9948542024013722,
"grad_norm": 4.590910911560059,
"learning_rate": 1.928936436551661e-05,
"loss": 6.2901,
"step": 145
},
{
"epoch": 1.0017152658662092,
"grad_norm": 6.709409236907959,
"learning_rate": 1.8641182076323148e-05,
"loss": 7.2746,
"step": 146
},
{
"epoch": 1.0085763293310464,
"grad_norm": 3.330141067504883,
"learning_rate": 1.800157297483417e-05,
"loss": 6.2361,
"step": 147
},
{
"epoch": 1.0154373927958833,
"grad_norm": 3.584874153137207,
"learning_rate": 1.7370711923791567e-05,
"loss": 6.5322,
"step": 148
},
{
"epoch": 1.0222984562607205,
"grad_norm": 3.6918985843658447,
"learning_rate": 1.6748771394307585e-05,
"loss": 6.3232,
"step": 149
},
{
"epoch": 1.0291595197255574,
"grad_norm": 3.692216396331787,
"learning_rate": 1.6135921418712956e-05,
"loss": 6.5346,
"step": 150
},
{
"epoch": 1.0291595197255574,
"eval_loss": 1.7711015939712524,
"eval_runtime": 3.3926,
"eval_samples_per_second": 72.512,
"eval_steps_per_second": 18.275,
"step": 150
},
{
"epoch": 1.0360205831903946,
"grad_norm": 3.892913341522217,
"learning_rate": 1.553232954407171e-05,
"loss": 6.1995,
"step": 151
},
{
"epoch": 1.0428816466552315,
"grad_norm": 3.7631001472473145,
"learning_rate": 1.4938160786375572e-05,
"loss": 5.9457,
"step": 152
},
{
"epoch": 1.0497427101200687,
"grad_norm": 3.9568772315979004,
"learning_rate": 1.435357758543015e-05,
"loss": 6.001,
"step": 153
},
{
"epoch": 1.0566037735849056,
"grad_norm": 4.111072063446045,
"learning_rate": 1.3778739760445552e-05,
"loss": 5.7899,
"step": 154
},
{
"epoch": 1.0634648370497426,
"grad_norm": 4.1294965744018555,
"learning_rate": 1.3213804466343421e-05,
"loss": 5.7028,
"step": 155
},
{
"epoch": 1.0703259005145798,
"grad_norm": 4.020060062408447,
"learning_rate": 1.2658926150792322e-05,
"loss": 5.6423,
"step": 156
},
{
"epoch": 1.0771869639794167,
"grad_norm": 4.024879455566406,
"learning_rate": 1.2114256511983274e-05,
"loss": 5.3037,
"step": 157
},
{
"epoch": 1.0840480274442539,
"grad_norm": 4.263257026672363,
"learning_rate": 1.157994445715706e-05,
"loss": 6.0707,
"step": 158
},
{
"epoch": 1.0909090909090908,
"grad_norm": 4.5629143714904785,
"learning_rate": 1.1056136061894384e-05,
"loss": 6.3808,
"step": 159
},
{
"epoch": 1.097770154373928,
"grad_norm": 4.353785514831543,
"learning_rate": 1.0542974530180327e-05,
"loss": 6.0241,
"step": 160
},
{
"epoch": 1.104631217838765,
"grad_norm": 4.443587303161621,
"learning_rate": 1.0040600155253765e-05,
"loss": 5.6189,
"step": 161
},
{
"epoch": 1.1114922813036021,
"grad_norm": 4.836148738861084,
"learning_rate": 9.549150281252633e-06,
"loss": 6.2618,
"step": 162
},
{
"epoch": 1.118353344768439,
"grad_norm": 4.852650165557861,
"learning_rate": 9.068759265665384e-06,
"loss": 5.983,
"step": 163
},
{
"epoch": 1.1252144082332762,
"grad_norm": 4.747347354888916,
"learning_rate": 8.599558442598998e-06,
"loss": 5.8938,
"step": 164
},
{
"epoch": 1.1320754716981132,
"grad_norm": 4.495707988739014,
"learning_rate": 8.141676086873572e-06,
"loss": 5.1317,
"step": 165
},
{
"epoch": 1.1389365351629503,
"grad_norm": 5.014614582061768,
"learning_rate": 7.695237378953223e-06,
"loss": 5.9791,
"step": 166
},
{
"epoch": 1.1457975986277873,
"grad_norm": 4.957954406738281,
"learning_rate": 7.260364370723044e-06,
"loss": 5.545,
"step": 167
},
{
"epoch": 1.1526586620926245,
"grad_norm": 4.905029773712158,
"learning_rate": 6.837175952121306e-06,
"loss": 5.2862,
"step": 168
},
{
"epoch": 1.1595197255574614,
"grad_norm": 5.664605140686035,
"learning_rate": 6.425787818636131e-06,
"loss": 5.6839,
"step": 169
},
{
"epoch": 1.1663807890222984,
"grad_norm": 5.323741436004639,
"learning_rate": 6.026312439675552e-06,
"loss": 5.2487,
"step": 170
},
{
"epoch": 1.1732418524871355,
"grad_norm": 5.226016521453857,
"learning_rate": 5.6388590278194096e-06,
"loss": 5.3256,
"step": 171
},
{
"epoch": 1.1801029159519725,
"grad_norm": 5.518038749694824,
"learning_rate": 5.263533508961827e-06,
"loss": 5.9731,
"step": 172
},
{
"epoch": 1.1869639794168096,
"grad_norm": 5.61544132232666,
"learning_rate": 4.900438493352055e-06,
"loss": 5.1816,
"step": 173
},
{
"epoch": 1.1938250428816466,
"grad_norm": 6.291421413421631,
"learning_rate": 4.549673247541875e-06,
"loss": 5.8492,
"step": 174
},
{
"epoch": 1.2006861063464838,
"grad_norm": 6.391525745391846,
"learning_rate": 4.2113336672471245e-06,
"loss": 6.0124,
"step": 175
},
{
"epoch": 1.2075471698113207,
"grad_norm": 6.810779094696045,
"learning_rate": 3.885512251130763e-06,
"loss": 6.3855,
"step": 176
},
{
"epoch": 1.2144082332761579,
"grad_norm": 6.760521411895752,
"learning_rate": 3.5722980755146517e-06,
"loss": 5.943,
"step": 177
},
{
"epoch": 1.2212692967409948,
"grad_norm": 7.761327266693115,
"learning_rate": 3.271776770026963e-06,
"loss": 6.1512,
"step": 178
},
{
"epoch": 1.228130360205832,
"grad_norm": 7.864924907684326,
"learning_rate": 2.9840304941919415e-06,
"loss": 6.0006,
"step": 179
},
{
"epoch": 1.234991423670669,
"grad_norm": 11.488492965698242,
"learning_rate": 2.7091379149682685e-06,
"loss": 7.2134,
"step": 180
},
{
"epoch": 1.241852487135506,
"grad_norm": 13.040229797363281,
"learning_rate": 2.4471741852423237e-06,
"loss": 7.9826,
"step": 181
},
{
"epoch": 1.248713550600343,
"grad_norm": 19.542373657226562,
"learning_rate": 2.1982109232821178e-06,
"loss": 10.1329,
"step": 182
},
{
"epoch": 1.2555746140651802,
"grad_norm": 3.2909622192382812,
"learning_rate": 1.962316193157593e-06,
"loss": 6.4409,
"step": 183
},
{
"epoch": 1.2624356775300172,
"grad_norm": 3.6267285346984863,
"learning_rate": 1.7395544861325718e-06,
"loss": 6.9252,
"step": 184
},
{
"epoch": 1.2692967409948541,
"grad_norm": 3.6771252155303955,
"learning_rate": 1.5299867030334814e-06,
"loss": 6.7405,
"step": 185
},
{
"epoch": 1.2761578044596913,
"grad_norm": 3.7336010932922363,
"learning_rate": 1.333670137599713e-06,
"loss": 6.1917,
"step": 186
},
{
"epoch": 1.2830188679245282,
"grad_norm": 3.6931679248809814,
"learning_rate": 1.1506584608200367e-06,
"loss": 6.2048,
"step": 187
},
{
"epoch": 1.2898799313893654,
"grad_norm": 3.749580144882202,
"learning_rate": 9.810017062595322e-07,
"loss": 5.8176,
"step": 188
},
{
"epoch": 1.2967409948542024,
"grad_norm": 3.754908800125122,
"learning_rate": 8.247462563808817e-07,
"loss": 5.7098,
"step": 189
},
{
"epoch": 1.3036020583190395,
"grad_norm": 3.99393367767334,
"learning_rate": 6.819348298638839e-07,
"loss": 6.2022,
"step": 190
},
{
"epoch": 1.3104631217838765,
"grad_norm": 4.034095287322998,
"learning_rate": 5.526064699265753e-07,
"loss": 5.9129,
"step": 191
},
{
"epoch": 1.3173241852487134,
"grad_norm": 4.365250587463379,
"learning_rate": 4.367965336512403e-07,
"loss": 6.5774,
"step": 192
},
{
"epoch": 1.3241852487135506,
"grad_norm": 4.243707656860352,
"learning_rate": 3.3453668231809286e-07,
"loss": 6.0522,
"step": 193
},
{
"epoch": 1.3310463121783878,
"grad_norm": 4.496011257171631,
"learning_rate": 2.458548727494292e-07,
"loss": 6.2079,
"step": 194
},
{
"epoch": 1.3379073756432247,
"grad_norm": 4.635636329650879,
"learning_rate": 1.7077534966650766e-07,
"loss": 6.341,
"step": 195
},
{
"epoch": 1.3447684391080617,
"grad_norm": 4.704632759094238,
"learning_rate": 1.0931863906127327e-07,
"loss": 6.0428,
"step": 196
},
{
"epoch": 1.3516295025728988,
"grad_norm": 4.597292900085449,
"learning_rate": 6.150154258476315e-08,
"loss": 6.0458,
"step": 197
},
{
"epoch": 1.3584905660377358,
"grad_norm": 4.808545112609863,
"learning_rate": 2.7337132953697554e-08,
"loss": 5.888,
"step": 198
},
{
"epoch": 1.365351629502573,
"grad_norm": 4.711483478546143,
"learning_rate": 6.834750376549792e-09,
"loss": 5.5179,
"step": 199
},
{
"epoch": 1.3722126929674099,
"grad_norm": 4.874032974243164,
"learning_rate": 0.0,
"loss": 5.8487,
"step": 200
},
{
"epoch": 1.3722126929674099,
"eval_loss": 1.746632695198059,
"eval_runtime": 3.3911,
"eval_samples_per_second": 72.543,
"eval_steps_per_second": 18.283,
"step": 200
}
],
"logging_steps": 1,
"max_steps": 200,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"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": 3.704748384662323e+16,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}