Files
b2_math_length_3k/trainer_state.json

1611 lines
39 KiB
JSON
Raw Permalink Normal View History

{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 6.850632911392405,
"eval_steps": 500,
"global_step": 224,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.030379746835443037,
"grad_norm": 6.067921922621284,
"learning_rate": 8.695652173913044e-07,
"loss": 0.7951,
"step": 1
},
{
"epoch": 0.060759493670886074,
"grad_norm": 6.108719114688138,
"learning_rate": 1.7391304347826088e-06,
"loss": 0.7914,
"step": 2
},
{
"epoch": 0.09113924050632911,
"grad_norm": 6.288511845899511,
"learning_rate": 2.6086956521739132e-06,
"loss": 0.8256,
"step": 3
},
{
"epoch": 0.12151898734177215,
"grad_norm": 5.700816735945768,
"learning_rate": 3.4782608695652175e-06,
"loss": 0.8069,
"step": 4
},
{
"epoch": 0.1518987341772152,
"grad_norm": 4.311989426038608,
"learning_rate": 4.347826086956522e-06,
"loss": 0.786,
"step": 5
},
{
"epoch": 0.18227848101265823,
"grad_norm": 2.3382574410736026,
"learning_rate": 5.2173913043478265e-06,
"loss": 0.7004,
"step": 6
},
{
"epoch": 0.21265822784810126,
"grad_norm": 2.0088413080231584,
"learning_rate": 6.086956521739132e-06,
"loss": 0.7298,
"step": 7
},
{
"epoch": 0.2430379746835443,
"grad_norm": 3.552758051426024,
"learning_rate": 6.956521739130435e-06,
"loss": 0.6914,
"step": 8
},
{
"epoch": 0.27341772151898736,
"grad_norm": 4.038019502514299,
"learning_rate": 7.82608695652174e-06,
"loss": 0.6995,
"step": 9
},
{
"epoch": 0.3037974683544304,
"grad_norm": 4.000235761498213,
"learning_rate": 8.695652173913044e-06,
"loss": 0.6807,
"step": 10
},
{
"epoch": 0.3341772151898734,
"grad_norm": 3.7947577560871983,
"learning_rate": 9.565217391304349e-06,
"loss": 0.6618,
"step": 11
},
{
"epoch": 0.36455696202531646,
"grad_norm": 3.0340250124666266,
"learning_rate": 1.0434782608695653e-05,
"loss": 0.6354,
"step": 12
},
{
"epoch": 0.3949367088607595,
"grad_norm": 1.811526349989083,
"learning_rate": 1.1304347826086957e-05,
"loss": 0.6327,
"step": 13
},
{
"epoch": 0.4253164556962025,
"grad_norm": 1.543063478406062,
"learning_rate": 1.2173913043478263e-05,
"loss": 0.6265,
"step": 14
},
{
"epoch": 0.45569620253164556,
"grad_norm": 1.6969256139364837,
"learning_rate": 1.3043478260869566e-05,
"loss": 0.6016,
"step": 15
},
{
"epoch": 0.4860759493670886,
"grad_norm": 1.4380631231045289,
"learning_rate": 1.391304347826087e-05,
"loss": 0.6376,
"step": 16
},
{
"epoch": 0.5164556962025316,
"grad_norm": 0.9527493252494761,
"learning_rate": 1.4782608695652174e-05,
"loss": 0.5866,
"step": 17
},
{
"epoch": 0.5468354430379747,
"grad_norm": 1.049530005726062,
"learning_rate": 1.565217391304348e-05,
"loss": 0.5703,
"step": 18
},
{
"epoch": 0.5772151898734177,
"grad_norm": 0.8038184411829851,
"learning_rate": 1.6521739130434785e-05,
"loss": 0.5505,
"step": 19
},
{
"epoch": 0.6075949367088608,
"grad_norm": 1.0595863114085984,
"learning_rate": 1.739130434782609e-05,
"loss": 0.5687,
"step": 20
},
{
"epoch": 0.6379746835443038,
"grad_norm": 0.7012774325635927,
"learning_rate": 1.8260869565217393e-05,
"loss": 0.5613,
"step": 21
},
{
"epoch": 0.6683544303797468,
"grad_norm": 0.8596606198426938,
"learning_rate": 1.9130434782608697e-05,
"loss": 0.5463,
"step": 22
},
{
"epoch": 0.6987341772151898,
"grad_norm": 0.9207507426851212,
"learning_rate": 2e-05,
"loss": 0.5378,
"step": 23
},
{
"epoch": 0.7291139240506329,
"grad_norm": 0.6942985100880201,
"learning_rate": 1.999877856940653e-05,
"loss": 0.5425,
"step": 24
},
{
"epoch": 0.759493670886076,
"grad_norm": 0.7862630342036437,
"learning_rate": 1.999511457600466e-05,
"loss": 0.5394,
"step": 25
},
{
"epoch": 0.789873417721519,
"grad_norm": 0.7309840240381787,
"learning_rate": 1.9989008914857115e-05,
"loss": 0.5322,
"step": 26
},
{
"epoch": 0.8202531645569621,
"grad_norm": 0.6423700302308096,
"learning_rate": 1.998046307749216e-05,
"loss": 0.5197,
"step": 27
},
{
"epoch": 0.850632911392405,
"grad_norm": 0.7886202803739318,
"learning_rate": 1.9969479151539238e-05,
"loss": 0.53,
"step": 28
},
{
"epoch": 0.8810126582278481,
"grad_norm": 0.6643658973489672,
"learning_rate": 1.9956059820218982e-05,
"loss": 0.5197,
"step": 29
},
{
"epoch": 0.9113924050632911,
"grad_norm": 0.7255018213056038,
"learning_rate": 1.9940208361687762e-05,
"loss": 0.5176,
"step": 30
},
{
"epoch": 0.9417721518987342,
"grad_norm": 0.5986990641409521,
"learning_rate": 1.9921928648236855e-05,
"loss": 0.5096,
"step": 31
},
{
"epoch": 0.9721518987341772,
"grad_norm": 0.5547261207246215,
"learning_rate": 1.990122514534651e-05,
"loss": 0.5026,
"step": 32
},
{
"epoch": 1.010126582278481,
"grad_norm": 0.5703822220148793,
"learning_rate": 1.9878102910595097e-05,
"loss": 0.4982,
"step": 33
},
{
"epoch": 1.040506329113924,
"grad_norm": 0.48757959755404084,
"learning_rate": 1.985256759242359e-05,
"loss": 0.4825,
"step": 34
},
{
"epoch": 1.070886075949367,
"grad_norm": 0.4571840939681635,
"learning_rate": 1.982462542875576e-05,
"loss": 0.4927,
"step": 35
},
{
"epoch": 1.1012658227848102,
"grad_norm": 0.4851528882434097,
"learning_rate": 1.979428324547432e-05,
"loss": 0.4858,
"step": 36
},
{
"epoch": 1.1316455696202532,
"grad_norm": 0.4530089506674034,
"learning_rate": 1.9761548454753455e-05,
"loss": 0.4878,
"step": 37
},
{
"epoch": 1.1620253164556962,
"grad_norm": 0.496411669356003,
"learning_rate": 1.972642905324813e-05,
"loss": 0.48,
"step": 38
},
{
"epoch": 1.1924050632911392,
"grad_norm": 0.44634933875095034,
"learning_rate": 1.9688933620140638e-05,
"loss": 0.4767,
"step": 39
},
{
"epoch": 1.2227848101265824,
"grad_norm": 0.5369930695150029,
"learning_rate": 1.96490713150448e-05,
"loss": 0.4748,
"step": 40
},
{
"epoch": 1.2531645569620253,
"grad_norm": 0.4507840293800539,
"learning_rate": 1.9606851875768404e-05,
"loss": 0.4827,
"step": 41
},
{
"epoch": 1.2835443037974683,
"grad_norm": 0.5588361622849422,
"learning_rate": 1.956228561593441e-05,
"loss": 0.4745,
"step": 42
},
{
"epoch": 1.3139240506329113,
"grad_norm": 0.608399361925596,
"learning_rate": 1.9515383422461457e-05,
"loss": 0.4794,
"step": 43
},
{
"epoch": 1.3443037974683545,
"grad_norm": 0.4334891155195449,
"learning_rate": 1.9466156752904344e-05,
"loss": 0.458,
"step": 44
},
{
"epoch": 1.3746835443037975,
"grad_norm": 0.580642782282209,
"learning_rate": 1.9414617632655114e-05,
"loss": 0.455,
"step": 45
},
{
"epoch": 1.4050632911392404,
"grad_norm": 0.4038066014823156,
"learning_rate": 1.9360778652005416e-05,
"loss": 0.4572,
"step": 46
},
{
"epoch": 1.4354430379746836,
"grad_norm": 0.4321591845092396,
"learning_rate": 1.9304652963070868e-05,
"loss": 0.4616,
"step": 47
},
{
"epoch": 1.4658227848101266,
"grad_norm": 0.3917317337891077,
"learning_rate": 1.9246254276578175e-05,
"loss": 0.4527,
"step": 48
},
{
"epoch": 1.4962025316455696,
"grad_norm": 0.4266219789965417,
"learning_rate": 1.9185596858515797e-05,
"loss": 0.4671,
"step": 49
},
{
"epoch": 1.5265822784810128,
"grad_norm": 0.4619876958583412,
"learning_rate": 1.9122695526648968e-05,
"loss": 0.4615,
"step": 50
},
{
"epoch": 1.5569620253164556,
"grad_norm": 0.36985369127721734,
"learning_rate": 1.905756564689991e-05,
"loss": 0.4628,
"step": 51
},
{
"epoch": 1.5873417721518988,
"grad_norm": 0.37249584548900333,
"learning_rate": 1.8990223129594146e-05,
"loss": 0.4509,
"step": 52
},
{
"epoch": 1.6177215189873417,
"grad_norm": 0.4701788607695343,
"learning_rate": 1.8920684425573865e-05,
"loss": 0.458,
"step": 53
},
{
"epoch": 1.6481012658227847,
"grad_norm": 0.37289021513403814,
"learning_rate": 1.884896652217917e-05,
"loss": 0.4619,
"step": 54
},
{
"epoch": 1.678481012658228,
"grad_norm": 0.5209521079430768,
"learning_rate": 1.877508693909831e-05,
"loss": 0.474,
"step": 55
},
{
"epoch": 1.7088607594936709,
"grad_norm": 0.30920470819265755,
"learning_rate": 1.8699063724087905e-05,
"loss": 0.4236,
"step": 56
},
{
"epoch": 1.7392405063291139,
"grad_norm": 0.4802588274248861,
"learning_rate": 1.862091544856407e-05,
"loss": 0.4623,
"step": 57
},
{
"epoch": 1.769620253164557,
"grad_norm": 0.38441459092235875,
"learning_rate": 1.854066120306571e-05,
"loss": 0.4471,
"step": 58
},
{
"epoch": 1.8,
"grad_norm": 0.4134473195975686,
"learning_rate": 1.8458320592590976e-05,
"loss": 0.4508,
"step": 59
},
{
"epoch": 1.830379746835443,
"grad_norm": 0.4335578526849257,
"learning_rate": 1.837391373180801e-05,
"loss": 0.47,
"step": 60
},
{
"epoch": 1.8607594936708862,
"grad_norm": 0.4461138795120413,
"learning_rate": 1.8287461240141217e-05,
"loss": 0.455,
"step": 61
},
{
"epoch": 1.891139240506329,
"grad_norm": 0.48021321527595906,
"learning_rate": 1.8198984236734246e-05,
"loss": 0.4526,
"step": 62
},
{
"epoch": 1.9215189873417722,
"grad_norm": 0.4096614538832078,
"learning_rate": 1.8108504335290852e-05,
"loss": 0.4546,
"step": 63
},
{
"epoch": 1.9518987341772152,
"grad_norm": 0.45929687038488765,
"learning_rate": 1.8016043638794975e-05,
"loss": 0.4472,
"step": 64
},
{
"epoch": 1.9822784810126581,
"grad_norm": 0.40616815484896374,
"learning_rate": 1.7921624734111292e-05,
"loss": 0.4544,
"step": 65
},
{
"epoch": 2.020253164556962,
"grad_norm": 0.4920837093917966,
"learning_rate": 1.7825270686467567e-05,
"loss": 0.4304,
"step": 66
},
{
"epoch": 2.050632911392405,
"grad_norm": 0.3726499647984961,
"learning_rate": 1.7727005033820117e-05,
"loss": 0.4278,
"step": 67
},
{
"epoch": 2.081012658227848,
"grad_norm": 0.36093568352608935,
"learning_rate": 1.762685178110382e-05,
"loss": 0.389,
"step": 68
},
{
"epoch": 2.1113924050632913,
"grad_norm": 0.43247029125780884,
"learning_rate": 1.752483539436807e-05,
"loss": 0.4189,
"step": 69
},
{
"epoch": 2.141772151898734,
"grad_norm": 0.44335025902613256,
"learning_rate": 1.7420980794800013e-05,
"loss": 0.4029,
"step": 70
},
{
"epoch": 2.1721518987341772,
"grad_norm": 0.4418906316133121,
"learning_rate": 1.731531335263669e-05,
"loss": 0.4116,
"step": 71
},
{
"epoch": 2.2025316455696204,
"grad_norm": 0.4069117230843435,
"learning_rate": 1.720785888096743e-05,
"loss": 0.3974,
"step": 72
},
{
"epoch": 2.232911392405063,
"grad_norm": 0.445008660978375,
"learning_rate": 1.7098643629428035e-05,
"loss": 0.3987,
"step": 73
},
{
"epoch": 2.2632911392405064,
"grad_norm": 0.36740044746599393,
"learning_rate": 1.698769427778842e-05,
"loss": 0.4096,
"step": 74
},
{
"epoch": 2.293670886075949,
"grad_norm": 0.3684783160234449,
"learning_rate": 1.687503792943506e-05,
"loss": 0.4152,
"step": 75
},
{
"epoch": 2.3240506329113924,
"grad_norm": 0.481965691984026,
"learning_rate": 1.6760702104750046e-05,
"loss": 0.4082,
"step": 76
},
{
"epoch": 2.3544303797468356,
"grad_norm": 0.36805975207894315,
"learning_rate": 1.664471473438822e-05,
"loss": 0.398,
"step": 77
},
{
"epoch": 2.3848101265822783,
"grad_norm": 0.5228585260757673,
"learning_rate": 1.6527104152454096e-05,
"loss": 0.4009,
"step": 78
},
{
"epoch": 2.4151898734177215,
"grad_norm": 0.4289340149194523,
"learning_rate": 1.6407899089580263e-05,
"loss": 0.4108,
"step": 79
},
{
"epoch": 2.4455696202531647,
"grad_norm": 0.48588710455735235,
"learning_rate": 1.628712866590885e-05,
"loss": 0.4034,
"step": 80
},
{
"epoch": 2.4759493670886075,
"grad_norm": 0.490766132002736,
"learning_rate": 1.6164822383977912e-05,
"loss": 0.393,
"step": 81
},
{
"epoch": 2.5063291139240507,
"grad_norm": 0.3897044417826887,
"learning_rate": 1.604101012151436e-05,
"loss": 0.4089,
"step": 82
},
{
"epoch": 2.536708860759494,
"grad_norm": 0.32974841947264566,
"learning_rate": 1.5915722124135227e-05,
"loss": 0.3894,
"step": 83
},
{
"epoch": 2.5670886075949366,
"grad_norm": 0.5214368187998129,
"learning_rate": 1.5788988997959115e-05,
"loss": 0.4039,
"step": 84
},
{
"epoch": 2.59746835443038,
"grad_norm": 0.4163588840014874,
"learning_rate": 1.5660841702129533e-05,
"loss": 0.3961,
"step": 85
},
{
"epoch": 2.6278481012658226,
"grad_norm": 0.4620015817991619,
"learning_rate": 1.5531311541251995e-05,
"loss": 0.3863,
"step": 86
},
{
"epoch": 2.6582278481012658,
"grad_norm": 0.4288042872696744,
"learning_rate": 1.540043015774676e-05,
"loss": 0.3994,
"step": 87
},
{
"epoch": 2.688607594936709,
"grad_norm": 0.5738624734501034,
"learning_rate": 1.5268229524119007e-05,
"loss": 0.4003,
"step": 88
},
{
"epoch": 2.7189873417721517,
"grad_norm": 0.4136009449359169,
"learning_rate": 1.513474193514842e-05,
"loss": 0.3899,
"step": 89
},
{
"epoch": 2.749367088607595,
"grad_norm": 0.4157372466810592,
"learning_rate": 1.5000000000000002e-05,
"loss": 0.3932,
"step": 90
},
{
"epoch": 2.779746835443038,
"grad_norm": 0.46531189591695404,
"learning_rate": 1.4864036634258112e-05,
"loss": 0.4052,
"step": 91
},
{
"epoch": 2.810126582278481,
"grad_norm": 0.381210286644834,
"learning_rate": 1.4726885051885654e-05,
"loss": 0.4164,
"step": 92
},
{
"epoch": 2.840506329113924,
"grad_norm": 0.42653707231110455,
"learning_rate": 1.4588578757110359e-05,
"loss": 0.4016,
"step": 93
},
{
"epoch": 2.8708860759493673,
"grad_norm": 0.4479903736963084,
"learning_rate": 1.4449151536240167e-05,
"loss": 0.3997,
"step": 94
},
{
"epoch": 2.90126582278481,
"grad_norm": 0.35838575802541517,
"learning_rate": 1.4308637449409705e-05,
"loss": 0.4197,
"step": 95
},
{
"epoch": 2.9316455696202532,
"grad_norm": 0.41499135989555724,
"learning_rate": 1.4167070822259868e-05,
"loss": 0.4058,
"step": 96
},
{
"epoch": 2.962025316455696,
"grad_norm": 0.4129650784609616,
"learning_rate": 1.402448623755254e-05,
"loss": 0.3994,
"step": 97
},
{
"epoch": 2.992405063291139,
"grad_norm": 0.4490667742196104,
"learning_rate": 1.3880918526722497e-05,
"loss": 0.4089,
"step": 98
},
{
"epoch": 3.030379746835443,
"grad_norm": 0.5784142294696635,
"learning_rate": 1.3736402761368597e-05,
"loss": 0.3687,
"step": 99
},
{
"epoch": 3.060759493670886,
"grad_norm": 0.5643195174899467,
"learning_rate": 1.3590974244686248e-05,
"loss": 0.3729,
"step": 100
},
{
"epoch": 3.091139240506329,
"grad_norm": 0.4049869234854002,
"learning_rate": 1.344466850284333e-05,
"loss": 0.3558,
"step": 101
},
{
"epoch": 3.1215189873417724,
"grad_norm": 0.5840920302876608,
"learning_rate": 1.3297521276301666e-05,
"loss": 0.3496,
"step": 102
},
{
"epoch": 3.151898734177215,
"grad_norm": 0.4538191212825923,
"learning_rate": 1.3149568511086104e-05,
"loss": 0.3591,
"step": 103
},
{
"epoch": 3.1822784810126583,
"grad_norm": 0.5036756891102359,
"learning_rate": 1.300084635000341e-05,
"loss": 0.3646,
"step": 104
},
{
"epoch": 3.212658227848101,
"grad_norm": 0.466213881533505,
"learning_rate": 1.2851391123813075e-05,
"loss": 0.3495,
"step": 105
},
{
"epoch": 3.2430379746835443,
"grad_norm": 0.407698713564287,
"learning_rate": 1.2701239342352223e-05,
"loss": 0.3549,
"step": 106
},
{
"epoch": 3.2734177215189875,
"grad_norm": 0.4878510831726186,
"learning_rate": 1.2550427685616767e-05,
"loss": 0.3635,
"step": 107
},
{
"epoch": 3.3037974683544302,
"grad_norm": 0.4165176105620911,
"learning_rate": 1.239899299480098e-05,
"loss": 0.3397,
"step": 108
},
{
"epoch": 3.3341772151898734,
"grad_norm": 0.42165335329838155,
"learning_rate": 1.2246972263297718e-05,
"loss": 0.3396,
"step": 109
},
{
"epoch": 3.3645569620253166,
"grad_norm": 0.47611666495164706,
"learning_rate": 1.2094402627661447e-05,
"loss": 0.3487,
"step": 110
},
{
"epoch": 3.3949367088607594,
"grad_norm": 0.4829213715427168,
"learning_rate": 1.1941321358536278e-05,
"loss": 0.3619,
"step": 111
},
{
"epoch": 3.4253164556962026,
"grad_norm": 0.41437385083123546,
"learning_rate": 1.1787765851551296e-05,
"loss": 0.3476,
"step": 112
},
{
"epoch": 3.4556962025316453,
"grad_norm": 0.41106763258054235,
"learning_rate": 1.1633773618185302e-05,
"loss": 0.3562,
"step": 113
},
{
"epoch": 3.4860759493670885,
"grad_norm": 0.4006822868084472,
"learning_rate": 1.14793822766033e-05,
"loss": 0.3618,
"step": 114
},
{
"epoch": 3.5164556962025317,
"grad_norm": 0.387817224759027,
"learning_rate": 1.132462954246688e-05,
"loss": 0.3626,
"step": 115
},
{
"epoch": 3.546835443037975,
"grad_norm": 0.3582072618650669,
"learning_rate": 1.1169553219720828e-05,
"loss": 0.3615,
"step": 116
},
{
"epoch": 3.5772151898734177,
"grad_norm": 0.37015387053901344,
"learning_rate": 1.1014191191358118e-05,
"loss": 0.3533,
"step": 117
},
{
"epoch": 3.607594936708861,
"grad_norm": 0.4027919873847683,
"learning_rate": 1.085858141016566e-05,
"loss": 0.3636,
"step": 118
},
{
"epoch": 3.6379746835443036,
"grad_norm": 0.33521119639891833,
"learning_rate": 1.070276188945293e-05,
"loss": 0.3514,
"step": 119
},
{
"epoch": 3.668354430379747,
"grad_norm": 0.3539111639484977,
"learning_rate": 1.0546770693765859e-05,
"loss": 0.3366,
"step": 120
},
{
"epoch": 3.69873417721519,
"grad_norm": 0.3488627899974187,
"learning_rate": 1.0390645929588197e-05,
"loss": 0.3494,
"step": 121
},
{
"epoch": 3.729113924050633,
"grad_norm": 0.38250257327787435,
"learning_rate": 1.0234425736032607e-05,
"loss": 0.354,
"step": 122
},
{
"epoch": 3.759493670886076,
"grad_norm": 0.34944025864673983,
"learning_rate": 1.007814827552384e-05,
"loss": 0.3504,
"step": 123
},
{
"epoch": 3.7898734177215188,
"grad_norm": 0.3087070550267954,
"learning_rate": 9.92185172447616e-06,
"loss": 0.356,
"step": 124
},
{
"epoch": 3.820253164556962,
"grad_norm": 0.3422767347446246,
"learning_rate": 9.765574263967397e-06,
"loss": 0.3416,
"step": 125
},
{
"epoch": 3.850632911392405,
"grad_norm": 0.309728511754158,
"learning_rate": 9.609354070411807e-06,
"loss": 0.3424,
"step": 126
},
{
"epoch": 3.8810126582278484,
"grad_norm": 0.2985535404230483,
"learning_rate": 9.453229306234143e-06,
"loss": 0.3602,
"step": 127
},
{
"epoch": 3.911392405063291,
"grad_norm": 0.3438786292703158,
"learning_rate": 9.297238110547075e-06,
"loss": 0.3585,
"step": 128
},
{
"epoch": 3.9417721518987343,
"grad_norm": 0.28943888372307297,
"learning_rate": 9.14141858983434e-06,
"loss": 0.3554,
"step": 129
},
{
"epoch": 3.972151898734177,
"grad_norm": 0.28139763564755316,
"learning_rate": 8.985808808641883e-06,
"loss": 0.3497,
"step": 130
},
{
"epoch": 4.010126582278481,
"grad_norm": 0.3130377531298614,
"learning_rate": 8.830446780279175e-06,
"loss": 0.3368,
"step": 131
},
{
"epoch": 4.040506329113924,
"grad_norm": 0.37084300837367556,
"learning_rate": 8.675370457533122e-06,
"loss": 0.3178,
"step": 132
},
{
"epoch": 4.0708860759493675,
"grad_norm": 0.340370683571999,
"learning_rate": 8.520617723396702e-06,
"loss": 0.3166,
"step": 133
},
{
"epoch": 4.10126582278481,
"grad_norm": 0.39053519861039254,
"learning_rate": 8.366226381814698e-06,
"loss": 0.3157,
"step": 134
},
{
"epoch": 4.131645569620253,
"grad_norm": 0.3424208898897663,
"learning_rate": 8.212234148448708e-06,
"loss": 0.3051,
"step": 135
},
{
"epoch": 4.162025316455696,
"grad_norm": 0.3472075380900132,
"learning_rate": 8.058678641463724e-06,
"loss": 0.3134,
"step": 136
},
{
"epoch": 4.192405063291139,
"grad_norm": 0.38845974316760523,
"learning_rate": 7.905597372338558e-06,
"loss": 0.3179,
"step": 137
},
{
"epoch": 4.222784810126583,
"grad_norm": 0.32736323970265624,
"learning_rate": 7.753027736702283e-06,
"loss": 0.3074,
"step": 138
},
{
"epoch": 4.253164556962025,
"grad_norm": 0.43260729613331333,
"learning_rate": 7.601007005199022e-06,
"loss": 0.3273,
"step": 139
},
{
"epoch": 4.283544303797468,
"grad_norm": 0.30441823992136646,
"learning_rate": 7.449572314383237e-06,
"loss": 0.3194,
"step": 140
},
{
"epoch": 4.313924050632911,
"grad_norm": 0.35945804511954704,
"learning_rate": 7.298760657647779e-06,
"loss": 0.3199,
"step": 141
},
{
"epoch": 4.3443037974683545,
"grad_norm": 0.3505556029957157,
"learning_rate": 7.148608876186931e-06,
"loss": 0.3065,
"step": 142
},
{
"epoch": 4.374683544303798,
"grad_norm": 0.3076722128954047,
"learning_rate": 6.999153649996595e-06,
"loss": 0.3085,
"step": 143
},
{
"epoch": 4.405063291139241,
"grad_norm": 0.350601886678572,
"learning_rate": 6.8504314889138956e-06,
"loss": 0.3209,
"step": 144
},
{
"epoch": 4.435443037974683,
"grad_norm": 0.2903013438323301,
"learning_rate": 6.702478723698336e-06,
"loss": 0.3043,
"step": 145
},
{
"epoch": 4.465822784810126,
"grad_norm": 0.30676907674335474,
"learning_rate": 6.555331497156671e-06,
"loss": 0.3164,
"step": 146
},
{
"epoch": 4.49620253164557,
"grad_norm": 0.28036107811816996,
"learning_rate": 6.4090257553137566e-06,
"loss": 0.3199,
"step": 147
},
{
"epoch": 4.526582278481013,
"grad_norm": 0.2707303342834444,
"learning_rate": 6.263597238631405e-06,
"loss": 0.3267,
"step": 148
},
{
"epoch": 4.556962025316456,
"grad_norm": 0.3086836551240169,
"learning_rate": 6.119081473277502e-06,
"loss": 0.3021,
"step": 149
},
{
"epoch": 4.587341772151898,
"grad_norm": 0.27898623433963937,
"learning_rate": 5.975513762447465e-06,
"loss": 0.318,
"step": 150
},
{
"epoch": 4.6177215189873415,
"grad_norm": 0.30428541429225603,
"learning_rate": 5.832929177740134e-06,
"loss": 0.3095,
"step": 151
},
{
"epoch": 4.648101265822785,
"grad_norm": 0.31726158287862327,
"learning_rate": 5.6913625505902966e-06,
"loss": 0.311,
"step": 152
},
{
"epoch": 4.678481012658228,
"grad_norm": 0.2756912154726042,
"learning_rate": 5.550848463759835e-06,
"loss": 0.3068,
"step": 153
},
{
"epoch": 4.708860759493671,
"grad_norm": 0.2770882754501325,
"learning_rate": 5.411421242889643e-06,
"loss": 0.302,
"step": 154
},
{
"epoch": 4.739240506329114,
"grad_norm": 0.2747423932904734,
"learning_rate": 5.273114948114346e-06,
"loss": 0.3027,
"step": 155
},
{
"epoch": 4.769620253164557,
"grad_norm": 0.2775633679148173,
"learning_rate": 5.135963365741892e-06,
"loss": 0.3114,
"step": 156
},
{
"epoch": 4.8,
"grad_norm": 0.28562611590095843,
"learning_rate": 5.000000000000003e-06,
"loss": 0.3114,
"step": 157
},
{
"epoch": 4.830379746835443,
"grad_norm": 0.26087842913651943,
"learning_rate": 4.865258064851579e-06,
"loss": 0.309,
"step": 158
},
{
"epoch": 4.860759493670886,
"grad_norm": 0.2757384038987269,
"learning_rate": 4.731770475880995e-06,
"loss": 0.3146,
"step": 159
},
{
"epoch": 4.891139240506329,
"grad_norm": 0.2567292839533133,
"learning_rate": 4.599569842253244e-06,
"loss": 0.3094,
"step": 160
},
{
"epoch": 4.921518987341772,
"grad_norm": 0.2540951612506195,
"learning_rate": 4.468688458748006e-06,
"loss": 0.2999,
"step": 161
},
{
"epoch": 4.951898734177215,
"grad_norm": 0.25060879693518734,
"learning_rate": 4.339158297870469e-06,
"loss": 0.3264,
"step": 162
},
{
"epoch": 4.982278481012658,
"grad_norm": 0.26263232170275996,
"learning_rate": 4.211011002040885e-06,
"loss": 0.3122,
"step": 163
},
{
"epoch": 5.020253164556962,
"grad_norm": 0.3094174951283545,
"learning_rate": 4.084277875864776e-06,
"loss": 0.291,
"step": 164
},
{
"epoch": 5.050632911392405,
"grad_norm": 0.3232328732633652,
"learning_rate": 3.958989878485644e-06,
"loss": 0.2864,
"step": 165
},
{
"epoch": 5.0810126582278485,
"grad_norm": 0.2646970593866554,
"learning_rate": 3.83517761602209e-06,
"loss": 0.2827,
"step": 166
},
{
"epoch": 5.111392405063291,
"grad_norm": 0.3489102151281411,
"learning_rate": 3.712871334091154e-06,
"loss": 0.2725,
"step": 167
},
{
"epoch": 5.141772151898734,
"grad_norm": 0.4118511645348355,
"learning_rate": 3.592100910419738e-06,
"loss": 0.2927,
"step": 168
},
{
"epoch": 5.172151898734177,
"grad_norm": 0.2628332059581019,
"learning_rate": 3.4728958475459052e-06,
"loss": 0.2989,
"step": 169
},
{
"epoch": 5.2025316455696204,
"grad_norm": 0.31335998910614754,
"learning_rate": 3.355285265611784e-06,
"loss": 0.2802,
"step": 170
},
{
"epoch": 5.232911392405064,
"grad_norm": 0.35506477091834154,
"learning_rate": 3.2392978952499553e-06,
"loss": 0.2839,
"step": 171
},
{
"epoch": 5.263291139240506,
"grad_norm": 0.2875298433006408,
"learning_rate": 3.1249620705649417e-06,
"loss": 0.2878,
"step": 172
},
{
"epoch": 5.293670886075949,
"grad_norm": 0.2512296346135503,
"learning_rate": 3.0123057222115835e-06,
"loss": 0.2927,
"step": 173
},
{
"epoch": 5.324050632911392,
"grad_norm": 0.28559098314603587,
"learning_rate": 2.9013563705719673e-06,
"loss": 0.281,
"step": 174
},
{
"epoch": 5.3544303797468356,
"grad_norm": 0.2987464671959271,
"learning_rate": 2.7921411190325753e-06,
"loss": 0.2822,
"step": 175
},
{
"epoch": 5.384810126582279,
"grad_norm": 0.2630696633805724,
"learning_rate": 2.6846866473633126e-06,
"loss": 0.2708,
"step": 176
},
{
"epoch": 5.415189873417722,
"grad_norm": 0.25148475298631007,
"learning_rate": 2.579019205199992e-06,
"loss": 0.2884,
"step": 177
},
{
"epoch": 5.445569620253164,
"grad_norm": 0.24791427343662187,
"learning_rate": 2.4751646056319334e-06,
"loss": 0.2759,
"step": 178
},
{
"epoch": 5.4759493670886075,
"grad_norm": 0.2644303727414798,
"learning_rate": 2.373148218896182e-06,
"loss": 0.2855,
"step": 179
},
{
"epoch": 5.506329113924051,
"grad_norm": 0.24460715077957715,
"learning_rate": 2.2729949661798876e-06,
"loss": 0.2779,
"step": 180
},
{
"epoch": 5.536708860759494,
"grad_norm": 0.250298616944708,
"learning_rate": 2.174729313532433e-06,
"loss": 0.2778,
"step": 181
},
{
"epoch": 5.567088607594937,
"grad_norm": 0.23850685413325676,
"learning_rate": 2.078375265888707e-06,
"loss": 0.2973,
"step": 182
},
{
"epoch": 5.597468354430379,
"grad_norm": 0.23734233974205043,
"learning_rate": 1.9839563612050273e-06,
"loss": 0.2793,
"step": 183
},
{
"epoch": 5.627848101265823,
"grad_norm": 0.24453519469748344,
"learning_rate": 1.8914956647091497e-06,
"loss": 0.2822,
"step": 184
},
{
"epoch": 5.658227848101266,
"grad_norm": 0.2475431878691523,
"learning_rate": 1.8010157632657544e-06,
"loss": 0.2891,
"step": 185
},
{
"epoch": 5.688607594936709,
"grad_norm": 0.22949323879670416,
"learning_rate": 1.7125387598587862e-06,
"loss": 0.2735,
"step": 186
},
{
"epoch": 5.718987341772152,
"grad_norm": 0.22360704794779043,
"learning_rate": 1.6260862681919965e-06,
"loss": 0.2847,
"step": 187
},
{
"epoch": 5.749367088607595,
"grad_norm": 0.23219804978872555,
"learning_rate": 1.5416794074090258e-06,
"loss": 0.2918,
"step": 188
},
{
"epoch": 5.779746835443038,
"grad_norm": 0.22475454600346906,
"learning_rate": 1.459338796934293e-06,
"loss": 0.2896,
"step": 189
},
{
"epoch": 5.810126582278481,
"grad_norm": 0.2291577687688738,
"learning_rate": 1.3790845514359363e-06,
"loss": 0.2874,
"step": 190
},
{
"epoch": 5.840506329113924,
"grad_norm": 0.23958072835667898,
"learning_rate": 1.300936275912098e-06,
"loss": 0.277,
"step": 191
},
{
"epoch": 5.870886075949367,
"grad_norm": 0.22506176327780406,
"learning_rate": 1.224913060901688e-06,
"loss": 0.2854,
"step": 192
},
{
"epoch": 5.9012658227848105,
"grad_norm": 0.22062132927128247,
"learning_rate": 1.1510334778208332e-06,
"loss": 0.2803,
"step": 193
},
{
"epoch": 5.931645569620253,
"grad_norm": 0.2334249181736107,
"learning_rate": 1.0793155744261352e-06,
"loss": 0.2796,
"step": 194
},
{
"epoch": 5.962025316455696,
"grad_norm": 0.2177525118823978,
"learning_rate": 1.0097768704058542e-06,
"loss": 0.2872,
"step": 195
},
{
"epoch": 5.992405063291139,
"grad_norm": 0.216624291657226,
"learning_rate": 9.424343531000968e-07,
"loss": 0.2867,
"step": 196
},
{
"epoch": 6.030379746835443,
"grad_norm": 0.24241087019923124,
"learning_rate": 8.773044733510338e-07,
"loss": 0.2692,
"step": 197
},
{
"epoch": 6.060759493670886,
"grad_norm": 0.27338295496213594,
"learning_rate": 8.144031414842012e-07,
"loss": 0.2589,
"step": 198
},
{
"epoch": 6.091139240506329,
"grad_norm": 0.23317322887339842,
"learning_rate": 7.537457234218271e-07,
"loss": 0.2787,
"step": 199
},
{
"epoch": 6.121518987341772,
"grad_norm": 0.21578582228826892,
"learning_rate": 6.953470369291349e-07,
"loss": 0.2741,
"step": 200
},
{
"epoch": 6.151898734177215,
"grad_norm": 0.21483091024842083,
"learning_rate": 6.392213479945852e-07,
"loss": 0.2825,
"step": 201
},
{
"epoch": 6.182278481012658,
"grad_norm": 0.21870657271010965,
"learning_rate": 5.853823673448877e-07,
"loss": 0.2657,
"step": 202
},
{
"epoch": 6.2126582278481015,
"grad_norm": 0.22035338370492527,
"learning_rate": 5.33843247095659e-07,
"loss": 0.2633,
"step": 203
},
{
"epoch": 6.243037974683545,
"grad_norm": 0.23892692953335715,
"learning_rate": 4.846165775385459e-07,
"loss": 0.2826,
"step": 204
},
{
"epoch": 6.273417721518987,
"grad_norm": 0.22946293708757406,
"learning_rate": 4.3771438406559173e-07,
"loss": 0.2786,
"step": 205
},
{
"epoch": 6.30379746835443,
"grad_norm": 0.2357878713903133,
"learning_rate": 3.931481242315993e-07,
"loss": 0.2749,
"step": 206
},
{
"epoch": 6.334177215189873,
"grad_norm": 0.2196319657652353,
"learning_rate": 3.5092868495520294e-07,
"loss": 0.2747,
"step": 207
},
{
"epoch": 6.364556962025317,
"grad_norm": 0.21577494432043984,
"learning_rate": 3.110663798593616e-07,
"loss": 0.2729,
"step": 208
},
{
"epoch": 6.39493670886076,
"grad_norm": 0.20682933373257414,
"learning_rate": 2.735709467518699e-07,
"loss": 0.2712,
"step": 209
},
{
"epoch": 6.425316455696202,
"grad_norm": 0.22449726686073437,
"learning_rate": 2.384515452465475e-07,
"loss": 0.2785,
"step": 210
},
{
"epoch": 6.455696202531645,
"grad_norm": 0.20659056025279612,
"learning_rate": 2.0571675452567997e-07,
"loss": 0.268,
"step": 211
},
{
"epoch": 6.4860759493670885,
"grad_norm": 0.21092641055632277,
"learning_rate": 1.7537457124423896e-07,
"loss": 0.2702,
"step": 212
},
{
"epoch": 6.516455696202532,
"grad_norm": 0.22183566303186786,
"learning_rate": 1.474324075764111e-07,
"loss": 0.2747,
"step": 213
},
{
"epoch": 6.546835443037975,
"grad_norm": 0.21053018559625034,
"learning_rate": 1.2189708940490653e-07,
"loss": 0.2646,
"step": 214
},
{
"epoch": 6.577215189873417,
"grad_norm": 0.2098546670515347,
"learning_rate": 9.877485465349057e-08,
"loss": 0.2624,
"step": 215
},
{
"epoch": 6.6075949367088604,
"grad_norm": 0.2244125918350075,
"learning_rate": 7.807135176314707e-08,
"loss": 0.2684,
"step": 216
},
{
"epoch": 6.637974683544304,
"grad_norm": 0.23461598705545597,
"learning_rate": 5.979163831223988e-08,
"loss": 0.2686,
"step": 217
},
{
"epoch": 6.668354430379747,
"grad_norm": 0.20763228111493268,
"learning_rate": 4.394017978101905e-08,
"loss": 0.269,
"step": 218
},
{
"epoch": 6.69873417721519,
"grad_norm": 0.21208305953806367,
"learning_rate": 3.0520848460765525e-08,
"loss": 0.2693,
"step": 219
},
{
"epoch": 6.729113924050633,
"grad_norm": 0.2087727381299796,
"learning_rate": 1.9536922507841227e-08,
"loss": 0.269,
"step": 220
},
{
"epoch": 6.759493670886076,
"grad_norm": 0.20493326251203908,
"learning_rate": 1.099108514288627e-08,
"loss": 0.2632,
"step": 221
},
{
"epoch": 6.789873417721519,
"grad_norm": 0.20729986264516603,
"learning_rate": 4.885423995341088e-09,
"loss": 0.2705,
"step": 222
},
{
"epoch": 6.820253164556962,
"grad_norm": 0.20813063743714902,
"learning_rate": 1.2214305934699078e-09,
"loss": 0.2785,
"step": 223
},
{
"epoch": 6.850632911392405,
"grad_norm": 0.21114524583674207,
"learning_rate": 0.0,
"loss": 0.2724,
"step": 224
},
{
"epoch": 6.850632911392405,
"step": 224,
"total_flos": 9.289277301231452e+17,
"train_loss": 0.3896952799654433,
"train_runtime": 26951.6631,
"train_samples_per_second": 0.821,
"train_steps_per_second": 0.008
}
],
"logging_steps": 1,
"max_steps": 224,
"num_input_tokens_seen": 0,
"num_train_epochs": 7,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 9.289277301231452e+17,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}