13683 lines
356 KiB
JSON
13683 lines
356 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 1.0,
|
|
"eval_steps": 500,
|
|
"global_step": 1706,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.0005861664712778429,
|
|
"grad_norm": 9.75,
|
|
"learning_rate": 1.1627906976744187e-07,
|
|
"loss": 1.922473430633545,
|
|
"step": 1,
|
|
"token_acc": 0.5627705627705628
|
|
},
|
|
{
|
|
"epoch": 0.0011723329425556857,
|
|
"grad_norm": 9.0,
|
|
"learning_rate": 2.3255813953488374e-07,
|
|
"loss": 1.8388819694519043,
|
|
"step": 2,
|
|
"token_acc": 0.5748502994011976
|
|
},
|
|
{
|
|
"epoch": 0.0017584994138335288,
|
|
"grad_norm": 9.4375,
|
|
"learning_rate": 3.488372093023256e-07,
|
|
"loss": 1.777435302734375,
|
|
"step": 3,
|
|
"token_acc": 0.5815423514538559
|
|
},
|
|
{
|
|
"epoch": 0.0023446658851113715,
|
|
"grad_norm": 9.8125,
|
|
"learning_rate": 4.651162790697675e-07,
|
|
"loss": 1.8569388389587402,
|
|
"step": 4,
|
|
"token_acc": 0.5766656968286297
|
|
},
|
|
{
|
|
"epoch": 0.0029308323563892145,
|
|
"grad_norm": 9.9375,
|
|
"learning_rate": 5.813953488372094e-07,
|
|
"loss": 1.8740830421447754,
|
|
"step": 5,
|
|
"token_acc": 0.5732115677321157
|
|
},
|
|
{
|
|
"epoch": 0.0035169988276670576,
|
|
"grad_norm": 9.9375,
|
|
"learning_rate": 6.976744186046513e-07,
|
|
"loss": 1.8404747247695923,
|
|
"step": 6,
|
|
"token_acc": 0.583679525222552
|
|
},
|
|
{
|
|
"epoch": 0.004103165298944901,
|
|
"grad_norm": 9.6875,
|
|
"learning_rate": 8.139534883720931e-07,
|
|
"loss": 1.866418480873108,
|
|
"step": 7,
|
|
"token_acc": 0.5745682888540031
|
|
},
|
|
{
|
|
"epoch": 0.004689331770222743,
|
|
"grad_norm": 10.125,
|
|
"learning_rate": 9.30232558139535e-07,
|
|
"loss": 1.8587970733642578,
|
|
"step": 8,
|
|
"token_acc": 0.5828959555418339
|
|
},
|
|
{
|
|
"epoch": 0.005275498241500586,
|
|
"grad_norm": 10.0625,
|
|
"learning_rate": 1.0465116279069768e-06,
|
|
"loss": 1.866164207458496,
|
|
"step": 9,
|
|
"token_acc": 0.5775182481751825
|
|
},
|
|
{
|
|
"epoch": 0.005861664712778429,
|
|
"grad_norm": 9.8125,
|
|
"learning_rate": 1.1627906976744188e-06,
|
|
"loss": 1.9087703227996826,
|
|
"step": 10,
|
|
"token_acc": 0.5718896713615024
|
|
},
|
|
{
|
|
"epoch": 0.006447831184056272,
|
|
"grad_norm": 10.0625,
|
|
"learning_rate": 1.2790697674418605e-06,
|
|
"loss": 1.7912871837615967,
|
|
"step": 11,
|
|
"token_acc": 0.5721379310344827
|
|
},
|
|
{
|
|
"epoch": 0.007033997655334115,
|
|
"grad_norm": 11.0,
|
|
"learning_rate": 1.3953488372093025e-06,
|
|
"loss": 1.8134530782699585,
|
|
"step": 12,
|
|
"token_acc": 0.5797705943691345
|
|
},
|
|
{
|
|
"epoch": 0.0076201641266119575,
|
|
"grad_norm": 9.9375,
|
|
"learning_rate": 1.5116279069767443e-06,
|
|
"loss": 1.8804280757904053,
|
|
"step": 13,
|
|
"token_acc": 0.5820497630331753
|
|
},
|
|
{
|
|
"epoch": 0.008206330597889801,
|
|
"grad_norm": 9.1875,
|
|
"learning_rate": 1.6279069767441862e-06,
|
|
"loss": 1.882932186126709,
|
|
"step": 14,
|
|
"token_acc": 0.5743576433947574
|
|
},
|
|
{
|
|
"epoch": 0.008792497069167644,
|
|
"grad_norm": 10.3125,
|
|
"learning_rate": 1.7441860465116282e-06,
|
|
"loss": 1.7767353057861328,
|
|
"step": 15,
|
|
"token_acc": 0.5865902964959568
|
|
},
|
|
{
|
|
"epoch": 0.009378663540445486,
|
|
"grad_norm": 10.1875,
|
|
"learning_rate": 1.86046511627907e-06,
|
|
"loss": 1.910154938697815,
|
|
"step": 16,
|
|
"token_acc": 0.5636363636363636
|
|
},
|
|
{
|
|
"epoch": 0.00996483001172333,
|
|
"grad_norm": 9.0,
|
|
"learning_rate": 1.976744186046512e-06,
|
|
"loss": 1.8497133255004883,
|
|
"step": 17,
|
|
"token_acc": 0.5751738897806313
|
|
},
|
|
{
|
|
"epoch": 0.010550996483001172,
|
|
"grad_norm": 10.875,
|
|
"learning_rate": 2.0930232558139536e-06,
|
|
"loss": 1.8481004238128662,
|
|
"step": 18,
|
|
"token_acc": 0.5811676082862524
|
|
},
|
|
{
|
|
"epoch": 0.011137162954279016,
|
|
"grad_norm": 8.0625,
|
|
"learning_rate": 2.2093023255813954e-06,
|
|
"loss": 1.8258540630340576,
|
|
"step": 19,
|
|
"token_acc": 0.5775401069518716
|
|
},
|
|
{
|
|
"epoch": 0.011723329425556858,
|
|
"grad_norm": 8.1875,
|
|
"learning_rate": 2.3255813953488376e-06,
|
|
"loss": 1.886909008026123,
|
|
"step": 20,
|
|
"token_acc": 0.5719981238273921
|
|
},
|
|
{
|
|
"epoch": 0.0123094958968347,
|
|
"grad_norm": 8.0,
|
|
"learning_rate": 2.4418604651162793e-06,
|
|
"loss": 1.7775611877441406,
|
|
"step": 21,
|
|
"token_acc": 0.594777805674115
|
|
},
|
|
{
|
|
"epoch": 0.012895662368112544,
|
|
"grad_norm": 8.6875,
|
|
"learning_rate": 2.558139534883721e-06,
|
|
"loss": 1.81183922290802,
|
|
"step": 22,
|
|
"token_acc": 0.5750265111346765
|
|
},
|
|
{
|
|
"epoch": 0.013481828839390387,
|
|
"grad_norm": 7.71875,
|
|
"learning_rate": 2.674418604651163e-06,
|
|
"loss": 1.815468430519104,
|
|
"step": 23,
|
|
"token_acc": 0.5876315789473684
|
|
},
|
|
{
|
|
"epoch": 0.01406799531066823,
|
|
"grad_norm": 8.125,
|
|
"learning_rate": 2.790697674418605e-06,
|
|
"loss": 1.828597068786621,
|
|
"step": 24,
|
|
"token_acc": 0.5727166957533449
|
|
},
|
|
{
|
|
"epoch": 0.014654161781946073,
|
|
"grad_norm": 7.875,
|
|
"learning_rate": 2.9069767441860468e-06,
|
|
"loss": 1.784855604171753,
|
|
"step": 25,
|
|
"token_acc": 0.5812169312169312
|
|
},
|
|
{
|
|
"epoch": 0.015240328253223915,
|
|
"grad_norm": 6.5625,
|
|
"learning_rate": 3.0232558139534885e-06,
|
|
"loss": 1.7573776245117188,
|
|
"step": 26,
|
|
"token_acc": 0.5937282229965157
|
|
},
|
|
{
|
|
"epoch": 0.015826494724501757,
|
|
"grad_norm": 7.75,
|
|
"learning_rate": 3.1395348837209307e-06,
|
|
"loss": 1.7574375867843628,
|
|
"step": 27,
|
|
"token_acc": 0.5871247291860106
|
|
},
|
|
{
|
|
"epoch": 0.016412661195779603,
|
|
"grad_norm": 7.0625,
|
|
"learning_rate": 3.2558139534883724e-06,
|
|
"loss": 1.7369260787963867,
|
|
"step": 28,
|
|
"token_acc": 0.5990656773838967
|
|
},
|
|
{
|
|
"epoch": 0.016998827667057445,
|
|
"grad_norm": 6.96875,
|
|
"learning_rate": 3.372093023255814e-06,
|
|
"loss": 1.716120719909668,
|
|
"step": 29,
|
|
"token_acc": 0.5879593432369038
|
|
},
|
|
{
|
|
"epoch": 0.017584994138335287,
|
|
"grad_norm": 6.1875,
|
|
"learning_rate": 3.4883720930232564e-06,
|
|
"loss": 1.7300697565078735,
|
|
"step": 30,
|
|
"token_acc": 0.5906198444496819
|
|
},
|
|
{
|
|
"epoch": 0.01817116060961313,
|
|
"grad_norm": 6.65625,
|
|
"learning_rate": 3.6046511627906977e-06,
|
|
"loss": 1.7272441387176514,
|
|
"step": 31,
|
|
"token_acc": 0.5928934010152285
|
|
},
|
|
{
|
|
"epoch": 0.01875732708089097,
|
|
"grad_norm": 6.78125,
|
|
"learning_rate": 3.72093023255814e-06,
|
|
"loss": 1.7296082973480225,
|
|
"step": 32,
|
|
"token_acc": 0.5938412529864614
|
|
},
|
|
{
|
|
"epoch": 0.019343493552168817,
|
|
"grad_norm": 6.78125,
|
|
"learning_rate": 3.837209302325582e-06,
|
|
"loss": 1.6838080883026123,
|
|
"step": 33,
|
|
"token_acc": 0.608338804990151
|
|
},
|
|
{
|
|
"epoch": 0.01992966002344666,
|
|
"grad_norm": 6.375,
|
|
"learning_rate": 3.953488372093024e-06,
|
|
"loss": 1.7167341709136963,
|
|
"step": 34,
|
|
"token_acc": 0.5988238302224494
|
|
},
|
|
{
|
|
"epoch": 0.020515826494724502,
|
|
"grad_norm": 6.25,
|
|
"learning_rate": 4.0697674418604655e-06,
|
|
"loss": 1.744483232498169,
|
|
"step": 35,
|
|
"token_acc": 0.584875301689461
|
|
},
|
|
{
|
|
"epoch": 0.021101992966002344,
|
|
"grad_norm": 6.375,
|
|
"learning_rate": 4.186046511627907e-06,
|
|
"loss": 1.5869176387786865,
|
|
"step": 36,
|
|
"token_acc": 0.6092391304347826
|
|
},
|
|
{
|
|
"epoch": 0.021688159437280186,
|
|
"grad_norm": 6.5,
|
|
"learning_rate": 4.302325581395349e-06,
|
|
"loss": 1.6861618757247925,
|
|
"step": 37,
|
|
"token_acc": 0.5972222222222222
|
|
},
|
|
{
|
|
"epoch": 0.022274325908558032,
|
|
"grad_norm": 6.1875,
|
|
"learning_rate": 4.418604651162791e-06,
|
|
"loss": 1.6232504844665527,
|
|
"step": 38,
|
|
"token_acc": 0.6122846025569761
|
|
},
|
|
{
|
|
"epoch": 0.022860492379835874,
|
|
"grad_norm": 5.5625,
|
|
"learning_rate": 4.5348837209302326e-06,
|
|
"loss": 1.6254441738128662,
|
|
"step": 39,
|
|
"token_acc": 0.6134431455897981
|
|
},
|
|
{
|
|
"epoch": 0.023446658851113716,
|
|
"grad_norm": 5.375,
|
|
"learning_rate": 4.651162790697675e-06,
|
|
"loss": 1.6419270038604736,
|
|
"step": 40,
|
|
"token_acc": 0.6110240334378265
|
|
},
|
|
{
|
|
"epoch": 0.02403282532239156,
|
|
"grad_norm": 5.375,
|
|
"learning_rate": 4.767441860465117e-06,
|
|
"loss": 1.6731677055358887,
|
|
"step": 41,
|
|
"token_acc": 0.5994838709677419
|
|
},
|
|
{
|
|
"epoch": 0.0246189917936694,
|
|
"grad_norm": 5.1875,
|
|
"learning_rate": 4.883720930232559e-06,
|
|
"loss": 1.5852640867233276,
|
|
"step": 42,
|
|
"token_acc": 0.6199686356508103
|
|
},
|
|
{
|
|
"epoch": 0.025205158264947247,
|
|
"grad_norm": 5.09375,
|
|
"learning_rate": 5e-06,
|
|
"loss": 1.6455340385437012,
|
|
"step": 43,
|
|
"token_acc": 0.6105619881389438
|
|
},
|
|
{
|
|
"epoch": 0.02579132473622509,
|
|
"grad_norm": 5.21875,
|
|
"learning_rate": 5.116279069767442e-06,
|
|
"loss": 1.639405369758606,
|
|
"step": 44,
|
|
"token_acc": 0.6054664289958408
|
|
},
|
|
{
|
|
"epoch": 0.02637749120750293,
|
|
"grad_norm": 5.03125,
|
|
"learning_rate": 5.232558139534885e-06,
|
|
"loss": 1.5981941223144531,
|
|
"step": 45,
|
|
"token_acc": 0.6188944135712197
|
|
},
|
|
{
|
|
"epoch": 0.026963657678780773,
|
|
"grad_norm": 4.90625,
|
|
"learning_rate": 5.348837209302326e-06,
|
|
"loss": 1.6428463459014893,
|
|
"step": 46,
|
|
"token_acc": 0.6052557219553546
|
|
},
|
|
{
|
|
"epoch": 0.027549824150058615,
|
|
"grad_norm": 4.90625,
|
|
"learning_rate": 5.465116279069767e-06,
|
|
"loss": 1.5801525115966797,
|
|
"step": 47,
|
|
"token_acc": 0.617827595409661
|
|
},
|
|
{
|
|
"epoch": 0.02813599062133646,
|
|
"grad_norm": 5.0625,
|
|
"learning_rate": 5.58139534883721e-06,
|
|
"loss": 1.5199975967407227,
|
|
"step": 48,
|
|
"token_acc": 0.6232501521606817
|
|
},
|
|
{
|
|
"epoch": 0.028722157092614303,
|
|
"grad_norm": 4.78125,
|
|
"learning_rate": 5.697674418604652e-06,
|
|
"loss": 1.6161065101623535,
|
|
"step": 49,
|
|
"token_acc": 0.6044823232323232
|
|
},
|
|
{
|
|
"epoch": 0.029308323563892145,
|
|
"grad_norm": 4.46875,
|
|
"learning_rate": 5.8139534883720935e-06,
|
|
"loss": 1.558159351348877,
|
|
"step": 50,
|
|
"token_acc": 0.615921426725252
|
|
},
|
|
{
|
|
"epoch": 0.029894490035169988,
|
|
"grad_norm": 4.84375,
|
|
"learning_rate": 5.930232558139536e-06,
|
|
"loss": 1.6065258979797363,
|
|
"step": 51,
|
|
"token_acc": 0.6131708581799326
|
|
},
|
|
{
|
|
"epoch": 0.03048065650644783,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 6.046511627906977e-06,
|
|
"loss": 1.4921927452087402,
|
|
"step": 52,
|
|
"token_acc": 0.6295802798134577
|
|
},
|
|
{
|
|
"epoch": 0.031066822977725676,
|
|
"grad_norm": 4.40625,
|
|
"learning_rate": 6.162790697674419e-06,
|
|
"loss": 1.5109354257583618,
|
|
"step": 53,
|
|
"token_acc": 0.6332421000248818
|
|
},
|
|
{
|
|
"epoch": 0.031652989449003514,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 6.279069767441861e-06,
|
|
"loss": 1.6076841354370117,
|
|
"step": 54,
|
|
"token_acc": 0.6133828996282528
|
|
},
|
|
{
|
|
"epoch": 0.03223915592028136,
|
|
"grad_norm": 4.5625,
|
|
"learning_rate": 6.395348837209303e-06,
|
|
"loss": 1.5447957515716553,
|
|
"step": 55,
|
|
"token_acc": 0.6245221993531315
|
|
},
|
|
{
|
|
"epoch": 0.032825322391559206,
|
|
"grad_norm": 4.53125,
|
|
"learning_rate": 6.511627906976745e-06,
|
|
"loss": 1.6029130220413208,
|
|
"step": 56,
|
|
"token_acc": 0.6161026837806302
|
|
},
|
|
{
|
|
"epoch": 0.03341148886283705,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 6.627906976744186e-06,
|
|
"loss": 1.4752726554870605,
|
|
"step": 57,
|
|
"token_acc": 0.6299428002984332
|
|
},
|
|
{
|
|
"epoch": 0.03399765533411489,
|
|
"grad_norm": 4.375,
|
|
"learning_rate": 6.744186046511628e-06,
|
|
"loss": 1.4409394264221191,
|
|
"step": 58,
|
|
"token_acc": 0.6480574600065295
|
|
},
|
|
{
|
|
"epoch": 0.03458382180539273,
|
|
"grad_norm": 4.84375,
|
|
"learning_rate": 6.86046511627907e-06,
|
|
"loss": 1.4520986080169678,
|
|
"step": 59,
|
|
"token_acc": 0.644297635605007
|
|
},
|
|
{
|
|
"epoch": 0.035169988276670575,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 6.976744186046513e-06,
|
|
"loss": 1.5404014587402344,
|
|
"step": 60,
|
|
"token_acc": 0.6251706873008648
|
|
},
|
|
{
|
|
"epoch": 0.03575615474794842,
|
|
"grad_norm": 4.5625,
|
|
"learning_rate": 7.0930232558139545e-06,
|
|
"loss": 1.4637079238891602,
|
|
"step": 61,
|
|
"token_acc": 0.6348944591029023
|
|
},
|
|
{
|
|
"epoch": 0.03634232121922626,
|
|
"grad_norm": 4.90625,
|
|
"learning_rate": 7.209302325581395e-06,
|
|
"loss": 1.493554949760437,
|
|
"step": 62,
|
|
"token_acc": 0.6421096693177062
|
|
},
|
|
{
|
|
"epoch": 0.0369284876905041,
|
|
"grad_norm": 4.28125,
|
|
"learning_rate": 7.325581395348837e-06,
|
|
"loss": 1.517531394958496,
|
|
"step": 63,
|
|
"token_acc": 0.6320832192718314
|
|
},
|
|
{
|
|
"epoch": 0.03751465416178194,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 7.44186046511628e-06,
|
|
"loss": 1.3376133441925049,
|
|
"step": 64,
|
|
"token_acc": 0.6659367396593674
|
|
},
|
|
{
|
|
"epoch": 0.038100820633059786,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 7.5581395348837215e-06,
|
|
"loss": 1.466962456703186,
|
|
"step": 65,
|
|
"token_acc": 0.632258064516129
|
|
},
|
|
{
|
|
"epoch": 0.038686987104337635,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 7.674418604651164e-06,
|
|
"loss": 1.451467752456665,
|
|
"step": 66,
|
|
"token_acc": 0.6403980634749865
|
|
},
|
|
{
|
|
"epoch": 0.03927315357561548,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 7.790697674418605e-06,
|
|
"loss": 1.5034494400024414,
|
|
"step": 67,
|
|
"token_acc": 0.6256215960217855
|
|
},
|
|
{
|
|
"epoch": 0.03985932004689332,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 7.906976744186048e-06,
|
|
"loss": 1.4610660076141357,
|
|
"step": 68,
|
|
"token_acc": 0.63441905687895
|
|
},
|
|
{
|
|
"epoch": 0.04044548651817116,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 8.023255813953488e-06,
|
|
"loss": 1.4441676139831543,
|
|
"step": 69,
|
|
"token_acc": 0.6433886662850601
|
|
},
|
|
{
|
|
"epoch": 0.041031652989449004,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 8.139534883720931e-06,
|
|
"loss": 1.4255037307739258,
|
|
"step": 70,
|
|
"token_acc": 0.6393268001704303
|
|
},
|
|
{
|
|
"epoch": 0.041617819460726846,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 8.255813953488374e-06,
|
|
"loss": 1.3357102870941162,
|
|
"step": 71,
|
|
"token_acc": 0.6539003097718953
|
|
},
|
|
{
|
|
"epoch": 0.04220398593200469,
|
|
"grad_norm": 3.609375,
|
|
"learning_rate": 8.372093023255815e-06,
|
|
"loss": 1.428622841835022,
|
|
"step": 72,
|
|
"token_acc": 0.6404421326397919
|
|
},
|
|
{
|
|
"epoch": 0.04279015240328253,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 8.488372093023256e-06,
|
|
"loss": 1.4214017391204834,
|
|
"step": 73,
|
|
"token_acc": 0.6410328374964918
|
|
},
|
|
{
|
|
"epoch": 0.04337631887456037,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 8.604651162790698e-06,
|
|
"loss": 1.4486011266708374,
|
|
"step": 74,
|
|
"token_acc": 0.6403337969401948
|
|
},
|
|
{
|
|
"epoch": 0.043962485345838215,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 8.72093023255814e-06,
|
|
"loss": 1.4594101905822754,
|
|
"step": 75,
|
|
"token_acc": 0.6407103825136612
|
|
},
|
|
{
|
|
"epoch": 0.044548651817116064,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 8.837209302325582e-06,
|
|
"loss": 1.3584792613983154,
|
|
"step": 76,
|
|
"token_acc": 0.6537976280595509
|
|
},
|
|
{
|
|
"epoch": 0.045134818288393906,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 8.953488372093024e-06,
|
|
"loss": 1.4149620532989502,
|
|
"step": 77,
|
|
"token_acc": 0.6538111144852717
|
|
},
|
|
{
|
|
"epoch": 0.04572098475967175,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 9.069767441860465e-06,
|
|
"loss": 1.3747549057006836,
|
|
"step": 78,
|
|
"token_acc": 0.6533198175367461
|
|
},
|
|
{
|
|
"epoch": 0.04630715123094959,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 9.186046511627908e-06,
|
|
"loss": 1.4737353324890137,
|
|
"step": 79,
|
|
"token_acc": 0.6408869659275284
|
|
},
|
|
{
|
|
"epoch": 0.04689331770222743,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 9.30232558139535e-06,
|
|
"loss": 1.4953830242156982,
|
|
"step": 80,
|
|
"token_acc": 0.6361954108068097
|
|
},
|
|
{
|
|
"epoch": 0.047479484173505275,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 9.418604651162791e-06,
|
|
"loss": 1.3967689275741577,
|
|
"step": 81,
|
|
"token_acc": 0.6467046894803549
|
|
},
|
|
{
|
|
"epoch": 0.04806565064478312,
|
|
"grad_norm": 4.375,
|
|
"learning_rate": 9.534883720930234e-06,
|
|
"loss": 1.3326421976089478,
|
|
"step": 82,
|
|
"token_acc": 0.6585730295106462
|
|
},
|
|
{
|
|
"epoch": 0.04865181711606096,
|
|
"grad_norm": 3.921875,
|
|
"learning_rate": 9.651162790697676e-06,
|
|
"loss": 1.4291915893554688,
|
|
"step": 83,
|
|
"token_acc": 0.654592127780947
|
|
},
|
|
{
|
|
"epoch": 0.0492379835873388,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 9.767441860465117e-06,
|
|
"loss": 1.3752350807189941,
|
|
"step": 84,
|
|
"token_acc": 0.6489058039961941
|
|
},
|
|
{
|
|
"epoch": 0.049824150058616644,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 9.883720930232558e-06,
|
|
"loss": 1.334608793258667,
|
|
"step": 85,
|
|
"token_acc": 0.6538690476190476
|
|
},
|
|
{
|
|
"epoch": 0.05041031652989449,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 1e-05,
|
|
"loss": 1.343532919883728,
|
|
"step": 86,
|
|
"token_acc": 0.6705359347987158
|
|
},
|
|
{
|
|
"epoch": 0.050996483001172335,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 9.999990598230731e-06,
|
|
"loss": 1.4497328996658325,
|
|
"step": 87,
|
|
"token_acc": 0.635893011216566
|
|
},
|
|
{
|
|
"epoch": 0.05158264947245018,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 9.999962392958281e-06,
|
|
"loss": 1.360518217086792,
|
|
"step": 88,
|
|
"token_acc": 0.6606767794632439
|
|
},
|
|
{
|
|
"epoch": 0.05216881594372802,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 9.999915384288723e-06,
|
|
"loss": 1.3616094589233398,
|
|
"step": 89,
|
|
"token_acc": 0.6512483574244415
|
|
},
|
|
{
|
|
"epoch": 0.05275498241500586,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 9.99984957239884e-06,
|
|
"loss": 1.3962854146957397,
|
|
"step": 90,
|
|
"token_acc": 0.6558481322718922
|
|
},
|
|
{
|
|
"epoch": 0.053341148886283704,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 9.99976495753613e-06,
|
|
"loss": 1.2813897132873535,
|
|
"step": 91,
|
|
"token_acc": 0.6675907956750763
|
|
},
|
|
{
|
|
"epoch": 0.053927315357561546,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 9.999661540018812e-06,
|
|
"loss": 1.3500356674194336,
|
|
"step": 92,
|
|
"token_acc": 0.6567574319260555
|
|
},
|
|
{
|
|
"epoch": 0.05451348182883939,
|
|
"grad_norm": 3.46875,
|
|
"learning_rate": 9.999539320235803e-06,
|
|
"loss": 1.387803554534912,
|
|
"step": 93,
|
|
"token_acc": 0.6572171037089535
|
|
},
|
|
{
|
|
"epoch": 0.05509964830011723,
|
|
"grad_norm": 4.28125,
|
|
"learning_rate": 9.999398298646738e-06,
|
|
"loss": 1.3693256378173828,
|
|
"step": 94,
|
|
"token_acc": 0.650962962962963
|
|
},
|
|
{
|
|
"epoch": 0.05568581477139507,
|
|
"grad_norm": 3.46875,
|
|
"learning_rate": 9.999238475781957e-06,
|
|
"loss": 1.3446455001831055,
|
|
"step": 95,
|
|
"token_acc": 0.6540461826503017
|
|
},
|
|
{
|
|
"epoch": 0.05627198124267292,
|
|
"grad_norm": 4.3125,
|
|
"learning_rate": 9.999059852242508e-06,
|
|
"loss": 1.3423447608947754,
|
|
"step": 96,
|
|
"token_acc": 0.6636466591166478
|
|
},
|
|
{
|
|
"epoch": 0.056858147713950764,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 9.998862428700142e-06,
|
|
"loss": 1.332951307296753,
|
|
"step": 97,
|
|
"token_acc": 0.6726190476190477
|
|
},
|
|
{
|
|
"epoch": 0.05744431418522861,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 9.99864620589731e-06,
|
|
"loss": 1.3405780792236328,
|
|
"step": 98,
|
|
"token_acc": 0.6580246913580247
|
|
},
|
|
{
|
|
"epoch": 0.05803048065650645,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 9.998411184647163e-06,
|
|
"loss": 1.377410650253296,
|
|
"step": 99,
|
|
"token_acc": 0.6473282442748092
|
|
},
|
|
{
|
|
"epoch": 0.05861664712778429,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 9.998157365833548e-06,
|
|
"loss": 1.3857698440551758,
|
|
"step": 100,
|
|
"token_acc": 0.6483600305110603
|
|
},
|
|
{
|
|
"epoch": 0.05920281359906213,
|
|
"grad_norm": 3.546875,
|
|
"learning_rate": 9.997884750411004e-06,
|
|
"loss": 1.371281385421753,
|
|
"step": 101,
|
|
"token_acc": 0.6525857654889913
|
|
},
|
|
{
|
|
"epoch": 0.059788980070339975,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 9.997593339404757e-06,
|
|
"loss": 1.2607136964797974,
|
|
"step": 102,
|
|
"token_acc": 0.6734254992319508
|
|
},
|
|
{
|
|
"epoch": 0.06037514654161782,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 9.997283133910718e-06,
|
|
"loss": 1.4018760919570923,
|
|
"step": 103,
|
|
"token_acc": 0.6583050847457627
|
|
},
|
|
{
|
|
"epoch": 0.06096131301289566,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 9.99695413509548e-06,
|
|
"loss": 1.3288028240203857,
|
|
"step": 104,
|
|
"token_acc": 0.6633016627078385
|
|
},
|
|
{
|
|
"epoch": 0.0615474794841735,
|
|
"grad_norm": 3.46875,
|
|
"learning_rate": 9.996606344196312e-06,
|
|
"loss": 1.3722116947174072,
|
|
"step": 105,
|
|
"token_acc": 0.6468654613759098
|
|
},
|
|
{
|
|
"epoch": 0.06213364595545135,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 9.996239762521152e-06,
|
|
"loss": 1.3463213443756104,
|
|
"step": 106,
|
|
"token_acc": 0.6578507795100222
|
|
},
|
|
{
|
|
"epoch": 0.0627198124267292,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 9.995854391448607e-06,
|
|
"loss": 1.2787327766418457,
|
|
"step": 107,
|
|
"token_acc": 0.6712564543889845
|
|
},
|
|
{
|
|
"epoch": 0.06330597889800703,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 9.995450232427947e-06,
|
|
"loss": 1.3415515422821045,
|
|
"step": 108,
|
|
"token_acc": 0.6605791807983303
|
|
},
|
|
{
|
|
"epoch": 0.06389214536928488,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 9.995027286979095e-06,
|
|
"loss": 1.2784342765808105,
|
|
"step": 109,
|
|
"token_acc": 0.6693759832197168
|
|
},
|
|
{
|
|
"epoch": 0.06447831184056271,
|
|
"grad_norm": 4.40625,
|
|
"learning_rate": 9.994585556692624e-06,
|
|
"loss": 1.2885017395019531,
|
|
"step": 110,
|
|
"token_acc": 0.6666666666666666
|
|
},
|
|
{
|
|
"epoch": 0.06506447831184056,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 9.994125043229753e-06,
|
|
"loss": 1.3190783262252808,
|
|
"step": 111,
|
|
"token_acc": 0.6599110232328226
|
|
},
|
|
{
|
|
"epoch": 0.06565064478311841,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 9.99364574832234e-06,
|
|
"loss": 1.3005887269973755,
|
|
"step": 112,
|
|
"token_acc": 0.6679790026246719
|
|
},
|
|
{
|
|
"epoch": 0.06623681125439625,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 9.993147673772869e-06,
|
|
"loss": 1.3465793132781982,
|
|
"step": 113,
|
|
"token_acc": 0.6574889867841409
|
|
},
|
|
{
|
|
"epoch": 0.0668229777256741,
|
|
"grad_norm": 4.40625,
|
|
"learning_rate": 9.992630821454458e-06,
|
|
"loss": 1.4024863243103027,
|
|
"step": 114,
|
|
"token_acc": 0.6504854368932039
|
|
},
|
|
{
|
|
"epoch": 0.06740914419695193,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 9.992095193310835e-06,
|
|
"loss": 1.3919742107391357,
|
|
"step": 115,
|
|
"token_acc": 0.6501501501501501
|
|
},
|
|
{
|
|
"epoch": 0.06799531066822978,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 9.991540791356342e-06,
|
|
"loss": 1.3471615314483643,
|
|
"step": 116,
|
|
"token_acc": 0.6529761904761905
|
|
},
|
|
{
|
|
"epoch": 0.06858147713950762,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 9.99096761767592e-06,
|
|
"loss": 1.378331184387207,
|
|
"step": 117,
|
|
"token_acc": 0.6557826520438684
|
|
},
|
|
{
|
|
"epoch": 0.06916764361078546,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 9.99037567442511e-06,
|
|
"loss": 1.29890775680542,
|
|
"step": 118,
|
|
"token_acc": 0.6683734939759036
|
|
},
|
|
{
|
|
"epoch": 0.0697538100820633,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 9.989764963830038e-06,
|
|
"loss": 1.4052197933197021,
|
|
"step": 119,
|
|
"token_acc": 0.6460887006908848
|
|
},
|
|
{
|
|
"epoch": 0.07033997655334115,
|
|
"grad_norm": 3.375,
|
|
"learning_rate": 9.989135488187407e-06,
|
|
"loss": 1.292475700378418,
|
|
"step": 120,
|
|
"token_acc": 0.660359944623904
|
|
},
|
|
{
|
|
"epoch": 0.070926143024619,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 9.98848724986449e-06,
|
|
"loss": 1.3177354335784912,
|
|
"step": 121,
|
|
"token_acc": 0.6630209872893881
|
|
},
|
|
{
|
|
"epoch": 0.07151230949589683,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 9.987820251299121e-06,
|
|
"loss": 1.2800346612930298,
|
|
"step": 122,
|
|
"token_acc": 0.6670148863933142
|
|
},
|
|
{
|
|
"epoch": 0.07209847596717468,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 9.98713449499969e-06,
|
|
"loss": 1.3428847789764404,
|
|
"step": 123,
|
|
"token_acc": 0.6530936012691697
|
|
},
|
|
{
|
|
"epoch": 0.07268464243845252,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 9.986429983545127e-06,
|
|
"loss": 1.3118129968643188,
|
|
"step": 124,
|
|
"token_acc": 0.6743782533256217
|
|
},
|
|
{
|
|
"epoch": 0.07327080890973037,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 9.985706719584888e-06,
|
|
"loss": 1.2767138481140137,
|
|
"step": 125,
|
|
"token_acc": 0.6690344646145752
|
|
},
|
|
{
|
|
"epoch": 0.0738569753810082,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 9.98496470583896e-06,
|
|
"loss": 1.2570061683654785,
|
|
"step": 126,
|
|
"token_acc": 0.67611725132148
|
|
},
|
|
{
|
|
"epoch": 0.07444314185228605,
|
|
"grad_norm": 3.40625,
|
|
"learning_rate": 9.984203945097843e-06,
|
|
"loss": 1.3110942840576172,
|
|
"step": 127,
|
|
"token_acc": 0.664469036967271
|
|
},
|
|
{
|
|
"epoch": 0.07502930832356389,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 9.98342444022253e-06,
|
|
"loss": 1.2449908256530762,
|
|
"step": 128,
|
|
"token_acc": 0.670354609929078
|
|
},
|
|
{
|
|
"epoch": 0.07561547479484174,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 9.982626194144516e-06,
|
|
"loss": 1.2349244356155396,
|
|
"step": 129,
|
|
"token_acc": 0.6781449597334074
|
|
},
|
|
{
|
|
"epoch": 0.07620164126611957,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 9.98180920986577e-06,
|
|
"loss": 1.3239625692367554,
|
|
"step": 130,
|
|
"token_acc": 0.6641558441558442
|
|
},
|
|
{
|
|
"epoch": 0.07678780773739742,
|
|
"grad_norm": 3.40625,
|
|
"learning_rate": 9.980973490458728e-06,
|
|
"loss": 1.2866564989089966,
|
|
"step": 131,
|
|
"token_acc": 0.6691194505764042
|
|
},
|
|
{
|
|
"epoch": 0.07737397420867527,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 9.98011903906629e-06,
|
|
"loss": 1.3870444297790527,
|
|
"step": 132,
|
|
"token_acc": 0.6537607108854332
|
|
},
|
|
{
|
|
"epoch": 0.0779601406799531,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 9.979245858901796e-06,
|
|
"loss": 1.3021979331970215,
|
|
"step": 133,
|
|
"token_acc": 0.6663630843958713
|
|
},
|
|
{
|
|
"epoch": 0.07854630715123095,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 9.978353953249023e-06,
|
|
"loss": 1.3446669578552246,
|
|
"step": 134,
|
|
"token_acc": 0.6608391608391608
|
|
},
|
|
{
|
|
"epoch": 0.07913247362250879,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 9.977443325462166e-06,
|
|
"loss": 1.289560317993164,
|
|
"step": 135,
|
|
"token_acc": 0.6667455061494797
|
|
},
|
|
{
|
|
"epoch": 0.07971864009378664,
|
|
"grad_norm": 3.5,
|
|
"learning_rate": 9.976513978965829e-06,
|
|
"loss": 1.2349467277526855,
|
|
"step": 136,
|
|
"token_acc": 0.671203629570323
|
|
},
|
|
{
|
|
"epoch": 0.08030480656506447,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 9.975565917255017e-06,
|
|
"loss": 1.3854610919952393,
|
|
"step": 137,
|
|
"token_acc": 0.6613653046244189
|
|
},
|
|
{
|
|
"epoch": 0.08089097303634232,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 9.974599143895107e-06,
|
|
"loss": 1.3215563297271729,
|
|
"step": 138,
|
|
"token_acc": 0.6625967837998809
|
|
},
|
|
{
|
|
"epoch": 0.08147713950762016,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 9.973613662521856e-06,
|
|
"loss": 1.3215044736862183,
|
|
"step": 139,
|
|
"token_acc": 0.6678387882066541
|
|
},
|
|
{
|
|
"epoch": 0.08206330597889801,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 9.972609476841368e-06,
|
|
"loss": 1.2585272789001465,
|
|
"step": 140,
|
|
"token_acc": 0.6627906976744186
|
|
},
|
|
{
|
|
"epoch": 0.08264947245017586,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 9.971586590630094e-06,
|
|
"loss": 1.3464109897613525,
|
|
"step": 141,
|
|
"token_acc": 0.6579677956690727
|
|
},
|
|
{
|
|
"epoch": 0.08323563892145369,
|
|
"grad_norm": 4.34375,
|
|
"learning_rate": 9.970545007734807e-06,
|
|
"loss": 1.3905692100524902,
|
|
"step": 142,
|
|
"token_acc": 0.6418539325842697
|
|
},
|
|
{
|
|
"epoch": 0.08382180539273154,
|
|
"grad_norm": 4.4375,
|
|
"learning_rate": 9.9694847320726e-06,
|
|
"loss": 1.2954742908477783,
|
|
"step": 143,
|
|
"token_acc": 0.6672443674176777
|
|
},
|
|
{
|
|
"epoch": 0.08440797186400938,
|
|
"grad_norm": 4.375,
|
|
"learning_rate": 9.968405767630857e-06,
|
|
"loss": 1.3477437496185303,
|
|
"step": 144,
|
|
"token_acc": 0.6564978645515558
|
|
},
|
|
{
|
|
"epoch": 0.08499413833528723,
|
|
"grad_norm": 4.3125,
|
|
"learning_rate": 9.967308118467252e-06,
|
|
"loss": 1.3611384630203247,
|
|
"step": 145,
|
|
"token_acc": 0.6640178003814368
|
|
},
|
|
{
|
|
"epoch": 0.08558030480656506,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 9.966191788709716e-06,
|
|
"loss": 1.2704602479934692,
|
|
"step": 146,
|
|
"token_acc": 0.677688827010094
|
|
},
|
|
{
|
|
"epoch": 0.08616647127784291,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 9.965056782556444e-06,
|
|
"loss": 1.299825668334961,
|
|
"step": 147,
|
|
"token_acc": 0.675739247311828
|
|
},
|
|
{
|
|
"epoch": 0.08675263774912075,
|
|
"grad_norm": 3.609375,
|
|
"learning_rate": 9.963903104275859e-06,
|
|
"loss": 1.3310610055923462,
|
|
"step": 148,
|
|
"token_acc": 0.6599071585634009
|
|
},
|
|
{
|
|
"epoch": 0.0873388042203986,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 9.962730758206612e-06,
|
|
"loss": 1.404206395149231,
|
|
"step": 149,
|
|
"token_acc": 0.6534059945504087
|
|
},
|
|
{
|
|
"epoch": 0.08792497069167643,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 9.96153974875755e-06,
|
|
"loss": 1.3637495040893555,
|
|
"step": 150,
|
|
"token_acc": 0.6573875802997858
|
|
},
|
|
{
|
|
"epoch": 0.08851113716295428,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 9.960330080407712e-06,
|
|
"loss": 1.338754653930664,
|
|
"step": 151,
|
|
"token_acc": 0.6675567423230975
|
|
},
|
|
{
|
|
"epoch": 0.08909730363423213,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 9.959101757706308e-06,
|
|
"loss": 1.2259619235992432,
|
|
"step": 152,
|
|
"token_acc": 0.6795730431142738
|
|
},
|
|
{
|
|
"epoch": 0.08968347010550996,
|
|
"grad_norm": 4.84375,
|
|
"learning_rate": 9.957854785272703e-06,
|
|
"loss": 1.2699283361434937,
|
|
"step": 153,
|
|
"token_acc": 0.6725043782837128
|
|
},
|
|
{
|
|
"epoch": 0.09026963657678781,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 9.956589167796392e-06,
|
|
"loss": 1.3584980964660645,
|
|
"step": 154,
|
|
"token_acc": 0.6546035125066525
|
|
},
|
|
{
|
|
"epoch": 0.09085580304806565,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 9.955304910036993e-06,
|
|
"loss": 1.2366204261779785,
|
|
"step": 155,
|
|
"token_acc": 0.6840452261306532
|
|
},
|
|
{
|
|
"epoch": 0.0914419695193435,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 9.954002016824226e-06,
|
|
"loss": 1.3371152877807617,
|
|
"step": 156,
|
|
"token_acc": 0.6620654396728016
|
|
},
|
|
{
|
|
"epoch": 0.09202813599062133,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 9.952680493057891e-06,
|
|
"loss": 1.2933123111724854,
|
|
"step": 157,
|
|
"token_acc": 0.6592693752133834
|
|
},
|
|
{
|
|
"epoch": 0.09261430246189918,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 9.951340343707852e-06,
|
|
"loss": 1.260751485824585,
|
|
"step": 158,
|
|
"token_acc": 0.680874929893438
|
|
},
|
|
{
|
|
"epoch": 0.09320046893317702,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 9.94998157381402e-06,
|
|
"loss": 1.2553704977035522,
|
|
"step": 159,
|
|
"token_acc": 0.6739130434782609
|
|
},
|
|
{
|
|
"epoch": 0.09378663540445487,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 9.948604188486328e-06,
|
|
"loss": 1.2568773031234741,
|
|
"step": 160,
|
|
"token_acc": 0.6665569453587886
|
|
},
|
|
{
|
|
"epoch": 0.09437280187573271,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 9.947208192904722e-06,
|
|
"loss": 1.3142812252044678,
|
|
"step": 161,
|
|
"token_acc": 0.6552505666079074
|
|
},
|
|
{
|
|
"epoch": 0.09495896834701055,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 9.945793592319137e-06,
|
|
"loss": 1.2683542966842651,
|
|
"step": 162,
|
|
"token_acc": 0.6670216542421016
|
|
},
|
|
{
|
|
"epoch": 0.0955451348182884,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 9.944360392049465e-06,
|
|
"loss": 1.3156769275665283,
|
|
"step": 163,
|
|
"token_acc": 0.6710945802337939
|
|
},
|
|
{
|
|
"epoch": 0.09613130128956623,
|
|
"grad_norm": 4.34375,
|
|
"learning_rate": 9.942908597485558e-06,
|
|
"loss": 1.342336893081665,
|
|
"step": 164,
|
|
"token_acc": 0.6547582383738836
|
|
},
|
|
{
|
|
"epoch": 0.09671746776084408,
|
|
"grad_norm": 3.5,
|
|
"learning_rate": 9.941438214087191e-06,
|
|
"loss": 1.2350528240203857,
|
|
"step": 165,
|
|
"token_acc": 0.6785539814362481
|
|
},
|
|
{
|
|
"epoch": 0.09730363423212192,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 9.939949247384046e-06,
|
|
"loss": 1.281822919845581,
|
|
"step": 166,
|
|
"token_acc": 0.675065445026178
|
|
},
|
|
{
|
|
"epoch": 0.09788980070339977,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 9.938441702975689e-06,
|
|
"loss": 1.3656988143920898,
|
|
"step": 167,
|
|
"token_acc": 0.6565592084264283
|
|
},
|
|
{
|
|
"epoch": 0.0984759671746776,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 9.936915586531556e-06,
|
|
"loss": 1.3796906471252441,
|
|
"step": 168,
|
|
"token_acc": 0.6530496453900709
|
|
},
|
|
{
|
|
"epoch": 0.09906213364595545,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 9.935370903790927e-06,
|
|
"loss": 1.3244438171386719,
|
|
"step": 169,
|
|
"token_acc": 0.6572298325722983
|
|
},
|
|
{
|
|
"epoch": 0.09964830011723329,
|
|
"grad_norm": 3.546875,
|
|
"learning_rate": 9.933807660562898e-06,
|
|
"loss": 1.286813735961914,
|
|
"step": 170,
|
|
"token_acc": 0.6713814238566714
|
|
},
|
|
{
|
|
"epoch": 0.10023446658851114,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 9.93222586272637e-06,
|
|
"loss": 1.239850640296936,
|
|
"step": 171,
|
|
"token_acc": 0.678817451205511
|
|
},
|
|
{
|
|
"epoch": 0.10082063305978899,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 9.930625516230026e-06,
|
|
"loss": 1.2726346254348755,
|
|
"step": 172,
|
|
"token_acc": 0.671120246659815
|
|
},
|
|
{
|
|
"epoch": 0.10140679953106682,
|
|
"grad_norm": 3.921875,
|
|
"learning_rate": 9.929006627092298e-06,
|
|
"loss": 1.2398560047149658,
|
|
"step": 173,
|
|
"token_acc": 0.6785598443074927
|
|
},
|
|
{
|
|
"epoch": 0.10199296600234467,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 9.927369201401358e-06,
|
|
"loss": 1.3094849586486816,
|
|
"step": 174,
|
|
"token_acc": 0.665938864628821
|
|
},
|
|
{
|
|
"epoch": 0.1025791324736225,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 9.925713245315083e-06,
|
|
"loss": 1.2763338088989258,
|
|
"step": 175,
|
|
"token_acc": 0.6644100580270793
|
|
},
|
|
{
|
|
"epoch": 0.10316529894490035,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 9.924038765061042e-06,
|
|
"loss": 1.173354983329773,
|
|
"step": 176,
|
|
"token_acc": 0.6864087641218761
|
|
},
|
|
{
|
|
"epoch": 0.10375146541617819,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 9.922345766936463e-06,
|
|
"loss": 1.2049219608306885,
|
|
"step": 177,
|
|
"token_acc": 0.691696113074205
|
|
},
|
|
{
|
|
"epoch": 0.10433763188745604,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 9.920634257308217e-06,
|
|
"loss": 1.279412865638733,
|
|
"step": 178,
|
|
"token_acc": 0.6635130873541469
|
|
},
|
|
{
|
|
"epoch": 0.10492379835873387,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 9.918904242612794e-06,
|
|
"loss": 1.2482014894485474,
|
|
"step": 179,
|
|
"token_acc": 0.6722251172485669
|
|
},
|
|
{
|
|
"epoch": 0.10550996483001172,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 9.917155729356273e-06,
|
|
"loss": 1.2842085361480713,
|
|
"step": 180,
|
|
"token_acc": 0.6696944673823286
|
|
},
|
|
{
|
|
"epoch": 0.10609613130128957,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 9.915388724114301e-06,
|
|
"loss": 1.2450541257858276,
|
|
"step": 181,
|
|
"token_acc": 0.6761194029850747
|
|
},
|
|
{
|
|
"epoch": 0.10668229777256741,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 9.913603233532067e-06,
|
|
"loss": 1.2588789463043213,
|
|
"step": 182,
|
|
"token_acc": 0.6728615071283096
|
|
},
|
|
{
|
|
"epoch": 0.10726846424384526,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 9.911799264324281e-06,
|
|
"loss": 1.2103667259216309,
|
|
"step": 183,
|
|
"token_acc": 0.6852418860992039
|
|
},
|
|
{
|
|
"epoch": 0.10785463071512309,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 9.909976823275143e-06,
|
|
"loss": 1.1937763690948486,
|
|
"step": 184,
|
|
"token_acc": 0.6968190854870775
|
|
},
|
|
{
|
|
"epoch": 0.10844079718640094,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 9.908135917238321e-06,
|
|
"loss": 1.2718100547790527,
|
|
"step": 185,
|
|
"token_acc": 0.6820276497695853
|
|
},
|
|
{
|
|
"epoch": 0.10902696365767878,
|
|
"grad_norm": 4.40625,
|
|
"learning_rate": 9.906276553136924e-06,
|
|
"loss": 1.3094046115875244,
|
|
"step": 186,
|
|
"token_acc": 0.6687187379761447
|
|
},
|
|
{
|
|
"epoch": 0.10961313012895663,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 9.904398737963478e-06,
|
|
"loss": 1.3250747919082642,
|
|
"step": 187,
|
|
"token_acc": 0.6685842122083733
|
|
},
|
|
{
|
|
"epoch": 0.11019929660023446,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 9.902502478779897e-06,
|
|
"loss": 1.2901142835617065,
|
|
"step": 188,
|
|
"token_acc": 0.685098650051921
|
|
},
|
|
{
|
|
"epoch": 0.11078546307151231,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 9.900587782717457e-06,
|
|
"loss": 1.20682954788208,
|
|
"step": 189,
|
|
"token_acc": 0.6905465288035451
|
|
},
|
|
{
|
|
"epoch": 0.11137162954279015,
|
|
"grad_norm": 4.28125,
|
|
"learning_rate": 9.89865465697677e-06,
|
|
"loss": 1.2712223529815674,
|
|
"step": 190,
|
|
"token_acc": 0.667953667953668
|
|
},
|
|
{
|
|
"epoch": 0.111957796014068,
|
|
"grad_norm": 3.25,
|
|
"learning_rate": 9.896703108827758e-06,
|
|
"loss": 1.2782304286956787,
|
|
"step": 191,
|
|
"token_acc": 0.6740190880169671
|
|
},
|
|
{
|
|
"epoch": 0.11254396248534584,
|
|
"grad_norm": 3.5,
|
|
"learning_rate": 9.894733145609623e-06,
|
|
"loss": 1.306996464729309,
|
|
"step": 192,
|
|
"token_acc": 0.6684782608695652
|
|
},
|
|
{
|
|
"epoch": 0.11313012895662368,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 9.892744774730818e-06,
|
|
"loss": 1.2896482944488525,
|
|
"step": 193,
|
|
"token_acc": 0.6650246305418719
|
|
},
|
|
{
|
|
"epoch": 0.11371629542790153,
|
|
"grad_norm": 3.546875,
|
|
"learning_rate": 9.890738003669029e-06,
|
|
"loss": 1.2813833951950073,
|
|
"step": 194,
|
|
"token_acc": 0.6736918604651163
|
|
},
|
|
{
|
|
"epoch": 0.11430246189917936,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 9.888712839971132e-06,
|
|
"loss": 1.2345757484436035,
|
|
"step": 195,
|
|
"token_acc": 0.6772239052602589
|
|
},
|
|
{
|
|
"epoch": 0.11488862837045721,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 9.886669291253178e-06,
|
|
"loss": 1.319353699684143,
|
|
"step": 196,
|
|
"token_acc": 0.6601990049751244
|
|
},
|
|
{
|
|
"epoch": 0.11547479484173505,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 9.884607365200355e-06,
|
|
"loss": 1.2943089008331299,
|
|
"step": 197,
|
|
"token_acc": 0.6742404227212682
|
|
},
|
|
{
|
|
"epoch": 0.1160609613130129,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 9.882527069566965e-06,
|
|
"loss": 1.299247145652771,
|
|
"step": 198,
|
|
"token_acc": 0.6646358349487677
|
|
},
|
|
{
|
|
"epoch": 0.11664712778429073,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 9.880428412176393e-06,
|
|
"loss": 1.1567022800445557,
|
|
"step": 199,
|
|
"token_acc": 0.6886056191467221
|
|
},
|
|
{
|
|
"epoch": 0.11723329425556858,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 9.878311400921072e-06,
|
|
"loss": 1.37424635887146,
|
|
"step": 200,
|
|
"token_acc": 0.6550163439778728
|
|
},
|
|
{
|
|
"epoch": 0.11781946072684643,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 9.876176043762467e-06,
|
|
"loss": 1.2848453521728516,
|
|
"step": 201,
|
|
"token_acc": 0.6610619469026549
|
|
},
|
|
{
|
|
"epoch": 0.11840562719812427,
|
|
"grad_norm": 4.65625,
|
|
"learning_rate": 9.87402234873103e-06,
|
|
"loss": 1.1777148246765137,
|
|
"step": 202,
|
|
"token_acc": 0.6890444810543658
|
|
},
|
|
{
|
|
"epoch": 0.11899179366940212,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 9.871850323926178e-06,
|
|
"loss": 1.2698417901992798,
|
|
"step": 203,
|
|
"token_acc": 0.674496644295302
|
|
},
|
|
{
|
|
"epoch": 0.11957796014067995,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 9.869659977516261e-06,
|
|
"loss": 1.3126946687698364,
|
|
"step": 204,
|
|
"token_acc": 0.6673753838884952
|
|
},
|
|
{
|
|
"epoch": 0.1201641266119578,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 9.867451317738534e-06,
|
|
"loss": 1.2252904176712036,
|
|
"step": 205,
|
|
"token_acc": 0.6883484162895928
|
|
},
|
|
{
|
|
"epoch": 0.12075029308323564,
|
|
"grad_norm": 4.625,
|
|
"learning_rate": 9.86522435289912e-06,
|
|
"loss": 1.2910317182540894,
|
|
"step": 206,
|
|
"token_acc": 0.6736605470213564
|
|
},
|
|
{
|
|
"epoch": 0.12133645955451348,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 9.862979091372982e-06,
|
|
"loss": 1.2071962356567383,
|
|
"step": 207,
|
|
"token_acc": 0.6792207792207792
|
|
},
|
|
{
|
|
"epoch": 0.12192262602579132,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 9.860715541603893e-06,
|
|
"loss": 1.2307209968566895,
|
|
"step": 208,
|
|
"token_acc": 0.6886149854038275
|
|
},
|
|
{
|
|
"epoch": 0.12250879249706917,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 9.858433712104403e-06,
|
|
"loss": 1.322591781616211,
|
|
"step": 209,
|
|
"token_acc": 0.6763274336283186
|
|
},
|
|
{
|
|
"epoch": 0.123094958968347,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 9.856133611455802e-06,
|
|
"loss": 1.2609598636627197,
|
|
"step": 210,
|
|
"token_acc": 0.6705882352941176
|
|
},
|
|
{
|
|
"epoch": 0.12368112543962485,
|
|
"grad_norm": 4.40625,
|
|
"learning_rate": 9.853815248308101e-06,
|
|
"loss": 1.2948882579803467,
|
|
"step": 211,
|
|
"token_acc": 0.6666666666666666
|
|
},
|
|
{
|
|
"epoch": 0.1242672919109027,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 9.851478631379982e-06,
|
|
"loss": 1.254845380783081,
|
|
"step": 212,
|
|
"token_acc": 0.6775621382362955
|
|
},
|
|
{
|
|
"epoch": 0.12485345838218054,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 9.849123769458783e-06,
|
|
"loss": 1.2732309103012085,
|
|
"step": 213,
|
|
"token_acc": 0.6751896600168587
|
|
},
|
|
{
|
|
"epoch": 0.1254396248534584,
|
|
"grad_norm": 4.4375,
|
|
"learning_rate": 9.846750671400447e-06,
|
|
"loss": 1.258301019668579,
|
|
"step": 214,
|
|
"token_acc": 0.6664613489374808
|
|
},
|
|
{
|
|
"epoch": 0.12602579132473624,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 9.844359346129504e-06,
|
|
"loss": 1.2688188552856445,
|
|
"step": 215,
|
|
"token_acc": 0.6826511912034209
|
|
},
|
|
{
|
|
"epoch": 0.12661195779601406,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 9.841949802639031e-06,
|
|
"loss": 1.2013604640960693,
|
|
"step": 216,
|
|
"token_acc": 0.6838001732601791
|
|
},
|
|
{
|
|
"epoch": 0.1271981242672919,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 9.839522049990614e-06,
|
|
"loss": 1.2248526811599731,
|
|
"step": 217,
|
|
"token_acc": 0.6806678841638404
|
|
},
|
|
{
|
|
"epoch": 0.12778429073856976,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 9.83707609731432e-06,
|
|
"loss": 1.2760939598083496,
|
|
"step": 218,
|
|
"token_acc": 0.6656649135987979
|
|
},
|
|
{
|
|
"epoch": 0.1283704572098476,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 9.834611953808666e-06,
|
|
"loss": 1.2184778451919556,
|
|
"step": 219,
|
|
"token_acc": 0.6771961492178099
|
|
},
|
|
{
|
|
"epoch": 0.12895662368112543,
|
|
"grad_norm": 3.421875,
|
|
"learning_rate": 9.832129628740574e-06,
|
|
"loss": 1.2700254917144775,
|
|
"step": 220,
|
|
"token_acc": 0.6749329758713136
|
|
},
|
|
{
|
|
"epoch": 0.12954279015240328,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 9.829629131445342e-06,
|
|
"loss": 1.2778222560882568,
|
|
"step": 221,
|
|
"token_acc": 0.676656585097608
|
|
},
|
|
{
|
|
"epoch": 0.13012895662368112,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 9.827110471326612e-06,
|
|
"loss": 1.3339558839797974,
|
|
"step": 222,
|
|
"token_acc": 0.6525348354580492
|
|
},
|
|
{
|
|
"epoch": 0.13071512309495897,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 9.824573657856324e-06,
|
|
"loss": 1.2444734573364258,
|
|
"step": 223,
|
|
"token_acc": 0.6779201577131592
|
|
},
|
|
{
|
|
"epoch": 0.13130128956623682,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 9.822018700574696e-06,
|
|
"loss": 1.2936793565750122,
|
|
"step": 224,
|
|
"token_acc": 0.6770614692653674
|
|
},
|
|
{
|
|
"epoch": 0.13188745603751464,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 9.819445609090174e-06,
|
|
"loss": 1.2197020053863525,
|
|
"step": 225,
|
|
"token_acc": 0.6839131727624145
|
|
},
|
|
{
|
|
"epoch": 0.1324736225087925,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 9.816854393079402e-06,
|
|
"loss": 1.2761380672454834,
|
|
"step": 226,
|
|
"token_acc": 0.6791530944625407
|
|
},
|
|
{
|
|
"epoch": 0.13305978898007034,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 9.81424506228719e-06,
|
|
"loss": 1.1647887229919434,
|
|
"step": 227,
|
|
"token_acc": 0.6913762551683402
|
|
},
|
|
{
|
|
"epoch": 0.1336459554513482,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 9.811617626526462e-06,
|
|
"loss": 1.2947075366973877,
|
|
"step": 228,
|
|
"token_acc": 0.6590649942987458
|
|
},
|
|
{
|
|
"epoch": 0.134232121922626,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 9.808972095678242e-06,
|
|
"loss": 1.2353019714355469,
|
|
"step": 229,
|
|
"token_acc": 0.6823218997361478
|
|
},
|
|
{
|
|
"epoch": 0.13481828839390386,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 9.806308479691595e-06,
|
|
"loss": 1.319331407546997,
|
|
"step": 230,
|
|
"token_acc": 0.6628989361702128
|
|
},
|
|
{
|
|
"epoch": 0.1354044548651817,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 9.803626788583603e-06,
|
|
"loss": 1.2679548263549805,
|
|
"step": 231,
|
|
"token_acc": 0.6740062911066629
|
|
},
|
|
{
|
|
"epoch": 0.13599062133645956,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 9.800927032439322e-06,
|
|
"loss": 1.232433557510376,
|
|
"step": 232,
|
|
"token_acc": 0.6765717092337917
|
|
},
|
|
{
|
|
"epoch": 0.1365767878077374,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 9.798209221411748e-06,
|
|
"loss": 1.1995563507080078,
|
|
"step": 233,
|
|
"token_acc": 0.6763132295719845
|
|
},
|
|
{
|
|
"epoch": 0.13716295427901523,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 9.79547336572177e-06,
|
|
"loss": 1.2370085716247559,
|
|
"step": 234,
|
|
"token_acc": 0.6784500339904826
|
|
},
|
|
{
|
|
"epoch": 0.13774912075029308,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 9.792719475658144e-06,
|
|
"loss": 1.215732216835022,
|
|
"step": 235,
|
|
"token_acc": 0.6742372103693508
|
|
},
|
|
{
|
|
"epoch": 0.13833528722157093,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 9.789947561577445e-06,
|
|
"loss": 1.3819772005081177,
|
|
"step": 236,
|
|
"token_acc": 0.6586774473000289
|
|
},
|
|
{
|
|
"epoch": 0.13892145369284878,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 9.787157633904032e-06,
|
|
"loss": 1.2074400186538696,
|
|
"step": 237,
|
|
"token_acc": 0.6784905660377358
|
|
},
|
|
{
|
|
"epoch": 0.1395076201641266,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 9.784349703130008e-06,
|
|
"loss": 1.2738075256347656,
|
|
"step": 238,
|
|
"token_acc": 0.6647342995169082
|
|
},
|
|
{
|
|
"epoch": 0.14009378663540445,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 9.781523779815178e-06,
|
|
"loss": 1.2466193437576294,
|
|
"step": 239,
|
|
"token_acc": 0.6818299974005718
|
|
},
|
|
{
|
|
"epoch": 0.1406799531066823,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 9.778679874587016e-06,
|
|
"loss": 1.2840759754180908,
|
|
"step": 240,
|
|
"token_acc": 0.679560768654854
|
|
},
|
|
{
|
|
"epoch": 0.14126611957796015,
|
|
"grad_norm": 3.265625,
|
|
"learning_rate": 9.775817998140615e-06,
|
|
"loss": 1.215688705444336,
|
|
"step": 241,
|
|
"token_acc": 0.6817191523073876
|
|
},
|
|
{
|
|
"epoch": 0.141852286049238,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 9.77293816123866e-06,
|
|
"loss": 1.199409008026123,
|
|
"step": 242,
|
|
"token_acc": 0.6897381079636558
|
|
},
|
|
{
|
|
"epoch": 0.14243845252051582,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 9.770040374711374e-06,
|
|
"loss": 1.1880569458007812,
|
|
"step": 243,
|
|
"token_acc": 0.6855151045701007
|
|
},
|
|
{
|
|
"epoch": 0.14302461899179367,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 9.767124649456484e-06,
|
|
"loss": 1.2265934944152832,
|
|
"step": 244,
|
|
"token_acc": 0.6833046471600689
|
|
},
|
|
{
|
|
"epoch": 0.14361078546307152,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 9.764190996439181e-06,
|
|
"loss": 1.2407772541046143,
|
|
"step": 245,
|
|
"token_acc": 0.6796982167352538
|
|
},
|
|
{
|
|
"epoch": 0.14419695193434937,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 9.761239426692077e-06,
|
|
"loss": 1.2412476539611816,
|
|
"step": 246,
|
|
"token_acc": 0.6766556787533713
|
|
},
|
|
{
|
|
"epoch": 0.1447831184056272,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 9.758269951315164e-06,
|
|
"loss": 1.2553603649139404,
|
|
"step": 247,
|
|
"token_acc": 0.6823570432357043
|
|
},
|
|
{
|
|
"epoch": 0.14536928487690504,
|
|
"grad_norm": 3.53125,
|
|
"learning_rate": 9.755282581475769e-06,
|
|
"loss": 1.2583595514297485,
|
|
"step": 248,
|
|
"token_acc": 0.6749140893470791
|
|
},
|
|
{
|
|
"epoch": 0.14595545134818289,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 9.752277328408518e-06,
|
|
"loss": 1.2226881980895996,
|
|
"step": 249,
|
|
"token_acc": 0.6730310262529833
|
|
},
|
|
{
|
|
"epoch": 0.14654161781946073,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 9.749254203415288e-06,
|
|
"loss": 1.27619206905365,
|
|
"step": 250,
|
|
"token_acc": 0.6636388659510046
|
|
},
|
|
{
|
|
"epoch": 0.14712778429073858,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 9.74621321786517e-06,
|
|
"loss": 1.2808241844177246,
|
|
"step": 251,
|
|
"token_acc": 0.6712470832253047
|
|
},
|
|
{
|
|
"epoch": 0.1477139507620164,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 9.743154383194422e-06,
|
|
"loss": 1.2323708534240723,
|
|
"step": 252,
|
|
"token_acc": 0.6790823877651702
|
|
},
|
|
{
|
|
"epoch": 0.14830011723329425,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 9.740077710906425e-06,
|
|
"loss": 1.2157396078109741,
|
|
"step": 253,
|
|
"token_acc": 0.6833731105807478
|
|
},
|
|
{
|
|
"epoch": 0.1488862837045721,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 9.736983212571646e-06,
|
|
"loss": 1.2374944686889648,
|
|
"step": 254,
|
|
"token_acc": 0.6670284938941655
|
|
},
|
|
{
|
|
"epoch": 0.14947245017584995,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 9.73387089982759e-06,
|
|
"loss": 1.3121461868286133,
|
|
"step": 255,
|
|
"token_acc": 0.6640277090549233
|
|
},
|
|
{
|
|
"epoch": 0.15005861664712777,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 9.730740784378755e-06,
|
|
"loss": 1.1824970245361328,
|
|
"step": 256,
|
|
"token_acc": 0.68513195714659
|
|
},
|
|
{
|
|
"epoch": 0.15064478311840562,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 9.727592877996585e-06,
|
|
"loss": 1.2350056171417236,
|
|
"step": 257,
|
|
"token_acc": 0.6744875277846382
|
|
},
|
|
{
|
|
"epoch": 0.15123094958968347,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 9.72442719251944e-06,
|
|
"loss": 1.2560629844665527,
|
|
"step": 258,
|
|
"token_acc": 0.6722474460839954
|
|
},
|
|
{
|
|
"epoch": 0.15181711606096132,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 9.721243739852541e-06,
|
|
"loss": 1.2138497829437256,
|
|
"step": 259,
|
|
"token_acc": 0.6785515320334262
|
|
},
|
|
{
|
|
"epoch": 0.15240328253223914,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 9.718042531967918e-06,
|
|
"loss": 1.247100591659546,
|
|
"step": 260,
|
|
"token_acc": 0.679639029892837
|
|
},
|
|
{
|
|
"epoch": 0.152989449003517,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 9.71482358090438e-06,
|
|
"loss": 1.2345657348632812,
|
|
"step": 261,
|
|
"token_acc": 0.6820169046925094
|
|
},
|
|
{
|
|
"epoch": 0.15357561547479484,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 9.711586898767462e-06,
|
|
"loss": 1.2890663146972656,
|
|
"step": 262,
|
|
"token_acc": 0.6662515566625156
|
|
},
|
|
{
|
|
"epoch": 0.1541617819460727,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 9.708332497729378e-06,
|
|
"loss": 1.300486445426941,
|
|
"step": 263,
|
|
"token_acc": 0.6727622714148219
|
|
},
|
|
{
|
|
"epoch": 0.15474794841735054,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 9.705060390028979e-06,
|
|
"loss": 1.2290432453155518,
|
|
"step": 264,
|
|
"token_acc": 0.6757286995515696
|
|
},
|
|
{
|
|
"epoch": 0.15533411488862836,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 9.701770587971705e-06,
|
|
"loss": 1.3092460632324219,
|
|
"step": 265,
|
|
"token_acc": 0.6636015325670498
|
|
},
|
|
{
|
|
"epoch": 0.1559202813599062,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 9.698463103929542e-06,
|
|
"loss": 1.2968261241912842,
|
|
"step": 266,
|
|
"token_acc": 0.6685144124168514
|
|
},
|
|
{
|
|
"epoch": 0.15650644783118406,
|
|
"grad_norm": 3.484375,
|
|
"learning_rate": 9.695137950340971e-06,
|
|
"loss": 1.2271721363067627,
|
|
"step": 267,
|
|
"token_acc": 0.6880385200202737
|
|
},
|
|
{
|
|
"epoch": 0.1570926143024619,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 9.69179513971092e-06,
|
|
"loss": 1.2090380191802979,
|
|
"step": 268,
|
|
"token_acc": 0.6890125173852573
|
|
},
|
|
{
|
|
"epoch": 0.15767878077373973,
|
|
"grad_norm": 3.328125,
|
|
"learning_rate": 9.688434684610725e-06,
|
|
"loss": 1.2554593086242676,
|
|
"step": 269,
|
|
"token_acc": 0.6735661923242777
|
|
},
|
|
{
|
|
"epoch": 0.15826494724501758,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 9.685056597678075e-06,
|
|
"loss": 1.2064403295516968,
|
|
"step": 270,
|
|
"token_acc": 0.6806970509383378
|
|
},
|
|
{
|
|
"epoch": 0.15885111371629543,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 9.681660891616967e-06,
|
|
"loss": 1.2292215824127197,
|
|
"step": 271,
|
|
"token_acc": 0.6725957190232138
|
|
},
|
|
{
|
|
"epoch": 0.15943728018757328,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 9.678247579197658e-06,
|
|
"loss": 1.209936499595642,
|
|
"step": 272,
|
|
"token_acc": 0.6792452830188679
|
|
},
|
|
{
|
|
"epoch": 0.16002344665885113,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 9.67481667325662e-06,
|
|
"loss": 1.2524675130844116,
|
|
"step": 273,
|
|
"token_acc": 0.6846907739101048
|
|
},
|
|
{
|
|
"epoch": 0.16060961313012895,
|
|
"grad_norm": 3.46875,
|
|
"learning_rate": 9.671368186696488e-06,
|
|
"loss": 1.3006181716918945,
|
|
"step": 274,
|
|
"token_acc": 0.6592865928659286
|
|
},
|
|
{
|
|
"epoch": 0.1611957796014068,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 9.667902132486009e-06,
|
|
"loss": 1.3238104581832886,
|
|
"step": 275,
|
|
"token_acc": 0.6606606606606606
|
|
},
|
|
{
|
|
"epoch": 0.16178194607268465,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 9.664418523660004e-06,
|
|
"loss": 1.1422791481018066,
|
|
"step": 276,
|
|
"token_acc": 0.6995596995596995
|
|
},
|
|
{
|
|
"epoch": 0.1623681125439625,
|
|
"grad_norm": 4.375,
|
|
"learning_rate": 9.660917373319303e-06,
|
|
"loss": 1.254688024520874,
|
|
"step": 277,
|
|
"token_acc": 0.6766823161189358
|
|
},
|
|
{
|
|
"epoch": 0.16295427901524032,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 9.657398694630713e-06,
|
|
"loss": 1.2908952236175537,
|
|
"step": 278,
|
|
"token_acc": 0.677810650887574
|
|
},
|
|
{
|
|
"epoch": 0.16354044548651817,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 9.653862500826953e-06,
|
|
"loss": 1.3178751468658447,
|
|
"step": 279,
|
|
"token_acc": 0.6628825271470878
|
|
},
|
|
{
|
|
"epoch": 0.16412661195779601,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 9.650308805206616e-06,
|
|
"loss": 1.2194089889526367,
|
|
"step": 280,
|
|
"token_acc": 0.6848333852382436
|
|
},
|
|
{
|
|
"epoch": 0.16471277842907386,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 9.646737621134112e-06,
|
|
"loss": 1.175027847290039,
|
|
"step": 281,
|
|
"token_acc": 0.6945812807881774
|
|
},
|
|
{
|
|
"epoch": 0.1652989449003517,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 9.643148962039622e-06,
|
|
"loss": 1.2113852500915527,
|
|
"step": 282,
|
|
"token_acc": 0.6986348122866894
|
|
},
|
|
{
|
|
"epoch": 0.16588511137162953,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 9.639542841419041e-06,
|
|
"loss": 1.2631524801254272,
|
|
"step": 283,
|
|
"token_acc": 0.6784575239199768
|
|
},
|
|
{
|
|
"epoch": 0.16647127784290738,
|
|
"grad_norm": 4.28125,
|
|
"learning_rate": 9.635919272833938e-06,
|
|
"loss": 1.2649877071380615,
|
|
"step": 284,
|
|
"token_acc": 0.679974449057809
|
|
},
|
|
{
|
|
"epoch": 0.16705744431418523,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 9.632278269911493e-06,
|
|
"loss": 1.2161870002746582,
|
|
"step": 285,
|
|
"token_acc": 0.6879052720608966
|
|
},
|
|
{
|
|
"epoch": 0.16764361078546308,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 9.628619846344453e-06,
|
|
"loss": 1.230222463607788,
|
|
"step": 286,
|
|
"token_acc": 0.6853506634172759
|
|
},
|
|
{
|
|
"epoch": 0.1682297772567409,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 9.62494401589108e-06,
|
|
"loss": 1.2671175003051758,
|
|
"step": 287,
|
|
"token_acc": 0.6733906852566004
|
|
},
|
|
{
|
|
"epoch": 0.16881594372801875,
|
|
"grad_norm": 3.546875,
|
|
"learning_rate": 9.6212507923751e-06,
|
|
"loss": 1.2006876468658447,
|
|
"step": 288,
|
|
"token_acc": 0.6841036058913154
|
|
},
|
|
{
|
|
"epoch": 0.1694021101992966,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 9.617540189685648e-06,
|
|
"loss": 1.203233242034912,
|
|
"step": 289,
|
|
"token_acc": 0.6915610676732273
|
|
},
|
|
{
|
|
"epoch": 0.16998827667057445,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 9.613812221777212e-06,
|
|
"loss": 1.2433654069900513,
|
|
"step": 290,
|
|
"token_acc": 0.6806883365200764
|
|
},
|
|
{
|
|
"epoch": 0.17057444314185227,
|
|
"grad_norm": 3.4375,
|
|
"learning_rate": 9.610066902669593e-06,
|
|
"loss": 1.2236400842666626,
|
|
"step": 291,
|
|
"token_acc": 0.681420233463035
|
|
},
|
|
{
|
|
"epoch": 0.17116060961313012,
|
|
"grad_norm": 3.296875,
|
|
"learning_rate": 9.60630424644784e-06,
|
|
"loss": 1.267236351966858,
|
|
"step": 292,
|
|
"token_acc": 0.6745922578109649
|
|
},
|
|
{
|
|
"epoch": 0.17174677608440797,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 9.602524267262202e-06,
|
|
"loss": 1.2827626466751099,
|
|
"step": 293,
|
|
"token_acc": 0.6706154290667438
|
|
},
|
|
{
|
|
"epoch": 0.17233294255568582,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 9.598726979328079e-06,
|
|
"loss": 1.2633857727050781,
|
|
"step": 294,
|
|
"token_acc": 0.6807755324959038
|
|
},
|
|
{
|
|
"epoch": 0.17291910902696367,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 9.594912396925958e-06,
|
|
"loss": 1.2157801389694214,
|
|
"step": 295,
|
|
"token_acc": 0.6820977662674005
|
|
},
|
|
{
|
|
"epoch": 0.1735052754982415,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 9.591080534401371e-06,
|
|
"loss": 1.2819228172302246,
|
|
"step": 296,
|
|
"token_acc": 0.6716196136701337
|
|
},
|
|
{
|
|
"epoch": 0.17409144196951934,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 9.587231406164831e-06,
|
|
"loss": 1.2857537269592285,
|
|
"step": 297,
|
|
"token_acc": 0.6740794750273423
|
|
},
|
|
{
|
|
"epoch": 0.1746776084407972,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 9.583365026691785e-06,
|
|
"loss": 1.243815541267395,
|
|
"step": 298,
|
|
"token_acc": 0.6734800838574424
|
|
},
|
|
{
|
|
"epoch": 0.17526377491207504,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 9.579481410522556e-06,
|
|
"loss": 1.19548499584198,
|
|
"step": 299,
|
|
"token_acc": 0.6933132227249874
|
|
},
|
|
{
|
|
"epoch": 0.17584994138335286,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 9.57558057226229e-06,
|
|
"loss": 1.1392521858215332,
|
|
"step": 300,
|
|
"token_acc": 0.6948019801980198
|
|
},
|
|
{
|
|
"epoch": 0.1764361078546307,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 9.571662526580898e-06,
|
|
"loss": 1.268701195716858,
|
|
"step": 301,
|
|
"token_acc": 0.6679915209326974
|
|
},
|
|
{
|
|
"epoch": 0.17702227432590856,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 9.567727288213005e-06,
|
|
"loss": 1.2397103309631348,
|
|
"step": 302,
|
|
"token_acc": 0.684172918899607
|
|
},
|
|
{
|
|
"epoch": 0.1776084407971864,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 9.563774871957893e-06,
|
|
"loss": 1.2294676303863525,
|
|
"step": 303,
|
|
"token_acc": 0.6740694401000938
|
|
},
|
|
{
|
|
"epoch": 0.17819460726846426,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 9.559805292679445e-06,
|
|
"loss": 1.1616265773773193,
|
|
"step": 304,
|
|
"token_acc": 0.6931742876076872
|
|
},
|
|
{
|
|
"epoch": 0.17878077373974208,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 9.555818565306086e-06,
|
|
"loss": 1.1858564615249634,
|
|
"step": 305,
|
|
"token_acc": 0.6901041666666666
|
|
},
|
|
{
|
|
"epoch": 0.17936694021101993,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 9.551814704830734e-06,
|
|
"loss": 1.2298040390014648,
|
|
"step": 306,
|
|
"token_acc": 0.6807409425224734
|
|
},
|
|
{
|
|
"epoch": 0.17995310668229778,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 9.547793726310738e-06,
|
|
"loss": 1.1947698593139648,
|
|
"step": 307,
|
|
"token_acc": 0.6838269863657734
|
|
},
|
|
{
|
|
"epoch": 0.18053927315357562,
|
|
"grad_norm": 3.53125,
|
|
"learning_rate": 9.543755644867823e-06,
|
|
"loss": 1.2667665481567383,
|
|
"step": 308,
|
|
"token_acc": 0.6694763729246488
|
|
},
|
|
{
|
|
"epoch": 0.18112543962485345,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 9.539700475688031e-06,
|
|
"loss": 1.1842645406723022,
|
|
"step": 309,
|
|
"token_acc": 0.6830970065271214
|
|
},
|
|
{
|
|
"epoch": 0.1817116060961313,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 9.53562823402167e-06,
|
|
"loss": 1.131266474723816,
|
|
"step": 310,
|
|
"token_acc": 0.7028475711892798
|
|
},
|
|
{
|
|
"epoch": 0.18229777256740914,
|
|
"grad_norm": 4.375,
|
|
"learning_rate": 9.531538935183252e-06,
|
|
"loss": 1.2256965637207031,
|
|
"step": 311,
|
|
"token_acc": 0.6898509581263307
|
|
},
|
|
{
|
|
"epoch": 0.182883939038687,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 9.52743259455143e-06,
|
|
"loss": 1.2770997285842896,
|
|
"step": 312,
|
|
"token_acc": 0.6758700028760426
|
|
},
|
|
{
|
|
"epoch": 0.18347010550996484,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 9.523309227568956e-06,
|
|
"loss": 1.1665555238723755,
|
|
"step": 313,
|
|
"token_acc": 0.6955782312925171
|
|
},
|
|
{
|
|
"epoch": 0.18405627198124266,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 9.519168849742603e-06,
|
|
"loss": 1.214857816696167,
|
|
"step": 314,
|
|
"token_acc": 0.6749710312862109
|
|
},
|
|
{
|
|
"epoch": 0.1846424384525205,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 9.515011476643127e-06,
|
|
"loss": 1.2182379961013794,
|
|
"step": 315,
|
|
"token_acc": 0.6833428084138715
|
|
},
|
|
{
|
|
"epoch": 0.18522860492379836,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 9.51083712390519e-06,
|
|
"loss": 1.2550464868545532,
|
|
"step": 316,
|
|
"token_acc": 0.6718266253869969
|
|
},
|
|
{
|
|
"epoch": 0.1858147713950762,
|
|
"grad_norm": 4.34375,
|
|
"learning_rate": 9.506645807227311e-06,
|
|
"loss": 1.1618129014968872,
|
|
"step": 317,
|
|
"token_acc": 0.6932197110040755
|
|
},
|
|
{
|
|
"epoch": 0.18640093786635403,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 9.502437542371812e-06,
|
|
"loss": 1.2061501741409302,
|
|
"step": 318,
|
|
"token_acc": 0.6859122401847575
|
|
},
|
|
{
|
|
"epoch": 0.18698710433763188,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 9.498212345164743e-06,
|
|
"loss": 1.1590545177459717,
|
|
"step": 319,
|
|
"token_acc": 0.6927881885292447
|
|
},
|
|
{
|
|
"epoch": 0.18757327080890973,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 9.493970231495836e-06,
|
|
"loss": 1.3217151165008545,
|
|
"step": 320,
|
|
"token_acc": 0.667819214620894
|
|
},
|
|
{
|
|
"epoch": 0.18815943728018758,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 9.48971121731844e-06,
|
|
"loss": 1.2895207405090332,
|
|
"step": 321,
|
|
"token_acc": 0.6619929894842264
|
|
},
|
|
{
|
|
"epoch": 0.18874560375146543,
|
|
"grad_norm": 3.546875,
|
|
"learning_rate": 9.485435318649468e-06,
|
|
"loss": 1.2839877605438232,
|
|
"step": 322,
|
|
"token_acc": 0.6687413554633471
|
|
},
|
|
{
|
|
"epoch": 0.18933177022274325,
|
|
"grad_norm": 3.609375,
|
|
"learning_rate": 9.481142551569318e-06,
|
|
"loss": 1.3126130104064941,
|
|
"step": 323,
|
|
"token_acc": 0.6639044601511089
|
|
},
|
|
{
|
|
"epoch": 0.1899179366940211,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 9.476832932221835e-06,
|
|
"loss": 1.3199176788330078,
|
|
"step": 324,
|
|
"token_acc": 0.6602447279354334
|
|
},
|
|
{
|
|
"epoch": 0.19050410316529895,
|
|
"grad_norm": 5.0,
|
|
"learning_rate": 9.472506476814239e-06,
|
|
"loss": 1.1878896951675415,
|
|
"step": 325,
|
|
"token_acc": 0.6956521739130435
|
|
},
|
|
{
|
|
"epoch": 0.1910902696365768,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 9.468163201617063e-06,
|
|
"loss": 1.2245277166366577,
|
|
"step": 326,
|
|
"token_acc": 0.6785978770673907
|
|
},
|
|
{
|
|
"epoch": 0.19167643610785462,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 9.463803122964094e-06,
|
|
"loss": 1.2510501146316528,
|
|
"step": 327,
|
|
"token_acc": 0.6775878442545109
|
|
},
|
|
{
|
|
"epoch": 0.19226260257913247,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 9.459426257252316e-06,
|
|
"loss": 1.1546225547790527,
|
|
"step": 328,
|
|
"token_acc": 0.6927312775330396
|
|
},
|
|
{
|
|
"epoch": 0.19284876905041032,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 9.45503262094184e-06,
|
|
"loss": 1.2478933334350586,
|
|
"step": 329,
|
|
"token_acc": 0.6814457831325301
|
|
},
|
|
{
|
|
"epoch": 0.19343493552168817,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 9.450622230555849e-06,
|
|
"loss": 1.3278099298477173,
|
|
"step": 330,
|
|
"token_acc": 0.666923076923077
|
|
},
|
|
{
|
|
"epoch": 0.194021101992966,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 9.446195102680531e-06,
|
|
"loss": 1.2192333936691284,
|
|
"step": 331,
|
|
"token_acc": 0.68525
|
|
},
|
|
{
|
|
"epoch": 0.19460726846424384,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 9.441751253965022e-06,
|
|
"loss": 1.1951171159744263,
|
|
"step": 332,
|
|
"token_acc": 0.6875365710942072
|
|
},
|
|
{
|
|
"epoch": 0.1951934349355217,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 9.437290701121335e-06,
|
|
"loss": 1.2858715057373047,
|
|
"step": 333,
|
|
"token_acc": 0.6726638023630505
|
|
},
|
|
{
|
|
"epoch": 0.19577960140679954,
|
|
"grad_norm": 3.359375,
|
|
"learning_rate": 9.432813460924308e-06,
|
|
"loss": 1.2072978019714355,
|
|
"step": 334,
|
|
"token_acc": 0.6863030303030303
|
|
},
|
|
{
|
|
"epoch": 0.19636576787807739,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 9.428319550211531e-06,
|
|
"loss": 1.2071993350982666,
|
|
"step": 335,
|
|
"token_acc": 0.6848648648648649
|
|
},
|
|
{
|
|
"epoch": 0.1969519343493552,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 9.423808985883289e-06,
|
|
"loss": 1.2421560287475586,
|
|
"step": 336,
|
|
"token_acc": 0.6808923375363725
|
|
},
|
|
{
|
|
"epoch": 0.19753810082063306,
|
|
"grad_norm": 4.28125,
|
|
"learning_rate": 9.419281784902499e-06,
|
|
"loss": 1.2532129287719727,
|
|
"step": 337,
|
|
"token_acc": 0.6803621169916435
|
|
},
|
|
{
|
|
"epoch": 0.1981242672919109,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 9.414737964294636e-06,
|
|
"loss": 1.1703143119812012,
|
|
"step": 338,
|
|
"token_acc": 0.6875981161695447
|
|
},
|
|
{
|
|
"epoch": 0.19871043376318875,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 9.410177541147682e-06,
|
|
"loss": 1.2518219947814941,
|
|
"step": 339,
|
|
"token_acc": 0.6799292661361627
|
|
},
|
|
{
|
|
"epoch": 0.19929660023446658,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 9.405600532612061e-06,
|
|
"loss": 1.2657400369644165,
|
|
"step": 340,
|
|
"token_acc": 0.6744847192608386
|
|
},
|
|
{
|
|
"epoch": 0.19988276670574442,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 9.401006955900555e-06,
|
|
"loss": 1.2638294696807861,
|
|
"step": 341,
|
|
"token_acc": 0.6716798368595462
|
|
},
|
|
{
|
|
"epoch": 0.20046893317702227,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 9.396396828288272e-06,
|
|
"loss": 1.2558658123016357,
|
|
"step": 342,
|
|
"token_acc": 0.6783905458638154
|
|
},
|
|
{
|
|
"epoch": 0.20105509964830012,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 9.39177016711255e-06,
|
|
"loss": 1.239455223083496,
|
|
"step": 343,
|
|
"token_acc": 0.6770012706480305
|
|
},
|
|
{
|
|
"epoch": 0.20164126611957797,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 9.38712698977291e-06,
|
|
"loss": 1.2625657320022583,
|
|
"step": 344,
|
|
"token_acc": 0.6689593301435407
|
|
},
|
|
{
|
|
"epoch": 0.2022274325908558,
|
|
"grad_norm": 3.5,
|
|
"learning_rate": 9.382467313730985e-06,
|
|
"loss": 1.274172306060791,
|
|
"step": 345,
|
|
"token_acc": 0.676297862344374
|
|
},
|
|
{
|
|
"epoch": 0.20281359906213364,
|
|
"grad_norm": 3.390625,
|
|
"learning_rate": 9.377791156510456e-06,
|
|
"loss": 1.2571346759796143,
|
|
"step": 346,
|
|
"token_acc": 0.6716621253405994
|
|
},
|
|
{
|
|
"epoch": 0.2033997655334115,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 9.37309853569698e-06,
|
|
"loss": 1.18192720413208,
|
|
"step": 347,
|
|
"token_acc": 0.6947614866842259
|
|
},
|
|
{
|
|
"epoch": 0.20398593200468934,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 9.368389468938134e-06,
|
|
"loss": 1.2145590782165527,
|
|
"step": 348,
|
|
"token_acc": 0.6773899848254932
|
|
},
|
|
{
|
|
"epoch": 0.20457209847596716,
|
|
"grad_norm": 3.5,
|
|
"learning_rate": 9.363663973943345e-06,
|
|
"loss": 1.207729697227478,
|
|
"step": 349,
|
|
"token_acc": 0.6767140734451285
|
|
},
|
|
{
|
|
"epoch": 0.205158264947245,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 9.358922068483813e-06,
|
|
"loss": 1.2084240913391113,
|
|
"step": 350,
|
|
"token_acc": 0.6796525401421427
|
|
},
|
|
{
|
|
"epoch": 0.20574443141852286,
|
|
"grad_norm": 3.328125,
|
|
"learning_rate": 9.35416377039246e-06,
|
|
"loss": 1.207877278327942,
|
|
"step": 351,
|
|
"token_acc": 0.6776257722859664
|
|
},
|
|
{
|
|
"epoch": 0.2063305978898007,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 9.349389097563858e-06,
|
|
"loss": 1.238926649093628,
|
|
"step": 352,
|
|
"token_acc": 0.6757575757575758
|
|
},
|
|
{
|
|
"epoch": 0.20691676436107856,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 9.344598067954151e-06,
|
|
"loss": 1.1538476943969727,
|
|
"step": 353,
|
|
"token_acc": 0.6938318951741771
|
|
},
|
|
{
|
|
"epoch": 0.20750293083235638,
|
|
"grad_norm": 3.609375,
|
|
"learning_rate": 9.339790699581004e-06,
|
|
"loss": 1.2124214172363281,
|
|
"step": 354,
|
|
"token_acc": 0.680952380952381
|
|
},
|
|
{
|
|
"epoch": 0.20808909730363423,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 9.334967010523523e-06,
|
|
"loss": 1.3020644187927246,
|
|
"step": 355,
|
|
"token_acc": 0.6594775940570333
|
|
},
|
|
{
|
|
"epoch": 0.20867526377491208,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 9.330127018922195e-06,
|
|
"loss": 1.2660328149795532,
|
|
"step": 356,
|
|
"token_acc": 0.6695587506197322
|
|
},
|
|
{
|
|
"epoch": 0.20926143024618993,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 9.325270742978809e-06,
|
|
"loss": 1.2856448888778687,
|
|
"step": 357,
|
|
"token_acc": 0.6785922047702152
|
|
},
|
|
{
|
|
"epoch": 0.20984759671746775,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 9.320398200956403e-06,
|
|
"loss": 1.210078477859497,
|
|
"step": 358,
|
|
"token_acc": 0.6814374493380168
|
|
},
|
|
{
|
|
"epoch": 0.2104337631887456,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 9.315509411179182e-06,
|
|
"loss": 1.2807371616363525,
|
|
"step": 359,
|
|
"token_acc": 0.672882427307206
|
|
},
|
|
{
|
|
"epoch": 0.21101992966002345,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 9.310604392032457e-06,
|
|
"loss": 1.238588571548462,
|
|
"step": 360,
|
|
"token_acc": 0.6782387806943269
|
|
},
|
|
{
|
|
"epoch": 0.2116060961313013,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 9.305683161962569e-06,
|
|
"loss": 1.2255992889404297,
|
|
"step": 361,
|
|
"token_acc": 0.6777486910994764
|
|
},
|
|
{
|
|
"epoch": 0.21219226260257915,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 9.30074573947683e-06,
|
|
"loss": 1.1911859512329102,
|
|
"step": 362,
|
|
"token_acc": 0.6892039258451472
|
|
},
|
|
{
|
|
"epoch": 0.21277842907385697,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 9.295792143143435e-06,
|
|
"loss": 1.2250515222549438,
|
|
"step": 363,
|
|
"token_acc": 0.6728948679621325
|
|
},
|
|
{
|
|
"epoch": 0.21336459554513482,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 9.290822391591418e-06,
|
|
"loss": 1.2066843509674072,
|
|
"step": 364,
|
|
"token_acc": 0.6871216290588883
|
|
},
|
|
{
|
|
"epoch": 0.21395076201641267,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 9.285836503510562e-06,
|
|
"loss": 1.1916322708129883,
|
|
"step": 365,
|
|
"token_acc": 0.6783216783216783
|
|
},
|
|
{
|
|
"epoch": 0.21453692848769051,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 9.280834497651334e-06,
|
|
"loss": 1.2550278902053833,
|
|
"step": 366,
|
|
"token_acc": 0.6840127204394334
|
|
},
|
|
{
|
|
"epoch": 0.21512309495896834,
|
|
"grad_norm": 3.3125,
|
|
"learning_rate": 9.275816392824813e-06,
|
|
"loss": 1.2170839309692383,
|
|
"step": 367,
|
|
"token_acc": 0.684995340167754
|
|
},
|
|
{
|
|
"epoch": 0.21570926143024619,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 9.27078220790263e-06,
|
|
"loss": 1.3127620220184326,
|
|
"step": 368,
|
|
"token_acc": 0.6568062827225131
|
|
},
|
|
{
|
|
"epoch": 0.21629542790152403,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 9.265731961816877e-06,
|
|
"loss": 1.243112564086914,
|
|
"step": 369,
|
|
"token_acc": 0.6830028328611898
|
|
},
|
|
{
|
|
"epoch": 0.21688159437280188,
|
|
"grad_norm": 3.421875,
|
|
"learning_rate": 9.260665673560058e-06,
|
|
"loss": 1.2259280681610107,
|
|
"step": 370,
|
|
"token_acc": 0.6747228381374722
|
|
},
|
|
{
|
|
"epoch": 0.2174677608440797,
|
|
"grad_norm": 3.921875,
|
|
"learning_rate": 9.255583362184998e-06,
|
|
"loss": 1.228597640991211,
|
|
"step": 371,
|
|
"token_acc": 0.6799888981404385
|
|
},
|
|
{
|
|
"epoch": 0.21805392731535755,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 9.25048504680479e-06,
|
|
"loss": 1.2435294389724731,
|
|
"step": 372,
|
|
"token_acc": 0.6889468536028692
|
|
},
|
|
{
|
|
"epoch": 0.2186400937866354,
|
|
"grad_norm": 3.921875,
|
|
"learning_rate": 9.245370746592706e-06,
|
|
"loss": 1.2231773138046265,
|
|
"step": 373,
|
|
"token_acc": 0.6729678638941399
|
|
},
|
|
{
|
|
"epoch": 0.21922626025791325,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 9.24024048078213e-06,
|
|
"loss": 1.2382752895355225,
|
|
"step": 374,
|
|
"token_acc": 0.6740571428571429
|
|
},
|
|
{
|
|
"epoch": 0.2198124267291911,
|
|
"grad_norm": 4.40625,
|
|
"learning_rate": 9.235094268666499e-06,
|
|
"loss": 1.2933881282806396,
|
|
"step": 375,
|
|
"token_acc": 0.6680584551148225
|
|
},
|
|
{
|
|
"epoch": 0.22039859320046892,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 9.229932129599206e-06,
|
|
"loss": 1.3505302667617798,
|
|
"step": 376,
|
|
"token_acc": 0.664701547531319
|
|
},
|
|
{
|
|
"epoch": 0.22098475967174677,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 9.224754082993553e-06,
|
|
"loss": 1.2024602890014648,
|
|
"step": 377,
|
|
"token_acc": 0.6901041666666666
|
|
},
|
|
{
|
|
"epoch": 0.22157092614302462,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 9.219560148322655e-06,
|
|
"loss": 1.22776198387146,
|
|
"step": 378,
|
|
"token_acc": 0.6750409612233752
|
|
},
|
|
{
|
|
"epoch": 0.22215709261430247,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 9.214350345119384e-06,
|
|
"loss": 1.1821080446243286,
|
|
"step": 379,
|
|
"token_acc": 0.6862679585699967
|
|
},
|
|
{
|
|
"epoch": 0.2227432590855803,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 9.209124692976287e-06,
|
|
"loss": 1.1949726343154907,
|
|
"step": 380,
|
|
"token_acc": 0.678495197438634
|
|
},
|
|
{
|
|
"epoch": 0.22332942555685814,
|
|
"grad_norm": 3.34375,
|
|
"learning_rate": 9.203883211545517e-06,
|
|
"loss": 1.2482595443725586,
|
|
"step": 381,
|
|
"token_acc": 0.6783601876402203
|
|
},
|
|
{
|
|
"epoch": 0.223915592028136,
|
|
"grad_norm": 3.359375,
|
|
"learning_rate": 9.19862592053875e-06,
|
|
"loss": 1.2233350276947021,
|
|
"step": 382,
|
|
"token_acc": 0.6851211072664359
|
|
},
|
|
{
|
|
"epoch": 0.22450175849941384,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 9.193352839727122e-06,
|
|
"loss": 1.2985519170761108,
|
|
"step": 383,
|
|
"token_acc": 0.6668421052631579
|
|
},
|
|
{
|
|
"epoch": 0.2250879249706917,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 9.188063988941147e-06,
|
|
"loss": 1.235428810119629,
|
|
"step": 384,
|
|
"token_acc": 0.6788211788211789
|
|
},
|
|
{
|
|
"epoch": 0.2256740914419695,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 9.18275938807065e-06,
|
|
"loss": 1.1613552570343018,
|
|
"step": 385,
|
|
"token_acc": 0.6899038461538461
|
|
},
|
|
{
|
|
"epoch": 0.22626025791324736,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 9.177439057064684e-06,
|
|
"loss": 1.1961932182312012,
|
|
"step": 386,
|
|
"token_acc": 0.6807365439093485
|
|
},
|
|
{
|
|
"epoch": 0.2268464243845252,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 9.172103015931454e-06,
|
|
"loss": 1.1353485584259033,
|
|
"step": 387,
|
|
"token_acc": 0.6970039570378745
|
|
},
|
|
{
|
|
"epoch": 0.22743259085580306,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 9.166751284738258e-06,
|
|
"loss": 1.2647016048431396,
|
|
"step": 388,
|
|
"token_acc": 0.6758986446670595
|
|
},
|
|
{
|
|
"epoch": 0.22801875732708088,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 9.16138388361139e-06,
|
|
"loss": 1.1723607778549194,
|
|
"step": 389,
|
|
"token_acc": 0.6986911723753829
|
|
},
|
|
{
|
|
"epoch": 0.22860492379835873,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 9.156000832736073e-06,
|
|
"loss": 1.2202770709991455,
|
|
"step": 390,
|
|
"token_acc": 0.6735019973368842
|
|
},
|
|
{
|
|
"epoch": 0.22919109026963658,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 9.150602152356394e-06,
|
|
"loss": 1.2000805139541626,
|
|
"step": 391,
|
|
"token_acc": 0.6839222130923035
|
|
},
|
|
{
|
|
"epoch": 0.22977725674091443,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 9.145187862775208e-06,
|
|
"loss": 1.243488073348999,
|
|
"step": 392,
|
|
"token_acc": 0.6746210899709771
|
|
},
|
|
{
|
|
"epoch": 0.23036342321219228,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 9.139757984354079e-06,
|
|
"loss": 1.3046820163726807,
|
|
"step": 393,
|
|
"token_acc": 0.671971706454465
|
|
},
|
|
{
|
|
"epoch": 0.2309495896834701,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 9.134312537513188e-06,
|
|
"loss": 1.2032735347747803,
|
|
"step": 394,
|
|
"token_acc": 0.6842703009653606
|
|
},
|
|
{
|
|
"epoch": 0.23153575615474795,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 9.128851542731271e-06,
|
|
"loss": 1.146256446838379,
|
|
"step": 395,
|
|
"token_acc": 0.6974212034383954
|
|
},
|
|
{
|
|
"epoch": 0.2321219226260258,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 9.123375020545534e-06,
|
|
"loss": 1.1589492559432983,
|
|
"step": 396,
|
|
"token_acc": 0.6992481203007519
|
|
},
|
|
{
|
|
"epoch": 0.23270808909730364,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 9.117882991551576e-06,
|
|
"loss": 1.2052499055862427,
|
|
"step": 397,
|
|
"token_acc": 0.6853088480801336
|
|
},
|
|
{
|
|
"epoch": 0.23329425556858147,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 9.112375476403313e-06,
|
|
"loss": 1.2576318979263306,
|
|
"step": 398,
|
|
"token_acc": 0.6777356103731815
|
|
},
|
|
{
|
|
"epoch": 0.23388042203985931,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 9.106852495812897e-06,
|
|
"loss": 1.216879963874817,
|
|
"step": 399,
|
|
"token_acc": 0.680579855036241
|
|
},
|
|
{
|
|
"epoch": 0.23446658851113716,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 9.101314070550647e-06,
|
|
"loss": 1.1998136043548584,
|
|
"step": 400,
|
|
"token_acc": 0.6800656275635767
|
|
},
|
|
{
|
|
"epoch": 0.235052754982415,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 9.09576022144496e-06,
|
|
"loss": 1.2278746366500854,
|
|
"step": 401,
|
|
"token_acc": 0.6780995220691594
|
|
},
|
|
{
|
|
"epoch": 0.23563892145369286,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 9.09019096938224e-06,
|
|
"loss": 1.1651427745819092,
|
|
"step": 402,
|
|
"token_acc": 0.6936965473550839
|
|
},
|
|
{
|
|
"epoch": 0.23622508792497068,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 9.084606335306814e-06,
|
|
"loss": 1.187014102935791,
|
|
"step": 403,
|
|
"token_acc": 0.6868385345997287
|
|
},
|
|
{
|
|
"epoch": 0.23681125439624853,
|
|
"grad_norm": 3.53125,
|
|
"learning_rate": 9.079006340220862e-06,
|
|
"loss": 1.2268826961517334,
|
|
"step": 404,
|
|
"token_acc": 0.6794778825235678
|
|
},
|
|
{
|
|
"epoch": 0.23739742086752638,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 9.073391005184325e-06,
|
|
"loss": 1.2262380123138428,
|
|
"step": 405,
|
|
"token_acc": 0.68718441933157
|
|
},
|
|
{
|
|
"epoch": 0.23798358733880423,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 9.067760351314838e-06,
|
|
"loss": 1.2058700323104858,
|
|
"step": 406,
|
|
"token_acc": 0.6958424507658644
|
|
},
|
|
{
|
|
"epoch": 0.23856975381008205,
|
|
"grad_norm": 4.5,
|
|
"learning_rate": 9.062114399787648e-06,
|
|
"loss": 1.1918308734893799,
|
|
"step": 407,
|
|
"token_acc": 0.672209026128266
|
|
},
|
|
{
|
|
"epoch": 0.2391559202813599,
|
|
"grad_norm": 3.609375,
|
|
"learning_rate": 9.056453171835523e-06,
|
|
"loss": 1.1539957523345947,
|
|
"step": 408,
|
|
"token_acc": 0.700077101002313
|
|
},
|
|
{
|
|
"epoch": 0.23974208675263775,
|
|
"grad_norm": 3.53125,
|
|
"learning_rate": 9.050776688748687e-06,
|
|
"loss": 1.2668378353118896,
|
|
"step": 409,
|
|
"token_acc": 0.6701534170153417
|
|
},
|
|
{
|
|
"epoch": 0.2403282532239156,
|
|
"grad_norm": 3.4375,
|
|
"learning_rate": 9.045084971874738e-06,
|
|
"loss": 1.2269353866577148,
|
|
"step": 410,
|
|
"token_acc": 0.6875430144528561
|
|
},
|
|
{
|
|
"epoch": 0.24091441969519342,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 9.039378042618556e-06,
|
|
"loss": 1.229290246963501,
|
|
"step": 411,
|
|
"token_acc": 0.6905757478288839
|
|
},
|
|
{
|
|
"epoch": 0.24150058616647127,
|
|
"grad_norm": 3.609375,
|
|
"learning_rate": 9.033655922442235e-06,
|
|
"loss": 1.1830763816833496,
|
|
"step": 412,
|
|
"token_acc": 0.6923641703377387
|
|
},
|
|
{
|
|
"epoch": 0.24208675263774912,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 9.027918632864998e-06,
|
|
"loss": 1.1557458639144897,
|
|
"step": 413,
|
|
"token_acc": 0.7068888888888889
|
|
},
|
|
{
|
|
"epoch": 0.24267291910902697,
|
|
"grad_norm": 3.4375,
|
|
"learning_rate": 9.022166195463112e-06,
|
|
"loss": 1.268033504486084,
|
|
"step": 414,
|
|
"token_acc": 0.6747375187486608
|
|
},
|
|
{
|
|
"epoch": 0.24325908558030482,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 9.016398631869811e-06,
|
|
"loss": 1.189266562461853,
|
|
"step": 415,
|
|
"token_acc": 0.6883708467309754
|
|
},
|
|
{
|
|
"epoch": 0.24384525205158264,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 9.01061596377522e-06,
|
|
"loss": 1.1923234462738037,
|
|
"step": 416,
|
|
"token_acc": 0.6858316221765913
|
|
},
|
|
{
|
|
"epoch": 0.2444314185228605,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 9.004818212926261e-06,
|
|
"loss": 1.2292840480804443,
|
|
"step": 417,
|
|
"token_acc": 0.6883509833585476
|
|
},
|
|
{
|
|
"epoch": 0.24501758499413834,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 8.99900540112658e-06,
|
|
"loss": 1.193171501159668,
|
|
"step": 418,
|
|
"token_acc": 0.6781638346436724
|
|
},
|
|
{
|
|
"epoch": 0.2456037514654162,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 8.993177550236464e-06,
|
|
"loss": 1.1733283996582031,
|
|
"step": 419,
|
|
"token_acc": 0.6961883408071748
|
|
},
|
|
{
|
|
"epoch": 0.246189917936694,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 8.987334682172759e-06,
|
|
"loss": 1.1982325315475464,
|
|
"step": 420,
|
|
"token_acc": 0.6913888888888889
|
|
},
|
|
{
|
|
"epoch": 0.24677608440797186,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 8.981476818908778e-06,
|
|
"loss": 1.2011311054229736,
|
|
"step": 421,
|
|
"token_acc": 0.679657269209508
|
|
},
|
|
{
|
|
"epoch": 0.2473622508792497,
|
|
"grad_norm": 3.53125,
|
|
"learning_rate": 8.97560398247424e-06,
|
|
"loss": 1.1189249753952026,
|
|
"step": 422,
|
|
"token_acc": 0.7039079357468233
|
|
},
|
|
{
|
|
"epoch": 0.24794841735052756,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 8.96971619495516e-06,
|
|
"loss": 1.1869139671325684,
|
|
"step": 423,
|
|
"token_acc": 0.6923990498812351
|
|
},
|
|
{
|
|
"epoch": 0.2485345838218054,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 8.963813478493788e-06,
|
|
"loss": 1.2303062677383423,
|
|
"step": 424,
|
|
"token_acc": 0.6837324525185797
|
|
},
|
|
{
|
|
"epoch": 0.24912075029308323,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 8.957895855288517e-06,
|
|
"loss": 1.1716539859771729,
|
|
"step": 425,
|
|
"token_acc": 0.690961057520543
|
|
},
|
|
{
|
|
"epoch": 0.24970691676436108,
|
|
"grad_norm": 3.421875,
|
|
"learning_rate": 8.951963347593797e-06,
|
|
"loss": 1.257789134979248,
|
|
"step": 426,
|
|
"token_acc": 0.6734600412182277
|
|
},
|
|
{
|
|
"epoch": 0.2502930832356389,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 8.946015977720054e-06,
|
|
"loss": 1.255378246307373,
|
|
"step": 427,
|
|
"token_acc": 0.6767241379310345
|
|
},
|
|
{
|
|
"epoch": 0.2508792497069168,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 8.94005376803361e-06,
|
|
"loss": 1.2041829824447632,
|
|
"step": 428,
|
|
"token_acc": 0.6740576496674058
|
|
},
|
|
{
|
|
"epoch": 0.2514654161781946,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 8.934076740956593e-06,
|
|
"loss": 1.31052565574646,
|
|
"step": 429,
|
|
"token_acc": 0.6703867116532573
|
|
},
|
|
{
|
|
"epoch": 0.25205158264947247,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 8.92808491896685e-06,
|
|
"loss": 1.2253599166870117,
|
|
"step": 430,
|
|
"token_acc": 0.6732747804265997
|
|
},
|
|
{
|
|
"epoch": 0.25263774912075027,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 8.92207832459788e-06,
|
|
"loss": 1.2086076736450195,
|
|
"step": 431,
|
|
"token_acc": 0.6790941728804197
|
|
},
|
|
{
|
|
"epoch": 0.2532239155920281,
|
|
"grad_norm": 3.546875,
|
|
"learning_rate": 8.916056980438723e-06,
|
|
"loss": 1.2373461723327637,
|
|
"step": 432,
|
|
"token_acc": 0.6811862835959221
|
|
},
|
|
{
|
|
"epoch": 0.25381008206330596,
|
|
"grad_norm": 4.28125,
|
|
"learning_rate": 8.910020909133894e-06,
|
|
"loss": 1.153116226196289,
|
|
"step": 433,
|
|
"token_acc": 0.6868395773294909
|
|
},
|
|
{
|
|
"epoch": 0.2543962485345838,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 8.903970133383297e-06,
|
|
"loss": 1.2406597137451172,
|
|
"step": 434,
|
|
"token_acc": 0.6775059665871122
|
|
},
|
|
{
|
|
"epoch": 0.25498241500586166,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 8.897904675942129e-06,
|
|
"loss": 1.251798152923584,
|
|
"step": 435,
|
|
"token_acc": 0.677038626609442
|
|
},
|
|
{
|
|
"epoch": 0.2555685814771395,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 8.891824559620801e-06,
|
|
"loss": 1.1965774297714233,
|
|
"step": 436,
|
|
"token_acc": 0.6886157448053848
|
|
},
|
|
{
|
|
"epoch": 0.25615474794841736,
|
|
"grad_norm": 3.40625,
|
|
"learning_rate": 8.885729807284855e-06,
|
|
"loss": 1.2742735147476196,
|
|
"step": 437,
|
|
"token_acc": 0.6731905465288035
|
|
},
|
|
{
|
|
"epoch": 0.2567409144196952,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 8.879620441854873e-06,
|
|
"loss": 1.1572833061218262,
|
|
"step": 438,
|
|
"token_acc": 0.6976314980006152
|
|
},
|
|
{
|
|
"epoch": 0.25732708089097306,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 8.873496486306394e-06,
|
|
"loss": 1.241819143295288,
|
|
"step": 439,
|
|
"token_acc": 0.683684794672586
|
|
},
|
|
{
|
|
"epoch": 0.25791324736225085,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 8.867357963669821e-06,
|
|
"loss": 1.1480334997177124,
|
|
"step": 440,
|
|
"token_acc": 0.6859391965255157
|
|
},
|
|
{
|
|
"epoch": 0.2584994138335287,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 8.861204897030346e-06,
|
|
"loss": 1.198982834815979,
|
|
"step": 441,
|
|
"token_acc": 0.6882352941176471
|
|
},
|
|
{
|
|
"epoch": 0.25908558030480655,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 8.855037309527854e-06,
|
|
"loss": 1.1977558135986328,
|
|
"step": 442,
|
|
"token_acc": 0.6731892523364486
|
|
},
|
|
{
|
|
"epoch": 0.2596717467760844,
|
|
"grad_norm": 4.34375,
|
|
"learning_rate": 8.84885522435684e-06,
|
|
"loss": 1.258596420288086,
|
|
"step": 443,
|
|
"token_acc": 0.6750614250614251
|
|
},
|
|
{
|
|
"epoch": 0.26025791324736225,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 8.842658664766317e-06,
|
|
"loss": 1.187192440032959,
|
|
"step": 444,
|
|
"token_acc": 0.6927657040785948
|
|
},
|
|
{
|
|
"epoch": 0.2608440797186401,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 8.836447654059734e-06,
|
|
"loss": 1.2215509414672852,
|
|
"step": 445,
|
|
"token_acc": 0.686533595726736
|
|
},
|
|
{
|
|
"epoch": 0.26143024618991795,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 8.83022221559489e-06,
|
|
"loss": 1.1613328456878662,
|
|
"step": 446,
|
|
"token_acc": 0.6940229173118613
|
|
},
|
|
{
|
|
"epoch": 0.2620164126611958,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 8.823982372783838e-06,
|
|
"loss": 1.2427836656570435,
|
|
"step": 447,
|
|
"token_acc": 0.6842832469775475
|
|
},
|
|
{
|
|
"epoch": 0.26260257913247365,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 8.817728149092803e-06,
|
|
"loss": 1.1736946105957031,
|
|
"step": 448,
|
|
"token_acc": 0.6882090065314541
|
|
},
|
|
{
|
|
"epoch": 0.26318874560375144,
|
|
"grad_norm": 3.609375,
|
|
"learning_rate": 8.811459568042092e-06,
|
|
"loss": 1.2839813232421875,
|
|
"step": 449,
|
|
"token_acc": 0.6703485517918507
|
|
},
|
|
{
|
|
"epoch": 0.2637749120750293,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 8.805176653206004e-06,
|
|
"loss": 1.1547906398773193,
|
|
"step": 450,
|
|
"token_acc": 0.6950431034482759
|
|
},
|
|
{
|
|
"epoch": 0.26436107854630714,
|
|
"grad_norm": 3.921875,
|
|
"learning_rate": 8.798879428212748e-06,
|
|
"loss": 1.1829626560211182,
|
|
"step": 451,
|
|
"token_acc": 0.6831108531630876
|
|
},
|
|
{
|
|
"epoch": 0.264947245017585,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 8.792567916744346e-06,
|
|
"loss": 1.1935487985610962,
|
|
"step": 452,
|
|
"token_acc": 0.6887096774193548
|
|
},
|
|
{
|
|
"epoch": 0.26553341148886284,
|
|
"grad_norm": 3.375,
|
|
"learning_rate": 8.786242142536547e-06,
|
|
"loss": 1.2206745147705078,
|
|
"step": 453,
|
|
"token_acc": 0.6790870982766651
|
|
},
|
|
{
|
|
"epoch": 0.2661195779601407,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 8.77990212937874e-06,
|
|
"loss": 1.2869253158569336,
|
|
"step": 454,
|
|
"token_acc": 0.6783009548896938
|
|
},
|
|
{
|
|
"epoch": 0.26670574443141853,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 8.773547901113862e-06,
|
|
"loss": 1.1881251335144043,
|
|
"step": 455,
|
|
"token_acc": 0.6803966437833715
|
|
},
|
|
{
|
|
"epoch": 0.2672919109026964,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 8.767179481638303e-06,
|
|
"loss": 1.2581734657287598,
|
|
"step": 456,
|
|
"token_acc": 0.6635363563187551
|
|
},
|
|
{
|
|
"epoch": 0.26787807737397423,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 8.760796894901835e-06,
|
|
"loss": 1.2150182723999023,
|
|
"step": 457,
|
|
"token_acc": 0.6939182452642074
|
|
},
|
|
{
|
|
"epoch": 0.268464243845252,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 8.754400164907496e-06,
|
|
"loss": 1.1261703968048096,
|
|
"step": 458,
|
|
"token_acc": 0.7026841018582244
|
|
},
|
|
{
|
|
"epoch": 0.2690504103165299,
|
|
"grad_norm": 3.3125,
|
|
"learning_rate": 8.74798931571152e-06,
|
|
"loss": 1.2272446155548096,
|
|
"step": 459,
|
|
"token_acc": 0.6754567715200602
|
|
},
|
|
{
|
|
"epoch": 0.2696365767878077,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 8.741564371423235e-06,
|
|
"loss": 1.1877363920211792,
|
|
"step": 460,
|
|
"token_acc": 0.6891805377720871
|
|
},
|
|
{
|
|
"epoch": 0.2702227432590856,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 8.735125356204982e-06,
|
|
"loss": 1.2438206672668457,
|
|
"step": 461,
|
|
"token_acc": 0.6767106089139987
|
|
},
|
|
{
|
|
"epoch": 0.2708089097303634,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 8.728672294272009e-06,
|
|
"loss": 1.197620153427124,
|
|
"step": 462,
|
|
"token_acc": 0.6902310047314222
|
|
},
|
|
{
|
|
"epoch": 0.27139507620164127,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 8.722205209892402e-06,
|
|
"loss": 1.2284767627716064,
|
|
"step": 463,
|
|
"token_acc": 0.6844996417482685
|
|
},
|
|
{
|
|
"epoch": 0.2719812426729191,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 8.715724127386971e-06,
|
|
"loss": 1.2122802734375,
|
|
"step": 464,
|
|
"token_acc": 0.682092004381161
|
|
},
|
|
{
|
|
"epoch": 0.27256740914419697,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 8.709229071129177e-06,
|
|
"loss": 1.1509402990341187,
|
|
"step": 465,
|
|
"token_acc": 0.6953821226032946
|
|
},
|
|
{
|
|
"epoch": 0.2731535756154748,
|
|
"grad_norm": 4.34375,
|
|
"learning_rate": 8.702720065545024e-06,
|
|
"loss": 1.1547797918319702,
|
|
"step": 466,
|
|
"token_acc": 0.6952758745041472
|
|
},
|
|
{
|
|
"epoch": 0.2737397420867526,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 8.69619713511298e-06,
|
|
"loss": 1.1942332983016968,
|
|
"step": 467,
|
|
"token_acc": 0.683878055479264
|
|
},
|
|
{
|
|
"epoch": 0.27432590855803046,
|
|
"grad_norm": 3.484375,
|
|
"learning_rate": 8.689660304363883e-06,
|
|
"loss": 1.1988351345062256,
|
|
"step": 468,
|
|
"token_acc": 0.6864235055724417
|
|
},
|
|
{
|
|
"epoch": 0.2749120750293083,
|
|
"grad_norm": 3.53125,
|
|
"learning_rate": 8.683109597880841e-06,
|
|
"loss": 1.1634159088134766,
|
|
"step": 469,
|
|
"token_acc": 0.69258872651357
|
|
},
|
|
{
|
|
"epoch": 0.27549824150058616,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 8.676545040299145e-06,
|
|
"loss": 1.1974842548370361,
|
|
"step": 470,
|
|
"token_acc": 0.6897343391275829
|
|
},
|
|
{
|
|
"epoch": 0.276084407971864,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 8.669966656306176e-06,
|
|
"loss": 1.1809864044189453,
|
|
"step": 471,
|
|
"token_acc": 0.6887577139790716
|
|
},
|
|
{
|
|
"epoch": 0.27667057444314186,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 8.663374470641319e-06,
|
|
"loss": 1.192874789237976,
|
|
"step": 472,
|
|
"token_acc": 0.6830282861896838
|
|
},
|
|
{
|
|
"epoch": 0.2772567409144197,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 8.656768508095853e-06,
|
|
"loss": 1.200680136680603,
|
|
"step": 473,
|
|
"token_acc": 0.6917786808377618
|
|
},
|
|
{
|
|
"epoch": 0.27784290738569756,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 8.650148793512874e-06,
|
|
"loss": 1.1095136404037476,
|
|
"step": 474,
|
|
"token_acc": 0.7019449355897954
|
|
},
|
|
{
|
|
"epoch": 0.2784290738569754,
|
|
"grad_norm": 4.34375,
|
|
"learning_rate": 8.643515351787193e-06,
|
|
"loss": 1.2887043952941895,
|
|
"step": 475,
|
|
"token_acc": 0.6745398773006135
|
|
},
|
|
{
|
|
"epoch": 0.2790152403282532,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 8.636868207865244e-06,
|
|
"loss": 1.2056461572647095,
|
|
"step": 476,
|
|
"token_acc": 0.6830427892234548
|
|
},
|
|
{
|
|
"epoch": 0.27960140679953105,
|
|
"grad_norm": 3.453125,
|
|
"learning_rate": 8.630207386744995e-06,
|
|
"loss": 1.2486038208007812,
|
|
"step": 477,
|
|
"token_acc": 0.6657072368421053
|
|
},
|
|
{
|
|
"epoch": 0.2801875732708089,
|
|
"grad_norm": 3.609375,
|
|
"learning_rate": 8.623532913475847e-06,
|
|
"loss": 1.1927520036697388,
|
|
"step": 478,
|
|
"token_acc": 0.6937062937062937
|
|
},
|
|
{
|
|
"epoch": 0.28077373974208675,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 8.61684481315854e-06,
|
|
"loss": 1.1822110414505005,
|
|
"step": 479,
|
|
"token_acc": 0.6840934371523916
|
|
},
|
|
{
|
|
"epoch": 0.2813599062133646,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 8.610143110945068e-06,
|
|
"loss": 1.1741970777511597,
|
|
"step": 480,
|
|
"token_acc": 0.6892655367231638
|
|
},
|
|
{
|
|
"epoch": 0.28194607268464245,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 8.603427832038574e-06,
|
|
"loss": 1.218209981918335,
|
|
"step": 481,
|
|
"token_acc": 0.6790915823253517
|
|
},
|
|
{
|
|
"epoch": 0.2825322391559203,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 8.596699001693257e-06,
|
|
"loss": 1.1537561416625977,
|
|
"step": 482,
|
|
"token_acc": 0.6912165848336062
|
|
},
|
|
{
|
|
"epoch": 0.28311840562719814,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 8.589956645214282e-06,
|
|
"loss": 1.105603814125061,
|
|
"step": 483,
|
|
"token_acc": 0.7088912694161756
|
|
},
|
|
{
|
|
"epoch": 0.283704572098476,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 8.58320078795768e-06,
|
|
"loss": 1.2241215705871582,
|
|
"step": 484,
|
|
"token_acc": 0.6827871460388164
|
|
},
|
|
{
|
|
"epoch": 0.2842907385697538,
|
|
"grad_norm": 4.375,
|
|
"learning_rate": 8.576431455330258e-06,
|
|
"loss": 1.2244592905044556,
|
|
"step": 485,
|
|
"token_acc": 0.688912429378531
|
|
},
|
|
{
|
|
"epoch": 0.28487690504103164,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 8.569648672789496e-06,
|
|
"loss": 1.1658697128295898,
|
|
"step": 486,
|
|
"token_acc": 0.6895306859205776
|
|
},
|
|
{
|
|
"epoch": 0.2854630715123095,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 8.562852465843456e-06,
|
|
"loss": 1.2220423221588135,
|
|
"step": 487,
|
|
"token_acc": 0.6846473029045643
|
|
},
|
|
{
|
|
"epoch": 0.28604923798358733,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 8.556042860050686e-06,
|
|
"loss": 1.2313519716262817,
|
|
"step": 488,
|
|
"token_acc": 0.6777158774373259
|
|
},
|
|
{
|
|
"epoch": 0.2866354044548652,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 8.549219881020125e-06,
|
|
"loss": 1.1571648120880127,
|
|
"step": 489,
|
|
"token_acc": 0.6952654232424678
|
|
},
|
|
{
|
|
"epoch": 0.28722157092614303,
|
|
"grad_norm": 3.484375,
|
|
"learning_rate": 8.542383554411e-06,
|
|
"loss": 1.2273643016815186,
|
|
"step": 490,
|
|
"token_acc": 0.6787434074753497
|
|
},
|
|
{
|
|
"epoch": 0.2878077373974209,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 8.535533905932739e-06,
|
|
"loss": 1.140744924545288,
|
|
"step": 491,
|
|
"token_acc": 0.6947739660147483
|
|
},
|
|
{
|
|
"epoch": 0.28839390386869873,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 8.528670961344866e-06,
|
|
"loss": 1.2256813049316406,
|
|
"step": 492,
|
|
"token_acc": 0.6813186813186813
|
|
},
|
|
{
|
|
"epoch": 0.2889800703399766,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 8.521794746456911e-06,
|
|
"loss": 1.1442461013793945,
|
|
"step": 493,
|
|
"token_acc": 0.6958023379383634
|
|
},
|
|
{
|
|
"epoch": 0.2895662368112544,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 8.51490528712831e-06,
|
|
"loss": 1.1973885297775269,
|
|
"step": 494,
|
|
"token_acc": 0.6890149561629706
|
|
},
|
|
{
|
|
"epoch": 0.2901524032825322,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 8.508002609268303e-06,
|
|
"loss": 1.162795066833496,
|
|
"step": 495,
|
|
"token_acc": 0.6994328922495274
|
|
},
|
|
{
|
|
"epoch": 0.29073856975381007,
|
|
"grad_norm": 3.375,
|
|
"learning_rate": 8.501086738835843e-06,
|
|
"loss": 1.1965055465698242,
|
|
"step": 496,
|
|
"token_acc": 0.6913414087034545
|
|
},
|
|
{
|
|
"epoch": 0.2913247362250879,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 8.4941577018395e-06,
|
|
"loss": 1.1787352561950684,
|
|
"step": 497,
|
|
"token_acc": 0.6934579439252336
|
|
},
|
|
{
|
|
"epoch": 0.29191090269636577,
|
|
"grad_norm": 3.609375,
|
|
"learning_rate": 8.487215524337357e-06,
|
|
"loss": 1.1996502876281738,
|
|
"step": 498,
|
|
"token_acc": 0.685131195335277
|
|
},
|
|
{
|
|
"epoch": 0.2924970691676436,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 8.480260232436913e-06,
|
|
"loss": 1.2471606731414795,
|
|
"step": 499,
|
|
"token_acc": 0.6692935547424711
|
|
},
|
|
{
|
|
"epoch": 0.29308323563892147,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 8.473291852294986e-06,
|
|
"loss": 1.1648485660552979,
|
|
"step": 500,
|
|
"token_acc": 0.688504326328801
|
|
},
|
|
{
|
|
"epoch": 0.2936694021101993,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 8.466310410117622e-06,
|
|
"loss": 1.238483190536499,
|
|
"step": 501,
|
|
"token_acc": 0.6823683349689238
|
|
},
|
|
{
|
|
"epoch": 0.29425556858147717,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 8.45931593215998e-06,
|
|
"loss": 1.2317476272583008,
|
|
"step": 502,
|
|
"token_acc": 0.6769776876267748
|
|
},
|
|
{
|
|
"epoch": 0.29484173505275496,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 8.452308444726249e-06,
|
|
"loss": 1.1453964710235596,
|
|
"step": 503,
|
|
"token_acc": 0.6945071482317532
|
|
},
|
|
{
|
|
"epoch": 0.2954279015240328,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 8.44528797416954e-06,
|
|
"loss": 1.1846725940704346,
|
|
"step": 504,
|
|
"token_acc": 0.6827725050330745
|
|
},
|
|
{
|
|
"epoch": 0.29601406799531066,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 8.438254546891792e-06,
|
|
"loss": 1.192009687423706,
|
|
"step": 505,
|
|
"token_acc": 0.6925739458779107
|
|
},
|
|
{
|
|
"epoch": 0.2966002344665885,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 8.43120818934367e-06,
|
|
"loss": 1.175542950630188,
|
|
"step": 506,
|
|
"token_acc": 0.6913193356314635
|
|
},
|
|
{
|
|
"epoch": 0.29718640093786636,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 8.42414892802446e-06,
|
|
"loss": 1.2578275203704834,
|
|
"step": 507,
|
|
"token_acc": 0.6804809414172422
|
|
},
|
|
{
|
|
"epoch": 0.2977725674091442,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 8.417076789481985e-06,
|
|
"loss": 1.2627696990966797,
|
|
"step": 508,
|
|
"token_acc": 0.6876106194690266
|
|
},
|
|
{
|
|
"epoch": 0.29835873388042206,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 8.409991800312493e-06,
|
|
"loss": 1.1767578125,
|
|
"step": 509,
|
|
"token_acc": 0.6856607811663992
|
|
},
|
|
{
|
|
"epoch": 0.2989449003516999,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 8.402893987160553e-06,
|
|
"loss": 1.2128772735595703,
|
|
"step": 510,
|
|
"token_acc": 0.6803455723542117
|
|
},
|
|
{
|
|
"epoch": 0.2995310668229777,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 8.395783376718967e-06,
|
|
"loss": 1.2272933721542358,
|
|
"step": 511,
|
|
"token_acc": 0.6938343477110347
|
|
},
|
|
{
|
|
"epoch": 0.30011723329425555,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 8.388659995728662e-06,
|
|
"loss": 1.1831417083740234,
|
|
"step": 512,
|
|
"token_acc": 0.6992022524636321
|
|
},
|
|
{
|
|
"epoch": 0.3007033997655334,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 8.381523870978593e-06,
|
|
"loss": 1.266658067703247,
|
|
"step": 513,
|
|
"token_acc": 0.6742744435052127
|
|
},
|
|
{
|
|
"epoch": 0.30128956623681125,
|
|
"grad_norm": 3.5,
|
|
"learning_rate": 8.37437502930564e-06,
|
|
"loss": 1.235410451889038,
|
|
"step": 514,
|
|
"token_acc": 0.6758241758241759
|
|
},
|
|
{
|
|
"epoch": 0.3018757327080891,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 8.367213497594501e-06,
|
|
"loss": 1.1848678588867188,
|
|
"step": 515,
|
|
"token_acc": 0.6943532501641497
|
|
},
|
|
{
|
|
"epoch": 0.30246189917936694,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 8.360039302777614e-06,
|
|
"loss": 1.1902058124542236,
|
|
"step": 516,
|
|
"token_acc": 0.682024659312135
|
|
},
|
|
{
|
|
"epoch": 0.3030480656506448,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 8.352852471835019e-06,
|
|
"loss": 1.230810284614563,
|
|
"step": 517,
|
|
"token_acc": 0.6829590488771466
|
|
},
|
|
{
|
|
"epoch": 0.30363423212192264,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 8.345653031794292e-06,
|
|
"loss": 1.261621117591858,
|
|
"step": 518,
|
|
"token_acc": 0.6762868193989884
|
|
},
|
|
{
|
|
"epoch": 0.3042203985932005,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 8.33844100973042e-06,
|
|
"loss": 1.0876890420913696,
|
|
"step": 519,
|
|
"token_acc": 0.707958707958708
|
|
},
|
|
{
|
|
"epoch": 0.3048065650644783,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 8.331216432765714e-06,
|
|
"loss": 1.2490644454956055,
|
|
"step": 520,
|
|
"token_acc": 0.6702758442897654
|
|
},
|
|
{
|
|
"epoch": 0.30539273153575613,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 8.323979328069689e-06,
|
|
"loss": 1.2038012742996216,
|
|
"step": 521,
|
|
"token_acc": 0.6754874651810585
|
|
},
|
|
{
|
|
"epoch": 0.305978898007034,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 8.316729722858987e-06,
|
|
"loss": 1.2012929916381836,
|
|
"step": 522,
|
|
"token_acc": 0.6823188405797102
|
|
},
|
|
{
|
|
"epoch": 0.30656506447831183,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 8.309467644397254e-06,
|
|
"loss": 1.2452181577682495,
|
|
"step": 523,
|
|
"token_acc": 0.6724492125085597
|
|
},
|
|
{
|
|
"epoch": 0.3071512309495897,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 8.302193119995038e-06,
|
|
"loss": 1.1812323331832886,
|
|
"step": 524,
|
|
"token_acc": 0.6911377595788242
|
|
},
|
|
{
|
|
"epoch": 0.30773739742086753,
|
|
"grad_norm": 3.46875,
|
|
"learning_rate": 8.294906177009707e-06,
|
|
"loss": 1.194439172744751,
|
|
"step": 525,
|
|
"token_acc": 0.6825670027046964
|
|
},
|
|
{
|
|
"epoch": 0.3083235638921454,
|
|
"grad_norm": 3.46875,
|
|
"learning_rate": 8.28760684284532e-06,
|
|
"loss": 1.1675419807434082,
|
|
"step": 526,
|
|
"token_acc": 0.692021803766105
|
|
},
|
|
{
|
|
"epoch": 0.30890973036342323,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 8.280295144952537e-06,
|
|
"loss": 1.2324539422988892,
|
|
"step": 527,
|
|
"token_acc": 0.6846823095175323
|
|
},
|
|
{
|
|
"epoch": 0.3094958968347011,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 8.272971110828521e-06,
|
|
"loss": 1.163048505783081,
|
|
"step": 528,
|
|
"token_acc": 0.689242053789731
|
|
},
|
|
{
|
|
"epoch": 0.31008206330597887,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 8.265634768016819e-06,
|
|
"loss": 1.1591242551803589,
|
|
"step": 529,
|
|
"token_acc": 0.7027761013880507
|
|
},
|
|
{
|
|
"epoch": 0.3106682297772567,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 8.258286144107277e-06,
|
|
"loss": 1.1612153053283691,
|
|
"step": 530,
|
|
"token_acc": 0.6879377431906615
|
|
},
|
|
{
|
|
"epoch": 0.31125439624853457,
|
|
"grad_norm": 3.609375,
|
|
"learning_rate": 8.25092526673592e-06,
|
|
"loss": 1.245072603225708,
|
|
"step": 531,
|
|
"token_acc": 0.6805283757338552
|
|
},
|
|
{
|
|
"epoch": 0.3118405627198124,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 8.243552163584851e-06,
|
|
"loss": 1.2727348804473877,
|
|
"step": 532,
|
|
"token_acc": 0.6685190330647403
|
|
},
|
|
{
|
|
"epoch": 0.31242672919109027,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 8.236166862382163e-06,
|
|
"loss": 1.1164054870605469,
|
|
"step": 533,
|
|
"token_acc": 0.6987363584147042
|
|
},
|
|
{
|
|
"epoch": 0.3130128956623681,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 8.228769390901812e-06,
|
|
"loss": 1.2008144855499268,
|
|
"step": 534,
|
|
"token_acc": 0.6921385376580539
|
|
},
|
|
{
|
|
"epoch": 0.31359906213364597,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 8.221359776963526e-06,
|
|
"loss": 1.1829125881195068,
|
|
"step": 535,
|
|
"token_acc": 0.6931754089114496
|
|
},
|
|
{
|
|
"epoch": 0.3141852286049238,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 8.213938048432697e-06,
|
|
"loss": 1.2105216979980469,
|
|
"step": 536,
|
|
"token_acc": 0.6875973015049299
|
|
},
|
|
{
|
|
"epoch": 0.31477139507620167,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 8.206504233220277e-06,
|
|
"loss": 1.175944447517395,
|
|
"step": 537,
|
|
"token_acc": 0.6885141584473433
|
|
},
|
|
{
|
|
"epoch": 0.31535756154747946,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 8.199058359282675e-06,
|
|
"loss": 1.1936707496643066,
|
|
"step": 538,
|
|
"token_acc": 0.679979648944289
|
|
},
|
|
{
|
|
"epoch": 0.3159437280187573,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 8.191600454621642e-06,
|
|
"loss": 1.215691328048706,
|
|
"step": 539,
|
|
"token_acc": 0.6859525012632642
|
|
},
|
|
{
|
|
"epoch": 0.31652989449003516,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 8.18413054728418e-06,
|
|
"loss": 1.2238906621932983,
|
|
"step": 540,
|
|
"token_acc": 0.6776556776556777
|
|
},
|
|
{
|
|
"epoch": 0.317116060961313,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 8.176648665362426e-06,
|
|
"loss": 1.198019027709961,
|
|
"step": 541,
|
|
"token_acc": 0.676209561981105
|
|
},
|
|
{
|
|
"epoch": 0.31770222743259086,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 8.16915483699355e-06,
|
|
"loss": 1.186300277709961,
|
|
"step": 542,
|
|
"token_acc": 0.696979865771812
|
|
},
|
|
{
|
|
"epoch": 0.3182883939038687,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 8.161649090359656e-06,
|
|
"loss": 1.1815621852874756,
|
|
"step": 543,
|
|
"token_acc": 0.6916264090177133
|
|
},
|
|
{
|
|
"epoch": 0.31887456037514655,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 8.154131453687657e-06,
|
|
"loss": 1.2640305757522583,
|
|
"step": 544,
|
|
"token_acc": 0.6712882726747544
|
|
},
|
|
{
|
|
"epoch": 0.3194607268464244,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 8.146601955249187e-06,
|
|
"loss": 1.1560747623443604,
|
|
"step": 545,
|
|
"token_acc": 0.6967651195499297
|
|
},
|
|
{
|
|
"epoch": 0.32004689331770225,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 8.139060623360494e-06,
|
|
"loss": 1.2553794384002686,
|
|
"step": 546,
|
|
"token_acc": 0.6701735928458706
|
|
},
|
|
{
|
|
"epoch": 0.32063305978898005,
|
|
"grad_norm": 3.53125,
|
|
"learning_rate": 8.131507486382318e-06,
|
|
"loss": 1.2302329540252686,
|
|
"step": 547,
|
|
"token_acc": 0.6804848484848485
|
|
},
|
|
{
|
|
"epoch": 0.3212192262602579,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 8.123942572719801e-06,
|
|
"loss": 1.1851441860198975,
|
|
"step": 548,
|
|
"token_acc": 0.6886577010910936
|
|
},
|
|
{
|
|
"epoch": 0.32180539273153574,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 8.116365910822373e-06,
|
|
"loss": 1.1850552558898926,
|
|
"step": 549,
|
|
"token_acc": 0.685428392745466
|
|
},
|
|
{
|
|
"epoch": 0.3223915592028136,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 8.108777529183644e-06,
|
|
"loss": 1.2002326250076294,
|
|
"step": 550,
|
|
"token_acc": 0.6909894969596462
|
|
},
|
|
{
|
|
"epoch": 0.32297772567409144,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 8.101177456341301e-06,
|
|
"loss": 1.1915640830993652,
|
|
"step": 551,
|
|
"token_acc": 0.6891068115557911
|
|
},
|
|
{
|
|
"epoch": 0.3235638921453693,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 8.093565720876994e-06,
|
|
"loss": 1.2914780378341675,
|
|
"step": 552,
|
|
"token_acc": 0.6660929432013769
|
|
},
|
|
{
|
|
"epoch": 0.32415005861664714,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 8.085942351416236e-06,
|
|
"loss": 1.257450819015503,
|
|
"step": 553,
|
|
"token_acc": 0.669185282522996
|
|
},
|
|
{
|
|
"epoch": 0.324736225087925,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 8.078307376628292e-06,
|
|
"loss": 1.1996419429779053,
|
|
"step": 554,
|
|
"token_acc": 0.6876166770379589
|
|
},
|
|
{
|
|
"epoch": 0.32532239155920284,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 8.07066082522607e-06,
|
|
"loss": 1.2556778192520142,
|
|
"step": 555,
|
|
"token_acc": 0.6730665224445647
|
|
},
|
|
{
|
|
"epoch": 0.32590855803048063,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 8.063002725966014e-06,
|
|
"loss": 1.2518224716186523,
|
|
"step": 556,
|
|
"token_acc": 0.6699523052464229
|
|
},
|
|
{
|
|
"epoch": 0.3264947245017585,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 8.055333107648e-06,
|
|
"loss": 1.1040608882904053,
|
|
"step": 557,
|
|
"token_acc": 0.6992506811989101
|
|
},
|
|
{
|
|
"epoch": 0.32708089097303633,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 8.047651999115216e-06,
|
|
"loss": 1.1599043607711792,
|
|
"step": 558,
|
|
"token_acc": 0.6981807336713391
|
|
},
|
|
{
|
|
"epoch": 0.3276670574443142,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 8.039959429254072e-06,
|
|
"loss": 1.193272352218628,
|
|
"step": 559,
|
|
"token_acc": 0.6833467850417003
|
|
},
|
|
{
|
|
"epoch": 0.32825322391559203,
|
|
"grad_norm": 4.5,
|
|
"learning_rate": 8.032255426994069e-06,
|
|
"loss": 1.293489933013916,
|
|
"step": 560,
|
|
"token_acc": 0.666893039049236
|
|
},
|
|
{
|
|
"epoch": 0.3288393903868699,
|
|
"grad_norm": 3.5,
|
|
"learning_rate": 8.024540021307709e-06,
|
|
"loss": 1.1751644611358643,
|
|
"step": 561,
|
|
"token_acc": 0.6870892018779343
|
|
},
|
|
{
|
|
"epoch": 0.32942555685814773,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 8.01681324121038e-06,
|
|
"loss": 1.1604390144348145,
|
|
"step": 562,
|
|
"token_acc": 0.6893551688843398
|
|
},
|
|
{
|
|
"epoch": 0.3300117233294256,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 8.009075115760243e-06,
|
|
"loss": 1.2107281684875488,
|
|
"step": 563,
|
|
"token_acc": 0.6839237057220708
|
|
},
|
|
{
|
|
"epoch": 0.3305978898007034,
|
|
"grad_norm": 3.4375,
|
|
"learning_rate": 8.001325674058124e-06,
|
|
"loss": 1.2474877834320068,
|
|
"step": 564,
|
|
"token_acc": 0.6790211451651224
|
|
},
|
|
{
|
|
"epoch": 0.3311840562719812,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 7.99356494524741e-06,
|
|
"loss": 1.1715588569641113,
|
|
"step": 565,
|
|
"token_acc": 0.6852622814321399
|
|
},
|
|
{
|
|
"epoch": 0.33177022274325907,
|
|
"grad_norm": 4.3125,
|
|
"learning_rate": 7.985792958513932e-06,
|
|
"loss": 1.2332148551940918,
|
|
"step": 566,
|
|
"token_acc": 0.6862404447533009
|
|
},
|
|
{
|
|
"epoch": 0.3323563892145369,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 7.978009743085862e-06,
|
|
"loss": 1.152860164642334,
|
|
"step": 567,
|
|
"token_acc": 0.6994015233949945
|
|
},
|
|
{
|
|
"epoch": 0.33294255568581477,
|
|
"grad_norm": 3.921875,
|
|
"learning_rate": 7.970215328233597e-06,
|
|
"loss": 1.1638327836990356,
|
|
"step": 568,
|
|
"token_acc": 0.6922651933701658
|
|
},
|
|
{
|
|
"epoch": 0.3335287221570926,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 7.962409743269654e-06,
|
|
"loss": 1.1895761489868164,
|
|
"step": 569,
|
|
"token_acc": 0.6933744221879815
|
|
},
|
|
{
|
|
"epoch": 0.33411488862837047,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 7.954593017548557e-06,
|
|
"loss": 1.2543402910232544,
|
|
"step": 570,
|
|
"token_acc": 0.6688860435339309
|
|
},
|
|
{
|
|
"epoch": 0.3347010550996483,
|
|
"grad_norm": 3.546875,
|
|
"learning_rate": 7.946765180466725e-06,
|
|
"loss": 1.1724731922149658,
|
|
"step": 571,
|
|
"token_acc": 0.6876626756897449
|
|
},
|
|
{
|
|
"epoch": 0.33528722157092616,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 7.938926261462366e-06,
|
|
"loss": 1.2718970775604248,
|
|
"step": 572,
|
|
"token_acc": 0.6797583081570997
|
|
},
|
|
{
|
|
"epoch": 0.335873388042204,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 7.931076290015366e-06,
|
|
"loss": 1.1356829404830933,
|
|
"step": 573,
|
|
"token_acc": 0.693127962085308
|
|
},
|
|
{
|
|
"epoch": 0.3364595545134818,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 7.923215295647167e-06,
|
|
"loss": 1.170560359954834,
|
|
"step": 574,
|
|
"token_acc": 0.6887283236994219
|
|
},
|
|
{
|
|
"epoch": 0.33704572098475966,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 7.915343307920674e-06,
|
|
"loss": 1.216208815574646,
|
|
"step": 575,
|
|
"token_acc": 0.6801186943620178
|
|
},
|
|
{
|
|
"epoch": 0.3376318874560375,
|
|
"grad_norm": 4.375,
|
|
"learning_rate": 7.907460356440133e-06,
|
|
"loss": 1.0834894180297852,
|
|
"step": 576,
|
|
"token_acc": 0.7034009156311315
|
|
},
|
|
{
|
|
"epoch": 0.33821805392731535,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 7.899566470851021e-06,
|
|
"loss": 1.197946310043335,
|
|
"step": 577,
|
|
"token_acc": 0.6881766998700736
|
|
},
|
|
{
|
|
"epoch": 0.3388042203985932,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 7.891661680839932e-06,
|
|
"loss": 1.179692029953003,
|
|
"step": 578,
|
|
"token_acc": 0.6960063058328955
|
|
},
|
|
{
|
|
"epoch": 0.33939038686987105,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 7.883746016134475e-06,
|
|
"loss": 1.1562771797180176,
|
|
"step": 579,
|
|
"token_acc": 0.699418426691154
|
|
},
|
|
{
|
|
"epoch": 0.3399765533411489,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 7.875819506503145e-06,
|
|
"loss": 1.2903468608856201,
|
|
"step": 580,
|
|
"token_acc": 0.6729957805907173
|
|
},
|
|
{
|
|
"epoch": 0.34056271981242675,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 7.86788218175523e-06,
|
|
"loss": 1.1782652139663696,
|
|
"step": 581,
|
|
"token_acc": 0.6900473933649289
|
|
},
|
|
{
|
|
"epoch": 0.34114888628370454,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 7.859934071740693e-06,
|
|
"loss": 1.2334686517715454,
|
|
"step": 582,
|
|
"token_acc": 0.6837527182354769
|
|
},
|
|
{
|
|
"epoch": 0.3417350527549824,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 7.851975206350047e-06,
|
|
"loss": 1.236402153968811,
|
|
"step": 583,
|
|
"token_acc": 0.6779102206441796
|
|
},
|
|
{
|
|
"epoch": 0.34232121922626024,
|
|
"grad_norm": 4.53125,
|
|
"learning_rate": 7.84400561551426e-06,
|
|
"loss": 1.2310223579406738,
|
|
"step": 584,
|
|
"token_acc": 0.6786971830985915
|
|
},
|
|
{
|
|
"epoch": 0.3429073856975381,
|
|
"grad_norm": 4.28125,
|
|
"learning_rate": 7.836025329204635e-06,
|
|
"loss": 1.2782983779907227,
|
|
"step": 585,
|
|
"token_acc": 0.6660351826792964
|
|
},
|
|
{
|
|
"epoch": 0.34349355216881594,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 7.828034377432694e-06,
|
|
"loss": 1.157891869544983,
|
|
"step": 586,
|
|
"token_acc": 0.6913996627318718
|
|
},
|
|
{
|
|
"epoch": 0.3440797186400938,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 7.820032790250073e-06,
|
|
"loss": 1.2458059787750244,
|
|
"step": 587,
|
|
"token_acc": 0.6798513405893284
|
|
},
|
|
{
|
|
"epoch": 0.34466588511137164,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 7.8120205977484e-06,
|
|
"loss": 1.2562525272369385,
|
|
"step": 588,
|
|
"token_acc": 0.6799242424242424
|
|
},
|
|
{
|
|
"epoch": 0.3452520515826495,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 7.803997830059193e-06,
|
|
"loss": 1.1357375383377075,
|
|
"step": 589,
|
|
"token_acc": 0.6981993406036013
|
|
},
|
|
{
|
|
"epoch": 0.34583821805392734,
|
|
"grad_norm": 3.546875,
|
|
"learning_rate": 7.795964517353734e-06,
|
|
"loss": 1.2033920288085938,
|
|
"step": 590,
|
|
"token_acc": 0.6805730082935411
|
|
},
|
|
{
|
|
"epoch": 0.34642438452520513,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 7.787920689842965e-06,
|
|
"loss": 1.1842308044433594,
|
|
"step": 591,
|
|
"token_acc": 0.6932203389830508
|
|
},
|
|
{
|
|
"epoch": 0.347010550996483,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 7.779866377777367e-06,
|
|
"loss": 1.2197084426879883,
|
|
"step": 592,
|
|
"token_acc": 0.679361523382806
|
|
},
|
|
{
|
|
"epoch": 0.34759671746776083,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 7.771801611446859e-06,
|
|
"loss": 1.1565543413162231,
|
|
"step": 593,
|
|
"token_acc": 0.6868857062622858
|
|
},
|
|
{
|
|
"epoch": 0.3481828839390387,
|
|
"grad_norm": 3.546875,
|
|
"learning_rate": 7.763726421180664e-06,
|
|
"loss": 1.2201502323150635,
|
|
"step": 594,
|
|
"token_acc": 0.6836989032901296
|
|
},
|
|
{
|
|
"epoch": 0.34876905041031653,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 7.755640837347215e-06,
|
|
"loss": 1.2335025072097778,
|
|
"step": 595,
|
|
"token_acc": 0.6832020997375328
|
|
},
|
|
{
|
|
"epoch": 0.3493552168815944,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 7.747544890354031e-06,
|
|
"loss": 1.2139551639556885,
|
|
"step": 596,
|
|
"token_acc": 0.683676268861454
|
|
},
|
|
{
|
|
"epoch": 0.3499413833528722,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 7.7394386106476e-06,
|
|
"loss": 1.2006831169128418,
|
|
"step": 597,
|
|
"token_acc": 0.6850967579793918
|
|
},
|
|
{
|
|
"epoch": 0.3505275498241501,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 7.73132202871327e-06,
|
|
"loss": 1.1887176036834717,
|
|
"step": 598,
|
|
"token_acc": 0.6836299824517423
|
|
},
|
|
{
|
|
"epoch": 0.3511137162954279,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 7.723195175075136e-06,
|
|
"loss": 1.2089827060699463,
|
|
"step": 599,
|
|
"token_acc": 0.6927036027853467
|
|
},
|
|
{
|
|
"epoch": 0.3516998827667057,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 7.715058080295918e-06,
|
|
"loss": 1.1964595317840576,
|
|
"step": 600,
|
|
"token_acc": 0.6805293005671077
|
|
},
|
|
{
|
|
"epoch": 0.35228604923798357,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 7.706910774976849e-06,
|
|
"loss": 1.1966360807418823,
|
|
"step": 601,
|
|
"token_acc": 0.6835599505562423
|
|
},
|
|
{
|
|
"epoch": 0.3528722157092614,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 7.698753289757565e-06,
|
|
"loss": 1.2051692008972168,
|
|
"step": 602,
|
|
"token_acc": 0.6871794871794872
|
|
},
|
|
{
|
|
"epoch": 0.35345838218053927,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 7.690585655315984e-06,
|
|
"loss": 1.2108685970306396,
|
|
"step": 603,
|
|
"token_acc": 0.6886935434980297
|
|
},
|
|
{
|
|
"epoch": 0.3540445486518171,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 7.68240790236819e-06,
|
|
"loss": 1.0827715396881104,
|
|
"step": 604,
|
|
"token_acc": 0.7008608321377331
|
|
},
|
|
{
|
|
"epoch": 0.35463071512309496,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 7.674220061668323e-06,
|
|
"loss": 1.196467638015747,
|
|
"step": 605,
|
|
"token_acc": 0.6806376960658267
|
|
},
|
|
{
|
|
"epoch": 0.3552168815943728,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 7.666022164008458e-06,
|
|
"loss": 1.1545685529708862,
|
|
"step": 606,
|
|
"token_acc": 0.6917630398867123
|
|
},
|
|
{
|
|
"epoch": 0.35580304806565066,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 7.657814240218491e-06,
|
|
"loss": 1.2435544729232788,
|
|
"step": 607,
|
|
"token_acc": 0.6699507389162561
|
|
},
|
|
{
|
|
"epoch": 0.3563892145369285,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 7.649596321166024e-06,
|
|
"loss": 1.1648612022399902,
|
|
"step": 608,
|
|
"token_acc": 0.6969031943428432
|
|
},
|
|
{
|
|
"epoch": 0.3569753810082063,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 7.641368437756253e-06,
|
|
"loss": 1.156242847442627,
|
|
"step": 609,
|
|
"token_acc": 0.6817144306365711
|
|
},
|
|
{
|
|
"epoch": 0.35756154747948415,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 7.633130620931837e-06,
|
|
"loss": 1.2175867557525635,
|
|
"step": 610,
|
|
"token_acc": 0.6880882830249919
|
|
},
|
|
{
|
|
"epoch": 0.358147713950762,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 7.624882901672801e-06,
|
|
"loss": 1.223467469215393,
|
|
"step": 611,
|
|
"token_acc": 0.6767749699157641
|
|
},
|
|
{
|
|
"epoch": 0.35873388042203985,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 7.616625310996405e-06,
|
|
"loss": 1.2017526626586914,
|
|
"step": 612,
|
|
"token_acc": 0.6881648936170213
|
|
},
|
|
{
|
|
"epoch": 0.3593200468933177,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 7.608357879957034e-06,
|
|
"loss": 1.135535478591919,
|
|
"step": 613,
|
|
"token_acc": 0.6999222596527598
|
|
},
|
|
{
|
|
"epoch": 0.35990621336459555,
|
|
"grad_norm": 3.53125,
|
|
"learning_rate": 7.600080639646077e-06,
|
|
"loss": 1.2121710777282715,
|
|
"step": 614,
|
|
"token_acc": 0.6827163297751533
|
|
},
|
|
{
|
|
"epoch": 0.3604923798358734,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 7.5917936211918194e-06,
|
|
"loss": 1.2086728811264038,
|
|
"step": 615,
|
|
"token_acc": 0.6876028524410313
|
|
},
|
|
{
|
|
"epoch": 0.36107854630715125,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 7.5834968557593155e-06,
|
|
"loss": 1.2283265590667725,
|
|
"step": 616,
|
|
"token_acc": 0.6777563926312895
|
|
},
|
|
{
|
|
"epoch": 0.3616647127784291,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 7.575190374550272e-06,
|
|
"loss": 1.1980435848236084,
|
|
"step": 617,
|
|
"token_acc": 0.6805671392827356
|
|
},
|
|
{
|
|
"epoch": 0.3622508792497069,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 7.566874208802939e-06,
|
|
"loss": 1.1944475173950195,
|
|
"step": 618,
|
|
"token_acc": 0.6889860617655097
|
|
},
|
|
{
|
|
"epoch": 0.36283704572098474,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 7.558548389791983e-06,
|
|
"loss": 1.1982555389404297,
|
|
"step": 619,
|
|
"token_acc": 0.6933827723901882
|
|
},
|
|
{
|
|
"epoch": 0.3634232121922626,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 7.550212948828377e-06,
|
|
"loss": 1.181703805923462,
|
|
"step": 620,
|
|
"token_acc": 0.6868250539956804
|
|
},
|
|
{
|
|
"epoch": 0.36400937866354044,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 7.541867917259278e-06,
|
|
"loss": 1.0912702083587646,
|
|
"step": 621,
|
|
"token_acc": 0.7074500302846759
|
|
},
|
|
{
|
|
"epoch": 0.3645955451348183,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 7.533513326467911e-06,
|
|
"loss": 1.1650407314300537,
|
|
"step": 622,
|
|
"token_acc": 0.6913788261724235
|
|
},
|
|
{
|
|
"epoch": 0.36518171160609614,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 7.5251492078734515e-06,
|
|
"loss": 1.2009086608886719,
|
|
"step": 623,
|
|
"token_acc": 0.6846153846153846
|
|
},
|
|
{
|
|
"epoch": 0.365767878077374,
|
|
"grad_norm": 3.484375,
|
|
"learning_rate": 7.5167755929309e-06,
|
|
"loss": 1.1798378229141235,
|
|
"step": 624,
|
|
"token_acc": 0.6890600106213489
|
|
},
|
|
{
|
|
"epoch": 0.36635404454865184,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 7.5083925131309795e-06,
|
|
"loss": 1.201135516166687,
|
|
"step": 625,
|
|
"token_acc": 0.6927725404882765
|
|
},
|
|
{
|
|
"epoch": 0.3669402110199297,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 7.500000000000001e-06,
|
|
"loss": 1.1381371021270752,
|
|
"step": 626,
|
|
"token_acc": 0.6964864864864865
|
|
},
|
|
{
|
|
"epoch": 0.3675263774912075,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 7.491598085099753e-06,
|
|
"loss": 1.170154094696045,
|
|
"step": 627,
|
|
"token_acc": 0.6889250814332247
|
|
},
|
|
{
|
|
"epoch": 0.36811254396248533,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 7.483186800027381e-06,
|
|
"loss": 1.0878584384918213,
|
|
"step": 628,
|
|
"token_acc": 0.6973088196109779
|
|
},
|
|
{
|
|
"epoch": 0.3686987104337632,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 7.4747661764152716e-06,
|
|
"loss": 1.200566291809082,
|
|
"step": 629,
|
|
"token_acc": 0.6777469990766389
|
|
},
|
|
{
|
|
"epoch": 0.369284876905041,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 7.466336245930927e-06,
|
|
"loss": 1.1673719882965088,
|
|
"step": 630,
|
|
"token_acc": 0.6936058949370097
|
|
},
|
|
{
|
|
"epoch": 0.3698710433763189,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 7.457897040276853e-06,
|
|
"loss": 1.2170510292053223,
|
|
"step": 631,
|
|
"token_acc": 0.6799292661361627
|
|
},
|
|
{
|
|
"epoch": 0.3704572098475967,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 7.449448591190436e-06,
|
|
"loss": 1.1258821487426758,
|
|
"step": 632,
|
|
"token_acc": 0.7013930950938825
|
|
},
|
|
{
|
|
"epoch": 0.3710433763188746,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 7.440990930443821e-06,
|
|
"loss": 1.1849185228347778,
|
|
"step": 633,
|
|
"token_acc": 0.6826291079812207
|
|
},
|
|
{
|
|
"epoch": 0.3716295427901524,
|
|
"grad_norm": 4.5,
|
|
"learning_rate": 7.4325240898438e-06,
|
|
"loss": 1.2004804611206055,
|
|
"step": 634,
|
|
"token_acc": 0.6847027388109552
|
|
},
|
|
{
|
|
"epoch": 0.37221570926143027,
|
|
"grad_norm": 3.421875,
|
|
"learning_rate": 7.424048101231687e-06,
|
|
"loss": 1.1302443742752075,
|
|
"step": 635,
|
|
"token_acc": 0.7029588534442903
|
|
},
|
|
{
|
|
"epoch": 0.37280187573270807,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 7.415562996483193e-06,
|
|
"loss": 1.2015308141708374,
|
|
"step": 636,
|
|
"token_acc": 0.687862796833773
|
|
},
|
|
{
|
|
"epoch": 0.3733880422039859,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 7.407068807508322e-06,
|
|
"loss": 1.1459503173828125,
|
|
"step": 637,
|
|
"token_acc": 0.6931982633863966
|
|
},
|
|
{
|
|
"epoch": 0.37397420867526376,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 7.398565566251232e-06,
|
|
"loss": 1.141577959060669,
|
|
"step": 638,
|
|
"token_acc": 0.7022593624264933
|
|
},
|
|
{
|
|
"epoch": 0.3745603751465416,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 7.390053304690131e-06,
|
|
"loss": 1.1710712909698486,
|
|
"step": 639,
|
|
"token_acc": 0.6852642819006941
|
|
},
|
|
{
|
|
"epoch": 0.37514654161781946,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 7.381532054837145e-06,
|
|
"loss": 1.2461779117584229,
|
|
"step": 640,
|
|
"token_acc": 0.6751159196290571
|
|
},
|
|
{
|
|
"epoch": 0.3757327080890973,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 7.373001848738203e-06,
|
|
"loss": 1.219635248184204,
|
|
"step": 641,
|
|
"token_acc": 0.6825199203187251
|
|
},
|
|
{
|
|
"epoch": 0.37631887456037516,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 7.364462718472919e-06,
|
|
"loss": 1.2398990392684937,
|
|
"step": 642,
|
|
"token_acc": 0.6745814307458143
|
|
},
|
|
{
|
|
"epoch": 0.376905041031653,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 7.355914696154465e-06,
|
|
"loss": 1.166886568069458,
|
|
"step": 643,
|
|
"token_acc": 0.6875976257419556
|
|
},
|
|
{
|
|
"epoch": 0.37749120750293086,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 7.347357813929455e-06,
|
|
"loss": 1.199724555015564,
|
|
"step": 644,
|
|
"token_acc": 0.681910055583628
|
|
},
|
|
{
|
|
"epoch": 0.37807737397420865,
|
|
"grad_norm": 3.3125,
|
|
"learning_rate": 7.3387921039778205e-06,
|
|
"loss": 1.1853562593460083,
|
|
"step": 645,
|
|
"token_acc": 0.6888659793814433
|
|
},
|
|
{
|
|
"epoch": 0.3786635404454865,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 7.330217598512696e-06,
|
|
"loss": 1.2224304676055908,
|
|
"step": 646,
|
|
"token_acc": 0.6770222743259086
|
|
},
|
|
{
|
|
"epoch": 0.37924970691676435,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 7.321634329780286e-06,
|
|
"loss": 1.1967148780822754,
|
|
"step": 647,
|
|
"token_acc": 0.676628971459343
|
|
},
|
|
{
|
|
"epoch": 0.3798358733880422,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 7.3130423300597575e-06,
|
|
"loss": 1.1807299852371216,
|
|
"step": 648,
|
|
"token_acc": 0.6886717718510694
|
|
},
|
|
{
|
|
"epoch": 0.38042203985932005,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 7.3044416316631106e-06,
|
|
"loss": 1.1578552722930908,
|
|
"step": 649,
|
|
"token_acc": 0.6991173416407062
|
|
},
|
|
{
|
|
"epoch": 0.3810082063305979,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 7.295832266935059e-06,
|
|
"loss": 1.1712455749511719,
|
|
"step": 650,
|
|
"token_acc": 0.690675600223339
|
|
},
|
|
{
|
|
"epoch": 0.38159437280187575,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 7.2872142682529045e-06,
|
|
"loss": 1.1617231369018555,
|
|
"step": 651,
|
|
"token_acc": 0.701017130620985
|
|
},
|
|
{
|
|
"epoch": 0.3821805392731536,
|
|
"grad_norm": 3.375,
|
|
"learning_rate": 7.278587668026422e-06,
|
|
"loss": 1.1855719089508057,
|
|
"step": 652,
|
|
"token_acc": 0.6914728682170542
|
|
},
|
|
{
|
|
"epoch": 0.38276670574443145,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 7.269952498697734e-06,
|
|
"loss": 1.1850855350494385,
|
|
"step": 653,
|
|
"token_acc": 0.691156914893617
|
|
},
|
|
{
|
|
"epoch": 0.38335287221570924,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 7.2613087927411885e-06,
|
|
"loss": 1.1851643323898315,
|
|
"step": 654,
|
|
"token_acc": 0.6855134474327629
|
|
},
|
|
{
|
|
"epoch": 0.3839390386869871,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 7.252656582663236e-06,
|
|
"loss": 1.126469612121582,
|
|
"step": 655,
|
|
"token_acc": 0.7055888223552894
|
|
},
|
|
{
|
|
"epoch": 0.38452520515826494,
|
|
"grad_norm": 4.625,
|
|
"learning_rate": 7.243995901002312e-06,
|
|
"loss": 1.1160491704940796,
|
|
"step": 656,
|
|
"token_acc": 0.7132918344928597
|
|
},
|
|
{
|
|
"epoch": 0.3851113716295428,
|
|
"grad_norm": 3.40625,
|
|
"learning_rate": 7.2353267803287065e-06,
|
|
"loss": 1.2882425785064697,
|
|
"step": 657,
|
|
"token_acc": 0.6737427864798021
|
|
},
|
|
{
|
|
"epoch": 0.38569753810082064,
|
|
"grad_norm": 4.5625,
|
|
"learning_rate": 7.226649253244448e-06,
|
|
"loss": 1.2225627899169922,
|
|
"step": 658,
|
|
"token_acc": 0.6774788057500921
|
|
},
|
|
{
|
|
"epoch": 0.3862837045720985,
|
|
"grad_norm": 3.34375,
|
|
"learning_rate": 7.217963352383182e-06,
|
|
"loss": 1.2704238891601562,
|
|
"step": 659,
|
|
"token_acc": 0.6651020408163265
|
|
},
|
|
{
|
|
"epoch": 0.38686987104337633,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 7.20926911041004e-06,
|
|
"loss": 1.2243058681488037,
|
|
"step": 660,
|
|
"token_acc": 0.6824173091270828
|
|
},
|
|
{
|
|
"epoch": 0.3874560375146542,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 7.200566560021525e-06,
|
|
"loss": 1.191321849822998,
|
|
"step": 661,
|
|
"token_acc": 0.6844934157484547
|
|
},
|
|
{
|
|
"epoch": 0.388042203985932,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 7.191855733945388e-06,
|
|
"loss": 1.126918077468872,
|
|
"step": 662,
|
|
"token_acc": 0.7040082219938335
|
|
},
|
|
{
|
|
"epoch": 0.3886283704572098,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 7.183136664940498e-06,
|
|
"loss": 1.163767695426941,
|
|
"step": 663,
|
|
"token_acc": 0.691249026226954
|
|
},
|
|
{
|
|
"epoch": 0.3892145369284877,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 7.174409385796726e-06,
|
|
"loss": 1.1889209747314453,
|
|
"step": 664,
|
|
"token_acc": 0.6866764275256223
|
|
},
|
|
{
|
|
"epoch": 0.3898007033997655,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 7.165673929334816e-06,
|
|
"loss": 1.152442455291748,
|
|
"step": 665,
|
|
"token_acc": 0.7020556856622431
|
|
},
|
|
{
|
|
"epoch": 0.3903868698710434,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 7.156930328406268e-06,
|
|
"loss": 1.1957954168319702,
|
|
"step": 666,
|
|
"token_acc": 0.6822934232715009
|
|
},
|
|
{
|
|
"epoch": 0.3909730363423212,
|
|
"grad_norm": 4.4375,
|
|
"learning_rate": 7.148178615893209e-06,
|
|
"loss": 1.2302402257919312,
|
|
"step": 667,
|
|
"token_acc": 0.6809324982602645
|
|
},
|
|
{
|
|
"epoch": 0.39155920281359907,
|
|
"grad_norm": 3.5,
|
|
"learning_rate": 7.1394188247082715e-06,
|
|
"loss": 1.152170181274414,
|
|
"step": 668,
|
|
"token_acc": 0.6992462311557789
|
|
},
|
|
{
|
|
"epoch": 0.3921453692848769,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 7.130650987794471e-06,
|
|
"loss": 1.180640697479248,
|
|
"step": 669,
|
|
"token_acc": 0.6935963253174817
|
|
},
|
|
{
|
|
"epoch": 0.39273153575615477,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 7.121875138125077e-06,
|
|
"loss": 1.2142106294631958,
|
|
"step": 670,
|
|
"token_acc": 0.6832555728356662
|
|
},
|
|
{
|
|
"epoch": 0.39331770222743256,
|
|
"grad_norm": 4.40625,
|
|
"learning_rate": 7.113091308703498e-06,
|
|
"loss": 1.200120210647583,
|
|
"step": 671,
|
|
"token_acc": 0.6837349397590361
|
|
},
|
|
{
|
|
"epoch": 0.3939038686987104,
|
|
"grad_norm": 3.40625,
|
|
"learning_rate": 7.104299532563146e-06,
|
|
"loss": 1.226205587387085,
|
|
"step": 672,
|
|
"token_acc": 0.6738730450781969
|
|
},
|
|
{
|
|
"epoch": 0.39449003516998826,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 7.095499842767324e-06,
|
|
"loss": 1.2068631649017334,
|
|
"step": 673,
|
|
"token_acc": 0.683162153750389
|
|
},
|
|
{
|
|
"epoch": 0.3950762016412661,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 7.08669227240909e-06,
|
|
"loss": 1.14707612991333,
|
|
"step": 674,
|
|
"token_acc": 0.6967895362663495
|
|
},
|
|
{
|
|
"epoch": 0.39566236811254396,
|
|
"grad_norm": 3.53125,
|
|
"learning_rate": 7.077876854611145e-06,
|
|
"loss": 1.1861579418182373,
|
|
"step": 675,
|
|
"token_acc": 0.6846598041677128
|
|
},
|
|
{
|
|
"epoch": 0.3962485345838218,
|
|
"grad_norm": 4.34375,
|
|
"learning_rate": 7.069053622525697e-06,
|
|
"loss": 1.1526713371276855,
|
|
"step": 676,
|
|
"token_acc": 0.6897526501766784
|
|
},
|
|
{
|
|
"epoch": 0.39683470105509966,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 7.060222609334343e-06,
|
|
"loss": 1.2515499591827393,
|
|
"step": 677,
|
|
"token_acc": 0.6785429407072587
|
|
},
|
|
{
|
|
"epoch": 0.3974208675263775,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 7.0513838482479424e-06,
|
|
"loss": 1.1928811073303223,
|
|
"step": 678,
|
|
"token_acc": 0.6850089232599643
|
|
},
|
|
{
|
|
"epoch": 0.39800703399765536,
|
|
"grad_norm": 4.4375,
|
|
"learning_rate": 7.042537372506493e-06,
|
|
"loss": 1.2117549180984497,
|
|
"step": 679,
|
|
"token_acc": 0.6825455898771865
|
|
},
|
|
{
|
|
"epoch": 0.39859320046893315,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 7.033683215379002e-06,
|
|
"loss": 1.165283203125,
|
|
"step": 680,
|
|
"token_acc": 0.6951052775499043
|
|
},
|
|
{
|
|
"epoch": 0.399179366940211,
|
|
"grad_norm": 3.53125,
|
|
"learning_rate": 7.0248214101633685e-06,
|
|
"loss": 1.2527129650115967,
|
|
"step": 681,
|
|
"token_acc": 0.6736027098959594
|
|
},
|
|
{
|
|
"epoch": 0.39976553341148885,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 7.0159519901862515e-06,
|
|
"loss": 1.1540418863296509,
|
|
"step": 682,
|
|
"token_acc": 0.6968965517241379
|
|
},
|
|
{
|
|
"epoch": 0.4003516998827667,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 7.007074988802946e-06,
|
|
"loss": 1.1985664367675781,
|
|
"step": 683,
|
|
"token_acc": 0.6896162528216704
|
|
},
|
|
{
|
|
"epoch": 0.40093786635404455,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 6.998190439397262e-06,
|
|
"loss": 1.1613727807998657,
|
|
"step": 684,
|
|
"token_acc": 0.6888103056327355
|
|
},
|
|
{
|
|
"epoch": 0.4015240328253224,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 6.98929837538139e-06,
|
|
"loss": 1.1794378757476807,
|
|
"step": 685,
|
|
"token_acc": 0.6912195121951219
|
|
},
|
|
{
|
|
"epoch": 0.40211019929660025,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 6.980398830195785e-06,
|
|
"loss": 1.1883779764175415,
|
|
"step": 686,
|
|
"token_acc": 0.688689505165851
|
|
},
|
|
{
|
|
"epoch": 0.4026963657678781,
|
|
"grad_norm": 4.75,
|
|
"learning_rate": 6.971491837309035e-06,
|
|
"loss": 1.2326340675354004,
|
|
"step": 687,
|
|
"token_acc": 0.6784423926134082
|
|
},
|
|
{
|
|
"epoch": 0.40328253223915594,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 6.962577430217736e-06,
|
|
"loss": 1.177336573600769,
|
|
"step": 688,
|
|
"token_acc": 0.6938354397730204
|
|
},
|
|
{
|
|
"epoch": 0.40386869871043374,
|
|
"grad_norm": 3.296875,
|
|
"learning_rate": 6.953655642446368e-06,
|
|
"loss": 1.1813665628433228,
|
|
"step": 689,
|
|
"token_acc": 0.6906041711459902
|
|
},
|
|
{
|
|
"epoch": 0.4044548651817116,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 6.944726507547169e-06,
|
|
"loss": 1.1732573509216309,
|
|
"step": 690,
|
|
"token_acc": 0.6826053243732231
|
|
},
|
|
{
|
|
"epoch": 0.40504103165298944,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 6.9357900591000034e-06,
|
|
"loss": 1.1729614734649658,
|
|
"step": 691,
|
|
"token_acc": 0.6935571196534922
|
|
},
|
|
{
|
|
"epoch": 0.4056271981242673,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 6.9268463307122425e-06,
|
|
"loss": 1.2343885898590088,
|
|
"step": 692,
|
|
"token_acc": 0.6831005984611
|
|
},
|
|
{
|
|
"epoch": 0.40621336459554513,
|
|
"grad_norm": 3.375,
|
|
"learning_rate": 6.917895356018634e-06,
|
|
"loss": 1.2693147659301758,
|
|
"step": 693,
|
|
"token_acc": 0.6737798853981427
|
|
},
|
|
{
|
|
"epoch": 0.406799531066823,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 6.908937168681176e-06,
|
|
"loss": 1.1856402158737183,
|
|
"step": 694,
|
|
"token_acc": 0.6896720509055311
|
|
},
|
|
{
|
|
"epoch": 0.40738569753810083,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 6.8999718023889965e-06,
|
|
"loss": 1.0824476480484009,
|
|
"step": 695,
|
|
"token_acc": 0.7144355120587207
|
|
},
|
|
{
|
|
"epoch": 0.4079718640093787,
|
|
"grad_norm": 3.3125,
|
|
"learning_rate": 6.890999290858213e-06,
|
|
"loss": 1.130150556564331,
|
|
"step": 696,
|
|
"token_acc": 0.7017203779985461
|
|
},
|
|
{
|
|
"epoch": 0.40855803048065653,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 6.8820196678318226e-06,
|
|
"loss": 1.209965467453003,
|
|
"step": 697,
|
|
"token_acc": 0.6870932754880694
|
|
},
|
|
{
|
|
"epoch": 0.4091441969519343,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 6.873032967079562e-06,
|
|
"loss": 1.1860438585281372,
|
|
"step": 698,
|
|
"token_acc": 0.6873212583412774
|
|
},
|
|
{
|
|
"epoch": 0.4097303634232122,
|
|
"grad_norm": 3.234375,
|
|
"learning_rate": 6.864039222397784e-06,
|
|
"loss": 1.169981598854065,
|
|
"step": 699,
|
|
"token_acc": 0.6975470871660097
|
|
},
|
|
{
|
|
"epoch": 0.41031652989449,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 6.8550384676093355e-06,
|
|
"loss": 1.1120405197143555,
|
|
"step": 700,
|
|
"token_acc": 0.6989124212936463
|
|
},
|
|
{
|
|
"epoch": 0.41090269636576787,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 6.8460307365634225e-06,
|
|
"loss": 1.2330752611160278,
|
|
"step": 701,
|
|
"token_acc": 0.6834269662921348
|
|
},
|
|
{
|
|
"epoch": 0.4114888628370457,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 6.837016063135491e-06,
|
|
"loss": 1.1952791213989258,
|
|
"step": 702,
|
|
"token_acc": 0.6926166203274637
|
|
},
|
|
{
|
|
"epoch": 0.41207502930832357,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 6.827994481227092e-06,
|
|
"loss": 1.1868919134140015,
|
|
"step": 703,
|
|
"token_acc": 0.6905389221556886
|
|
},
|
|
{
|
|
"epoch": 0.4126611957796014,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 6.818966024765758e-06,
|
|
"loss": 1.2339987754821777,
|
|
"step": 704,
|
|
"token_acc": 0.6836865767501605
|
|
},
|
|
{
|
|
"epoch": 0.41324736225087927,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 6.8099307277048745e-06,
|
|
"loss": 1.2510945796966553,
|
|
"step": 705,
|
|
"token_acc": 0.6867965962119132
|
|
},
|
|
{
|
|
"epoch": 0.4138335287221571,
|
|
"grad_norm": 4.46875,
|
|
"learning_rate": 6.800888624023552e-06,
|
|
"loss": 1.2509201765060425,
|
|
"step": 706,
|
|
"token_acc": 0.6711047139258726
|
|
},
|
|
{
|
|
"epoch": 0.4144196951934349,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 6.7918397477265e-06,
|
|
"loss": 1.1668145656585693,
|
|
"step": 707,
|
|
"token_acc": 0.6859891465219536
|
|
},
|
|
{
|
|
"epoch": 0.41500586166471276,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 6.782784132843901e-06,
|
|
"loss": 1.229659080505371,
|
|
"step": 708,
|
|
"token_acc": 0.6745049504950495
|
|
},
|
|
{
|
|
"epoch": 0.4155920281359906,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 6.773721813431271e-06,
|
|
"loss": 1.1853140592575073,
|
|
"step": 709,
|
|
"token_acc": 0.6951940375224878
|
|
},
|
|
{
|
|
"epoch": 0.41617819460726846,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 6.7646528235693445e-06,
|
|
"loss": 1.2300974130630493,
|
|
"step": 710,
|
|
"token_acc": 0.6669167291822956
|
|
},
|
|
{
|
|
"epoch": 0.4167643610785463,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 6.755577197363945e-06,
|
|
"loss": 1.1925463676452637,
|
|
"step": 711,
|
|
"token_acc": 0.6861689679382167
|
|
},
|
|
{
|
|
"epoch": 0.41735052754982416,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 6.746494968945847e-06,
|
|
"loss": 1.2416661977767944,
|
|
"step": 712,
|
|
"token_acc": 0.6819271332694151
|
|
},
|
|
{
|
|
"epoch": 0.417936694021102,
|
|
"grad_norm": 3.921875,
|
|
"learning_rate": 6.737406172470657e-06,
|
|
"loss": 1.2150845527648926,
|
|
"step": 713,
|
|
"token_acc": 0.6833333333333333
|
|
},
|
|
{
|
|
"epoch": 0.41852286049237986,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 6.7283108421186835e-06,
|
|
"loss": 1.2101713418960571,
|
|
"step": 714,
|
|
"token_acc": 0.6818045112781955
|
|
},
|
|
{
|
|
"epoch": 0.4191090269636577,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 6.719209012094806e-06,
|
|
"loss": 1.1589527130126953,
|
|
"step": 715,
|
|
"token_acc": 0.7015667744342203
|
|
},
|
|
{
|
|
"epoch": 0.4196951934349355,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 6.710100716628345e-06,
|
|
"loss": 1.1380584239959717,
|
|
"step": 716,
|
|
"token_acc": 0.6902985074626866
|
|
},
|
|
{
|
|
"epoch": 0.42028135990621335,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 6.7009859899729365e-06,
|
|
"loss": 1.1407101154327393,
|
|
"step": 717,
|
|
"token_acc": 0.6972866026003391
|
|
},
|
|
{
|
|
"epoch": 0.4208675263774912,
|
|
"grad_norm": 4.375,
|
|
"learning_rate": 6.691864866406407e-06,
|
|
"loss": 1.1770849227905273,
|
|
"step": 718,
|
|
"token_acc": 0.6821036106750392
|
|
},
|
|
{
|
|
"epoch": 0.42145369284876905,
|
|
"grad_norm": 3.484375,
|
|
"learning_rate": 6.682737380230633e-06,
|
|
"loss": 1.0616233348846436,
|
|
"step": 719,
|
|
"token_acc": 0.7191408374116368
|
|
},
|
|
{
|
|
"epoch": 0.4220398593200469,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 6.6736035657714235e-06,
|
|
"loss": 1.1659215688705444,
|
|
"step": 720,
|
|
"token_acc": 0.6897081413210445
|
|
},
|
|
{
|
|
"epoch": 0.42262602579132474,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 6.6644634573783825e-06,
|
|
"loss": 1.1330894231796265,
|
|
"step": 721,
|
|
"token_acc": 0.6951142131979695
|
|
},
|
|
{
|
|
"epoch": 0.4232121922626026,
|
|
"grad_norm": 4.375,
|
|
"learning_rate": 6.655317089424791e-06,
|
|
"loss": 1.2074017524719238,
|
|
"step": 722,
|
|
"token_acc": 0.6759868421052632
|
|
},
|
|
{
|
|
"epoch": 0.42379835873388044,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 6.646164496307461e-06,
|
|
"loss": 1.1283739805221558,
|
|
"step": 723,
|
|
"token_acc": 0.7009582863585119
|
|
},
|
|
{
|
|
"epoch": 0.4243845252051583,
|
|
"grad_norm": 3.609375,
|
|
"learning_rate": 6.637005712446622e-06,
|
|
"loss": 1.2149906158447266,
|
|
"step": 724,
|
|
"token_acc": 0.6843488649940263
|
|
},
|
|
{
|
|
"epoch": 0.4249706916764361,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 6.627840772285784e-06,
|
|
"loss": 1.1606683731079102,
|
|
"step": 725,
|
|
"token_acc": 0.6971830985915493
|
|
},
|
|
{
|
|
"epoch": 0.42555685814771393,
|
|
"grad_norm": 4.5,
|
|
"learning_rate": 6.618669710291607e-06,
|
|
"loss": 1.1949771642684937,
|
|
"step": 726,
|
|
"token_acc": 0.6825568797399784
|
|
},
|
|
{
|
|
"epoch": 0.4261430246189918,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 6.609492560953772e-06,
|
|
"loss": 1.183776617050171,
|
|
"step": 727,
|
|
"token_acc": 0.6915113871635611
|
|
},
|
|
{
|
|
"epoch": 0.42672919109026963,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 6.600309358784858e-06,
|
|
"loss": 1.2284566164016724,
|
|
"step": 728,
|
|
"token_acc": 0.6811411259782884
|
|
},
|
|
{
|
|
"epoch": 0.4273153575615475,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 6.591120138320204e-06,
|
|
"loss": 1.1667931079864502,
|
|
"step": 729,
|
|
"token_acc": 0.6900129701686122
|
|
},
|
|
{
|
|
"epoch": 0.42790152403282533,
|
|
"grad_norm": 4.65625,
|
|
"learning_rate": 6.581924934117783e-06,
|
|
"loss": 1.1865568161010742,
|
|
"step": 730,
|
|
"token_acc": 0.6815881666017906
|
|
},
|
|
{
|
|
"epoch": 0.4284876905041032,
|
|
"grad_norm": 3.921875,
|
|
"learning_rate": 6.572723780758069e-06,
|
|
"loss": 1.146211862564087,
|
|
"step": 731,
|
|
"token_acc": 0.7008408796895214
|
|
},
|
|
{
|
|
"epoch": 0.42907385697538103,
|
|
"grad_norm": 3.484375,
|
|
"learning_rate": 6.56351671284391e-06,
|
|
"loss": 1.2373653650283813,
|
|
"step": 732,
|
|
"token_acc": 0.6765365634813818
|
|
},
|
|
{
|
|
"epoch": 0.4296600234466589,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 6.554303765000398e-06,
|
|
"loss": 1.2165528535842896,
|
|
"step": 733,
|
|
"token_acc": 0.6842886842886843
|
|
},
|
|
{
|
|
"epoch": 0.43024618991793667,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 6.545084971874738e-06,
|
|
"loss": 1.1972196102142334,
|
|
"step": 734,
|
|
"token_acc": 0.6898439638653162
|
|
},
|
|
{
|
|
"epoch": 0.4308323563892145,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 6.535860368136115e-06,
|
|
"loss": 1.203583002090454,
|
|
"step": 735,
|
|
"token_acc": 0.6855409504550051
|
|
},
|
|
{
|
|
"epoch": 0.43141852286049237,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 6.526629988475567e-06,
|
|
"loss": 1.197810173034668,
|
|
"step": 736,
|
|
"token_acc": 0.682640586797066
|
|
},
|
|
{
|
|
"epoch": 0.4320046893317702,
|
|
"grad_norm": 4.28125,
|
|
"learning_rate": 6.517393867605854e-06,
|
|
"loss": 1.1633309125900269,
|
|
"step": 737,
|
|
"token_acc": 0.699238158330573
|
|
},
|
|
{
|
|
"epoch": 0.43259085580304807,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 6.508152040261329e-06,
|
|
"loss": 1.1446728706359863,
|
|
"step": 738,
|
|
"token_acc": 0.6970899470899471
|
|
},
|
|
{
|
|
"epoch": 0.4331770222743259,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 6.4989045411978e-06,
|
|
"loss": 1.2283661365509033,
|
|
"step": 739,
|
|
"token_acc": 0.6847168774051676
|
|
},
|
|
{
|
|
"epoch": 0.43376318874560377,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 6.48965140519241e-06,
|
|
"loss": 1.2017767429351807,
|
|
"step": 740,
|
|
"token_acc": 0.6822045516863176
|
|
},
|
|
{
|
|
"epoch": 0.4343493552168816,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 6.4803926670435e-06,
|
|
"loss": 1.1417598724365234,
|
|
"step": 741,
|
|
"token_acc": 0.6856720827178729
|
|
},
|
|
{
|
|
"epoch": 0.4349355216881594,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 6.4711283615704755e-06,
|
|
"loss": 1.236156940460205,
|
|
"step": 742,
|
|
"token_acc": 0.6888177208433414
|
|
},
|
|
{
|
|
"epoch": 0.43552168815943726,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 6.461858523613684e-06,
|
|
"loss": 1.1748905181884766,
|
|
"step": 743,
|
|
"token_acc": 0.694333843797856
|
|
},
|
|
{
|
|
"epoch": 0.4361078546307151,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 6.452583188034275e-06,
|
|
"loss": 1.2122077941894531,
|
|
"step": 744,
|
|
"token_acc": 0.6756675667566757
|
|
},
|
|
{
|
|
"epoch": 0.43669402110199296,
|
|
"grad_norm": 4.46875,
|
|
"learning_rate": 6.443302389714075e-06,
|
|
"loss": 1.152127981185913,
|
|
"step": 745,
|
|
"token_acc": 0.7067718606180144
|
|
},
|
|
{
|
|
"epoch": 0.4372801875732708,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 6.434016163555452e-06,
|
|
"loss": 1.123063087463379,
|
|
"step": 746,
|
|
"token_acc": 0.703838017527954
|
|
},
|
|
{
|
|
"epoch": 0.43786635404454866,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 6.42472454448119e-06,
|
|
"loss": 1.1732043027877808,
|
|
"step": 747,
|
|
"token_acc": 0.6907651715039578
|
|
},
|
|
{
|
|
"epoch": 0.4384525205158265,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 6.415427567434353e-06,
|
|
"loss": 1.167546033859253,
|
|
"step": 748,
|
|
"token_acc": 0.6874566874566874
|
|
},
|
|
{
|
|
"epoch": 0.43903868698710435,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 6.4061252673781534e-06,
|
|
"loss": 1.2351329326629639,
|
|
"step": 749,
|
|
"token_acc": 0.6848958333333334
|
|
},
|
|
{
|
|
"epoch": 0.4396248534583822,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 6.396817679295823e-06,
|
|
"loss": 1.2400213479995728,
|
|
"step": 750,
|
|
"token_acc": 0.6701530612244898
|
|
},
|
|
{
|
|
"epoch": 0.44021101992966,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 6.387504838190479e-06,
|
|
"loss": 1.17219877243042,
|
|
"step": 751,
|
|
"token_acc": 0.687903699354081
|
|
},
|
|
{
|
|
"epoch": 0.44079718640093785,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 6.378186779084996e-06,
|
|
"loss": 1.1502467393875122,
|
|
"step": 752,
|
|
"token_acc": 0.7001882226404947
|
|
},
|
|
{
|
|
"epoch": 0.4413833528722157,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 6.368863537021871e-06,
|
|
"loss": 1.2025468349456787,
|
|
"step": 753,
|
|
"token_acc": 0.6829542263409155
|
|
},
|
|
{
|
|
"epoch": 0.44196951934349354,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 6.359535147063092e-06,
|
|
"loss": 1.154723882675171,
|
|
"step": 754,
|
|
"token_acc": 0.6917697686928456
|
|
},
|
|
{
|
|
"epoch": 0.4425556858147714,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 6.350201644290005e-06,
|
|
"loss": 1.2090145349502563,
|
|
"step": 755,
|
|
"token_acc": 0.6840182648401827
|
|
},
|
|
{
|
|
"epoch": 0.44314185228604924,
|
|
"grad_norm": 3.40625,
|
|
"learning_rate": 6.340863063803187e-06,
|
|
"loss": 1.1780719757080078,
|
|
"step": 756,
|
|
"token_acc": 0.686489058039962
|
|
},
|
|
{
|
|
"epoch": 0.4437280187573271,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 6.331519440722311e-06,
|
|
"loss": 1.181138515472412,
|
|
"step": 757,
|
|
"token_acc": 0.6875973015049299
|
|
},
|
|
{
|
|
"epoch": 0.44431418522860494,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 6.322170810186013e-06,
|
|
"loss": 1.223347783088684,
|
|
"step": 758,
|
|
"token_acc": 0.6955602536997886
|
|
},
|
|
{
|
|
"epoch": 0.4449003516998828,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 6.312817207351756e-06,
|
|
"loss": 1.1368157863616943,
|
|
"step": 759,
|
|
"token_acc": 0.7018557955582598
|
|
},
|
|
{
|
|
"epoch": 0.4454865181711606,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 6.3034586673957075e-06,
|
|
"loss": 1.2401881217956543,
|
|
"step": 760,
|
|
"token_acc": 0.6738799146601646
|
|
},
|
|
{
|
|
"epoch": 0.44607268464243843,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 6.294095225512604e-06,
|
|
"loss": 1.1425551176071167,
|
|
"step": 761,
|
|
"token_acc": 0.6984810126582278
|
|
},
|
|
{
|
|
"epoch": 0.4466588511137163,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 6.284726916915611e-06,
|
|
"loss": 1.1765165328979492,
|
|
"step": 762,
|
|
"token_acc": 0.6858771374287523
|
|
},
|
|
{
|
|
"epoch": 0.44724501758499413,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 6.2753537768361995e-06,
|
|
"loss": 1.1849387884140015,
|
|
"step": 763,
|
|
"token_acc": 0.6851373561868983
|
|
},
|
|
{
|
|
"epoch": 0.447831184056272,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 6.26597584052401e-06,
|
|
"loss": 1.159532904624939,
|
|
"step": 764,
|
|
"token_acc": 0.6979289940828403
|
|
},
|
|
{
|
|
"epoch": 0.44841735052754983,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 6.256593143246718e-06,
|
|
"loss": 1.1583362817764282,
|
|
"step": 765,
|
|
"token_acc": 0.6986301369863014
|
|
},
|
|
{
|
|
"epoch": 0.4490035169988277,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 6.247205720289907e-06,
|
|
"loss": 1.2346231937408447,
|
|
"step": 766,
|
|
"token_acc": 0.6752280364858377
|
|
},
|
|
{
|
|
"epoch": 0.44958968347010553,
|
|
"grad_norm": 3.46875,
|
|
"learning_rate": 6.237813606956931e-06,
|
|
"loss": 1.1810780763626099,
|
|
"step": 767,
|
|
"token_acc": 0.69040392248424
|
|
},
|
|
{
|
|
"epoch": 0.4501758499413834,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 6.228416838568782e-06,
|
|
"loss": 1.2385624647140503,
|
|
"step": 768,
|
|
"token_acc": 0.6744297719087635
|
|
},
|
|
{
|
|
"epoch": 0.45076201641266117,
|
|
"grad_norm": 3.328125,
|
|
"learning_rate": 6.2190154504639615e-06,
|
|
"loss": 1.1726287603378296,
|
|
"step": 769,
|
|
"token_acc": 0.6822132711469154
|
|
},
|
|
{
|
|
"epoch": 0.451348182883939,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 6.209609477998339e-06,
|
|
"loss": 1.2084970474243164,
|
|
"step": 770,
|
|
"token_acc": 0.6814773673979451
|
|
},
|
|
{
|
|
"epoch": 0.45193434935521687,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 6.2001989565450305e-06,
|
|
"loss": 1.2035040855407715,
|
|
"step": 771,
|
|
"token_acc": 0.6835363948766011
|
|
},
|
|
{
|
|
"epoch": 0.4525205158264947,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 6.190783921494255e-06,
|
|
"loss": 1.2563278675079346,
|
|
"step": 772,
|
|
"token_acc": 0.6765086765086765
|
|
},
|
|
{
|
|
"epoch": 0.45310668229777257,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 6.181364408253209e-06,
|
|
"loss": 1.2018795013427734,
|
|
"step": 773,
|
|
"token_acc": 0.6919216061185468
|
|
},
|
|
{
|
|
"epoch": 0.4536928487690504,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 6.171940452245923e-06,
|
|
"loss": 1.2115321159362793,
|
|
"step": 774,
|
|
"token_acc": 0.6839961667465261
|
|
},
|
|
{
|
|
"epoch": 0.45427901524032827,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 6.1625120889131495e-06,
|
|
"loss": 1.1642186641693115,
|
|
"step": 775,
|
|
"token_acc": 0.6975412596833951
|
|
},
|
|
{
|
|
"epoch": 0.4548651817116061,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 6.153079353712201e-06,
|
|
"loss": 1.1551928520202637,
|
|
"step": 776,
|
|
"token_acc": 0.7016927499201533
|
|
},
|
|
{
|
|
"epoch": 0.45545134818288396,
|
|
"grad_norm": 3.4375,
|
|
"learning_rate": 6.14364228211684e-06,
|
|
"loss": 1.1889935731887817,
|
|
"step": 777,
|
|
"token_acc": 0.6868987631699496
|
|
},
|
|
{
|
|
"epoch": 0.45603751465416176,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 6.134200909617135e-06,
|
|
"loss": 1.2079780101776123,
|
|
"step": 778,
|
|
"token_acc": 0.6817711030554117
|
|
},
|
|
{
|
|
"epoch": 0.4566236811254396,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 6.124755271719326e-06,
|
|
"loss": 1.0973297357559204,
|
|
"step": 779,
|
|
"token_acc": 0.7017241379310345
|
|
},
|
|
{
|
|
"epoch": 0.45720984759671746,
|
|
"grad_norm": 3.359375,
|
|
"learning_rate": 6.115305403945697e-06,
|
|
"loss": 1.208069920539856,
|
|
"step": 780,
|
|
"token_acc": 0.6928442844284428
|
|
},
|
|
{
|
|
"epoch": 0.4577960140679953,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 6.105851341834439e-06,
|
|
"loss": 1.2523008584976196,
|
|
"step": 781,
|
|
"token_acc": 0.6790053687482339
|
|
},
|
|
{
|
|
"epoch": 0.45838218053927315,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 6.0963931209395165e-06,
|
|
"loss": 1.1895978450775146,
|
|
"step": 782,
|
|
"token_acc": 0.7026352288488211
|
|
},
|
|
{
|
|
"epoch": 0.458968347010551,
|
|
"grad_norm": 3.421875,
|
|
"learning_rate": 6.086930776830534e-06,
|
|
"loss": 1.1328290700912476,
|
|
"step": 783,
|
|
"token_acc": 0.6996544916090819
|
|
},
|
|
{
|
|
"epoch": 0.45955451348182885,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 6.077464345092601e-06,
|
|
"loss": 1.187293291091919,
|
|
"step": 784,
|
|
"token_acc": 0.6921663220767287
|
|
},
|
|
{
|
|
"epoch": 0.4601406799531067,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 6.0679938613262015e-06,
|
|
"loss": 1.1979875564575195,
|
|
"step": 785,
|
|
"token_acc": 0.6833416708354177
|
|
},
|
|
{
|
|
"epoch": 0.46072684642438455,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 6.058519361147055e-06,
|
|
"loss": 1.0934845209121704,
|
|
"step": 786,
|
|
"token_acc": 0.7105176393010221
|
|
},
|
|
{
|
|
"epoch": 0.46131301289566234,
|
|
"grad_norm": 3.46875,
|
|
"learning_rate": 6.0490408801859865e-06,
|
|
"loss": 1.1497573852539062,
|
|
"step": 787,
|
|
"token_acc": 0.6916666666666667
|
|
},
|
|
{
|
|
"epoch": 0.4618991793669402,
|
|
"grad_norm": 3.265625,
|
|
"learning_rate": 6.039558454088796e-06,
|
|
"loss": 1.1589206457138062,
|
|
"step": 788,
|
|
"token_acc": 0.6926634433466471
|
|
},
|
|
{
|
|
"epoch": 0.46248534583821804,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 6.030072118516116e-06,
|
|
"loss": 1.093768835067749,
|
|
"step": 789,
|
|
"token_acc": 0.7063304439190964
|
|
},
|
|
{
|
|
"epoch": 0.4630715123094959,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 6.020581909143279e-06,
|
|
"loss": 1.1778533458709717,
|
|
"step": 790,
|
|
"token_acc": 0.6925008970218873
|
|
},
|
|
{
|
|
"epoch": 0.46365767878077374,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 6.011087861660191e-06,
|
|
"loss": 1.1090742349624634,
|
|
"step": 791,
|
|
"token_acc": 0.7017726798748697
|
|
},
|
|
{
|
|
"epoch": 0.4642438452520516,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 6.001590011771188e-06,
|
|
"loss": 1.2339239120483398,
|
|
"step": 792,
|
|
"token_acc": 0.6798984537272098
|
|
},
|
|
{
|
|
"epoch": 0.46483001172332944,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 5.9920883951949075e-06,
|
|
"loss": 1.1771843433380127,
|
|
"step": 793,
|
|
"token_acc": 0.6879039704524469
|
|
},
|
|
{
|
|
"epoch": 0.4654161781946073,
|
|
"grad_norm": 4.3125,
|
|
"learning_rate": 5.982583047664151e-06,
|
|
"loss": 1.2606897354125977,
|
|
"step": 794,
|
|
"token_acc": 0.6702538631346578
|
|
},
|
|
{
|
|
"epoch": 0.46600234466588514,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 5.973074004925755e-06,
|
|
"loss": 1.2090387344360352,
|
|
"step": 795,
|
|
"token_acc": 0.680983606557377
|
|
},
|
|
{
|
|
"epoch": 0.46658851113716293,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 5.9635613027404495e-06,
|
|
"loss": 1.1153841018676758,
|
|
"step": 796,
|
|
"token_acc": 0.7017430600387347
|
|
},
|
|
{
|
|
"epoch": 0.4671746776084408,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 5.954044976882725e-06,
|
|
"loss": 1.1666665077209473,
|
|
"step": 797,
|
|
"token_acc": 0.6921854304635762
|
|
},
|
|
{
|
|
"epoch": 0.46776084407971863,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 5.944525063140703e-06,
|
|
"loss": 1.154627799987793,
|
|
"step": 798,
|
|
"token_acc": 0.690638405248549
|
|
},
|
|
{
|
|
"epoch": 0.4683470105509965,
|
|
"grad_norm": 4.4375,
|
|
"learning_rate": 5.9350015973159955e-06,
|
|
"loss": 1.206421136856079,
|
|
"step": 799,
|
|
"token_acc": 0.6809770540340488
|
|
},
|
|
{
|
|
"epoch": 0.46893317702227433,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 5.925474615223573e-06,
|
|
"loss": 1.058230996131897,
|
|
"step": 800,
|
|
"token_acc": 0.7109929078014184
|
|
},
|
|
{
|
|
"epoch": 0.4695193434935522,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 5.915944152691634e-06,
|
|
"loss": 1.1197408437728882,
|
|
"step": 801,
|
|
"token_acc": 0.6991250352808355
|
|
},
|
|
{
|
|
"epoch": 0.47010550996483,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 5.906410245561459e-06,
|
|
"loss": 1.1799342632293701,
|
|
"step": 802,
|
|
"token_acc": 0.6776567559611422
|
|
},
|
|
{
|
|
"epoch": 0.4706916764361079,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 5.896872929687287e-06,
|
|
"loss": 1.1675238609313965,
|
|
"step": 803,
|
|
"token_acc": 0.6977280210734277
|
|
},
|
|
{
|
|
"epoch": 0.4712778429073857,
|
|
"grad_norm": 3.609375,
|
|
"learning_rate": 5.887332240936177e-06,
|
|
"loss": 1.1091777086257935,
|
|
"step": 804,
|
|
"token_acc": 0.7107623318385651
|
|
},
|
|
{
|
|
"epoch": 0.4718640093786635,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 5.877788215187868e-06,
|
|
"loss": 1.1467045545578003,
|
|
"step": 805,
|
|
"token_acc": 0.6862236628849271
|
|
},
|
|
{
|
|
"epoch": 0.47245017584994137,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 5.8682408883346535e-06,
|
|
"loss": 1.1714551448822021,
|
|
"step": 806,
|
|
"token_acc": 0.6931229972886369
|
|
},
|
|
{
|
|
"epoch": 0.4730363423212192,
|
|
"grad_norm": 3.53125,
|
|
"learning_rate": 5.858690296281235e-06,
|
|
"loss": 1.2163138389587402,
|
|
"step": 807,
|
|
"token_acc": 0.6811251314405888
|
|
},
|
|
{
|
|
"epoch": 0.47362250879249707,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 5.849136474944603e-06,
|
|
"loss": 1.23085355758667,
|
|
"step": 808,
|
|
"token_acc": 0.6779372874952777
|
|
},
|
|
{
|
|
"epoch": 0.4742086752637749,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 5.839579460253887e-06,
|
|
"loss": 1.1153181791305542,
|
|
"step": 809,
|
|
"token_acc": 0.7026012335746848
|
|
},
|
|
{
|
|
"epoch": 0.47479484173505276,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 5.830019288150222e-06,
|
|
"loss": 1.1448118686676025,
|
|
"step": 810,
|
|
"token_acc": 0.6954898911353032
|
|
},
|
|
{
|
|
"epoch": 0.4753810082063306,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 5.820455994586621e-06,
|
|
"loss": 1.0969486236572266,
|
|
"step": 811,
|
|
"token_acc": 0.7133261687264911
|
|
},
|
|
{
|
|
"epoch": 0.47596717467760846,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 5.810889615527839e-06,
|
|
"loss": 1.2024489641189575,
|
|
"step": 812,
|
|
"token_acc": 0.6905343097369134
|
|
},
|
|
{
|
|
"epoch": 0.4765533411488863,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 5.801320186950229e-06,
|
|
"loss": 1.0883691310882568,
|
|
"step": 813,
|
|
"token_acc": 0.6977462179685088
|
|
},
|
|
{
|
|
"epoch": 0.4771395076201641,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 5.791747744841615e-06,
|
|
"loss": 1.2142512798309326,
|
|
"step": 814,
|
|
"token_acc": 0.6833482275841525
|
|
},
|
|
{
|
|
"epoch": 0.47772567409144195,
|
|
"grad_norm": 3.609375,
|
|
"learning_rate": 5.782172325201155e-06,
|
|
"loss": 1.166691780090332,
|
|
"step": 815,
|
|
"token_acc": 0.6789640850232104
|
|
},
|
|
{
|
|
"epoch": 0.4783118405627198,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 5.772593964039203e-06,
|
|
"loss": 1.0866860151290894,
|
|
"step": 816,
|
|
"token_acc": 0.7071246819338423
|
|
},
|
|
{
|
|
"epoch": 0.47889800703399765,
|
|
"grad_norm": 3.453125,
|
|
"learning_rate": 5.763012697377176e-06,
|
|
"loss": 1.136025309562683,
|
|
"step": 817,
|
|
"token_acc": 0.695545108573987
|
|
},
|
|
{
|
|
"epoch": 0.4794841735052755,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 5.753428561247416e-06,
|
|
"loss": 1.1498796939849854,
|
|
"step": 818,
|
|
"token_acc": 0.6931597895319856
|
|
},
|
|
{
|
|
"epoch": 0.48007033997655335,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 5.74384159169306e-06,
|
|
"loss": 1.1558449268341064,
|
|
"step": 819,
|
|
"token_acc": 0.6926632004621606
|
|
},
|
|
{
|
|
"epoch": 0.4806565064478312,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 5.734251824767895e-06,
|
|
"loss": 1.2522790431976318,
|
|
"step": 820,
|
|
"token_acc": 0.6824644549763034
|
|
},
|
|
{
|
|
"epoch": 0.48124267291910905,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 5.724659296536234e-06,
|
|
"loss": 1.1139061450958252,
|
|
"step": 821,
|
|
"token_acc": 0.6997357674753784
|
|
},
|
|
{
|
|
"epoch": 0.48182883939038684,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 5.715064043072771e-06,
|
|
"loss": 1.213309645652771,
|
|
"step": 822,
|
|
"token_acc": 0.6833238149628783
|
|
},
|
|
{
|
|
"epoch": 0.4824150058616647,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 5.7054661004624505e-06,
|
|
"loss": 1.1384762525558472,
|
|
"step": 823,
|
|
"token_acc": 0.6955945677376615
|
|
},
|
|
{
|
|
"epoch": 0.48300117233294254,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 5.695865504800328e-06,
|
|
"loss": 1.1774663925170898,
|
|
"step": 824,
|
|
"token_acc": 0.691245440333507
|
|
},
|
|
{
|
|
"epoch": 0.4835873388042204,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 5.686262292191438e-06,
|
|
"loss": 1.1979947090148926,
|
|
"step": 825,
|
|
"token_acc": 0.6916812100058173
|
|
},
|
|
{
|
|
"epoch": 0.48417350527549824,
|
|
"grad_norm": 3.53125,
|
|
"learning_rate": 5.6766564987506564e-06,
|
|
"loss": 1.1178882122039795,
|
|
"step": 826,
|
|
"token_acc": 0.7011230219499744
|
|
},
|
|
{
|
|
"epoch": 0.4847596717467761,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 5.667048160602564e-06,
|
|
"loss": 1.2735307216644287,
|
|
"step": 827,
|
|
"token_acc": 0.6735582154515778
|
|
},
|
|
{
|
|
"epoch": 0.48534583821805394,
|
|
"grad_norm": 3.328125,
|
|
"learning_rate": 5.657437313881314e-06,
|
|
"loss": 1.162387490272522,
|
|
"step": 828,
|
|
"token_acc": 0.6935329111494914
|
|
},
|
|
{
|
|
"epoch": 0.4859320046893318,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 5.647823994730489e-06,
|
|
"loss": 1.166453242301941,
|
|
"step": 829,
|
|
"token_acc": 0.690426638917794
|
|
},
|
|
{
|
|
"epoch": 0.48651817116060964,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 5.638208239302975e-06,
|
|
"loss": 1.1924514770507812,
|
|
"step": 830,
|
|
"token_acc": 0.6889843355229914
|
|
},
|
|
{
|
|
"epoch": 0.48710433763188743,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 5.628590083760815e-06,
|
|
"loss": 1.2094618082046509,
|
|
"step": 831,
|
|
"token_acc": 0.6846875
|
|
},
|
|
{
|
|
"epoch": 0.4876905041031653,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 5.618969564275083e-06,
|
|
"loss": 1.228240966796875,
|
|
"step": 832,
|
|
"token_acc": 0.6792846497764531
|
|
},
|
|
{
|
|
"epoch": 0.48827667057444313,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 5.609346717025738e-06,
|
|
"loss": 1.0798802375793457,
|
|
"step": 833,
|
|
"token_acc": 0.7059846903270702
|
|
},
|
|
{
|
|
"epoch": 0.488862837045721,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 5.599721578201499e-06,
|
|
"loss": 1.254061222076416,
|
|
"step": 834,
|
|
"token_acc": 0.6864935064935065
|
|
},
|
|
{
|
|
"epoch": 0.4894490035169988,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 5.590094183999698e-06,
|
|
"loss": 1.1165828704833984,
|
|
"step": 835,
|
|
"token_acc": 0.7058497366232326
|
|
},
|
|
{
|
|
"epoch": 0.4900351699882767,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 5.5804645706261515e-06,
|
|
"loss": 1.1523222923278809,
|
|
"step": 836,
|
|
"token_acc": 0.6899509803921569
|
|
},
|
|
{
|
|
"epoch": 0.4906213364595545,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 5.5708327742950205e-06,
|
|
"loss": 1.1334465742111206,
|
|
"step": 837,
|
|
"token_acc": 0.6940998487140696
|
|
},
|
|
{
|
|
"epoch": 0.4912075029308324,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 5.561198831228676e-06,
|
|
"loss": 1.1505519151687622,
|
|
"step": 838,
|
|
"token_acc": 0.6999184561022017
|
|
},
|
|
{
|
|
"epoch": 0.4917936694021102,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 5.55156277765756e-06,
|
|
"loss": 1.1992018222808838,
|
|
"step": 839,
|
|
"token_acc": 0.6928860613071139
|
|
},
|
|
{
|
|
"epoch": 0.492379835873388,
|
|
"grad_norm": 3.453125,
|
|
"learning_rate": 5.541924649820054e-06,
|
|
"loss": 1.1375513076782227,
|
|
"step": 840,
|
|
"token_acc": 0.7007792207792207
|
|
},
|
|
{
|
|
"epoch": 0.49296600234466587,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 5.532284483962341e-06,
|
|
"loss": 1.1178703308105469,
|
|
"step": 841,
|
|
"token_acc": 0.7015494030988062
|
|
},
|
|
{
|
|
"epoch": 0.4935521688159437,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 5.522642316338268e-06,
|
|
"loss": 1.140783667564392,
|
|
"step": 842,
|
|
"token_acc": 0.6955536675162843
|
|
},
|
|
{
|
|
"epoch": 0.49413833528722156,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 5.512998183209206e-06,
|
|
"loss": 1.1850268840789795,
|
|
"step": 843,
|
|
"token_acc": 0.6925007844367744
|
|
},
|
|
{
|
|
"epoch": 0.4947245017584994,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 5.503352120843923e-06,
|
|
"loss": 1.1783547401428223,
|
|
"step": 844,
|
|
"token_acc": 0.6941792423775793
|
|
},
|
|
{
|
|
"epoch": 0.49531066822977726,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 5.493704165518439e-06,
|
|
"loss": 1.2007172107696533,
|
|
"step": 845,
|
|
"token_acc": 0.6865227103499628
|
|
},
|
|
{
|
|
"epoch": 0.4958968347010551,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 5.484054353515896e-06,
|
|
"loss": 1.2056479454040527,
|
|
"step": 846,
|
|
"token_acc": 0.6866606982990152
|
|
},
|
|
{
|
|
"epoch": 0.49648300117233296,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 5.474402721126414e-06,
|
|
"loss": 1.1122303009033203,
|
|
"step": 847,
|
|
"token_acc": 0.7053226286054118
|
|
},
|
|
{
|
|
"epoch": 0.4970691676436108,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 5.464749304646963e-06,
|
|
"loss": 1.1381810903549194,
|
|
"step": 848,
|
|
"token_acc": 0.7002272727272727
|
|
},
|
|
{
|
|
"epoch": 0.4976553341148886,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 5.455094140381217e-06,
|
|
"loss": 1.1669211387634277,
|
|
"step": 849,
|
|
"token_acc": 0.6903508771929825
|
|
},
|
|
{
|
|
"epoch": 0.49824150058616645,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 5.445437264639433e-06,
|
|
"loss": 1.222468376159668,
|
|
"step": 850,
|
|
"token_acc": 0.6934566145092461
|
|
},
|
|
{
|
|
"epoch": 0.4988276670574443,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 5.435778713738292e-06,
|
|
"loss": 1.253871202468872,
|
|
"step": 851,
|
|
"token_acc": 0.6850538481744155
|
|
},
|
|
{
|
|
"epoch": 0.49941383352872215,
|
|
"grad_norm": 4.3125,
|
|
"learning_rate": 5.426118524000784e-06,
|
|
"loss": 1.175839900970459,
|
|
"step": 852,
|
|
"token_acc": 0.6873641726171996
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 5.416456731756058e-06,
|
|
"loss": 1.1561051607131958,
|
|
"step": 853,
|
|
"token_acc": 0.6844248388191381
|
|
},
|
|
{
|
|
"epoch": 0.5005861664712778,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 5.406793373339292e-06,
|
|
"loss": 1.1493860483169556,
|
|
"step": 854,
|
|
"token_acc": 0.7005390835579515
|
|
},
|
|
{
|
|
"epoch": 0.5011723329425557,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 5.397128485091552e-06,
|
|
"loss": 1.1760578155517578,
|
|
"step": 855,
|
|
"token_acc": 0.6907273736812882
|
|
},
|
|
{
|
|
"epoch": 0.5017584994138335,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 5.387462103359655e-06,
|
|
"loss": 1.209730863571167,
|
|
"step": 856,
|
|
"token_acc": 0.6859975961538461
|
|
},
|
|
{
|
|
"epoch": 0.5023446658851114,
|
|
"grad_norm": 4.40625,
|
|
"learning_rate": 5.377794264496041e-06,
|
|
"loss": 1.0938944816589355,
|
|
"step": 857,
|
|
"token_acc": 0.7095152603231598
|
|
},
|
|
{
|
|
"epoch": 0.5029308323563892,
|
|
"grad_norm": 3.453125,
|
|
"learning_rate": 5.3681250048586246e-06,
|
|
"loss": 1.124606728553772,
|
|
"step": 858,
|
|
"token_acc": 0.6949192782526116
|
|
},
|
|
{
|
|
"epoch": 0.5035169988276671,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 5.358454360810666e-06,
|
|
"loss": 1.215639591217041,
|
|
"step": 859,
|
|
"token_acc": 0.6781299168352439
|
|
},
|
|
{
|
|
"epoch": 0.5041031652989449,
|
|
"grad_norm": 3.5,
|
|
"learning_rate": 5.348782368720627e-06,
|
|
"loss": 1.1586073637008667,
|
|
"step": 860,
|
|
"token_acc": 0.6919147045757441
|
|
},
|
|
{
|
|
"epoch": 0.5046893317702228,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 5.339109064962047e-06,
|
|
"loss": 1.164958119392395,
|
|
"step": 861,
|
|
"token_acc": 0.6885288065843621
|
|
},
|
|
{
|
|
"epoch": 0.5052754982415005,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 5.329434485913393e-06,
|
|
"loss": 1.2011687755584717,
|
|
"step": 862,
|
|
"token_acc": 0.6808306709265176
|
|
},
|
|
{
|
|
"epoch": 0.5058616647127784,
|
|
"grad_norm": 4.3125,
|
|
"learning_rate": 5.319758667957929e-06,
|
|
"loss": 1.1608021259307861,
|
|
"step": 863,
|
|
"token_acc": 0.6947941476692753
|
|
},
|
|
{
|
|
"epoch": 0.5064478311840562,
|
|
"grad_norm": 4.6875,
|
|
"learning_rate": 5.310081647483577e-06,
|
|
"loss": 1.2065818309783936,
|
|
"step": 864,
|
|
"token_acc": 0.6909937888198758
|
|
},
|
|
{
|
|
"epoch": 0.5070339976553341,
|
|
"grad_norm": 4.3125,
|
|
"learning_rate": 5.300403460882784e-06,
|
|
"loss": 1.1881102323532104,
|
|
"step": 865,
|
|
"token_acc": 0.6849034523112931
|
|
},
|
|
{
|
|
"epoch": 0.5076201641266119,
|
|
"grad_norm": 3.4375,
|
|
"learning_rate": 5.290724144552379e-06,
|
|
"loss": 1.1066943407058716,
|
|
"step": 866,
|
|
"token_acc": 0.6970991679784124
|
|
},
|
|
{
|
|
"epoch": 0.5082063305978898,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 5.281043734893445e-06,
|
|
"loss": 1.1395900249481201,
|
|
"step": 867,
|
|
"token_acc": 0.7047465808527755
|
|
},
|
|
{
|
|
"epoch": 0.5087924970691676,
|
|
"grad_norm": 3.453125,
|
|
"learning_rate": 5.27136226831117e-06,
|
|
"loss": 1.1316787004470825,
|
|
"step": 868,
|
|
"token_acc": 0.7042117930204573
|
|
},
|
|
{
|
|
"epoch": 0.5093786635404455,
|
|
"grad_norm": 3.5,
|
|
"learning_rate": 5.2616797812147205e-06,
|
|
"loss": 1.2069311141967773,
|
|
"step": 869,
|
|
"token_acc": 0.685539403213466
|
|
},
|
|
{
|
|
"epoch": 0.5099648300117233,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 5.251996310017101e-06,
|
|
"loss": 1.1751840114593506,
|
|
"step": 870,
|
|
"token_acc": 0.6852848811126065
|
|
},
|
|
{
|
|
"epoch": 0.5105509964830012,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 5.242311891135016e-06,
|
|
"loss": 1.0958333015441895,
|
|
"step": 871,
|
|
"token_acc": 0.7014631401238042
|
|
},
|
|
{
|
|
"epoch": 0.511137162954279,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 5.232626560988735e-06,
|
|
"loss": 1.1636955738067627,
|
|
"step": 872,
|
|
"token_acc": 0.7003891050583657
|
|
},
|
|
{
|
|
"epoch": 0.5117233294255569,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 5.2229403560019506e-06,
|
|
"loss": 1.1904213428497314,
|
|
"step": 873,
|
|
"token_acc": 0.6887677868604299
|
|
},
|
|
{
|
|
"epoch": 0.5123094958968347,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 5.213253312601654e-06,
|
|
"loss": 1.1545310020446777,
|
|
"step": 874,
|
|
"token_acc": 0.7020634121791646
|
|
},
|
|
{
|
|
"epoch": 0.5128956623681126,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 5.20356546721798e-06,
|
|
"loss": 1.148392915725708,
|
|
"step": 875,
|
|
"token_acc": 0.6921450151057402
|
|
},
|
|
{
|
|
"epoch": 0.5134818288393904,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 5.193876856284085e-06,
|
|
"loss": 1.179274559020996,
|
|
"step": 876,
|
|
"token_acc": 0.6961951786232936
|
|
},
|
|
{
|
|
"epoch": 0.5140679953106683,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 5.1841875162360015e-06,
|
|
"loss": 1.173039436340332,
|
|
"step": 877,
|
|
"token_acc": 0.6837080381890976
|
|
},
|
|
{
|
|
"epoch": 0.5146541617819461,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 5.174497483512506e-06,
|
|
"loss": 1.2456729412078857,
|
|
"step": 878,
|
|
"token_acc": 0.6812953995157385
|
|
},
|
|
{
|
|
"epoch": 0.515240328253224,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 5.164806794554979e-06,
|
|
"loss": 1.1726534366607666,
|
|
"step": 879,
|
|
"token_acc": 0.6855909694555112
|
|
},
|
|
{
|
|
"epoch": 0.5158264947245017,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 5.155115485807269e-06,
|
|
"loss": 1.193587303161621,
|
|
"step": 880,
|
|
"token_acc": 0.6920935412026726
|
|
},
|
|
{
|
|
"epoch": 0.5164126611957796,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 5.145423593715558e-06,
|
|
"loss": 1.1590113639831543,
|
|
"step": 881,
|
|
"token_acc": 0.6977621322604979
|
|
},
|
|
{
|
|
"epoch": 0.5169988276670574,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 5.135731154728215e-06,
|
|
"loss": 1.1173095703125,
|
|
"step": 882,
|
|
"token_acc": 0.7032612205360026
|
|
},
|
|
{
|
|
"epoch": 0.5175849941383353,
|
|
"grad_norm": 3.453125,
|
|
"learning_rate": 5.126038205295674e-06,
|
|
"loss": 1.1815435886383057,
|
|
"step": 883,
|
|
"token_acc": 0.687394247038917
|
|
},
|
|
{
|
|
"epoch": 0.5181711606096131,
|
|
"grad_norm": 3.921875,
|
|
"learning_rate": 5.116344781870282e-06,
|
|
"loss": 1.1894080638885498,
|
|
"step": 884,
|
|
"token_acc": 0.6860364267129229
|
|
},
|
|
{
|
|
"epoch": 0.518757327080891,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 5.106650920906172e-06,
|
|
"loss": 1.1938481330871582,
|
|
"step": 885,
|
|
"token_acc": 0.6864349011588275
|
|
},
|
|
{
|
|
"epoch": 0.5193434935521688,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 5.096956658859122e-06,
|
|
"loss": 1.151395559310913,
|
|
"step": 886,
|
|
"token_acc": 0.6975228161668839
|
|
},
|
|
{
|
|
"epoch": 0.5199296600234466,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 5.087262032186418e-06,
|
|
"loss": 1.165186882019043,
|
|
"step": 887,
|
|
"token_acc": 0.6889217134416543
|
|
},
|
|
{
|
|
"epoch": 0.5205158264947245,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 5.077567077346717e-06,
|
|
"loss": 1.1306114196777344,
|
|
"step": 888,
|
|
"token_acc": 0.6908747936807357
|
|
},
|
|
{
|
|
"epoch": 0.5211019929660023,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 5.067871830799911e-06,
|
|
"loss": 1.1212241649627686,
|
|
"step": 889,
|
|
"token_acc": 0.700963523743978
|
|
},
|
|
{
|
|
"epoch": 0.5216881594372802,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 5.0581763290069865e-06,
|
|
"loss": 1.1786991357803345,
|
|
"step": 890,
|
|
"token_acc": 0.6882190520018407
|
|
},
|
|
{
|
|
"epoch": 0.522274325908558,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 5.048480608429893e-06,
|
|
"loss": 1.1433749198913574,
|
|
"step": 891,
|
|
"token_acc": 0.696
|
|
},
|
|
{
|
|
"epoch": 0.5228604923798359,
|
|
"grad_norm": 3.34375,
|
|
"learning_rate": 5.038784705531402e-06,
|
|
"loss": 1.1172782182693481,
|
|
"step": 892,
|
|
"token_acc": 0.6986654179349099
|
|
},
|
|
{
|
|
"epoch": 0.5234466588511137,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 5.02908865677497e-06,
|
|
"loss": 1.2127292156219482,
|
|
"step": 893,
|
|
"token_acc": 0.6803861788617886
|
|
},
|
|
{
|
|
"epoch": 0.5240328253223916,
|
|
"grad_norm": 3.390625,
|
|
"learning_rate": 5.019392498624602e-06,
|
|
"loss": 1.2446504831314087,
|
|
"step": 894,
|
|
"token_acc": 0.6667366652666946
|
|
},
|
|
{
|
|
"epoch": 0.5246189917936694,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 5.009696267544716e-06,
|
|
"loss": 1.1869213581085205,
|
|
"step": 895,
|
|
"token_acc": 0.6902457185405808
|
|
},
|
|
{
|
|
"epoch": 0.5252051582649473,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 5e-06,
|
|
"loss": 1.123297095298767,
|
|
"step": 896,
|
|
"token_acc": 0.7097833223900197
|
|
},
|
|
{
|
|
"epoch": 0.5257913247362251,
|
|
"grad_norm": 3.359375,
|
|
"learning_rate": 4.990303732455285e-06,
|
|
"loss": 1.1604033708572388,
|
|
"step": 897,
|
|
"token_acc": 0.7000451059990979
|
|
},
|
|
{
|
|
"epoch": 0.5263774912075029,
|
|
"grad_norm": 3.921875,
|
|
"learning_rate": 4.980607501375399e-06,
|
|
"loss": 1.1788723468780518,
|
|
"step": 898,
|
|
"token_acc": 0.6897369044797346
|
|
},
|
|
{
|
|
"epoch": 0.5269636576787807,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 4.970911343225031e-06,
|
|
"loss": 1.1918590068817139,
|
|
"step": 899,
|
|
"token_acc": 0.680728113262063
|
|
},
|
|
{
|
|
"epoch": 0.5275498241500586,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 4.9612152944686e-06,
|
|
"loss": 1.1371216773986816,
|
|
"step": 900,
|
|
"token_acc": 0.698441093789931
|
|
},
|
|
{
|
|
"epoch": 0.5281359906213364,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 4.951519391570108e-06,
|
|
"loss": 1.2306256294250488,
|
|
"step": 901,
|
|
"token_acc": 0.6759425493716338
|
|
},
|
|
{
|
|
"epoch": 0.5287221570926143,
|
|
"grad_norm": 3.53125,
|
|
"learning_rate": 4.941823670993016e-06,
|
|
"loss": 1.1767598390579224,
|
|
"step": 902,
|
|
"token_acc": 0.6916552040348464
|
|
},
|
|
{
|
|
"epoch": 0.5293083235638921,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 4.932128169200091e-06,
|
|
"loss": 1.1815407276153564,
|
|
"step": 903,
|
|
"token_acc": 0.6887714663143989
|
|
},
|
|
{
|
|
"epoch": 0.52989449003517,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 4.922432922653284e-06,
|
|
"loss": 1.1603407859802246,
|
|
"step": 904,
|
|
"token_acc": 0.689965903555772
|
|
},
|
|
{
|
|
"epoch": 0.5304806565064478,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 4.9127379678135825e-06,
|
|
"loss": 1.1624188423156738,
|
|
"step": 905,
|
|
"token_acc": 0.6827880512091038
|
|
},
|
|
{
|
|
"epoch": 0.5310668229777257,
|
|
"grad_norm": 4.375,
|
|
"learning_rate": 4.903043341140879e-06,
|
|
"loss": 1.2003625631332397,
|
|
"step": 906,
|
|
"token_acc": 0.6911577608142494
|
|
},
|
|
{
|
|
"epoch": 0.5316529894490035,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 4.8933490790938284e-06,
|
|
"loss": 1.1941111087799072,
|
|
"step": 907,
|
|
"token_acc": 0.690084985835694
|
|
},
|
|
{
|
|
"epoch": 0.5322391559202814,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 4.883655218129719e-06,
|
|
"loss": 1.193389654159546,
|
|
"step": 908,
|
|
"token_acc": 0.6948480845442536
|
|
},
|
|
{
|
|
"epoch": 0.5328253223915592,
|
|
"grad_norm": 3.46875,
|
|
"learning_rate": 4.873961794704328e-06,
|
|
"loss": 1.1248512268066406,
|
|
"step": 909,
|
|
"token_acc": 0.6967881112176414
|
|
},
|
|
{
|
|
"epoch": 0.5334114888628371,
|
|
"grad_norm": 4.34375,
|
|
"learning_rate": 4.864268845271786e-06,
|
|
"loss": 1.1147956848144531,
|
|
"step": 910,
|
|
"token_acc": 0.7054481546572935
|
|
},
|
|
{
|
|
"epoch": 0.5339976553341149,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 4.854576406284443e-06,
|
|
"loss": 1.170578956604004,
|
|
"step": 911,
|
|
"token_acc": 0.690804897108622
|
|
},
|
|
{
|
|
"epoch": 0.5345838218053928,
|
|
"grad_norm": 4.28125,
|
|
"learning_rate": 4.844884514192732e-06,
|
|
"loss": 1.2386834621429443,
|
|
"step": 912,
|
|
"token_acc": 0.6741011984021305
|
|
},
|
|
{
|
|
"epoch": 0.5351699882766706,
|
|
"grad_norm": 4.5,
|
|
"learning_rate": 4.8351932054450215e-06,
|
|
"loss": 1.1312248706817627,
|
|
"step": 913,
|
|
"token_acc": 0.6957561267184699
|
|
},
|
|
{
|
|
"epoch": 0.5357561547479485,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 4.825502516487497e-06,
|
|
"loss": 1.1794991493225098,
|
|
"step": 914,
|
|
"token_acc": 0.6903536977491962
|
|
},
|
|
{
|
|
"epoch": 0.5363423212192263,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 4.815812483764e-06,
|
|
"loss": 1.2045207023620605,
|
|
"step": 915,
|
|
"token_acc": 0.6906696553656486
|
|
},
|
|
{
|
|
"epoch": 0.536928487690504,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 4.806123143715916e-06,
|
|
"loss": 1.1778253316879272,
|
|
"step": 916,
|
|
"token_acc": 0.6969001148105626
|
|
},
|
|
{
|
|
"epoch": 0.5375146541617819,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 4.796434532782021e-06,
|
|
"loss": 1.0932819843292236,
|
|
"step": 917,
|
|
"token_acc": 0.7048040455120101
|
|
},
|
|
{
|
|
"epoch": 0.5381008206330598,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 4.786746687398347e-06,
|
|
"loss": 1.1150025129318237,
|
|
"step": 918,
|
|
"token_acc": 0.7042866711319491
|
|
},
|
|
{
|
|
"epoch": 0.5386869871043376,
|
|
"grad_norm": 4.3125,
|
|
"learning_rate": 4.77705964399805e-06,
|
|
"loss": 1.2451002597808838,
|
|
"step": 919,
|
|
"token_acc": 0.6725195094760312
|
|
},
|
|
{
|
|
"epoch": 0.5392731535756154,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 4.767373439011267e-06,
|
|
"loss": 1.19478440284729,
|
|
"step": 920,
|
|
"token_acc": 0.6932440657334145
|
|
},
|
|
{
|
|
"epoch": 0.5398593200468933,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 4.7576881088649865e-06,
|
|
"loss": 1.1451679468154907,
|
|
"step": 921,
|
|
"token_acc": 0.6878321678321678
|
|
},
|
|
{
|
|
"epoch": 0.5404454865181711,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 4.748003689982901e-06,
|
|
"loss": 1.1987621784210205,
|
|
"step": 922,
|
|
"token_acc": 0.6820072802912116
|
|
},
|
|
{
|
|
"epoch": 0.541031652989449,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 4.738320218785281e-06,
|
|
"loss": 1.1490012407302856,
|
|
"step": 923,
|
|
"token_acc": 0.6972147349505841
|
|
},
|
|
{
|
|
"epoch": 0.5416178194607268,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 4.728637731688832e-06,
|
|
"loss": 1.1598198413848877,
|
|
"step": 924,
|
|
"token_acc": 0.7038755458515283
|
|
},
|
|
{
|
|
"epoch": 0.5422039859320047,
|
|
"grad_norm": 4.4375,
|
|
"learning_rate": 4.718956265106557e-06,
|
|
"loss": 1.114462971687317,
|
|
"step": 925,
|
|
"token_acc": 0.6963797814207651
|
|
},
|
|
{
|
|
"epoch": 0.5427901524032825,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 4.7092758554476215e-06,
|
|
"loss": 1.145419955253601,
|
|
"step": 926,
|
|
"token_acc": 0.6974223784417106
|
|
},
|
|
{
|
|
"epoch": 0.5433763188745604,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 4.699596539117218e-06,
|
|
"loss": 1.1182832717895508,
|
|
"step": 927,
|
|
"token_acc": 0.701972080655883
|
|
},
|
|
{
|
|
"epoch": 0.5439624853458382,
|
|
"grad_norm": 3.34375,
|
|
"learning_rate": 4.689918352516424e-06,
|
|
"loss": 1.16782546043396,
|
|
"step": 928,
|
|
"token_acc": 0.6916383701188455
|
|
},
|
|
{
|
|
"epoch": 0.5445486518171161,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 4.680241332042072e-06,
|
|
"loss": 1.0931568145751953,
|
|
"step": 929,
|
|
"token_acc": 0.7054054054054054
|
|
},
|
|
{
|
|
"epoch": 0.5451348182883939,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 4.670565514086607e-06,
|
|
"loss": 1.144572138786316,
|
|
"step": 930,
|
|
"token_acc": 0.6940489229645856
|
|
},
|
|
{
|
|
"epoch": 0.5457209847596718,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 4.660890935037954e-06,
|
|
"loss": 1.1359310150146484,
|
|
"step": 931,
|
|
"token_acc": 0.7016045548654244
|
|
},
|
|
{
|
|
"epoch": 0.5463071512309496,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 4.651217631279374e-06,
|
|
"loss": 1.2488981485366821,
|
|
"step": 932,
|
|
"token_acc": 0.6723930078360458
|
|
},
|
|
{
|
|
"epoch": 0.5468933177022274,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 4.641545639189338e-06,
|
|
"loss": 1.2285038232803345,
|
|
"step": 933,
|
|
"token_acc": 0.6862518089725036
|
|
},
|
|
{
|
|
"epoch": 0.5474794841735052,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 4.631874995141376e-06,
|
|
"loss": 1.162295937538147,
|
|
"step": 934,
|
|
"token_acc": 0.6871914028463549
|
|
},
|
|
{
|
|
"epoch": 0.5480656506447831,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 4.6222057355039614e-06,
|
|
"loss": 1.2656493186950684,
|
|
"step": 935,
|
|
"token_acc": 0.6744884038199181
|
|
},
|
|
{
|
|
"epoch": 0.5486518171160609,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 4.6125378966403465e-06,
|
|
"loss": 1.1668286323547363,
|
|
"step": 936,
|
|
"token_acc": 0.6875961045617632
|
|
},
|
|
{
|
|
"epoch": 0.5492379835873388,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 4.602871514908449e-06,
|
|
"loss": 1.2113261222839355,
|
|
"step": 937,
|
|
"token_acc": 0.693815987933635
|
|
},
|
|
{
|
|
"epoch": 0.5498241500586166,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 4.59320662666071e-06,
|
|
"loss": 1.12813401222229,
|
|
"step": 938,
|
|
"token_acc": 0.7043269230769231
|
|
},
|
|
{
|
|
"epoch": 0.5504103165298945,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 4.583543268243942e-06,
|
|
"loss": 1.1386587619781494,
|
|
"step": 939,
|
|
"token_acc": 0.6918405069976235
|
|
},
|
|
{
|
|
"epoch": 0.5509964830011723,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 4.573881475999218e-06,
|
|
"loss": 1.1705043315887451,
|
|
"step": 940,
|
|
"token_acc": 0.6921785834172541
|
|
},
|
|
{
|
|
"epoch": 0.5515826494724502,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 4.564221286261709e-06,
|
|
"loss": 1.132240653038025,
|
|
"step": 941,
|
|
"token_acc": 0.7023844568737121
|
|
},
|
|
{
|
|
"epoch": 0.552168815943728,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 4.5545627353605705e-06,
|
|
"loss": 1.1395361423492432,
|
|
"step": 942,
|
|
"token_acc": 0.6978947368421052
|
|
},
|
|
{
|
|
"epoch": 0.5527549824150059,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 4.5449058596187835e-06,
|
|
"loss": 1.1810059547424316,
|
|
"step": 943,
|
|
"token_acc": 0.6960218645611904
|
|
},
|
|
{
|
|
"epoch": 0.5533411488862837,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 4.53525069535304e-06,
|
|
"loss": 1.235945224761963,
|
|
"step": 944,
|
|
"token_acc": 0.6857054646495832
|
|
},
|
|
{
|
|
"epoch": 0.5539273153575616,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 4.5255972788735874e-06,
|
|
"loss": 1.1820869445800781,
|
|
"step": 945,
|
|
"token_acc": 0.6844041196665032
|
|
},
|
|
{
|
|
"epoch": 0.5545134818288394,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 4.515945646484105e-06,
|
|
"loss": 1.1741750240325928,
|
|
"step": 946,
|
|
"token_acc": 0.6880513231756215
|
|
},
|
|
{
|
|
"epoch": 0.5550996483001173,
|
|
"grad_norm": 4.375,
|
|
"learning_rate": 4.506295834481561e-06,
|
|
"loss": 1.2387683391571045,
|
|
"step": 947,
|
|
"token_acc": 0.6730013106159896
|
|
},
|
|
{
|
|
"epoch": 0.5556858147713951,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 4.496647879156078e-06,
|
|
"loss": 1.0567104816436768,
|
|
"step": 948,
|
|
"token_acc": 0.7229856339787633
|
|
},
|
|
{
|
|
"epoch": 0.556271981242673,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 4.487001816790796e-06,
|
|
"loss": 1.1551339626312256,
|
|
"step": 949,
|
|
"token_acc": 0.6943521594684385
|
|
},
|
|
{
|
|
"epoch": 0.5568581477139508,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 4.477357683661734e-06,
|
|
"loss": 1.1739327907562256,
|
|
"step": 950,
|
|
"token_acc": 0.6904129437333977
|
|
},
|
|
{
|
|
"epoch": 0.5574443141852286,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 4.467715516037659e-06,
|
|
"loss": 1.189483880996704,
|
|
"step": 951,
|
|
"token_acc": 0.6945464820570783
|
|
},
|
|
{
|
|
"epoch": 0.5580304806565064,
|
|
"grad_norm": 3.421875,
|
|
"learning_rate": 4.458075350179948e-06,
|
|
"loss": 1.1898963451385498,
|
|
"step": 952,
|
|
"token_acc": 0.6881082364111085
|
|
},
|
|
{
|
|
"epoch": 0.5586166471277842,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 4.448437222342441e-06,
|
|
"loss": 1.1983251571655273,
|
|
"step": 953,
|
|
"token_acc": 0.6901279707495429
|
|
},
|
|
{
|
|
"epoch": 0.5592028135990621,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 4.4388011687713274e-06,
|
|
"loss": 1.1533105373382568,
|
|
"step": 954,
|
|
"token_acc": 0.6856060606060606
|
|
},
|
|
{
|
|
"epoch": 0.55978898007034,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 4.42916722570498e-06,
|
|
"loss": 1.0980952978134155,
|
|
"step": 955,
|
|
"token_acc": 0.7060842433697347
|
|
},
|
|
{
|
|
"epoch": 0.5603751465416178,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 4.4195354293738484e-06,
|
|
"loss": 1.1456965208053589,
|
|
"step": 956,
|
|
"token_acc": 0.6945193171608266
|
|
},
|
|
{
|
|
"epoch": 0.5609613130128956,
|
|
"grad_norm": 4.59375,
|
|
"learning_rate": 4.409905816000303e-06,
|
|
"loss": 1.2196109294891357,
|
|
"step": 957,
|
|
"token_acc": 0.6781231320980275
|
|
},
|
|
{
|
|
"epoch": 0.5615474794841735,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 4.400278421798501e-06,
|
|
"loss": 1.3013489246368408,
|
|
"step": 958,
|
|
"token_acc": 0.6638655462184874
|
|
},
|
|
{
|
|
"epoch": 0.5621336459554513,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 4.390653282974264e-06,
|
|
"loss": 1.2017362117767334,
|
|
"step": 959,
|
|
"token_acc": 0.6780722891566265
|
|
},
|
|
{
|
|
"epoch": 0.5627198124267292,
|
|
"grad_norm": 3.328125,
|
|
"learning_rate": 4.381030435724919e-06,
|
|
"loss": 1.1932332515716553,
|
|
"step": 960,
|
|
"token_acc": 0.6805313588850174
|
|
},
|
|
{
|
|
"epoch": 0.563305978898007,
|
|
"grad_norm": 3.421875,
|
|
"learning_rate": 4.371409916239188e-06,
|
|
"loss": 1.1480777263641357,
|
|
"step": 961,
|
|
"token_acc": 0.6906302420321112
|
|
},
|
|
{
|
|
"epoch": 0.5638921453692849,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 4.361791760697027e-06,
|
|
"loss": 1.2381106615066528,
|
|
"step": 962,
|
|
"token_acc": 0.6765929494197503
|
|
},
|
|
{
|
|
"epoch": 0.5644783118405627,
|
|
"grad_norm": 4.40625,
|
|
"learning_rate": 4.352176005269513e-06,
|
|
"loss": 1.1430294513702393,
|
|
"step": 963,
|
|
"token_acc": 0.7056107954545454
|
|
},
|
|
{
|
|
"epoch": 0.5650644783118406,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 4.342562686118687e-06,
|
|
"loss": 1.1622049808502197,
|
|
"step": 964,
|
|
"token_acc": 0.6923650833126088
|
|
},
|
|
{
|
|
"epoch": 0.5656506447831184,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 4.3329518393974365e-06,
|
|
"loss": 1.2432630062103271,
|
|
"step": 965,
|
|
"token_acc": 0.6776296296296296
|
|
},
|
|
{
|
|
"epoch": 0.5662368112543963,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 4.323343501249346e-06,
|
|
"loss": 1.1531585454940796,
|
|
"step": 966,
|
|
"token_acc": 0.6988564167725541
|
|
},
|
|
{
|
|
"epoch": 0.5668229777256741,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 4.313737707808564e-06,
|
|
"loss": 1.1484179496765137,
|
|
"step": 967,
|
|
"token_acc": 0.6853789231946925
|
|
},
|
|
{
|
|
"epoch": 0.567409144196952,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 4.304134495199675e-06,
|
|
"loss": 1.1696326732635498,
|
|
"step": 968,
|
|
"token_acc": 0.6973496722713024
|
|
},
|
|
{
|
|
"epoch": 0.5679953106682297,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 4.294533899537551e-06,
|
|
"loss": 1.256885051727295,
|
|
"step": 969,
|
|
"token_acc": 0.662915728932233
|
|
},
|
|
{
|
|
"epoch": 0.5685814771395076,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 4.284935956927229e-06,
|
|
"loss": 1.2050038576126099,
|
|
"step": 970,
|
|
"token_acc": 0.6881746810598626
|
|
},
|
|
{
|
|
"epoch": 0.5691676436107854,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 4.275340703463767e-06,
|
|
"loss": 1.1681671142578125,
|
|
"step": 971,
|
|
"token_acc": 0.6871965244058267
|
|
},
|
|
{
|
|
"epoch": 0.5697538100820633,
|
|
"grad_norm": 3.296875,
|
|
"learning_rate": 4.265748175232105e-06,
|
|
"loss": 1.1690319776535034,
|
|
"step": 972,
|
|
"token_acc": 0.6942011834319527
|
|
},
|
|
{
|
|
"epoch": 0.5703399765533411,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 4.256158408306943e-06,
|
|
"loss": 1.1374669075012207,
|
|
"step": 973,
|
|
"token_acc": 0.6949796472184532
|
|
},
|
|
{
|
|
"epoch": 0.570926143024619,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 4.246571438752585e-06,
|
|
"loss": 1.1309425830841064,
|
|
"step": 974,
|
|
"token_acc": 0.6817365269461078
|
|
},
|
|
{
|
|
"epoch": 0.5715123094958968,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 4.236987302622826e-06,
|
|
"loss": 1.1443626880645752,
|
|
"step": 975,
|
|
"token_acc": 0.6898064516129032
|
|
},
|
|
{
|
|
"epoch": 0.5720984759671747,
|
|
"grad_norm": 3.34375,
|
|
"learning_rate": 4.227406035960798e-06,
|
|
"loss": 1.109255075454712,
|
|
"step": 976,
|
|
"token_acc": 0.6972083035075161
|
|
},
|
|
{
|
|
"epoch": 0.5726846424384525,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 4.217827674798845e-06,
|
|
"loss": 1.1192266941070557,
|
|
"step": 977,
|
|
"token_acc": 0.7026942355889725
|
|
},
|
|
{
|
|
"epoch": 0.5732708089097304,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 4.208252255158387e-06,
|
|
"loss": 1.1843724250793457,
|
|
"step": 978,
|
|
"token_acc": 0.6851851851851852
|
|
},
|
|
{
|
|
"epoch": 0.5738569753810082,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 4.198679813049772e-06,
|
|
"loss": 1.1476900577545166,
|
|
"step": 979,
|
|
"token_acc": 0.6921359464152815
|
|
},
|
|
{
|
|
"epoch": 0.5744431418522861,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 4.189110384472164e-06,
|
|
"loss": 1.1631526947021484,
|
|
"step": 980,
|
|
"token_acc": 0.7005559968228753
|
|
},
|
|
{
|
|
"epoch": 0.5750293083235639,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 4.17954400541338e-06,
|
|
"loss": 1.1357121467590332,
|
|
"step": 981,
|
|
"token_acc": 0.7036634022488212
|
|
},
|
|
{
|
|
"epoch": 0.5756154747948418,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 4.1699807118497815e-06,
|
|
"loss": 1.131561517715454,
|
|
"step": 982,
|
|
"token_acc": 0.6892230576441103
|
|
},
|
|
{
|
|
"epoch": 0.5762016412661196,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 4.160420539746115e-06,
|
|
"loss": 1.145232915878296,
|
|
"step": 983,
|
|
"token_acc": 0.6902461826114055
|
|
},
|
|
{
|
|
"epoch": 0.5767878077373975,
|
|
"grad_norm": 3.484375,
|
|
"learning_rate": 4.150863525055397e-06,
|
|
"loss": 1.1100873947143555,
|
|
"step": 984,
|
|
"token_acc": 0.7010126582278481
|
|
},
|
|
{
|
|
"epoch": 0.5773739742086753,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 4.141309703718766e-06,
|
|
"loss": 1.2016353607177734,
|
|
"step": 985,
|
|
"token_acc": 0.6842553191489362
|
|
},
|
|
{
|
|
"epoch": 0.5779601406799532,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 4.131759111665349e-06,
|
|
"loss": 1.2143893241882324,
|
|
"step": 986,
|
|
"token_acc": 0.6837803855825649
|
|
},
|
|
{
|
|
"epoch": 0.5785463071512309,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 4.122211784812135e-06,
|
|
"loss": 1.1792724132537842,
|
|
"step": 987,
|
|
"token_acc": 0.6851674641148325
|
|
},
|
|
{
|
|
"epoch": 0.5791324736225087,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 4.112667759063825e-06,
|
|
"loss": 1.1582739353179932,
|
|
"step": 988,
|
|
"token_acc": 0.6905529382593603
|
|
},
|
|
{
|
|
"epoch": 0.5797186400937866,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 4.103127070312713e-06,
|
|
"loss": 1.1247036457061768,
|
|
"step": 989,
|
|
"token_acc": 0.6986559139784946
|
|
},
|
|
{
|
|
"epoch": 0.5803048065650644,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 4.093589754438543e-06,
|
|
"loss": 1.2095725536346436,
|
|
"step": 990,
|
|
"token_acc": 0.6786795048143054
|
|
},
|
|
{
|
|
"epoch": 0.5808909730363423,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 4.084055847308367e-06,
|
|
"loss": 1.1352102756500244,
|
|
"step": 991,
|
|
"token_acc": 0.6913174472747192
|
|
},
|
|
{
|
|
"epoch": 0.5814771395076201,
|
|
"grad_norm": 3.5,
|
|
"learning_rate": 4.074525384776428e-06,
|
|
"loss": 1.1675437688827515,
|
|
"step": 992,
|
|
"token_acc": 0.6962594696969697
|
|
},
|
|
{
|
|
"epoch": 0.582063305978898,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 4.064998402684006e-06,
|
|
"loss": 1.18882417678833,
|
|
"step": 993,
|
|
"token_acc": 0.6882923191648024
|
|
},
|
|
{
|
|
"epoch": 0.5826494724501758,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 4.0554749368593e-06,
|
|
"loss": 1.2319111824035645,
|
|
"step": 994,
|
|
"token_acc": 0.6815428753586229
|
|
},
|
|
{
|
|
"epoch": 0.5832356389214537,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 4.045955023117276e-06,
|
|
"loss": 1.1313731670379639,
|
|
"step": 995,
|
|
"token_acc": 0.6916938701487124
|
|
},
|
|
{
|
|
"epoch": 0.5838218053927315,
|
|
"grad_norm": 3.546875,
|
|
"learning_rate": 4.036438697259551e-06,
|
|
"loss": 1.207520842552185,
|
|
"step": 996,
|
|
"token_acc": 0.6854171590640511
|
|
},
|
|
{
|
|
"epoch": 0.5844079718640094,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 4.026925995074246e-06,
|
|
"loss": 1.1935958862304688,
|
|
"step": 997,
|
|
"token_acc": 0.6847797414930098
|
|
},
|
|
{
|
|
"epoch": 0.5849941383352872,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 4.017416952335849e-06,
|
|
"loss": 1.13088059425354,
|
|
"step": 998,
|
|
"token_acc": 0.7055306427503737
|
|
},
|
|
{
|
|
"epoch": 0.5855803048065651,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 4.007911604805095e-06,
|
|
"loss": 1.092301607131958,
|
|
"step": 999,
|
|
"token_acc": 0.7128712871287128
|
|
},
|
|
{
|
|
"epoch": 0.5861664712778429,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 3.998409988228813e-06,
|
|
"loss": 1.2197643518447876,
|
|
"step": 1000,
|
|
"token_acc": 0.6938946194591581
|
|
},
|
|
{
|
|
"epoch": 0.5867526377491208,
|
|
"grad_norm": 3.921875,
|
|
"learning_rate": 3.988912138339812e-06,
|
|
"loss": 1.249880075454712,
|
|
"step": 1001,
|
|
"token_acc": 0.6682375117813384
|
|
},
|
|
{
|
|
"epoch": 0.5873388042203986,
|
|
"grad_norm": 3.421875,
|
|
"learning_rate": 3.979418090856723e-06,
|
|
"loss": 1.1469327211380005,
|
|
"step": 1002,
|
|
"token_acc": 0.6889655172413793
|
|
},
|
|
{
|
|
"epoch": 0.5879249706916765,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 3.969927881483885e-06,
|
|
"loss": 1.2545855045318604,
|
|
"step": 1003,
|
|
"token_acc": 0.6801319043693322
|
|
},
|
|
{
|
|
"epoch": 0.5885111371629543,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 3.960441545911205e-06,
|
|
"loss": 1.1388708353042603,
|
|
"step": 1004,
|
|
"token_acc": 0.6982999238771885
|
|
},
|
|
{
|
|
"epoch": 0.5890973036342321,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 3.9509591198140135e-06,
|
|
"loss": 1.1593632698059082,
|
|
"step": 1005,
|
|
"token_acc": 0.687441570582736
|
|
},
|
|
{
|
|
"epoch": 0.5896834701055099,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 3.941480638852948e-06,
|
|
"loss": 1.1745730638504028,
|
|
"step": 1006,
|
|
"token_acc": 0.6912712360867018
|
|
},
|
|
{
|
|
"epoch": 0.5902696365767878,
|
|
"grad_norm": 3.296875,
|
|
"learning_rate": 3.932006138673801e-06,
|
|
"loss": 1.254392147064209,
|
|
"step": 1007,
|
|
"token_acc": 0.6774193548387096
|
|
},
|
|
{
|
|
"epoch": 0.5908558030480656,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 3.922535654907401e-06,
|
|
"loss": 1.2114301919937134,
|
|
"step": 1008,
|
|
"token_acc": 0.6767879548306148
|
|
},
|
|
{
|
|
"epoch": 0.5914419695193435,
|
|
"grad_norm": 3.484375,
|
|
"learning_rate": 3.913069223169467e-06,
|
|
"loss": 1.1584281921386719,
|
|
"step": 1009,
|
|
"token_acc": 0.6922543352601156
|
|
},
|
|
{
|
|
"epoch": 0.5920281359906213,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 3.903606879060483e-06,
|
|
"loss": 1.2197861671447754,
|
|
"step": 1010,
|
|
"token_acc": 0.6821833648393195
|
|
},
|
|
{
|
|
"epoch": 0.5926143024618992,
|
|
"grad_norm": 3.484375,
|
|
"learning_rate": 3.894148658165562e-06,
|
|
"loss": 1.1586461067199707,
|
|
"step": 1011,
|
|
"token_acc": 0.6994205697730564
|
|
},
|
|
{
|
|
"epoch": 0.593200468933177,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 3.884694596054304e-06,
|
|
"loss": 1.1011208295822144,
|
|
"step": 1012,
|
|
"token_acc": 0.7141396933560477
|
|
},
|
|
{
|
|
"epoch": 0.5937866354044549,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 3.875244728280676e-06,
|
|
"loss": 1.129084825515747,
|
|
"step": 1013,
|
|
"token_acc": 0.6982265446224256
|
|
},
|
|
{
|
|
"epoch": 0.5943728018757327,
|
|
"grad_norm": 3.453125,
|
|
"learning_rate": 3.865799090382866e-06,
|
|
"loss": 1.1885218620300293,
|
|
"step": 1014,
|
|
"token_acc": 0.6944196428571429
|
|
},
|
|
{
|
|
"epoch": 0.5949589683470106,
|
|
"grad_norm": 3.296875,
|
|
"learning_rate": 3.856357717883161e-06,
|
|
"loss": 1.1913628578186035,
|
|
"step": 1015,
|
|
"token_acc": 0.6858387799564271
|
|
},
|
|
{
|
|
"epoch": 0.5955451348182884,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 3.8469206462878e-06,
|
|
"loss": 1.227684497833252,
|
|
"step": 1016,
|
|
"token_acc": 0.6738744075829384
|
|
},
|
|
{
|
|
"epoch": 0.5961313012895663,
|
|
"grad_norm": 3.3125,
|
|
"learning_rate": 3.8374879110868505e-06,
|
|
"loss": 1.1344852447509766,
|
|
"step": 1017,
|
|
"token_acc": 0.7017861180194438
|
|
},
|
|
{
|
|
"epoch": 0.5967174677608441,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 3.828059547754078e-06,
|
|
"loss": 1.1291170120239258,
|
|
"step": 1018,
|
|
"token_acc": 0.6942219679633868
|
|
},
|
|
{
|
|
"epoch": 0.597303634232122,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 3.818635591746794e-06,
|
|
"loss": 1.1983873844146729,
|
|
"step": 1019,
|
|
"token_acc": 0.6829399141630901
|
|
},
|
|
{
|
|
"epoch": 0.5978898007033998,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 3.809216078505747e-06,
|
|
"loss": 1.1919068098068237,
|
|
"step": 1020,
|
|
"token_acc": 0.6800736067297581
|
|
},
|
|
{
|
|
"epoch": 0.5984759671746777,
|
|
"grad_norm": 3.5,
|
|
"learning_rate": 3.7998010434549716e-06,
|
|
"loss": 1.1524500846862793,
|
|
"step": 1021,
|
|
"token_acc": 0.6993355481727574
|
|
},
|
|
{
|
|
"epoch": 0.5990621336459554,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 3.790390522001662e-06,
|
|
"loss": 1.1999340057373047,
|
|
"step": 1022,
|
|
"token_acc": 0.6857712434587591
|
|
},
|
|
{
|
|
"epoch": 0.5996483001172332,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 3.780984549536041e-06,
|
|
"loss": 1.1909232139587402,
|
|
"step": 1023,
|
|
"token_acc": 0.6882080329557158
|
|
},
|
|
{
|
|
"epoch": 0.6002344665885111,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 3.7715831614312184e-06,
|
|
"loss": 1.2531805038452148,
|
|
"step": 1024,
|
|
"token_acc": 0.6665867114415928
|
|
},
|
|
{
|
|
"epoch": 0.6008206330597889,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 3.7621863930430715e-06,
|
|
"loss": 1.1164977550506592,
|
|
"step": 1025,
|
|
"token_acc": 0.6979003447195237
|
|
},
|
|
{
|
|
"epoch": 0.6014067995310668,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 3.752794279710094e-06,
|
|
"loss": 1.2498589754104614,
|
|
"step": 1026,
|
|
"token_acc": 0.6718306222566486
|
|
},
|
|
{
|
|
"epoch": 0.6019929660023446,
|
|
"grad_norm": 3.53125,
|
|
"learning_rate": 3.7434068567532845e-06,
|
|
"loss": 1.2044942378997803,
|
|
"step": 1027,
|
|
"token_acc": 0.6777083333333334
|
|
},
|
|
{
|
|
"epoch": 0.6025791324736225,
|
|
"grad_norm": 3.609375,
|
|
"learning_rate": 3.7340241594759917e-06,
|
|
"loss": 1.1824272871017456,
|
|
"step": 1028,
|
|
"token_acc": 0.6952180767209669
|
|
},
|
|
{
|
|
"epoch": 0.6031652989449003,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 3.7246462231638004e-06,
|
|
"loss": 1.2450122833251953,
|
|
"step": 1029,
|
|
"token_acc": 0.6786524133462909
|
|
},
|
|
{
|
|
"epoch": 0.6037514654161782,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 3.7152730830843904e-06,
|
|
"loss": 1.1963801383972168,
|
|
"step": 1030,
|
|
"token_acc": 0.6899781454886045
|
|
},
|
|
{
|
|
"epoch": 0.604337631887456,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 3.705904774487396e-06,
|
|
"loss": 1.213850736618042,
|
|
"step": 1031,
|
|
"token_acc": 0.6911548633558625
|
|
},
|
|
{
|
|
"epoch": 0.6049237983587339,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 3.6965413326042933e-06,
|
|
"loss": 1.1682238578796387,
|
|
"step": 1032,
|
|
"token_acc": 0.6959442004649962
|
|
},
|
|
{
|
|
"epoch": 0.6055099648300117,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 3.687182792648245e-06,
|
|
"loss": 1.156261682510376,
|
|
"step": 1033,
|
|
"token_acc": 0.6962380685008422
|
|
},
|
|
{
|
|
"epoch": 0.6060961313012896,
|
|
"grad_norm": 3.53125,
|
|
"learning_rate": 3.6778291898139907e-06,
|
|
"loss": 1.1249921321868896,
|
|
"step": 1034,
|
|
"token_acc": 0.6997565593724642
|
|
},
|
|
{
|
|
"epoch": 0.6066822977725674,
|
|
"grad_norm": 4.34375,
|
|
"learning_rate": 3.6684805592776896e-06,
|
|
"loss": 1.1711900234222412,
|
|
"step": 1035,
|
|
"token_acc": 0.6934766493699036
|
|
},
|
|
{
|
|
"epoch": 0.6072684642438453,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 3.6591369361968127e-06,
|
|
"loss": 1.1827120780944824,
|
|
"step": 1036,
|
|
"token_acc": 0.6751815677435512
|
|
},
|
|
{
|
|
"epoch": 0.6078546307151231,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 3.649798355709997e-06,
|
|
"loss": 1.2052336931228638,
|
|
"step": 1037,
|
|
"token_acc": 0.6828988828415926
|
|
},
|
|
{
|
|
"epoch": 0.608440797186401,
|
|
"grad_norm": 3.5,
|
|
"learning_rate": 3.640464852936909e-06,
|
|
"loss": 1.1443045139312744,
|
|
"step": 1038,
|
|
"token_acc": 0.6964510445507174
|
|
},
|
|
{
|
|
"epoch": 0.6090269636576788,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 3.6311364629781305e-06,
|
|
"loss": 1.1923774480819702,
|
|
"step": 1039,
|
|
"token_acc": 0.6985230234578628
|
|
},
|
|
{
|
|
"epoch": 0.6096131301289566,
|
|
"grad_norm": 3.921875,
|
|
"learning_rate": 3.6218132209150047e-06,
|
|
"loss": 1.1877858638763428,
|
|
"step": 1040,
|
|
"token_acc": 0.6960174237710018
|
|
},
|
|
{
|
|
"epoch": 0.6101992966002344,
|
|
"grad_norm": 4.53125,
|
|
"learning_rate": 3.6124951618095224e-06,
|
|
"loss": 1.1935055255889893,
|
|
"step": 1041,
|
|
"token_acc": 0.6925403225806451
|
|
},
|
|
{
|
|
"epoch": 0.6107854630715123,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 3.603182320704179e-06,
|
|
"loss": 1.1833957433700562,
|
|
"step": 1042,
|
|
"token_acc": 0.6886090440755581
|
|
},
|
|
{
|
|
"epoch": 0.6113716295427901,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 3.593874732621847e-06,
|
|
"loss": 1.1963509321212769,
|
|
"step": 1043,
|
|
"token_acc": 0.6832648870636551
|
|
},
|
|
{
|
|
"epoch": 0.611957796014068,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 3.5845724325656485e-06,
|
|
"loss": 1.2229799032211304,
|
|
"step": 1044,
|
|
"token_acc": 0.6926943699731903
|
|
},
|
|
{
|
|
"epoch": 0.6125439624853458,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 3.5752754555188105e-06,
|
|
"loss": 1.2934041023254395,
|
|
"step": 1045,
|
|
"token_acc": 0.6679074714171763
|
|
},
|
|
{
|
|
"epoch": 0.6131301289566237,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 3.5659838364445505e-06,
|
|
"loss": 1.2350175380706787,
|
|
"step": 1046,
|
|
"token_acc": 0.6820018650917004
|
|
},
|
|
{
|
|
"epoch": 0.6137162954279015,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 3.5566976102859276e-06,
|
|
"loss": 1.1548781394958496,
|
|
"step": 1047,
|
|
"token_acc": 0.6983615363953801
|
|
},
|
|
{
|
|
"epoch": 0.6143024618991794,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 3.5474168119657275e-06,
|
|
"loss": 1.1922781467437744,
|
|
"step": 1048,
|
|
"token_acc": 0.6747276109487111
|
|
},
|
|
{
|
|
"epoch": 0.6148886283704572,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 3.538141476386317e-06,
|
|
"loss": 1.2030515670776367,
|
|
"step": 1049,
|
|
"token_acc": 0.6869719875500222
|
|
},
|
|
{
|
|
"epoch": 0.6154747948417351,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 3.528871638429524e-06,
|
|
"loss": 1.2457503080368042,
|
|
"step": 1050,
|
|
"token_acc": 0.6754300551768906
|
|
},
|
|
{
|
|
"epoch": 0.6160609613130129,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 3.519607332956502e-06,
|
|
"loss": 1.122985601425171,
|
|
"step": 1051,
|
|
"token_acc": 0.7069693769799367
|
|
},
|
|
{
|
|
"epoch": 0.6166471277842908,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 3.51034859480759e-06,
|
|
"loss": 1.0941150188446045,
|
|
"step": 1052,
|
|
"token_acc": 0.7062130177514793
|
|
},
|
|
{
|
|
"epoch": 0.6172332942555686,
|
|
"grad_norm": 4.90625,
|
|
"learning_rate": 3.5010954588022017e-06,
|
|
"loss": 1.293731927871704,
|
|
"step": 1053,
|
|
"token_acc": 0.6670896785109983
|
|
},
|
|
{
|
|
"epoch": 0.6178194607268465,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 3.491847959738673e-06,
|
|
"loss": 1.235663652420044,
|
|
"step": 1054,
|
|
"token_acc": 0.6740654205607477
|
|
},
|
|
{
|
|
"epoch": 0.6184056271981243,
|
|
"grad_norm": 4.46875,
|
|
"learning_rate": 3.4826061323941483e-06,
|
|
"loss": 1.2375608682632446,
|
|
"step": 1055,
|
|
"token_acc": 0.6779862972892463
|
|
},
|
|
{
|
|
"epoch": 0.6189917936694022,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 3.473370011524435e-06,
|
|
"loss": 1.2215361595153809,
|
|
"step": 1056,
|
|
"token_acc": 0.677024200776815
|
|
},
|
|
{
|
|
"epoch": 0.61957796014068,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 3.464139631863887e-06,
|
|
"loss": 1.217350959777832,
|
|
"step": 1057,
|
|
"token_acc": 0.6846284741917187
|
|
},
|
|
{
|
|
"epoch": 0.6201641266119577,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 3.4549150281252635e-06,
|
|
"loss": 1.1973509788513184,
|
|
"step": 1058,
|
|
"token_acc": 0.6849465313956676
|
|
},
|
|
{
|
|
"epoch": 0.6207502930832356,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 3.4456962349996026e-06,
|
|
"loss": 1.1094391345977783,
|
|
"step": 1059,
|
|
"token_acc": 0.7022058823529411
|
|
},
|
|
{
|
|
"epoch": 0.6213364595545134,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 3.436483287156091e-06,
|
|
"loss": 1.1208590269088745,
|
|
"step": 1060,
|
|
"token_acc": 0.7002997002997003
|
|
},
|
|
{
|
|
"epoch": 0.6219226260257913,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 3.427276219241933e-06,
|
|
"loss": 1.1616228818893433,
|
|
"step": 1061,
|
|
"token_acc": 0.6871825515386913
|
|
},
|
|
{
|
|
"epoch": 0.6225087924970691,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 3.418075065882217e-06,
|
|
"loss": 1.1883854866027832,
|
|
"step": 1062,
|
|
"token_acc": 0.6908860759493671
|
|
},
|
|
{
|
|
"epoch": 0.623094958968347,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 3.408879861679797e-06,
|
|
"loss": 1.1669220924377441,
|
|
"step": 1063,
|
|
"token_acc": 0.6965204236006052
|
|
},
|
|
{
|
|
"epoch": 0.6236811254396248,
|
|
"grad_norm": 3.453125,
|
|
"learning_rate": 3.399690641215142e-06,
|
|
"loss": 1.2013435363769531,
|
|
"step": 1064,
|
|
"token_acc": 0.6896871378910776
|
|
},
|
|
{
|
|
"epoch": 0.6242672919109027,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 3.39050743904623e-06,
|
|
"loss": 1.2416810989379883,
|
|
"step": 1065,
|
|
"token_acc": 0.6796320630749014
|
|
},
|
|
{
|
|
"epoch": 0.6248534583821805,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 3.3813302897083955e-06,
|
|
"loss": 1.1853712797164917,
|
|
"step": 1066,
|
|
"token_acc": 0.6950334821428571
|
|
},
|
|
{
|
|
"epoch": 0.6254396248534584,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 3.372159227714218e-06,
|
|
"loss": 1.0983827114105225,
|
|
"step": 1067,
|
|
"token_acc": 0.708405296488198
|
|
},
|
|
{
|
|
"epoch": 0.6260257913247362,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 3.3629942875533784e-06,
|
|
"loss": 1.12241792678833,
|
|
"step": 1068,
|
|
"token_acc": 0.696681007990166
|
|
},
|
|
{
|
|
"epoch": 0.6266119577960141,
|
|
"grad_norm": 3.4375,
|
|
"learning_rate": 3.3538355036925385e-06,
|
|
"loss": 1.2027463912963867,
|
|
"step": 1069,
|
|
"token_acc": 0.6886621315192744
|
|
},
|
|
{
|
|
"epoch": 0.6271981242672919,
|
|
"grad_norm": 3.921875,
|
|
"learning_rate": 3.3446829105752103e-06,
|
|
"loss": 1.3219525814056396,
|
|
"step": 1070,
|
|
"token_acc": 0.6596796985398021
|
|
},
|
|
{
|
|
"epoch": 0.6277842907385698,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 3.335536542621617e-06,
|
|
"loss": 1.2298333644866943,
|
|
"step": 1071,
|
|
"token_acc": 0.6757610568638713
|
|
},
|
|
{
|
|
"epoch": 0.6283704572098476,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 3.3263964342285795e-06,
|
|
"loss": 1.2087805271148682,
|
|
"step": 1072,
|
|
"token_acc": 0.6840200445434298
|
|
},
|
|
{
|
|
"epoch": 0.6289566236811255,
|
|
"grad_norm": 3.46875,
|
|
"learning_rate": 3.317262619769368e-06,
|
|
"loss": 1.1750450134277344,
|
|
"step": 1073,
|
|
"token_acc": 0.6948844455985458
|
|
},
|
|
{
|
|
"epoch": 0.6295427901524033,
|
|
"grad_norm": 3.453125,
|
|
"learning_rate": 3.308135133593595e-06,
|
|
"loss": 1.2884511947631836,
|
|
"step": 1074,
|
|
"token_acc": 0.671130276895143
|
|
},
|
|
{
|
|
"epoch": 0.6301289566236812,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 3.299014010027064e-06,
|
|
"loss": 1.2029030323028564,
|
|
"step": 1075,
|
|
"token_acc": 0.686896981014628
|
|
},
|
|
{
|
|
"epoch": 0.6307151230949589,
|
|
"grad_norm": 3.40625,
|
|
"learning_rate": 3.289899283371657e-06,
|
|
"loss": 1.1618454456329346,
|
|
"step": 1076,
|
|
"token_acc": 0.6932966023875115
|
|
},
|
|
{
|
|
"epoch": 0.6313012895662368,
|
|
"grad_norm": 3.40625,
|
|
"learning_rate": 3.2807909879051958e-06,
|
|
"loss": 1.1941276788711548,
|
|
"step": 1077,
|
|
"token_acc": 0.6863679694947569
|
|
},
|
|
{
|
|
"epoch": 0.6318874560375146,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 3.271689157881317e-06,
|
|
"loss": 1.2034378051757812,
|
|
"step": 1078,
|
|
"token_acc": 0.683312101910828
|
|
},
|
|
{
|
|
"epoch": 0.6324736225087925,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 3.2625938275293436e-06,
|
|
"loss": 1.1802480220794678,
|
|
"step": 1079,
|
|
"token_acc": 0.6873786407766991
|
|
},
|
|
{
|
|
"epoch": 0.6330597889800703,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 3.253505031054155e-06,
|
|
"loss": 1.1583318710327148,
|
|
"step": 1080,
|
|
"token_acc": 0.6938309215536939
|
|
},
|
|
{
|
|
"epoch": 0.6336459554513482,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 3.244422802636057e-06,
|
|
"loss": 1.1310129165649414,
|
|
"step": 1081,
|
|
"token_acc": 0.7036272485992333
|
|
},
|
|
{
|
|
"epoch": 0.634232121922626,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 3.2353471764306567e-06,
|
|
"loss": 1.1003400087356567,
|
|
"step": 1082,
|
|
"token_acc": 0.6982551143200962
|
|
},
|
|
{
|
|
"epoch": 0.6348182883939039,
|
|
"grad_norm": 3.375,
|
|
"learning_rate": 3.22627818656873e-06,
|
|
"loss": 1.1751258373260498,
|
|
"step": 1083,
|
|
"token_acc": 0.6980083371931449
|
|
},
|
|
{
|
|
"epoch": 0.6354044548651817,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 3.2172158671561005e-06,
|
|
"loss": 1.164064645767212,
|
|
"step": 1084,
|
|
"token_acc": 0.6917378917378917
|
|
},
|
|
{
|
|
"epoch": 0.6359906213364596,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 3.2081602522734987e-06,
|
|
"loss": 1.1496959924697876,
|
|
"step": 1085,
|
|
"token_acc": 0.6889131026968128
|
|
},
|
|
{
|
|
"epoch": 0.6365767878077374,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 3.1991113759764493e-06,
|
|
"loss": 1.099763035774231,
|
|
"step": 1086,
|
|
"token_acc": 0.703394877903514
|
|
},
|
|
{
|
|
"epoch": 0.6371629542790153,
|
|
"grad_norm": 3.546875,
|
|
"learning_rate": 3.1900692722951267e-06,
|
|
"loss": 1.1811811923980713,
|
|
"step": 1087,
|
|
"token_acc": 0.6844547563805105
|
|
},
|
|
{
|
|
"epoch": 0.6377491207502931,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 3.1810339752342446e-06,
|
|
"loss": 1.2169749736785889,
|
|
"step": 1088,
|
|
"token_acc": 0.6778599886169607
|
|
},
|
|
{
|
|
"epoch": 0.638335287221571,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 3.1720055187729093e-06,
|
|
"loss": 1.168398141860962,
|
|
"step": 1089,
|
|
"token_acc": 0.696578631452581
|
|
},
|
|
{
|
|
"epoch": 0.6389214536928488,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 3.1629839368645087e-06,
|
|
"loss": 1.2339184284210205,
|
|
"step": 1090,
|
|
"token_acc": 0.6807181889149102
|
|
},
|
|
{
|
|
"epoch": 0.6395076201641267,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 3.1539692634365788e-06,
|
|
"loss": 1.1307259798049927,
|
|
"step": 1091,
|
|
"token_acc": 0.6983050847457627
|
|
},
|
|
{
|
|
"epoch": 0.6400937866354045,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 3.1449615323906657e-06,
|
|
"loss": 1.1411519050598145,
|
|
"step": 1092,
|
|
"token_acc": 0.6964024568587306
|
|
},
|
|
{
|
|
"epoch": 0.6406799531066822,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 3.1359607776022184e-06,
|
|
"loss": 1.199252724647522,
|
|
"step": 1093,
|
|
"token_acc": 0.6960663798401967
|
|
},
|
|
{
|
|
"epoch": 0.6412661195779601,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 3.12696703292044e-06,
|
|
"loss": 1.183197021484375,
|
|
"step": 1094,
|
|
"token_acc": 0.6876310272536688
|
|
},
|
|
{
|
|
"epoch": 0.6418522860492379,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 3.1179803321681795e-06,
|
|
"loss": 1.183380126953125,
|
|
"step": 1095,
|
|
"token_acc": 0.6894372554920253
|
|
},
|
|
{
|
|
"epoch": 0.6424384525205158,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 3.1090007091417884e-06,
|
|
"loss": 1.2236629724502563,
|
|
"step": 1096,
|
|
"token_acc": 0.6835145459888334
|
|
},
|
|
{
|
|
"epoch": 0.6430246189917936,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 3.100028197611006e-06,
|
|
"loss": 1.1678136587142944,
|
|
"step": 1097,
|
|
"token_acc": 0.6925894618202311
|
|
},
|
|
{
|
|
"epoch": 0.6436107854630715,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 3.091062831318825e-06,
|
|
"loss": 1.1211228370666504,
|
|
"step": 1098,
|
|
"token_acc": 0.7029871011541072
|
|
},
|
|
{
|
|
"epoch": 0.6441969519343493,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 3.0821046439813682e-06,
|
|
"loss": 1.2567938566207886,
|
|
"step": 1099,
|
|
"token_acc": 0.6730136342266102
|
|
},
|
|
{
|
|
"epoch": 0.6447831184056272,
|
|
"grad_norm": 3.53125,
|
|
"learning_rate": 3.0731536692877596e-06,
|
|
"loss": 1.1506843566894531,
|
|
"step": 1100,
|
|
"token_acc": 0.6965620328849028
|
|
},
|
|
{
|
|
"epoch": 0.645369284876905,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 3.0642099408999982e-06,
|
|
"loss": 1.1254549026489258,
|
|
"step": 1101,
|
|
"token_acc": 0.7072337783297112
|
|
},
|
|
{
|
|
"epoch": 0.6459554513481829,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 3.0552734924528304e-06,
|
|
"loss": 1.1821115016937256,
|
|
"step": 1102,
|
|
"token_acc": 0.6706666666666666
|
|
},
|
|
{
|
|
"epoch": 0.6465416178194607,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 3.0463443575536324e-06,
|
|
"loss": 1.1486269235610962,
|
|
"step": 1103,
|
|
"token_acc": 0.6919176856546376
|
|
},
|
|
{
|
|
"epoch": 0.6471277842907386,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 3.0374225697822645e-06,
|
|
"loss": 1.1598789691925049,
|
|
"step": 1104,
|
|
"token_acc": 0.695322376738306
|
|
},
|
|
{
|
|
"epoch": 0.6477139507620164,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 3.028508162690967e-06,
|
|
"loss": 1.2249096632003784,
|
|
"step": 1105,
|
|
"token_acc": 0.6722769896609762
|
|
},
|
|
{
|
|
"epoch": 0.6483001172332943,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 3.019601169804216e-06,
|
|
"loss": 1.1214826107025146,
|
|
"step": 1106,
|
|
"token_acc": 0.6989422294548413
|
|
},
|
|
{
|
|
"epoch": 0.6488862837045721,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 3.010701624618612e-06,
|
|
"loss": 1.1703617572784424,
|
|
"step": 1107,
|
|
"token_acc": 0.694170403587444
|
|
},
|
|
{
|
|
"epoch": 0.64947245017585,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 3.00180956060274e-06,
|
|
"loss": 1.1371263265609741,
|
|
"step": 1108,
|
|
"token_acc": 0.6887853029887578
|
|
},
|
|
{
|
|
"epoch": 0.6500586166471278,
|
|
"grad_norm": 3.609375,
|
|
"learning_rate": 2.9929250111970533e-06,
|
|
"loss": 1.21773099899292,
|
|
"step": 1109,
|
|
"token_acc": 0.6849208409031923
|
|
},
|
|
{
|
|
"epoch": 0.6506447831184057,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 2.9840480098137498e-06,
|
|
"loss": 1.1996715068817139,
|
|
"step": 1110,
|
|
"token_acc": 0.6756756756756757
|
|
},
|
|
{
|
|
"epoch": 0.6512309495896834,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 2.975178589836632e-06,
|
|
"loss": 1.1779634952545166,
|
|
"step": 1111,
|
|
"token_acc": 0.6866604186191815
|
|
},
|
|
{
|
|
"epoch": 0.6518171160609613,
|
|
"grad_norm": 4.59375,
|
|
"learning_rate": 2.966316784621e-06,
|
|
"loss": 1.1783099174499512,
|
|
"step": 1112,
|
|
"token_acc": 0.6972555746140652
|
|
},
|
|
{
|
|
"epoch": 0.6524032825322391,
|
|
"grad_norm": 3.390625,
|
|
"learning_rate": 2.957462627493509e-06,
|
|
"loss": 1.2079603672027588,
|
|
"step": 1113,
|
|
"token_acc": 0.6862367661212705
|
|
},
|
|
{
|
|
"epoch": 0.652989449003517,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 2.94861615175206e-06,
|
|
"loss": 1.1581699848175049,
|
|
"step": 1114,
|
|
"token_acc": 0.694019471488178
|
|
},
|
|
{
|
|
"epoch": 0.6535756154747948,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 2.9397773906656584e-06,
|
|
"loss": 1.1566981077194214,
|
|
"step": 1115,
|
|
"token_acc": 0.6941371681415929
|
|
},
|
|
{
|
|
"epoch": 0.6541617819460727,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 2.9309463774743047e-06,
|
|
"loss": 1.201323390007019,
|
|
"step": 1116,
|
|
"token_acc": 0.682067932067932
|
|
},
|
|
{
|
|
"epoch": 0.6547479484173505,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 2.9221231453888567e-06,
|
|
"loss": 1.1512397527694702,
|
|
"step": 1117,
|
|
"token_acc": 0.6906512605042017
|
|
},
|
|
{
|
|
"epoch": 0.6553341148886284,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 2.9133077275909112e-06,
|
|
"loss": 1.0943059921264648,
|
|
"step": 1118,
|
|
"token_acc": 0.7098039215686275
|
|
},
|
|
{
|
|
"epoch": 0.6559202813599062,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 2.904500157232678e-06,
|
|
"loss": 1.2505214214324951,
|
|
"step": 1119,
|
|
"token_acc": 0.6852589641434262
|
|
},
|
|
{
|
|
"epoch": 0.6565064478311841,
|
|
"grad_norm": 3.390625,
|
|
"learning_rate": 2.895700467436855e-06,
|
|
"loss": 1.1579997539520264,
|
|
"step": 1120,
|
|
"token_acc": 0.6904870277651343
|
|
},
|
|
{
|
|
"epoch": 0.6570926143024619,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 2.886908691296504e-06,
|
|
"loss": 1.1385352611541748,
|
|
"step": 1121,
|
|
"token_acc": 0.684044233807267
|
|
},
|
|
{
|
|
"epoch": 0.6576787807737398,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 2.8781248618749235e-06,
|
|
"loss": 1.1437532901763916,
|
|
"step": 1122,
|
|
"token_acc": 0.6987128972944576
|
|
},
|
|
{
|
|
"epoch": 0.6582649472450176,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 2.869349012205529e-06,
|
|
"loss": 1.1518923044204712,
|
|
"step": 1123,
|
|
"token_acc": 0.6959479443951494
|
|
},
|
|
{
|
|
"epoch": 0.6588511137162955,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 2.86058117529173e-06,
|
|
"loss": 1.1966581344604492,
|
|
"step": 1124,
|
|
"token_acc": 0.6918347270044269
|
|
},
|
|
{
|
|
"epoch": 0.6594372801875733,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 2.851821384106791e-06,
|
|
"loss": 1.2111376523971558,
|
|
"step": 1125,
|
|
"token_acc": 0.6802908726178536
|
|
},
|
|
{
|
|
"epoch": 0.6600234466588512,
|
|
"grad_norm": 3.5,
|
|
"learning_rate": 2.843069671593734e-06,
|
|
"loss": 1.21970534324646,
|
|
"step": 1126,
|
|
"token_acc": 0.6864468864468865
|
|
},
|
|
{
|
|
"epoch": 0.660609613130129,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 2.8343260706651864e-06,
|
|
"loss": 1.1392496824264526,
|
|
"step": 1127,
|
|
"token_acc": 0.6980819322756334
|
|
},
|
|
{
|
|
"epoch": 0.6611957796014069,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 2.825590614203277e-06,
|
|
"loss": 1.2180278301239014,
|
|
"step": 1128,
|
|
"token_acc": 0.6879861711322386
|
|
},
|
|
{
|
|
"epoch": 0.6617819460726846,
|
|
"grad_norm": 4.34375,
|
|
"learning_rate": 2.8168633350595042e-06,
|
|
"loss": 1.2393032312393188,
|
|
"step": 1129,
|
|
"token_acc": 0.6793085655314758
|
|
},
|
|
{
|
|
"epoch": 0.6623681125439624,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 2.8081442660546126e-06,
|
|
"loss": 1.1584525108337402,
|
|
"step": 1130,
|
|
"token_acc": 0.6976635514018692
|
|
},
|
|
{
|
|
"epoch": 0.6629542790152403,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 2.7994334399784773e-06,
|
|
"loss": 1.121964931488037,
|
|
"step": 1131,
|
|
"token_acc": 0.6996129032258065
|
|
},
|
|
{
|
|
"epoch": 0.6635404454865181,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 2.790730889589962e-06,
|
|
"loss": 1.1321778297424316,
|
|
"step": 1132,
|
|
"token_acc": 0.6979351032448378
|
|
},
|
|
{
|
|
"epoch": 0.664126611957796,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 2.7820366476168224e-06,
|
|
"loss": 1.1787943840026855,
|
|
"step": 1133,
|
|
"token_acc": 0.6895252449133383
|
|
},
|
|
{
|
|
"epoch": 0.6647127784290738,
|
|
"grad_norm": 3.109375,
|
|
"learning_rate": 2.7733507467555532e-06,
|
|
"loss": 1.1260957717895508,
|
|
"step": 1134,
|
|
"token_acc": 0.6966955879638176
|
|
},
|
|
{
|
|
"epoch": 0.6652989449003517,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 2.7646732196712977e-06,
|
|
"loss": 1.203115463256836,
|
|
"step": 1135,
|
|
"token_acc": 0.6849278460716195
|
|
},
|
|
{
|
|
"epoch": 0.6658851113716295,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 2.7560040989976894e-06,
|
|
"loss": 1.2261807918548584,
|
|
"step": 1136,
|
|
"token_acc": 0.68077900084674
|
|
},
|
|
{
|
|
"epoch": 0.6664712778429074,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 2.747343417336764e-06,
|
|
"loss": 1.2260699272155762,
|
|
"step": 1137,
|
|
"token_acc": 0.6842737094837935
|
|
},
|
|
{
|
|
"epoch": 0.6670574443141852,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 2.7386912072588123e-06,
|
|
"loss": 1.1160529851913452,
|
|
"step": 1138,
|
|
"token_acc": 0.6926357691214103
|
|
},
|
|
{
|
|
"epoch": 0.6676436107854631,
|
|
"grad_norm": 3.390625,
|
|
"learning_rate": 2.7300475013022666e-06,
|
|
"loss": 1.1243956089019775,
|
|
"step": 1139,
|
|
"token_acc": 0.6972727272727273
|
|
},
|
|
{
|
|
"epoch": 0.6682297772567409,
|
|
"grad_norm": 3.546875,
|
|
"learning_rate": 2.7214123319735787e-06,
|
|
"loss": 1.178177833557129,
|
|
"step": 1140,
|
|
"token_acc": 0.6888888888888889
|
|
},
|
|
{
|
|
"epoch": 0.6688159437280188,
|
|
"grad_norm": 3.46875,
|
|
"learning_rate": 2.7127857317470967e-06,
|
|
"loss": 1.2184196710586548,
|
|
"step": 1141,
|
|
"token_acc": 0.6805301645338209
|
|
},
|
|
{
|
|
"epoch": 0.6694021101992966,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 2.7041677330649408e-06,
|
|
"loss": 1.1360690593719482,
|
|
"step": 1142,
|
|
"token_acc": 0.6949152542372882
|
|
},
|
|
{
|
|
"epoch": 0.6699882766705745,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 2.6955583683368902e-06,
|
|
"loss": 1.2742327451705933,
|
|
"step": 1143,
|
|
"token_acc": 0.6852127343052663
|
|
},
|
|
{
|
|
"epoch": 0.6705744431418523,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 2.686957669940242e-06,
|
|
"loss": 1.2304381132125854,
|
|
"step": 1144,
|
|
"token_acc": 0.6837222378408772
|
|
},
|
|
{
|
|
"epoch": 0.6711606096131302,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 2.678365670219716e-06,
|
|
"loss": 1.171257495880127,
|
|
"step": 1145,
|
|
"token_acc": 0.6870807815689706
|
|
},
|
|
{
|
|
"epoch": 0.671746776084408,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 2.6697824014873076e-06,
|
|
"loss": 1.140916347503662,
|
|
"step": 1146,
|
|
"token_acc": 0.6986831913245546
|
|
},
|
|
{
|
|
"epoch": 0.6723329425556858,
|
|
"grad_norm": 4.5625,
|
|
"learning_rate": 2.661207896022181e-06,
|
|
"loss": 1.1356415748596191,
|
|
"step": 1147,
|
|
"token_acc": 0.7057698437998088
|
|
},
|
|
{
|
|
"epoch": 0.6729191090269636,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 2.6526421860705474e-06,
|
|
"loss": 1.1673271656036377,
|
|
"step": 1148,
|
|
"token_acc": 0.690947666195191
|
|
},
|
|
{
|
|
"epoch": 0.6735052754982415,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 2.6440853038455354e-06,
|
|
"loss": 1.1352019309997559,
|
|
"step": 1149,
|
|
"token_acc": 0.6897081413210445
|
|
},
|
|
{
|
|
"epoch": 0.6740914419695193,
|
|
"grad_norm": 3.546875,
|
|
"learning_rate": 2.6355372815270837e-06,
|
|
"loss": 1.138607382774353,
|
|
"step": 1150,
|
|
"token_acc": 0.7007794820216243
|
|
},
|
|
{
|
|
"epoch": 0.6746776084407972,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 2.626998151261798e-06,
|
|
"loss": 1.1647425889968872,
|
|
"step": 1151,
|
|
"token_acc": 0.6945981554677206
|
|
},
|
|
{
|
|
"epoch": 0.675263774912075,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 2.6184679451628587e-06,
|
|
"loss": 1.211193323135376,
|
|
"step": 1152,
|
|
"token_acc": 0.6820212171970966
|
|
},
|
|
{
|
|
"epoch": 0.6758499413833529,
|
|
"grad_norm": 4.46875,
|
|
"learning_rate": 2.6099466953098708e-06,
|
|
"loss": 1.1127341985702515,
|
|
"step": 1153,
|
|
"token_acc": 0.7038475114719379
|
|
},
|
|
{
|
|
"epoch": 0.6764361078546307,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 2.601434433748771e-06,
|
|
"loss": 1.1639795303344727,
|
|
"step": 1154,
|
|
"token_acc": 0.683011391778108
|
|
},
|
|
{
|
|
"epoch": 0.6770222743259086,
|
|
"grad_norm": 4.5,
|
|
"learning_rate": 2.59293119249168e-06,
|
|
"loss": 1.2003720998764038,
|
|
"step": 1155,
|
|
"token_acc": 0.686922430691683
|
|
},
|
|
{
|
|
"epoch": 0.6776084407971864,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 2.5844370035168077e-06,
|
|
"loss": 1.1505885124206543,
|
|
"step": 1156,
|
|
"token_acc": 0.6853303471444568
|
|
},
|
|
{
|
|
"epoch": 0.6781946072684643,
|
|
"grad_norm": 3.484375,
|
|
"learning_rate": 2.5759518987683154e-06,
|
|
"loss": 1.1576824188232422,
|
|
"step": 1157,
|
|
"token_acc": 0.6933848382423949
|
|
},
|
|
{
|
|
"epoch": 0.6787807737397421,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 2.567475910156201e-06,
|
|
"loss": 1.2298462390899658,
|
|
"step": 1158,
|
|
"token_acc": 0.6824631101021567
|
|
},
|
|
{
|
|
"epoch": 0.67936694021102,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 2.5590090695561798e-06,
|
|
"loss": 1.1015112400054932,
|
|
"step": 1159,
|
|
"token_acc": 0.7033785581271614
|
|
},
|
|
{
|
|
"epoch": 0.6799531066822978,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 2.550551408809566e-06,
|
|
"loss": 1.2408682107925415,
|
|
"step": 1160,
|
|
"token_acc": 0.6821241585639491
|
|
},
|
|
{
|
|
"epoch": 0.6805392731535757,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 2.5421029597231476e-06,
|
|
"loss": 1.1837949752807617,
|
|
"step": 1161,
|
|
"token_acc": 0.6935823250920569
|
|
},
|
|
{
|
|
"epoch": 0.6811254396248535,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 2.533663754069074e-06,
|
|
"loss": 1.164008617401123,
|
|
"step": 1162,
|
|
"token_acc": 0.6909825033647375
|
|
},
|
|
{
|
|
"epoch": 0.6817116060961314,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 2.52523382358473e-06,
|
|
"loss": 1.1895369291305542,
|
|
"step": 1163,
|
|
"token_acc": 0.6763832658569501
|
|
},
|
|
{
|
|
"epoch": 0.6822977725674091,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 2.5168131999726203e-06,
|
|
"loss": 1.1027013063430786,
|
|
"step": 1164,
|
|
"token_acc": 0.7080250426863973
|
|
},
|
|
{
|
|
"epoch": 0.6828839390386869,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 2.508401914900249e-06,
|
|
"loss": 1.1388211250305176,
|
|
"step": 1165,
|
|
"token_acc": 0.6930208601596703
|
|
},
|
|
{
|
|
"epoch": 0.6834701055099648,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 2.5000000000000015e-06,
|
|
"loss": 1.1886847019195557,
|
|
"step": 1166,
|
|
"token_acc": 0.6804260985352862
|
|
},
|
|
{
|
|
"epoch": 0.6840562719812426,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 2.491607486869022e-06,
|
|
"loss": 1.2125225067138672,
|
|
"step": 1167,
|
|
"token_acc": 0.6977989292088043
|
|
},
|
|
{
|
|
"epoch": 0.6846424384525205,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 2.4832244070691013e-06,
|
|
"loss": 1.1562386751174927,
|
|
"step": 1168,
|
|
"token_acc": 0.6897985289414774
|
|
},
|
|
{
|
|
"epoch": 0.6852286049237983,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 2.4748507921265514e-06,
|
|
"loss": 1.1087639331817627,
|
|
"step": 1169,
|
|
"token_acc": 0.7011294526498697
|
|
},
|
|
{
|
|
"epoch": 0.6858147713950762,
|
|
"grad_norm": 4.3125,
|
|
"learning_rate": 2.4664866735320886e-06,
|
|
"loss": 1.1648273468017578,
|
|
"step": 1170,
|
|
"token_acc": 0.6961501749920458
|
|
},
|
|
{
|
|
"epoch": 0.686400937866354,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 2.458132082740724e-06,
|
|
"loss": 1.2443337440490723,
|
|
"step": 1171,
|
|
"token_acc": 0.6720554272517321
|
|
},
|
|
{
|
|
"epoch": 0.6869871043376319,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 2.4497870511716237e-06,
|
|
"loss": 1.1700353622436523,
|
|
"step": 1172,
|
|
"token_acc": 0.6902384914032169
|
|
},
|
|
{
|
|
"epoch": 0.6875732708089097,
|
|
"grad_norm": 4.3125,
|
|
"learning_rate": 2.4414516102080197e-06,
|
|
"loss": 1.1348958015441895,
|
|
"step": 1173,
|
|
"token_acc": 0.6941095418532552
|
|
},
|
|
{
|
|
"epoch": 0.6881594372801876,
|
|
"grad_norm": 3.609375,
|
|
"learning_rate": 2.4331257911970628e-06,
|
|
"loss": 1.1829365491867065,
|
|
"step": 1174,
|
|
"token_acc": 0.7000717531690983
|
|
},
|
|
{
|
|
"epoch": 0.6887456037514654,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 2.424809625449729e-06,
|
|
"loss": 1.2013264894485474,
|
|
"step": 1175,
|
|
"token_acc": 0.6831022085259374
|
|
},
|
|
{
|
|
"epoch": 0.6893317702227433,
|
|
"grad_norm": 3.265625,
|
|
"learning_rate": 2.4165031442406857e-06,
|
|
"loss": 1.1194162368774414,
|
|
"step": 1176,
|
|
"token_acc": 0.7010958265329914
|
|
},
|
|
{
|
|
"epoch": 0.6899179366940211,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 2.4082063788081805e-06,
|
|
"loss": 1.1707444190979004,
|
|
"step": 1177,
|
|
"token_acc": 0.6975652684071575
|
|
},
|
|
{
|
|
"epoch": 0.690504103165299,
|
|
"grad_norm": 4.4375,
|
|
"learning_rate": 2.3999193603539234e-06,
|
|
"loss": 1.1821787357330322,
|
|
"step": 1178,
|
|
"token_acc": 0.6864014801110083
|
|
},
|
|
{
|
|
"epoch": 0.6910902696365768,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 2.391642120042968e-06,
|
|
"loss": 1.2690048217773438,
|
|
"step": 1179,
|
|
"token_acc": 0.6811145510835913
|
|
},
|
|
{
|
|
"epoch": 0.6916764361078547,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 2.3833746890035964e-06,
|
|
"loss": 1.1957205533981323,
|
|
"step": 1180,
|
|
"token_acc": 0.6872620790629576
|
|
},
|
|
{
|
|
"epoch": 0.6922626025791325,
|
|
"grad_norm": 3.453125,
|
|
"learning_rate": 2.3751170983272e-06,
|
|
"loss": 1.2394673824310303,
|
|
"step": 1181,
|
|
"token_acc": 0.6761596548004315
|
|
},
|
|
{
|
|
"epoch": 0.6928487690504103,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 2.3668693790681634e-06,
|
|
"loss": 1.1906461715698242,
|
|
"step": 1182,
|
|
"token_acc": 0.6844194508766126
|
|
},
|
|
{
|
|
"epoch": 0.6934349355216881,
|
|
"grad_norm": 3.921875,
|
|
"learning_rate": 2.3586315622437482e-06,
|
|
"loss": 1.1574299335479736,
|
|
"step": 1183,
|
|
"token_acc": 0.6874820968203953
|
|
},
|
|
{
|
|
"epoch": 0.694021101992966,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 2.3504036788339763e-06,
|
|
"loss": 1.2216331958770752,
|
|
"step": 1184,
|
|
"token_acc": 0.6823882224645583
|
|
},
|
|
{
|
|
"epoch": 0.6946072684642438,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 2.342185759781511e-06,
|
|
"loss": 1.2370855808258057,
|
|
"step": 1185,
|
|
"token_acc": 0.6779760163790582
|
|
},
|
|
{
|
|
"epoch": 0.6951934349355217,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 2.333977835991545e-06,
|
|
"loss": 1.126565933227539,
|
|
"step": 1186,
|
|
"token_acc": 0.7046603032004491
|
|
},
|
|
{
|
|
"epoch": 0.6957796014067995,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 2.32577993833168e-06,
|
|
"loss": 1.1537145376205444,
|
|
"step": 1187,
|
|
"token_acc": 0.7001716106888943
|
|
},
|
|
{
|
|
"epoch": 0.6963657678780774,
|
|
"grad_norm": 4.65625,
|
|
"learning_rate": 2.317592097631812e-06,
|
|
"loss": 1.1783006191253662,
|
|
"step": 1188,
|
|
"token_acc": 0.692354740061162
|
|
},
|
|
{
|
|
"epoch": 0.6969519343493552,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 2.3094143446840166e-06,
|
|
"loss": 1.1735305786132812,
|
|
"step": 1189,
|
|
"token_acc": 0.6859890824018716
|
|
},
|
|
{
|
|
"epoch": 0.6975381008206331,
|
|
"grad_norm": 4.34375,
|
|
"learning_rate": 2.3012467102424373e-06,
|
|
"loss": 1.1380336284637451,
|
|
"step": 1190,
|
|
"token_acc": 0.6932981181908221
|
|
},
|
|
{
|
|
"epoch": 0.6981242672919109,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 2.293089225023152e-06,
|
|
"loss": 1.1001231670379639,
|
|
"step": 1191,
|
|
"token_acc": 0.7030365414307771
|
|
},
|
|
{
|
|
"epoch": 0.6987104337631888,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 2.284941919704085e-06,
|
|
"loss": 1.1293675899505615,
|
|
"step": 1192,
|
|
"token_acc": 0.7039954980303883
|
|
},
|
|
{
|
|
"epoch": 0.6992966002344666,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 2.2768048249248648e-06,
|
|
"loss": 1.1959903240203857,
|
|
"step": 1193,
|
|
"token_acc": 0.6883259911894273
|
|
},
|
|
{
|
|
"epoch": 0.6998827667057445,
|
|
"grad_norm": 3.4375,
|
|
"learning_rate": 2.268677971286732e-06,
|
|
"loss": 1.1193506717681885,
|
|
"step": 1194,
|
|
"token_acc": 0.6956944782291413
|
|
},
|
|
{
|
|
"epoch": 0.7004689331770223,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 2.260561389352401e-06,
|
|
"loss": 1.114840030670166,
|
|
"step": 1195,
|
|
"token_acc": 0.7055174665360757
|
|
},
|
|
{
|
|
"epoch": 0.7010550996483002,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 2.2524551096459703e-06,
|
|
"loss": 1.2654098272323608,
|
|
"step": 1196,
|
|
"token_acc": 0.6832361146335277
|
|
},
|
|
{
|
|
"epoch": 0.701641266119578,
|
|
"grad_norm": 3.46875,
|
|
"learning_rate": 2.244359162652785e-06,
|
|
"loss": 1.1436595916748047,
|
|
"step": 1197,
|
|
"token_acc": 0.6997629707663945
|
|
},
|
|
{
|
|
"epoch": 0.7022274325908558,
|
|
"grad_norm": 4.34375,
|
|
"learning_rate": 2.236273578819337e-06,
|
|
"loss": 1.1252301931381226,
|
|
"step": 1198,
|
|
"token_acc": 0.6938086303939962
|
|
},
|
|
{
|
|
"epoch": 0.7028135990621337,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 2.2281983885531428e-06,
|
|
"loss": 1.1912730932235718,
|
|
"step": 1199,
|
|
"token_acc": 0.6825571782860292
|
|
},
|
|
{
|
|
"epoch": 0.7033997655334114,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 2.2201336222226332e-06,
|
|
"loss": 1.162748098373413,
|
|
"step": 1200,
|
|
"token_acc": 0.6823498139287613
|
|
},
|
|
{
|
|
"epoch": 0.7039859320046893,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 2.2120793101570366e-06,
|
|
"loss": 1.1506905555725098,
|
|
"step": 1201,
|
|
"token_acc": 0.6939795153634774
|
|
},
|
|
{
|
|
"epoch": 0.7045720984759671,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 2.204035482646267e-06,
|
|
"loss": 1.1333820819854736,
|
|
"step": 1202,
|
|
"token_acc": 0.7015267175572519
|
|
},
|
|
{
|
|
"epoch": 0.705158264947245,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 2.1960021699408075e-06,
|
|
"loss": 1.1860252618789673,
|
|
"step": 1203,
|
|
"token_acc": 0.6850072780203784
|
|
},
|
|
{
|
|
"epoch": 0.7057444314185228,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 2.1879794022516006e-06,
|
|
"loss": 1.199240803718567,
|
|
"step": 1204,
|
|
"token_acc": 0.6882882882882883
|
|
},
|
|
{
|
|
"epoch": 0.7063305978898007,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 2.1799672097499293e-06,
|
|
"loss": 1.199594497680664,
|
|
"step": 1205,
|
|
"token_acc": 0.6855804268610098
|
|
},
|
|
{
|
|
"epoch": 0.7069167643610785,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 2.171965622567308e-06,
|
|
"loss": 1.1478431224822998,
|
|
"step": 1206,
|
|
"token_acc": 0.6928450975668513
|
|
},
|
|
{
|
|
"epoch": 0.7075029308323564,
|
|
"grad_norm": 3.390625,
|
|
"learning_rate": 2.163974670795368e-06,
|
|
"loss": 1.1674165725708008,
|
|
"step": 1207,
|
|
"token_acc": 0.6798917944093779
|
|
},
|
|
{
|
|
"epoch": 0.7080890973036342,
|
|
"grad_norm": 3.390625,
|
|
"learning_rate": 2.155994384485742e-06,
|
|
"loss": 1.164170265197754,
|
|
"step": 1208,
|
|
"token_acc": 0.6849652699977593
|
|
},
|
|
{
|
|
"epoch": 0.7086752637749121,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 2.148024793649956e-06,
|
|
"loss": 1.208821177482605,
|
|
"step": 1209,
|
|
"token_acc": 0.6842243186582809
|
|
},
|
|
{
|
|
"epoch": 0.7092614302461899,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 2.1400659282593083e-06,
|
|
"loss": 1.272471308708191,
|
|
"step": 1210,
|
|
"token_acc": 0.6671523730224813
|
|
},
|
|
{
|
|
"epoch": 0.7098475967174678,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 2.132117818244771e-06,
|
|
"loss": 1.217325210571289,
|
|
"step": 1211,
|
|
"token_acc": 0.6803566603620643
|
|
},
|
|
{
|
|
"epoch": 0.7104337631887456,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 2.1241804934968558e-06,
|
|
"loss": 1.2030888795852661,
|
|
"step": 1212,
|
|
"token_acc": 0.6874640598044853
|
|
},
|
|
{
|
|
"epoch": 0.7110199296600235,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 2.1162539838655288e-06,
|
|
"loss": 1.1421886682510376,
|
|
"step": 1213,
|
|
"token_acc": 0.6976800976800976
|
|
},
|
|
{
|
|
"epoch": 0.7116060961313013,
|
|
"grad_norm": 3.46875,
|
|
"learning_rate": 2.1083383191600676e-06,
|
|
"loss": 1.1834850311279297,
|
|
"step": 1214,
|
|
"token_acc": 0.6829429960229784
|
|
},
|
|
{
|
|
"epoch": 0.7121922626025792,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 2.1004335291489793e-06,
|
|
"loss": 1.1928448677062988,
|
|
"step": 1215,
|
|
"token_acc": 0.6827804107424961
|
|
},
|
|
{
|
|
"epoch": 0.712778429073857,
|
|
"grad_norm": 3.421875,
|
|
"learning_rate": 2.0925396435598665e-06,
|
|
"loss": 1.1650149822235107,
|
|
"step": 1216,
|
|
"token_acc": 0.6941612604263206
|
|
},
|
|
{
|
|
"epoch": 0.7133645955451349,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 2.0846566920793265e-06,
|
|
"loss": 1.1702353954315186,
|
|
"step": 1217,
|
|
"token_acc": 0.6988505747126437
|
|
},
|
|
{
|
|
"epoch": 0.7139507620164126,
|
|
"grad_norm": 3.546875,
|
|
"learning_rate": 2.076784704352835e-06,
|
|
"loss": 1.1235432624816895,
|
|
"step": 1218,
|
|
"token_acc": 0.6912669126691267
|
|
},
|
|
{
|
|
"epoch": 0.7145369284876905,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 2.068923709984637e-06,
|
|
"loss": 1.0548449754714966,
|
|
"step": 1219,
|
|
"token_acc": 0.7156731413261889
|
|
},
|
|
{
|
|
"epoch": 0.7151230949589683,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 2.061073738537635e-06,
|
|
"loss": 1.1902203559875488,
|
|
"step": 1220,
|
|
"token_acc": 0.6810596026490067
|
|
},
|
|
{
|
|
"epoch": 0.7157092614302462,
|
|
"grad_norm": 3.53125,
|
|
"learning_rate": 2.053234819533276e-06,
|
|
"loss": 1.0891600847244263,
|
|
"step": 1221,
|
|
"token_acc": 0.7060320692288115
|
|
},
|
|
{
|
|
"epoch": 0.716295427901524,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 2.0454069824514445e-06,
|
|
"loss": 1.1338796615600586,
|
|
"step": 1222,
|
|
"token_acc": 0.7021733297558358
|
|
},
|
|
{
|
|
"epoch": 0.7168815943728019,
|
|
"grad_norm": 3.390625,
|
|
"learning_rate": 2.0375902567303474e-06,
|
|
"loss": 1.1477164030075073,
|
|
"step": 1223,
|
|
"token_acc": 0.6888463379607468
|
|
},
|
|
{
|
|
"epoch": 0.7174677608440797,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 2.0297846717664043e-06,
|
|
"loss": 1.1622651815414429,
|
|
"step": 1224,
|
|
"token_acc": 0.6925490196078431
|
|
},
|
|
{
|
|
"epoch": 0.7180539273153576,
|
|
"grad_norm": 3.5,
|
|
"learning_rate": 2.0219902569141405e-06,
|
|
"loss": 1.2012808322906494,
|
|
"step": 1225,
|
|
"token_acc": 0.6851716581446311
|
|
},
|
|
{
|
|
"epoch": 0.7186400937866354,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 2.0142070414860704e-06,
|
|
"loss": 1.131361484527588,
|
|
"step": 1226,
|
|
"token_acc": 0.7001789976133651
|
|
},
|
|
{
|
|
"epoch": 0.7192262602579133,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 2.006435054752593e-06,
|
|
"loss": 1.167328119277954,
|
|
"step": 1227,
|
|
"token_acc": 0.6954691259021651
|
|
},
|
|
{
|
|
"epoch": 0.7198124267291911,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 1.9986743259418786e-06,
|
|
"loss": 1.1792668104171753,
|
|
"step": 1228,
|
|
"token_acc": 0.6879699248120301
|
|
},
|
|
{
|
|
"epoch": 0.720398593200469,
|
|
"grad_norm": 3.5,
|
|
"learning_rate": 1.990924884239758e-06,
|
|
"loss": 1.1464402675628662,
|
|
"step": 1229,
|
|
"token_acc": 0.6912114014251781
|
|
},
|
|
{
|
|
"epoch": 0.7209847596717468,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 1.983186758789622e-06,
|
|
"loss": 1.1427706480026245,
|
|
"step": 1230,
|
|
"token_acc": 0.6978557504873294
|
|
},
|
|
{
|
|
"epoch": 0.7215709261430246,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 1.9754599786922913e-06,
|
|
"loss": 1.1599314212799072,
|
|
"step": 1231,
|
|
"token_acc": 0.6835755415409926
|
|
},
|
|
{
|
|
"epoch": 0.7221570926143025,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 1.9677445730059348e-06,
|
|
"loss": 1.1890777349472046,
|
|
"step": 1232,
|
|
"token_acc": 0.6848849372384938
|
|
},
|
|
{
|
|
"epoch": 0.7227432590855803,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 1.96004057074593e-06,
|
|
"loss": 1.152843713760376,
|
|
"step": 1233,
|
|
"token_acc": 0.6829804784162772
|
|
},
|
|
{
|
|
"epoch": 0.7233294255568582,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 1.9523480008847856e-06,
|
|
"loss": 1.1408559083938599,
|
|
"step": 1234,
|
|
"token_acc": 0.6998435054773083
|
|
},
|
|
{
|
|
"epoch": 0.723915592028136,
|
|
"grad_norm": 3.484375,
|
|
"learning_rate": 1.9446668923520014e-06,
|
|
"loss": 1.1791422367095947,
|
|
"step": 1235,
|
|
"token_acc": 0.6922155688622754
|
|
},
|
|
{
|
|
"epoch": 0.7245017584994138,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 1.936997274033986e-06,
|
|
"loss": 1.1258341073989868,
|
|
"step": 1236,
|
|
"token_acc": 0.7120146744114949
|
|
},
|
|
{
|
|
"epoch": 0.7250879249706916,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 1.929339174773931e-06,
|
|
"loss": 1.1504864692687988,
|
|
"step": 1237,
|
|
"token_acc": 0.7015613002303558
|
|
},
|
|
{
|
|
"epoch": 0.7256740914419695,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 1.9216926233717087e-06,
|
|
"loss": 1.1897146701812744,
|
|
"step": 1238,
|
|
"token_acc": 0.6875387476751394
|
|
},
|
|
{
|
|
"epoch": 0.7262602579132473,
|
|
"grad_norm": 4.28125,
|
|
"learning_rate": 1.9140576485837646e-06,
|
|
"loss": 1.1358810663223267,
|
|
"step": 1239,
|
|
"token_acc": 0.6957486136783734
|
|
},
|
|
{
|
|
"epoch": 0.7268464243845252,
|
|
"grad_norm": 3.3125,
|
|
"learning_rate": 1.9064342791230072e-06,
|
|
"loss": 1.1267626285552979,
|
|
"step": 1240,
|
|
"token_acc": 0.6985074626865672
|
|
},
|
|
{
|
|
"epoch": 0.727432590855803,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 1.8988225436587005e-06,
|
|
"loss": 1.1859606504440308,
|
|
"step": 1241,
|
|
"token_acc": 0.6939804601323668
|
|
},
|
|
{
|
|
"epoch": 0.7280187573270809,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 1.8912224708163561e-06,
|
|
"loss": 1.2938754558563232,
|
|
"step": 1242,
|
|
"token_acc": 0.6840361445783133
|
|
},
|
|
{
|
|
"epoch": 0.7286049237983587,
|
|
"grad_norm": 3.421875,
|
|
"learning_rate": 1.8836340891776278e-06,
|
|
"loss": 1.1167583465576172,
|
|
"step": 1243,
|
|
"token_acc": 0.6932485322896281
|
|
},
|
|
{
|
|
"epoch": 0.7291910902696366,
|
|
"grad_norm": 3.46875,
|
|
"learning_rate": 1.8760574272802002e-06,
|
|
"loss": 1.194901943206787,
|
|
"step": 1244,
|
|
"token_acc": 0.6846725185685347
|
|
},
|
|
{
|
|
"epoch": 0.7297772567409144,
|
|
"grad_norm": 3.375,
|
|
"learning_rate": 1.8684925136176836e-06,
|
|
"loss": 1.1578341722488403,
|
|
"step": 1245,
|
|
"token_acc": 0.7008304195804196
|
|
},
|
|
{
|
|
"epoch": 0.7303634232121923,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 1.8609393766395083e-06,
|
|
"loss": 1.2012755870819092,
|
|
"step": 1246,
|
|
"token_acc": 0.6877323420074349
|
|
},
|
|
{
|
|
"epoch": 0.7309495896834701,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 1.8533980447508138e-06,
|
|
"loss": 1.272649884223938,
|
|
"step": 1247,
|
|
"token_acc": 0.6703397612488522
|
|
},
|
|
{
|
|
"epoch": 0.731535756154748,
|
|
"grad_norm": 3.34375,
|
|
"learning_rate": 1.8458685463123438e-06,
|
|
"loss": 1.2019352912902832,
|
|
"step": 1248,
|
|
"token_acc": 0.6837002625732175
|
|
},
|
|
{
|
|
"epoch": 0.7321219226260258,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 1.8383509096403463e-06,
|
|
"loss": 1.1311109066009521,
|
|
"step": 1249,
|
|
"token_acc": 0.694199411921946
|
|
},
|
|
{
|
|
"epoch": 0.7327080890973037,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 1.8308451630064484e-06,
|
|
"loss": 1.1709685325622559,
|
|
"step": 1250,
|
|
"token_acc": 0.6866477272727273
|
|
},
|
|
{
|
|
"epoch": 0.7332942555685815,
|
|
"grad_norm": 3.296875,
|
|
"learning_rate": 1.823351334637576e-06,
|
|
"loss": 1.1879963874816895,
|
|
"step": 1251,
|
|
"token_acc": 0.6880894933099364
|
|
},
|
|
{
|
|
"epoch": 0.7338804220398594,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 1.8158694527158205e-06,
|
|
"loss": 1.175192952156067,
|
|
"step": 1252,
|
|
"token_acc": 0.69271699368141
|
|
},
|
|
{
|
|
"epoch": 0.7344665885111371,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 1.8083995453783604e-06,
|
|
"loss": 1.203125,
|
|
"step": 1253,
|
|
"token_acc": 0.6890600106213489
|
|
},
|
|
{
|
|
"epoch": 0.735052754982415,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 1.8009416407173258e-06,
|
|
"loss": 1.2011289596557617,
|
|
"step": 1254,
|
|
"token_acc": 0.6884210526315789
|
|
},
|
|
{
|
|
"epoch": 0.7356389214536928,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 1.7934957667797226e-06,
|
|
"loss": 1.1292072534561157,
|
|
"step": 1255,
|
|
"token_acc": 0.6998153521498286
|
|
},
|
|
{
|
|
"epoch": 0.7362250879249707,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 1.7860619515673034e-06,
|
|
"loss": 1.146385669708252,
|
|
"step": 1256,
|
|
"token_acc": 0.6939917295062029
|
|
},
|
|
{
|
|
"epoch": 0.7368112543962485,
|
|
"grad_norm": 3.359375,
|
|
"learning_rate": 1.7786402230364753e-06,
|
|
"loss": 1.1327096223831177,
|
|
"step": 1257,
|
|
"token_acc": 0.6958110082805651
|
|
},
|
|
{
|
|
"epoch": 0.7373974208675264,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 1.7712306090981896e-06,
|
|
"loss": 1.1948044300079346,
|
|
"step": 1258,
|
|
"token_acc": 0.6816460489775309
|
|
},
|
|
{
|
|
"epoch": 0.7379835873388042,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 1.7638331376178385e-06,
|
|
"loss": 1.1925272941589355,
|
|
"step": 1259,
|
|
"token_acc": 0.6870574907958086
|
|
},
|
|
{
|
|
"epoch": 0.738569753810082,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 1.75644783641515e-06,
|
|
"loss": 1.1774451732635498,
|
|
"step": 1260,
|
|
"token_acc": 0.6901811248808389
|
|
},
|
|
{
|
|
"epoch": 0.7391559202813599,
|
|
"grad_norm": 4.3125,
|
|
"learning_rate": 1.7490747332640833e-06,
|
|
"loss": 1.1047813892364502,
|
|
"step": 1261,
|
|
"token_acc": 0.7107583774250441
|
|
},
|
|
{
|
|
"epoch": 0.7397420867526378,
|
|
"grad_norm": 3.53125,
|
|
"learning_rate": 1.7417138558927244e-06,
|
|
"loss": 1.1908293962478638,
|
|
"step": 1262,
|
|
"token_acc": 0.6852667645619187
|
|
},
|
|
{
|
|
"epoch": 0.7403282532239156,
|
|
"grad_norm": 3.390625,
|
|
"learning_rate": 1.7343652319831817e-06,
|
|
"loss": 1.178828477859497,
|
|
"step": 1263,
|
|
"token_acc": 0.6972235519387266
|
|
},
|
|
{
|
|
"epoch": 0.7409144196951934,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 1.7270288891714814e-06,
|
|
"loss": 1.1311687231063843,
|
|
"step": 1264,
|
|
"token_acc": 0.6926475228778795
|
|
},
|
|
{
|
|
"epoch": 0.7415005861664713,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 1.7197048550474643e-06,
|
|
"loss": 1.1658614873886108,
|
|
"step": 1265,
|
|
"token_acc": 0.6876222408493993
|
|
},
|
|
{
|
|
"epoch": 0.7420867526377491,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 1.7123931571546826e-06,
|
|
"loss": 1.2029447555541992,
|
|
"step": 1266,
|
|
"token_acc": 0.6873638344226579
|
|
},
|
|
{
|
|
"epoch": 0.742672919109027,
|
|
"grad_norm": 3.609375,
|
|
"learning_rate": 1.7050938229902946e-06,
|
|
"loss": 1.1587637662887573,
|
|
"step": 1267,
|
|
"token_acc": 0.6992481203007519
|
|
},
|
|
{
|
|
"epoch": 0.7432590855803048,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 1.6978068800049624e-06,
|
|
"loss": 1.2535085678100586,
|
|
"step": 1268,
|
|
"token_acc": 0.6924697209057399
|
|
},
|
|
{
|
|
"epoch": 0.7438452520515827,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 1.6905323556027475e-06,
|
|
"loss": 1.1496511697769165,
|
|
"step": 1269,
|
|
"token_acc": 0.6860819505587538
|
|
},
|
|
{
|
|
"epoch": 0.7444314185228605,
|
|
"grad_norm": 3.3125,
|
|
"learning_rate": 1.6832702771410142e-06,
|
|
"loss": 1.1488885879516602,
|
|
"step": 1270,
|
|
"token_acc": 0.6906906906906907
|
|
},
|
|
{
|
|
"epoch": 0.7450175849941383,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 1.6760206719303107e-06,
|
|
"loss": 1.2161635160446167,
|
|
"step": 1271,
|
|
"token_acc": 0.6901779828609097
|
|
},
|
|
{
|
|
"epoch": 0.7456037514654161,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 1.6687835672342895e-06,
|
|
"loss": 1.1316810846328735,
|
|
"step": 1272,
|
|
"token_acc": 0.7051282051282052
|
|
},
|
|
{
|
|
"epoch": 0.746189917936694,
|
|
"grad_norm": 3.921875,
|
|
"learning_rate": 1.66155899026958e-06,
|
|
"loss": 1.164171814918518,
|
|
"step": 1273,
|
|
"token_acc": 0.6980168269230769
|
|
},
|
|
{
|
|
"epoch": 0.7467760844079718,
|
|
"grad_norm": 4.34375,
|
|
"learning_rate": 1.6543469682057105e-06,
|
|
"loss": 1.116528034210205,
|
|
"step": 1274,
|
|
"token_acc": 0.7083010085337471
|
|
},
|
|
{
|
|
"epoch": 0.7473622508792497,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 1.6471475281649818e-06,
|
|
"loss": 1.1532490253448486,
|
|
"step": 1275,
|
|
"token_acc": 0.6980898040188539
|
|
},
|
|
{
|
|
"epoch": 0.7479484173505275,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 1.639960697222388e-06,
|
|
"loss": 1.1486763954162598,
|
|
"step": 1276,
|
|
"token_acc": 0.6967071057192374
|
|
},
|
|
{
|
|
"epoch": 0.7485345838218054,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 1.6327865024054984e-06,
|
|
"loss": 1.2091307640075684,
|
|
"step": 1277,
|
|
"token_acc": 0.678146121660766
|
|
},
|
|
{
|
|
"epoch": 0.7491207502930832,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 1.6256249706943628e-06,
|
|
"loss": 1.1342439651489258,
|
|
"step": 1278,
|
|
"token_acc": 0.6975622823466381
|
|
},
|
|
{
|
|
"epoch": 0.7497069167643611,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 1.6184761290214079e-06,
|
|
"loss": 1.1738226413726807,
|
|
"step": 1279,
|
|
"token_acc": 0.6877133105802048
|
|
},
|
|
{
|
|
"epoch": 0.7502930832356389,
|
|
"grad_norm": 3.609375,
|
|
"learning_rate": 1.611340004271339e-06,
|
|
"loss": 1.169606328010559,
|
|
"step": 1280,
|
|
"token_acc": 0.6880668257756564
|
|
},
|
|
{
|
|
"epoch": 0.7508792497069168,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 1.6042166232810346e-06,
|
|
"loss": 1.2163209915161133,
|
|
"step": 1281,
|
|
"token_acc": 0.6759191624056489
|
|
},
|
|
{
|
|
"epoch": 0.7514654161781946,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 1.5971060128394483e-06,
|
|
"loss": 1.1251789331436157,
|
|
"step": 1282,
|
|
"token_acc": 0.6923272357723578
|
|
},
|
|
{
|
|
"epoch": 0.7520515826494725,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 1.5900081996875083e-06,
|
|
"loss": 1.1192710399627686,
|
|
"step": 1283,
|
|
"token_acc": 0.7018633540372671
|
|
},
|
|
{
|
|
"epoch": 0.7526377491207503,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 1.5829232105180143e-06,
|
|
"loss": 1.1664652824401855,
|
|
"step": 1284,
|
|
"token_acc": 0.6933539412673879
|
|
},
|
|
{
|
|
"epoch": 0.7532239155920282,
|
|
"grad_norm": 4.46875,
|
|
"learning_rate": 1.575851071975541e-06,
|
|
"loss": 1.2514925003051758,
|
|
"step": 1285,
|
|
"token_acc": 0.6723007933770265
|
|
},
|
|
{
|
|
"epoch": 0.753810082063306,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 1.5687918106563326e-06,
|
|
"loss": 1.1698015928268433,
|
|
"step": 1286,
|
|
"token_acc": 0.6963337547408344
|
|
},
|
|
{
|
|
"epoch": 0.7543962485345839,
|
|
"grad_norm": 4.28125,
|
|
"learning_rate": 1.5617454531082093e-06,
|
|
"loss": 1.167072057723999,
|
|
"step": 1287,
|
|
"token_acc": 0.6807054488023164
|
|
},
|
|
{
|
|
"epoch": 0.7549824150058617,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 1.55471202583046e-06,
|
|
"loss": 1.0630364418029785,
|
|
"step": 1288,
|
|
"token_acc": 0.7089201877934272
|
|
},
|
|
{
|
|
"epoch": 0.7555685814771395,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 1.5476915552737532e-06,
|
|
"loss": 1.235093116760254,
|
|
"step": 1289,
|
|
"token_acc": 0.6780341023069207
|
|
},
|
|
{
|
|
"epoch": 0.7561547479484173,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 1.5406840678400204e-06,
|
|
"loss": 1.2358726263046265,
|
|
"step": 1290,
|
|
"token_acc": 0.6792672288455947
|
|
},
|
|
{
|
|
"epoch": 0.7567409144196952,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 1.5336895898823801e-06,
|
|
"loss": 1.1916831731796265,
|
|
"step": 1291,
|
|
"token_acc": 0.6888368004687958
|
|
},
|
|
{
|
|
"epoch": 0.757327080890973,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 1.5267081477050132e-06,
|
|
"loss": 1.1831395626068115,
|
|
"step": 1292,
|
|
"token_acc": 0.6969891095451634
|
|
},
|
|
{
|
|
"epoch": 0.7579132473622509,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 1.51973976756309e-06,
|
|
"loss": 1.175769329071045,
|
|
"step": 1293,
|
|
"token_acc": 0.692980264801399
|
|
},
|
|
{
|
|
"epoch": 0.7584994138335287,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 1.5127844756626437e-06,
|
|
"loss": 1.201106071472168,
|
|
"step": 1294,
|
|
"token_acc": 0.6828478964401294
|
|
},
|
|
{
|
|
"epoch": 0.7590855803048066,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 1.5058422981604998e-06,
|
|
"loss": 1.2084473371505737,
|
|
"step": 1295,
|
|
"token_acc": 0.6897353881942425
|
|
},
|
|
{
|
|
"epoch": 0.7596717467760844,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 1.4989132611641576e-06,
|
|
"loss": 1.1355628967285156,
|
|
"step": 1296,
|
|
"token_acc": 0.7005920496193967
|
|
},
|
|
{
|
|
"epoch": 0.7602579132473622,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 1.4919973907316987e-06,
|
|
"loss": 1.11936616897583,
|
|
"step": 1297,
|
|
"token_acc": 0.6978680879413725
|
|
},
|
|
{
|
|
"epoch": 0.7608440797186401,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 1.4850947128716914e-06,
|
|
"loss": 1.1855089664459229,
|
|
"step": 1298,
|
|
"token_acc": 0.687462326702833
|
|
},
|
|
{
|
|
"epoch": 0.761430246189918,
|
|
"grad_norm": 3.5,
|
|
"learning_rate": 1.4782052535430896e-06,
|
|
"loss": 1.167141318321228,
|
|
"step": 1299,
|
|
"token_acc": 0.6878651685393259
|
|
},
|
|
{
|
|
"epoch": 0.7620164126611958,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 1.471329038655135e-06,
|
|
"loss": 1.2189459800720215,
|
|
"step": 1300,
|
|
"token_acc": 0.6868250539956804
|
|
},
|
|
{
|
|
"epoch": 0.7626025791324736,
|
|
"grad_norm": 3.203125,
|
|
"learning_rate": 1.4644660940672628e-06,
|
|
"loss": 1.2017531394958496,
|
|
"step": 1301,
|
|
"token_acc": 0.6795935647756138
|
|
},
|
|
{
|
|
"epoch": 0.7631887456037515,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 1.4576164455890014e-06,
|
|
"loss": 1.1633294820785522,
|
|
"step": 1302,
|
|
"token_acc": 0.6994480682388359
|
|
},
|
|
{
|
|
"epoch": 0.7637749120750293,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 1.4507801189798765e-06,
|
|
"loss": 1.2420661449432373,
|
|
"step": 1303,
|
|
"token_acc": 0.6744042092231507
|
|
},
|
|
{
|
|
"epoch": 0.7643610785463072,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 1.4439571399493146e-06,
|
|
"loss": 1.1757030487060547,
|
|
"step": 1304,
|
|
"token_acc": 0.7008875739644971
|
|
},
|
|
{
|
|
"epoch": 0.764947245017585,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 1.4371475341565455e-06,
|
|
"loss": 1.1364424228668213,
|
|
"step": 1305,
|
|
"token_acc": 0.6993639102778708
|
|
},
|
|
{
|
|
"epoch": 0.7655334114888629,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 1.4303513272105057e-06,
|
|
"loss": 1.1546838283538818,
|
|
"step": 1306,
|
|
"token_acc": 0.6872586872586872
|
|
},
|
|
{
|
|
"epoch": 0.7661195779601406,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 1.4235685446697433e-06,
|
|
"loss": 1.1485613584518433,
|
|
"step": 1307,
|
|
"token_acc": 0.6898591549295775
|
|
},
|
|
{
|
|
"epoch": 0.7667057444314185,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 1.4167992120423212e-06,
|
|
"loss": 1.1324533224105835,
|
|
"step": 1308,
|
|
"token_acc": 0.6994986729578295
|
|
},
|
|
{
|
|
"epoch": 0.7672919109026963,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 1.410043354785719e-06,
|
|
"loss": 1.1110223531723022,
|
|
"step": 1309,
|
|
"token_acc": 0.7076526225279449
|
|
},
|
|
{
|
|
"epoch": 0.7678780773739742,
|
|
"grad_norm": 4.65625,
|
|
"learning_rate": 1.4033009983067454e-06,
|
|
"loss": 1.2463672161102295,
|
|
"step": 1310,
|
|
"token_acc": 0.6817196531791907
|
|
},
|
|
{
|
|
"epoch": 0.768464243845252,
|
|
"grad_norm": 5.0625,
|
|
"learning_rate": 1.396572167961427e-06,
|
|
"loss": 1.2277967929840088,
|
|
"step": 1311,
|
|
"token_acc": 0.6886380189366351
|
|
},
|
|
{
|
|
"epoch": 0.7690504103165299,
|
|
"grad_norm": 3.40625,
|
|
"learning_rate": 1.3898568890549335e-06,
|
|
"loss": 1.1662789583206177,
|
|
"step": 1312,
|
|
"token_acc": 0.6930980567344204
|
|
},
|
|
{
|
|
"epoch": 0.7696365767878077,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 1.38315518684146e-06,
|
|
"loss": 1.2122944593429565,
|
|
"step": 1313,
|
|
"token_acc": 0.6842105263157895
|
|
},
|
|
{
|
|
"epoch": 0.7702227432590856,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 1.3764670865241557e-06,
|
|
"loss": 1.244516134262085,
|
|
"step": 1314,
|
|
"token_acc": 0.6802059496567505
|
|
},
|
|
{
|
|
"epoch": 0.7708089097303634,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 1.3697926132550054e-06,
|
|
"loss": 1.2574474811553955,
|
|
"step": 1315,
|
|
"token_acc": 0.6763496143958869
|
|
},
|
|
{
|
|
"epoch": 0.7713950762016413,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 1.3631317921347564e-06,
|
|
"loss": 1.1313972473144531,
|
|
"step": 1316,
|
|
"token_acc": 0.6993661334138244
|
|
},
|
|
{
|
|
"epoch": 0.7719812426729191,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 1.3564846482128086e-06,
|
|
"loss": 1.2453927993774414,
|
|
"step": 1317,
|
|
"token_acc": 0.6780541131456682
|
|
},
|
|
{
|
|
"epoch": 0.772567409144197,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 1.3498512064871272e-06,
|
|
"loss": 1.154325008392334,
|
|
"step": 1318,
|
|
"token_acc": 0.6894618834080718
|
|
},
|
|
{
|
|
"epoch": 0.7731535756154748,
|
|
"grad_norm": 3.453125,
|
|
"learning_rate": 1.3432314919041478e-06,
|
|
"loss": 1.1832101345062256,
|
|
"step": 1319,
|
|
"token_acc": 0.6847904191616766
|
|
},
|
|
{
|
|
"epoch": 0.7737397420867527,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 1.3366255293586822e-06,
|
|
"loss": 1.2077072858810425,
|
|
"step": 1320,
|
|
"token_acc": 0.6833192209991532
|
|
},
|
|
{
|
|
"epoch": 0.7743259085580305,
|
|
"grad_norm": 3.5,
|
|
"learning_rate": 1.330033343693824e-06,
|
|
"loss": 1.1287164688110352,
|
|
"step": 1321,
|
|
"token_acc": 0.6950284442245857
|
|
},
|
|
{
|
|
"epoch": 0.7749120750293084,
|
|
"grad_norm": 3.453125,
|
|
"learning_rate": 1.3234549597008572e-06,
|
|
"loss": 1.1538439989089966,
|
|
"step": 1322,
|
|
"token_acc": 0.6900198281559815
|
|
},
|
|
{
|
|
"epoch": 0.7754982415005862,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 1.3168904021191614e-06,
|
|
"loss": 1.180354356765747,
|
|
"step": 1323,
|
|
"token_acc": 0.6950146627565983
|
|
},
|
|
{
|
|
"epoch": 0.776084407971864,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 1.310339695636118e-06,
|
|
"loss": 1.2135355472564697,
|
|
"step": 1324,
|
|
"token_acc": 0.685226019845645
|
|
},
|
|
{
|
|
"epoch": 0.7766705744431418,
|
|
"grad_norm": 4.71875,
|
|
"learning_rate": 1.3038028648870204e-06,
|
|
"loss": 1.1843690872192383,
|
|
"step": 1325,
|
|
"token_acc": 0.7008733624454149
|
|
},
|
|
{
|
|
"epoch": 0.7772567409144197,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 1.297279934454978e-06,
|
|
"loss": 1.1952319145202637,
|
|
"step": 1326,
|
|
"token_acc": 0.6818901198281709
|
|
},
|
|
{
|
|
"epoch": 0.7778429073856975,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 1.2907709288708252e-06,
|
|
"loss": 1.1513004302978516,
|
|
"step": 1327,
|
|
"token_acc": 0.6983880967141971
|
|
},
|
|
{
|
|
"epoch": 0.7784290738569754,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 1.2842758726130283e-06,
|
|
"loss": 1.1462223529815674,
|
|
"step": 1328,
|
|
"token_acc": 0.6938303341902313
|
|
},
|
|
{
|
|
"epoch": 0.7790152403282532,
|
|
"grad_norm": 3.484375,
|
|
"learning_rate": 1.2777947901075999e-06,
|
|
"loss": 1.1672712564468384,
|
|
"step": 1329,
|
|
"token_acc": 0.6960408684546615
|
|
},
|
|
{
|
|
"epoch": 0.779601406799531,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 1.271327705727991e-06,
|
|
"loss": 1.130542278289795,
|
|
"step": 1330,
|
|
"token_acc": 0.6956037840845855
|
|
},
|
|
{
|
|
"epoch": 0.7801875732708089,
|
|
"grad_norm": 4.375,
|
|
"learning_rate": 1.264874643795021e-06,
|
|
"loss": 1.1141135692596436,
|
|
"step": 1331,
|
|
"token_acc": 0.7003125
|
|
},
|
|
{
|
|
"epoch": 0.7807737397420867,
|
|
"grad_norm": 4.28125,
|
|
"learning_rate": 1.2584356285767652e-06,
|
|
"loss": 1.1702884435653687,
|
|
"step": 1332,
|
|
"token_acc": 0.6837712519319938
|
|
},
|
|
{
|
|
"epoch": 0.7813599062133646,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 1.252010684288482e-06,
|
|
"loss": 1.1452285051345825,
|
|
"step": 1333,
|
|
"token_acc": 0.6945346320346321
|
|
},
|
|
{
|
|
"epoch": 0.7819460726846424,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 1.2455998350925042e-06,
|
|
"loss": 1.1990966796875,
|
|
"step": 1334,
|
|
"token_acc": 0.6829465731246627
|
|
},
|
|
{
|
|
"epoch": 0.7825322391559203,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 1.239203105098165e-06,
|
|
"loss": 1.2190043926239014,
|
|
"step": 1335,
|
|
"token_acc": 0.6880026809651475
|
|
},
|
|
{
|
|
"epoch": 0.7831184056271981,
|
|
"grad_norm": 3.921875,
|
|
"learning_rate": 1.2328205183616964e-06,
|
|
"loss": 1.1903975009918213,
|
|
"step": 1336,
|
|
"token_acc": 0.6826252377932783
|
|
},
|
|
{
|
|
"epoch": 0.783704572098476,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 1.22645209888614e-06,
|
|
"loss": 1.206367015838623,
|
|
"step": 1337,
|
|
"token_acc": 0.679304347826087
|
|
},
|
|
{
|
|
"epoch": 0.7842907385697538,
|
|
"grad_norm": 4.4375,
|
|
"learning_rate": 1.2200978706212606e-06,
|
|
"loss": 1.2379748821258545,
|
|
"step": 1338,
|
|
"token_acc": 0.6862680683311432
|
|
},
|
|
{
|
|
"epoch": 0.7848769050410317,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 1.213757857463454e-06,
|
|
"loss": 1.2319223880767822,
|
|
"step": 1339,
|
|
"token_acc": 0.6776232616940582
|
|
},
|
|
{
|
|
"epoch": 0.7854630715123095,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 1.2074320832556558e-06,
|
|
"loss": 1.2280304431915283,
|
|
"step": 1340,
|
|
"token_acc": 0.6880170172536043
|
|
},
|
|
{
|
|
"epoch": 0.7860492379835874,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 1.2011205717872538e-06,
|
|
"loss": 1.1342203617095947,
|
|
"step": 1341,
|
|
"token_acc": 0.7115268557634279
|
|
},
|
|
{
|
|
"epoch": 0.7866354044548651,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 1.1948233467939978e-06,
|
|
"loss": 1.172709584236145,
|
|
"step": 1342,
|
|
"token_acc": 0.692812602559895
|
|
},
|
|
{
|
|
"epoch": 0.787221570926143,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 1.1885404319579108e-06,
|
|
"loss": 1.0814075469970703,
|
|
"step": 1343,
|
|
"token_acc": 0.7192419825072887
|
|
},
|
|
{
|
|
"epoch": 0.7878077373974208,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 1.182271850907199e-06,
|
|
"loss": 1.198106050491333,
|
|
"step": 1344,
|
|
"token_acc": 0.6834811529933481
|
|
},
|
|
{
|
|
"epoch": 0.7883939038686987,
|
|
"grad_norm": 3.4375,
|
|
"learning_rate": 1.1760176272161627e-06,
|
|
"loss": 1.1613229513168335,
|
|
"step": 1345,
|
|
"token_acc": 0.6876267748478702
|
|
},
|
|
{
|
|
"epoch": 0.7889800703399765,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 1.1697777844051105e-06,
|
|
"loss": 1.2328696250915527,
|
|
"step": 1346,
|
|
"token_acc": 0.6805009823182712
|
|
},
|
|
{
|
|
"epoch": 0.7895662368112544,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 1.1635523459402665e-06,
|
|
"loss": 1.1123777627944946,
|
|
"step": 1347,
|
|
"token_acc": 0.7021732476277931
|
|
},
|
|
{
|
|
"epoch": 0.7901524032825322,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 1.1573413352336848e-06,
|
|
"loss": 1.1592592000961304,
|
|
"step": 1348,
|
|
"token_acc": 0.6992969172525689
|
|
},
|
|
{
|
|
"epoch": 0.7907385697538101,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 1.1511447756431604e-06,
|
|
"loss": 1.1294673681259155,
|
|
"step": 1349,
|
|
"token_acc": 0.7047795088460523
|
|
},
|
|
{
|
|
"epoch": 0.7913247362250879,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 1.1449626904721472e-06,
|
|
"loss": 1.1190552711486816,
|
|
"step": 1350,
|
|
"token_acc": 0.699135446685879
|
|
},
|
|
{
|
|
"epoch": 0.7919109026963658,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 1.1387951029696543e-06,
|
|
"loss": 1.1819028854370117,
|
|
"step": 1351,
|
|
"token_acc": 0.687636761487965
|
|
},
|
|
{
|
|
"epoch": 0.7924970691676436,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 1.132642036330181e-06,
|
|
"loss": 1.177093744277954,
|
|
"step": 1352,
|
|
"token_acc": 0.6918694976696013
|
|
},
|
|
{
|
|
"epoch": 0.7930832356389215,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 1.1265035136936076e-06,
|
|
"loss": 1.2067594528198242,
|
|
"step": 1353,
|
|
"token_acc": 0.6835334476843911
|
|
},
|
|
{
|
|
"epoch": 0.7936694021101993,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 1.1203795581451288e-06,
|
|
"loss": 1.1559319496154785,
|
|
"step": 1354,
|
|
"token_acc": 0.6984913307813555
|
|
},
|
|
{
|
|
"epoch": 0.7942555685814772,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 1.1142701927151456e-06,
|
|
"loss": 1.228494644165039,
|
|
"step": 1355,
|
|
"token_acc": 0.6728238866396761
|
|
},
|
|
{
|
|
"epoch": 0.794841735052755,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 1.1081754403792e-06,
|
|
"loss": 1.1023423671722412,
|
|
"step": 1356,
|
|
"token_acc": 0.7045222713362802
|
|
},
|
|
{
|
|
"epoch": 0.7954279015240329,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 1.1020953240578724e-06,
|
|
"loss": 1.1845364570617676,
|
|
"step": 1357,
|
|
"token_acc": 0.6885321100917431
|
|
},
|
|
{
|
|
"epoch": 0.7960140679953107,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 1.096029866616704e-06,
|
|
"loss": 1.1747312545776367,
|
|
"step": 1358,
|
|
"token_acc": 0.6868194842406877
|
|
},
|
|
{
|
|
"epoch": 0.7966002344665886,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 1.0899790908661061e-06,
|
|
"loss": 1.1109538078308105,
|
|
"step": 1359,
|
|
"token_acc": 0.7081585081585081
|
|
},
|
|
{
|
|
"epoch": 0.7971864009378663,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 1.0839430195612794e-06,
|
|
"loss": 1.1851210594177246,
|
|
"step": 1360,
|
|
"token_acc": 0.6785046728971963
|
|
},
|
|
{
|
|
"epoch": 0.7977725674091442,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 1.0779216754021215e-06,
|
|
"loss": 1.208925724029541,
|
|
"step": 1361,
|
|
"token_acc": 0.6902138690861958
|
|
},
|
|
{
|
|
"epoch": 0.798358733880422,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 1.0719150810331497e-06,
|
|
"loss": 1.1703481674194336,
|
|
"step": 1362,
|
|
"token_acc": 0.6918819188191881
|
|
},
|
|
{
|
|
"epoch": 0.7989449003516998,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 1.0659232590434094e-06,
|
|
"loss": 1.1322376728057861,
|
|
"step": 1363,
|
|
"token_acc": 0.7017881705639615
|
|
},
|
|
{
|
|
"epoch": 0.7995310668229777,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 1.0599462319663906e-06,
|
|
"loss": 1.251816987991333,
|
|
"step": 1364,
|
|
"token_acc": 0.6712439418416801
|
|
},
|
|
{
|
|
"epoch": 0.8001172332942555,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 1.0539840222799463e-06,
|
|
"loss": 1.1773637533187866,
|
|
"step": 1365,
|
|
"token_acc": 0.6837782340862423
|
|
},
|
|
{
|
|
"epoch": 0.8007033997655334,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 1.0480366524062041e-06,
|
|
"loss": 1.1951593160629272,
|
|
"step": 1366,
|
|
"token_acc": 0.6737270875763748
|
|
},
|
|
{
|
|
"epoch": 0.8012895662368112,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 1.042104144711484e-06,
|
|
"loss": 1.0868040323257446,
|
|
"step": 1367,
|
|
"token_acc": 0.6985155195681512
|
|
},
|
|
{
|
|
"epoch": 0.8018757327080891,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 1.036186521506211e-06,
|
|
"loss": 1.1127233505249023,
|
|
"step": 1368,
|
|
"token_acc": 0.7087188113260443
|
|
},
|
|
{
|
|
"epoch": 0.8024618991793669,
|
|
"grad_norm": 3.5,
|
|
"learning_rate": 1.0302838050448416e-06,
|
|
"loss": 1.1603217124938965,
|
|
"step": 1369,
|
|
"token_acc": 0.696276357110812
|
|
},
|
|
{
|
|
"epoch": 0.8030480656506448,
|
|
"grad_norm": 3.421875,
|
|
"learning_rate": 1.0243960175257605e-06,
|
|
"loss": 1.1386816501617432,
|
|
"step": 1370,
|
|
"token_acc": 0.7071065989847716
|
|
},
|
|
{
|
|
"epoch": 0.8036342321219226,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 1.0185231810912223e-06,
|
|
"loss": 1.163072109222412,
|
|
"step": 1371,
|
|
"token_acc": 0.6945618484080023
|
|
},
|
|
{
|
|
"epoch": 0.8042203985932005,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 1.0126653178272422e-06,
|
|
"loss": 1.162095546722412,
|
|
"step": 1372,
|
|
"token_acc": 0.6942601672916066
|
|
},
|
|
{
|
|
"epoch": 0.8048065650644783,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 1.006822449763537e-06,
|
|
"loss": 1.1558542251586914,
|
|
"step": 1373,
|
|
"token_acc": 0.6954373728567277
|
|
},
|
|
{
|
|
"epoch": 0.8053927315357562,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 1.0009945988734205e-06,
|
|
"loss": 1.179213047027588,
|
|
"step": 1374,
|
|
"token_acc": 0.6888400219901044
|
|
},
|
|
{
|
|
"epoch": 0.805978898007034,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 9.9518178707374e-07,
|
|
"loss": 1.1593331098556519,
|
|
"step": 1375,
|
|
"token_acc": 0.6813301911495155
|
|
},
|
|
{
|
|
"epoch": 0.8065650644783119,
|
|
"grad_norm": 3.453125,
|
|
"learning_rate": 9.893840362247809e-07,
|
|
"loss": 1.137618064880371,
|
|
"step": 1376,
|
|
"token_acc": 0.7071644367716875
|
|
},
|
|
{
|
|
"epoch": 0.8071512309495897,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 9.8360136813019e-07,
|
|
"loss": 1.1997452974319458,
|
|
"step": 1377,
|
|
"token_acc": 0.6895522388059702
|
|
},
|
|
{
|
|
"epoch": 0.8077373974208675,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 9.778338045368901e-07,
|
|
"loss": 1.1998045444488525,
|
|
"step": 1378,
|
|
"token_acc": 0.6828509181450358
|
|
},
|
|
{
|
|
"epoch": 0.8083235638921453,
|
|
"grad_norm": 3.609375,
|
|
"learning_rate": 9.720813671350033e-07,
|
|
"loss": 1.1690711975097656,
|
|
"step": 1379,
|
|
"token_acc": 0.6833374597672691
|
|
},
|
|
{
|
|
"epoch": 0.8089097303634232,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 9.663440775577653e-07,
|
|
"loss": 1.161434531211853,
|
|
"step": 1380,
|
|
"token_acc": 0.6966176973009908
|
|
},
|
|
{
|
|
"epoch": 0.809495896834701,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 9.606219573814447e-07,
|
|
"loss": 1.2299973964691162,
|
|
"step": 1381,
|
|
"token_acc": 0.6788491446345256
|
|
},
|
|
{
|
|
"epoch": 0.8100820633059789,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 9.549150281252633e-07,
|
|
"loss": 1.2072006464004517,
|
|
"step": 1382,
|
|
"token_acc": 0.6825827455236029
|
|
},
|
|
{
|
|
"epoch": 0.8106682297772567,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 9.492233112513138e-07,
|
|
"loss": 1.2507808208465576,
|
|
"step": 1383,
|
|
"token_acc": 0.6778332145402709
|
|
},
|
|
{
|
|
"epoch": 0.8112543962485346,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 9.435468281644799e-07,
|
|
"loss": 1.1039612293243408,
|
|
"step": 1384,
|
|
"token_acc": 0.6984969053934571
|
|
},
|
|
{
|
|
"epoch": 0.8118405627198124,
|
|
"grad_norm": 3.484375,
|
|
"learning_rate": 9.378856002123549e-07,
|
|
"loss": 1.1911215782165527,
|
|
"step": 1385,
|
|
"token_acc": 0.6792989104689721
|
|
},
|
|
{
|
|
"epoch": 0.8124267291910903,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 9.322396486851626e-07,
|
|
"loss": 1.0605854988098145,
|
|
"step": 1386,
|
|
"token_acc": 0.7098159509202454
|
|
},
|
|
{
|
|
"epoch": 0.8130128956623681,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 9.26608994815677e-07,
|
|
"loss": 1.164738655090332,
|
|
"step": 1387,
|
|
"token_acc": 0.6910994764397905
|
|
},
|
|
{
|
|
"epoch": 0.813599062133646,
|
|
"grad_norm": 4.34375,
|
|
"learning_rate": 9.209936597791407e-07,
|
|
"loss": 1.196162462234497,
|
|
"step": 1388,
|
|
"token_acc": 0.6905071521456437
|
|
},
|
|
{
|
|
"epoch": 0.8141852286049238,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 9.15393664693186e-07,
|
|
"loss": 1.2342137098312378,
|
|
"step": 1389,
|
|
"token_acc": 0.6805138015851325
|
|
},
|
|
{
|
|
"epoch": 0.8147713950762017,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 9.098090306177626e-07,
|
|
"loss": 1.1733989715576172,
|
|
"step": 1390,
|
|
"token_acc": 0.6986058301647655
|
|
},
|
|
{
|
|
"epoch": 0.8153575615474795,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 9.042397785550405e-07,
|
|
"loss": 1.1252615451812744,
|
|
"step": 1391,
|
|
"token_acc": 0.6948801420538621
|
|
},
|
|
{
|
|
"epoch": 0.8159437280187574,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 8.98685929449355e-07,
|
|
"loss": 1.2019615173339844,
|
|
"step": 1392,
|
|
"token_acc": 0.6896726719106826
|
|
},
|
|
{
|
|
"epoch": 0.8165298944900352,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 8.931475041871035e-07,
|
|
"loss": 1.1389646530151367,
|
|
"step": 1393,
|
|
"token_acc": 0.6958131586442609
|
|
},
|
|
{
|
|
"epoch": 0.8171160609613131,
|
|
"grad_norm": 3.5,
|
|
"learning_rate": 8.876245235966884e-07,
|
|
"loss": 1.146538496017456,
|
|
"step": 1394,
|
|
"token_acc": 0.692367601246106
|
|
},
|
|
{
|
|
"epoch": 0.8177022274325909,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 8.821170084484248e-07,
|
|
"loss": 1.1713523864746094,
|
|
"step": 1395,
|
|
"token_acc": 0.6829871414441148
|
|
},
|
|
{
|
|
"epoch": 0.8182883939038686,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 8.766249794544662e-07,
|
|
"loss": 1.1876342296600342,
|
|
"step": 1396,
|
|
"token_acc": 0.6883078730904818
|
|
},
|
|
{
|
|
"epoch": 0.8188745603751465,
|
|
"grad_norm": 3.453125,
|
|
"learning_rate": 8.711484572687295e-07,
|
|
"loss": 1.1850242614746094,
|
|
"step": 1397,
|
|
"token_acc": 0.691921726041144
|
|
},
|
|
{
|
|
"epoch": 0.8194607268464243,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 8.656874624868133e-07,
|
|
"loss": 1.1436007022857666,
|
|
"step": 1398,
|
|
"token_acc": 0.6909578357299537
|
|
},
|
|
{
|
|
"epoch": 0.8200468933177022,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 8.602420156459223e-07,
|
|
"loss": 1.1556098461151123,
|
|
"step": 1399,
|
|
"token_acc": 0.6996789330698938
|
|
},
|
|
{
|
|
"epoch": 0.82063305978898,
|
|
"grad_norm": 3.546875,
|
|
"learning_rate": 8.54812137224792e-07,
|
|
"loss": 1.189967393875122,
|
|
"step": 1400,
|
|
"token_acc": 0.6873818525519849
|
|
},
|
|
{
|
|
"epoch": 0.8212192262602579,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 8.49397847643606e-07,
|
|
"loss": 1.1459509134292603,
|
|
"step": 1401,
|
|
"token_acc": 0.7027376033057852
|
|
},
|
|
{
|
|
"epoch": 0.8218053927315357,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 8.439991672639264e-07,
|
|
"loss": 1.142472505569458,
|
|
"step": 1402,
|
|
"token_acc": 0.6988065326633166
|
|
},
|
|
{
|
|
"epoch": 0.8223915592028136,
|
|
"grad_norm": 3.375,
|
|
"learning_rate": 8.38616116388612e-07,
|
|
"loss": 1.1672914028167725,
|
|
"step": 1403,
|
|
"token_acc": 0.6955658504301787
|
|
},
|
|
{
|
|
"epoch": 0.8229777256740914,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 8.332487152617424e-07,
|
|
"loss": 1.1082758903503418,
|
|
"step": 1404,
|
|
"token_acc": 0.7054898392478011
|
|
},
|
|
{
|
|
"epoch": 0.8235638921453693,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 8.278969840685458e-07,
|
|
"loss": 1.1692967414855957,
|
|
"step": 1405,
|
|
"token_acc": 0.6841794569067297
|
|
},
|
|
{
|
|
"epoch": 0.8241500586166471,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 8.225609429353187e-07,
|
|
"loss": 1.275005578994751,
|
|
"step": 1406,
|
|
"token_acc": 0.6663832199546486
|
|
},
|
|
{
|
|
"epoch": 0.824736225087925,
|
|
"grad_norm": 3.390625,
|
|
"learning_rate": 8.172406119293513e-07,
|
|
"loss": 1.1971495151519775,
|
|
"step": 1407,
|
|
"token_acc": 0.6851769911504425
|
|
},
|
|
{
|
|
"epoch": 0.8253223915592028,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 8.119360110588531e-07,
|
|
"loss": 1.128738284111023,
|
|
"step": 1408,
|
|
"token_acc": 0.70249653259362
|
|
},
|
|
{
|
|
"epoch": 0.8259085580304807,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 8.066471602728804e-07,
|
|
"loss": 1.1290843486785889,
|
|
"step": 1409,
|
|
"token_acc": 0.7013513513513514
|
|
},
|
|
{
|
|
"epoch": 0.8264947245017585,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 8.013740794612512e-07,
|
|
"loss": 1.2341251373291016,
|
|
"step": 1410,
|
|
"token_acc": 0.6743755781683626
|
|
},
|
|
{
|
|
"epoch": 0.8270808909730364,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 7.961167884544852e-07,
|
|
"loss": 1.0788683891296387,
|
|
"step": 1411,
|
|
"token_acc": 0.7043991416309013
|
|
},
|
|
{
|
|
"epoch": 0.8276670574443142,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 7.908753070237124e-07,
|
|
"loss": 1.1375706195831299,
|
|
"step": 1412,
|
|
"token_acc": 0.7
|
|
},
|
|
{
|
|
"epoch": 0.828253223915592,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 7.856496548806181e-07,
|
|
"loss": 1.2880496978759766,
|
|
"step": 1413,
|
|
"token_acc": 0.6654666297424536
|
|
},
|
|
{
|
|
"epoch": 0.8288393903868698,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 7.804398516773465e-07,
|
|
"loss": 1.1564209461212158,
|
|
"step": 1414,
|
|
"token_acc": 0.6939135548368127
|
|
},
|
|
{
|
|
"epoch": 0.8294255568581477,
|
|
"grad_norm": 3.484375,
|
|
"learning_rate": 7.752459170064491e-07,
|
|
"loss": 1.1494587659835815,
|
|
"step": 1415,
|
|
"token_acc": 0.6928119878248654
|
|
},
|
|
{
|
|
"epoch": 0.8300117233294255,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 7.700678704007947e-07,
|
|
"loss": 1.1363188028335571,
|
|
"step": 1416,
|
|
"token_acc": 0.7065602836879432
|
|
},
|
|
{
|
|
"epoch": 0.8305978898007034,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 7.649057313335029e-07,
|
|
"loss": 1.1553529500961304,
|
|
"step": 1417,
|
|
"token_acc": 0.6817587019104946
|
|
},
|
|
{
|
|
"epoch": 0.8311840562719812,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 7.597595192178702e-07,
|
|
"loss": 1.2219898700714111,
|
|
"step": 1418,
|
|
"token_acc": 0.6813888888888889
|
|
},
|
|
{
|
|
"epoch": 0.8317702227432591,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 7.546292534072958e-07,
|
|
"loss": 1.1678904294967651,
|
|
"step": 1419,
|
|
"token_acc": 0.6846071044133477
|
|
},
|
|
{
|
|
"epoch": 0.8323563892145369,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 7.495149531952101e-07,
|
|
"loss": 1.1134357452392578,
|
|
"step": 1420,
|
|
"token_acc": 0.7084384636949818
|
|
},
|
|
{
|
|
"epoch": 0.8329425556858148,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 7.444166378150014e-07,
|
|
"loss": 1.1394962072372437,
|
|
"step": 1421,
|
|
"token_acc": 0.6872556684910086
|
|
},
|
|
{
|
|
"epoch": 0.8335287221570926,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 7.393343264399439e-07,
|
|
"loss": 1.1421490907669067,
|
|
"step": 1422,
|
|
"token_acc": 0.6985924112607099
|
|
},
|
|
{
|
|
"epoch": 0.8341148886283705,
|
|
"grad_norm": 3.375,
|
|
"learning_rate": 7.342680381831241e-07,
|
|
"loss": 1.1853200197219849,
|
|
"step": 1423,
|
|
"token_acc": 0.6883423977299598
|
|
},
|
|
{
|
|
"epoch": 0.8347010550996483,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 7.292177920973726e-07,
|
|
"loss": 1.16133451461792,
|
|
"step": 1424,
|
|
"token_acc": 0.6926889714993805
|
|
},
|
|
{
|
|
"epoch": 0.8352872215709262,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 7.241836071751878e-07,
|
|
"loss": 1.1419845819473267,
|
|
"step": 1425,
|
|
"token_acc": 0.6901936289818863
|
|
},
|
|
{
|
|
"epoch": 0.835873388042204,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 7.191655023486682e-07,
|
|
"loss": 1.2097097635269165,
|
|
"step": 1426,
|
|
"token_acc": 0.6806564433683078
|
|
},
|
|
{
|
|
"epoch": 0.8364595545134819,
|
|
"grad_norm": 3.921875,
|
|
"learning_rate": 7.141634964894389e-07,
|
|
"loss": 1.169245719909668,
|
|
"step": 1427,
|
|
"token_acc": 0.691358024691358
|
|
},
|
|
{
|
|
"epoch": 0.8370457209847597,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 7.091776084085828e-07,
|
|
"loss": 1.144958257675171,
|
|
"step": 1428,
|
|
"token_acc": 0.6849874371859297
|
|
},
|
|
{
|
|
"epoch": 0.8376318874560376,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 7.042078568565658e-07,
|
|
"loss": 1.1531081199645996,
|
|
"step": 1429,
|
|
"token_acc": 0.6920099529997236
|
|
},
|
|
{
|
|
"epoch": 0.8382180539273154,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 6.992542605231739e-07,
|
|
"loss": 1.1104117631912231,
|
|
"step": 1430,
|
|
"token_acc": 0.7014877102199224
|
|
},
|
|
{
|
|
"epoch": 0.8388042203985931,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 6.94316838037431e-07,
|
|
"loss": 1.2800004482269287,
|
|
"step": 1431,
|
|
"token_acc": 0.6755126658624849
|
|
},
|
|
{
|
|
"epoch": 0.839390386869871,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 6.893956079675452e-07,
|
|
"loss": 1.1435908079147339,
|
|
"step": 1432,
|
|
"token_acc": 0.6969875959834613
|
|
},
|
|
{
|
|
"epoch": 0.8399765533411488,
|
|
"grad_norm": 3.921875,
|
|
"learning_rate": 6.84490588820818e-07,
|
|
"loss": 1.2215361595153809,
|
|
"step": 1433,
|
|
"token_acc": 0.6803153611393693
|
|
},
|
|
{
|
|
"epoch": 0.8405627198124267,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 6.796017990435977e-07,
|
|
"loss": 1.1726887226104736,
|
|
"step": 1434,
|
|
"token_acc": 0.687129947023995
|
|
},
|
|
{
|
|
"epoch": 0.8411488862837045,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 6.747292570211916e-07,
|
|
"loss": 1.111106276512146,
|
|
"step": 1435,
|
|
"token_acc": 0.6971312515866971
|
|
},
|
|
{
|
|
"epoch": 0.8417350527549824,
|
|
"grad_norm": 4.40625,
|
|
"learning_rate": 6.698729810778065e-07,
|
|
"loss": 1.2144769430160522,
|
|
"step": 1436,
|
|
"token_acc": 0.6869476546566962
|
|
},
|
|
{
|
|
"epoch": 0.8423212192262602,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 6.650329894764768e-07,
|
|
"loss": 1.1752411127090454,
|
|
"step": 1437,
|
|
"token_acc": 0.6906818181818182
|
|
},
|
|
{
|
|
"epoch": 0.8429073856975381,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 6.602093004189963e-07,
|
|
"loss": 1.1916290521621704,
|
|
"step": 1438,
|
|
"token_acc": 0.6998799519807923
|
|
},
|
|
{
|
|
"epoch": 0.8434935521688159,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 6.554019320458494e-07,
|
|
"loss": 1.1797103881835938,
|
|
"step": 1439,
|
|
"token_acc": 0.6877382465057179
|
|
},
|
|
{
|
|
"epoch": 0.8440797186400938,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 6.506109024361429e-07,
|
|
"loss": 1.193385362625122,
|
|
"step": 1440,
|
|
"token_acc": 0.6847070506454817
|
|
},
|
|
{
|
|
"epoch": 0.8446658851113716,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 6.458362296075399e-07,
|
|
"loss": 1.234968900680542,
|
|
"step": 1441,
|
|
"token_acc": 0.6831395348837209
|
|
},
|
|
{
|
|
"epoch": 0.8452520515826495,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 6.410779315161885e-07,
|
|
"loss": 1.1995550394058228,
|
|
"step": 1442,
|
|
"token_acc": 0.6868404717566728
|
|
},
|
|
{
|
|
"epoch": 0.8458382180539273,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 6.363360260566571e-07,
|
|
"loss": 1.1251697540283203,
|
|
"step": 1443,
|
|
"token_acc": 0.6962982977573628
|
|
},
|
|
{
|
|
"epoch": 0.8464243845252052,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 6.316105310618664e-07,
|
|
"loss": 1.180797815322876,
|
|
"step": 1444,
|
|
"token_acc": 0.6908165737824086
|
|
},
|
|
{
|
|
"epoch": 0.847010550996483,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 6.269014643030214e-07,
|
|
"loss": 1.2021948099136353,
|
|
"step": 1445,
|
|
"token_acc": 0.6865422123154082
|
|
},
|
|
{
|
|
"epoch": 0.8475967174677609,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 6.222088434895462e-07,
|
|
"loss": 1.230437159538269,
|
|
"step": 1446,
|
|
"token_acc": 0.68363246718457
|
|
},
|
|
{
|
|
"epoch": 0.8481828839390387,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 6.175326862690162e-07,
|
|
"loss": 1.1017515659332275,
|
|
"step": 1447,
|
|
"token_acc": 0.7064321055525014
|
|
},
|
|
{
|
|
"epoch": 0.8487690504103166,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 6.128730102270897e-07,
|
|
"loss": 1.2121186256408691,
|
|
"step": 1448,
|
|
"token_acc": 0.6845601913366994
|
|
},
|
|
{
|
|
"epoch": 0.8493552168815943,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 6.082298328874509e-07,
|
|
"loss": 1.2109477519989014,
|
|
"step": 1449,
|
|
"token_acc": 0.6762568140520896
|
|
},
|
|
{
|
|
"epoch": 0.8499413833528722,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 6.03603171711728e-07,
|
|
"loss": 1.1513304710388184,
|
|
"step": 1450,
|
|
"token_acc": 0.6900328587075575
|
|
},
|
|
{
|
|
"epoch": 0.85052754982415,
|
|
"grad_norm": 3.28125,
|
|
"learning_rate": 5.989930440994451e-07,
|
|
"loss": 1.1527934074401855,
|
|
"step": 1451,
|
|
"token_acc": 0.6914489841482474
|
|
},
|
|
{
|
|
"epoch": 0.8511137162954279,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 5.943994673879405e-07,
|
|
"loss": 1.1706149578094482,
|
|
"step": 1452,
|
|
"token_acc": 0.6912038267339888
|
|
},
|
|
{
|
|
"epoch": 0.8516998827667057,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 5.898224588523182e-07,
|
|
"loss": 1.225450038909912,
|
|
"step": 1453,
|
|
"token_acc": 0.6869948567229978
|
|
},
|
|
{
|
|
"epoch": 0.8522860492379836,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 5.852620357053651e-07,
|
|
"loss": 1.1760717630386353,
|
|
"step": 1454,
|
|
"token_acc": 0.68649468892261
|
|
},
|
|
{
|
|
"epoch": 0.8528722157092614,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 5.807182150975027e-07,
|
|
"loss": 1.1644572019577026,
|
|
"step": 1455,
|
|
"token_acc": 0.6935071574642127
|
|
},
|
|
{
|
|
"epoch": 0.8534583821805393,
|
|
"grad_norm": 3.484375,
|
|
"learning_rate": 5.76191014116711e-07,
|
|
"loss": 1.1690869331359863,
|
|
"step": 1456,
|
|
"token_acc": 0.681849710982659
|
|
},
|
|
{
|
|
"epoch": 0.8540445486518171,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 5.716804497884698e-07,
|
|
"loss": 1.1293895244598389,
|
|
"step": 1457,
|
|
"token_acc": 0.6983571236197145
|
|
},
|
|
{
|
|
"epoch": 0.854630715123095,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 5.671865390756948e-07,
|
|
"loss": 1.176634669303894,
|
|
"step": 1458,
|
|
"token_acc": 0.683929931454684
|
|
},
|
|
{
|
|
"epoch": 0.8552168815943728,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 5.627092988786659e-07,
|
|
"loss": 1.1980212926864624,
|
|
"step": 1459,
|
|
"token_acc": 0.6919191919191919
|
|
},
|
|
{
|
|
"epoch": 0.8558030480656507,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 5.582487460349806e-07,
|
|
"loss": 1.1963926553726196,
|
|
"step": 1460,
|
|
"token_acc": 0.6995998768851954
|
|
},
|
|
{
|
|
"epoch": 0.8563892145369285,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 5.538048973194699e-07,
|
|
"loss": 1.1226892471313477,
|
|
"step": 1461,
|
|
"token_acc": 0.7045964125560538
|
|
},
|
|
{
|
|
"epoch": 0.8569753810082064,
|
|
"grad_norm": 3.453125,
|
|
"learning_rate": 5.493777694441521e-07,
|
|
"loss": 1.1744701862335205,
|
|
"step": 1462,
|
|
"token_acc": 0.6877777777777778
|
|
},
|
|
{
|
|
"epoch": 0.8575615474794842,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 5.449673790581611e-07,
|
|
"loss": 1.1884660720825195,
|
|
"step": 1463,
|
|
"token_acc": 0.6849719515795689
|
|
},
|
|
{
|
|
"epoch": 0.8581477139507621,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 5.405737427476854e-07,
|
|
"loss": 1.1927860975265503,
|
|
"step": 1464,
|
|
"token_acc": 0.6757105943152455
|
|
},
|
|
{
|
|
"epoch": 0.8587338804220399,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 5.361968770359072e-07,
|
|
"loss": 1.1518969535827637,
|
|
"step": 1465,
|
|
"token_acc": 0.698090692124105
|
|
},
|
|
{
|
|
"epoch": 0.8593200468933178,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 5.318367983829393e-07,
|
|
"loss": 1.1760423183441162,
|
|
"step": 1466,
|
|
"token_acc": 0.6941701178719337
|
|
},
|
|
{
|
|
"epoch": 0.8599062133645955,
|
|
"grad_norm": 4.375,
|
|
"learning_rate": 5.274935231857615e-07,
|
|
"loss": 1.181771993637085,
|
|
"step": 1467,
|
|
"token_acc": 0.6887393767705382
|
|
},
|
|
{
|
|
"epoch": 0.8604923798358733,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 5.231670677781659e-07,
|
|
"loss": 1.2127717733383179,
|
|
"step": 1468,
|
|
"token_acc": 0.6828895010713193
|
|
},
|
|
{
|
|
"epoch": 0.8610785463071512,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 5.188574484306829e-07,
|
|
"loss": 1.1513279676437378,
|
|
"step": 1469,
|
|
"token_acc": 0.6869616869616869
|
|
},
|
|
{
|
|
"epoch": 0.861664712778429,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 5.145646813505339e-07,
|
|
"loss": 1.2217496633529663,
|
|
"step": 1470,
|
|
"token_acc": 0.6771325368157821
|
|
},
|
|
{
|
|
"epoch": 0.8622508792497069,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 5.102887826815589e-07,
|
|
"loss": 1.1267294883728027,
|
|
"step": 1471,
|
|
"token_acc": 0.7002128306476133
|
|
},
|
|
{
|
|
"epoch": 0.8628370457209847,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 5.06029768504166e-07,
|
|
"loss": 1.1682792901992798,
|
|
"step": 1472,
|
|
"token_acc": 0.6952908587257618
|
|
},
|
|
{
|
|
"epoch": 0.8634232121922626,
|
|
"grad_norm": 3.34375,
|
|
"learning_rate": 5.017876548352585e-07,
|
|
"loss": 1.24690842628479,
|
|
"step": 1473,
|
|
"token_acc": 0.685799907791609
|
|
},
|
|
{
|
|
"epoch": 0.8640093786635404,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 4.97562457628189e-07,
|
|
"loss": 1.1945253610610962,
|
|
"step": 1474,
|
|
"token_acc": 0.6843086966658894
|
|
},
|
|
{
|
|
"epoch": 0.8645955451348183,
|
|
"grad_norm": 3.4375,
|
|
"learning_rate": 4.933541927726887e-07,
|
|
"loss": 1.152950644493103,
|
|
"step": 1475,
|
|
"token_acc": 0.7033919286952216
|
|
},
|
|
{
|
|
"epoch": 0.8651817116060961,
|
|
"grad_norm": 3.46875,
|
|
"learning_rate": 4.891628760948114e-07,
|
|
"loss": 1.1074192523956299,
|
|
"step": 1476,
|
|
"token_acc": 0.70225
|
|
},
|
|
{
|
|
"epoch": 0.865767878077374,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 4.849885233568752e-07,
|
|
"loss": 1.1491429805755615,
|
|
"step": 1477,
|
|
"token_acc": 0.6931044687777449
|
|
},
|
|
{
|
|
"epoch": 0.8663540445486518,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 4.808311502573976e-07,
|
|
"loss": 1.1059802770614624,
|
|
"step": 1478,
|
|
"token_acc": 0.6945883707541739
|
|
},
|
|
{
|
|
"epoch": 0.8669402110199297,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 4.7669077243104714e-07,
|
|
"loss": 1.1917452812194824,
|
|
"step": 1479,
|
|
"token_acc": 0.6826893824155765
|
|
},
|
|
{
|
|
"epoch": 0.8675263774912075,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 4.7256740544857124e-07,
|
|
"loss": 1.153546929359436,
|
|
"step": 1480,
|
|
"token_acc": 0.6912811387900356
|
|
},
|
|
{
|
|
"epoch": 0.8681125439624854,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 4.6846106481675035e-07,
|
|
"loss": 1.195996642112732,
|
|
"step": 1481,
|
|
"token_acc": 0.6875497743562516
|
|
},
|
|
{
|
|
"epoch": 0.8686987104337632,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 4.643717659783309e-07,
|
|
"loss": 1.202422857284546,
|
|
"step": 1482,
|
|
"token_acc": 0.6921908234579155
|
|
},
|
|
{
|
|
"epoch": 0.8692848769050411,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 4.602995243119701e-07,
|
|
"loss": 1.205765724182129,
|
|
"step": 1483,
|
|
"token_acc": 0.6795969773299748
|
|
},
|
|
{
|
|
"epoch": 0.8698710433763188,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 4.562443551321788e-07,
|
|
"loss": 1.1622669696807861,
|
|
"step": 1484,
|
|
"token_acc": 0.6910616141162857
|
|
},
|
|
{
|
|
"epoch": 0.8704572098475967,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 4.522062736892635e-07,
|
|
"loss": 1.1235719919204712,
|
|
"step": 1485,
|
|
"token_acc": 0.7069807526572824
|
|
},
|
|
{
|
|
"epoch": 0.8710433763188745,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 4.481852951692672e-07,
|
|
"loss": 1.1482852697372437,
|
|
"step": 1486,
|
|
"token_acc": 0.6908838500694123
|
|
},
|
|
{
|
|
"epoch": 0.8716295427901524,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 4.4418143469391497e-07,
|
|
"loss": 1.1985642910003662,
|
|
"step": 1487,
|
|
"token_acc": 0.6855345911949685
|
|
},
|
|
{
|
|
"epoch": 0.8722157092614302,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 4.401947073205559e-07,
|
|
"loss": 1.1419129371643066,
|
|
"step": 1488,
|
|
"token_acc": 0.6936400541271989
|
|
},
|
|
{
|
|
"epoch": 0.8728018757327081,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 4.362251280421076e-07,
|
|
"loss": 1.123828411102295,
|
|
"step": 1489,
|
|
"token_acc": 0.6991048224083165
|
|
},
|
|
{
|
|
"epoch": 0.8733880422039859,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 4.322727117869951e-07,
|
|
"loss": 1.2080843448638916,
|
|
"step": 1490,
|
|
"token_acc": 0.6835205992509363
|
|
},
|
|
{
|
|
"epoch": 0.8739742086752638,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 4.283374734191037e-07,
|
|
"loss": 1.217576503753662,
|
|
"step": 1491,
|
|
"token_acc": 0.6808330987897552
|
|
},
|
|
{
|
|
"epoch": 0.8745603751465416,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 4.2441942773771114e-07,
|
|
"loss": 1.2114224433898926,
|
|
"step": 1492,
|
|
"token_acc": 0.6881491973559962
|
|
},
|
|
{
|
|
"epoch": 0.8751465416178195,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 4.205185894774455e-07,
|
|
"loss": 1.2101552486419678,
|
|
"step": 1493,
|
|
"token_acc": 0.6755555555555556
|
|
},
|
|
{
|
|
"epoch": 0.8757327080890973,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 4.1663497330821536e-07,
|
|
"loss": 1.1756892204284668,
|
|
"step": 1494,
|
|
"token_acc": 0.68603515625
|
|
},
|
|
{
|
|
"epoch": 0.8763188745603752,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 4.1276859383516944e-07,
|
|
"loss": 1.1735472679138184,
|
|
"step": 1495,
|
|
"token_acc": 0.6922119930330928
|
|
},
|
|
{
|
|
"epoch": 0.876905041031653,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 4.089194655986306e-07,
|
|
"loss": 1.1850225925445557,
|
|
"step": 1496,
|
|
"token_acc": 0.6946929039952295
|
|
},
|
|
{
|
|
"epoch": 0.8774912075029309,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 4.050876030740425e-07,
|
|
"loss": 1.2005321979522705,
|
|
"step": 1497,
|
|
"token_acc": 0.6879432624113475
|
|
},
|
|
{
|
|
"epoch": 0.8780773739742087,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 4.0127302067192285e-07,
|
|
"loss": 1.2494759559631348,
|
|
"step": 1498,
|
|
"token_acc": 0.6847612818835841
|
|
},
|
|
{
|
|
"epoch": 0.8786635404454866,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 3.9747573273779816e-07,
|
|
"loss": 1.183923363685608,
|
|
"step": 1499,
|
|
"token_acc": 0.6894176581089543
|
|
},
|
|
{
|
|
"epoch": 0.8792497069167644,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 3.936957535521624e-07,
|
|
"loss": 1.1769791841506958,
|
|
"step": 1500,
|
|
"token_acc": 0.68574732581671
|
|
},
|
|
{
|
|
"epoch": 0.8798358733880423,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 3.899330973304083e-07,
|
|
"loss": 1.1955506801605225,
|
|
"step": 1501,
|
|
"token_acc": 0.6873056994818653
|
|
},
|
|
{
|
|
"epoch": 0.88042203985932,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 3.8618777822278854e-07,
|
|
"loss": 1.1323904991149902,
|
|
"step": 1502,
|
|
"token_acc": 0.6976608187134503
|
|
},
|
|
{
|
|
"epoch": 0.8810082063305978,
|
|
"grad_norm": 4.34375,
|
|
"learning_rate": 3.824598103143529e-07,
|
|
"loss": 1.2752876281738281,
|
|
"step": 1503,
|
|
"token_acc": 0.6820083682008368
|
|
},
|
|
{
|
|
"epoch": 0.8815943728018757,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 3.787492076248994e-07,
|
|
"loss": 1.2019588947296143,
|
|
"step": 1504,
|
|
"token_acc": 0.6855305466237942
|
|
},
|
|
{
|
|
"epoch": 0.8821805392731535,
|
|
"grad_norm": 4.4375,
|
|
"learning_rate": 3.750559841089196e-07,
|
|
"loss": 1.16989266872406,
|
|
"step": 1505,
|
|
"token_acc": 0.6952837140751658
|
|
},
|
|
{
|
|
"epoch": 0.8827667057444314,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 3.7138015365554834e-07,
|
|
"loss": 1.1521722078323364,
|
|
"step": 1506,
|
|
"token_acc": 0.6962848297213622
|
|
},
|
|
{
|
|
"epoch": 0.8833528722157092,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 3.677217300885083e-07,
|
|
"loss": 1.074339509010315,
|
|
"step": 1507,
|
|
"token_acc": 0.7159386751518658
|
|
},
|
|
{
|
|
"epoch": 0.8839390386869871,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 3.6408072716606346e-07,
|
|
"loss": 1.160008430480957,
|
|
"step": 1508,
|
|
"token_acc": 0.6940720881739648
|
|
},
|
|
{
|
|
"epoch": 0.8845252051582649,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 3.604571585809585e-07,
|
|
"loss": 1.2223683595657349,
|
|
"step": 1509,
|
|
"token_acc": 0.679364278506559
|
|
},
|
|
{
|
|
"epoch": 0.8851113716295428,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 3.56851037960379e-07,
|
|
"loss": 1.134746789932251,
|
|
"step": 1510,
|
|
"token_acc": 0.7000281928390188
|
|
},
|
|
{
|
|
"epoch": 0.8856975381008206,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 3.5326237886588734e-07,
|
|
"loss": 1.1651989221572876,
|
|
"step": 1511,
|
|
"token_acc": 0.6915938864628821
|
|
},
|
|
{
|
|
"epoch": 0.8862837045720985,
|
|
"grad_norm": 3.46875,
|
|
"learning_rate": 3.496911947933845e-07,
|
|
"loss": 1.183305263519287,
|
|
"step": 1512,
|
|
"token_acc": 0.681124303368064
|
|
},
|
|
{
|
|
"epoch": 0.8868698710433763,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 3.461374991730482e-07,
|
|
"loss": 1.193648338317871,
|
|
"step": 1513,
|
|
"token_acc": 0.6845177664974619
|
|
},
|
|
{
|
|
"epoch": 0.8874560375146542,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 3.426013053692878e-07,
|
|
"loss": 1.1667733192443848,
|
|
"step": 1514,
|
|
"token_acc": 0.6926217556138816
|
|
},
|
|
{
|
|
"epoch": 0.888042203985932,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 3.390826266806985e-07,
|
|
"loss": 1.1104120016098022,
|
|
"step": 1515,
|
|
"token_acc": 0.695264765784114
|
|
},
|
|
{
|
|
"epoch": 0.8886283704572099,
|
|
"grad_norm": 3.5,
|
|
"learning_rate": 3.355814763399973e-07,
|
|
"loss": 1.117077350616455,
|
|
"step": 1516,
|
|
"token_acc": 0.6989637305699482
|
|
},
|
|
{
|
|
"epoch": 0.8892145369284877,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 3.320978675139919e-07,
|
|
"loss": 1.1796401739120483,
|
|
"step": 1517,
|
|
"token_acc": 0.6928645294725957
|
|
},
|
|
{
|
|
"epoch": 0.8898007033997656,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 3.2863181330351325e-07,
|
|
"loss": 1.239884853363037,
|
|
"step": 1518,
|
|
"token_acc": 0.6804268463914631
|
|
},
|
|
{
|
|
"epoch": 0.8903868698710434,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 3.2518332674338126e-07,
|
|
"loss": 1.178505539894104,
|
|
"step": 1519,
|
|
"token_acc": 0.6878543718293046
|
|
},
|
|
{
|
|
"epoch": 0.8909730363423212,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 3.2175242080234314e-07,
|
|
"loss": 1.204371452331543,
|
|
"step": 1520,
|
|
"token_acc": 0.6883461640562782
|
|
},
|
|
{
|
|
"epoch": 0.891559202813599,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 3.183391083830345e-07,
|
|
"loss": 1.129250407218933,
|
|
"step": 1521,
|
|
"token_acc": 0.6967294350842418
|
|
},
|
|
{
|
|
"epoch": 0.8921453692848769,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 3.1494340232192667e-07,
|
|
"loss": 1.229207158088684,
|
|
"step": 1522,
|
|
"token_acc": 0.6866156344235677
|
|
},
|
|
{
|
|
"epoch": 0.8927315357561547,
|
|
"grad_norm": 3.3125,
|
|
"learning_rate": 3.1156531538927615e-07,
|
|
"loss": 1.150852918624878,
|
|
"step": 1523,
|
|
"token_acc": 0.696535500427716
|
|
},
|
|
{
|
|
"epoch": 0.8933177022274326,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 3.082048602890808e-07,
|
|
"loss": 1.1254236698150635,
|
|
"step": 1524,
|
|
"token_acc": 0.7087542087542088
|
|
},
|
|
{
|
|
"epoch": 0.8939038686987104,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 3.0486204965903045e-07,
|
|
"loss": 1.166311502456665,
|
|
"step": 1525,
|
|
"token_acc": 0.6855541718555417
|
|
},
|
|
{
|
|
"epoch": 0.8944900351699883,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 3.015368960704584e-07,
|
|
"loss": 1.081763744354248,
|
|
"step": 1526,
|
|
"token_acc": 0.704474505723205
|
|
},
|
|
{
|
|
"epoch": 0.8950762016412661,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 2.982294120282958e-07,
|
|
"loss": 1.1462355852127075,
|
|
"step": 1527,
|
|
"token_acc": 0.691480996068152
|
|
},
|
|
{
|
|
"epoch": 0.895662368112544,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 2.9493960997102224e-07,
|
|
"loss": 1.19098699092865,
|
|
"step": 1528,
|
|
"token_acc": 0.6870513161393247
|
|
},
|
|
{
|
|
"epoch": 0.8962485345838218,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 2.916675022706239e-07,
|
|
"loss": 1.1769344806671143,
|
|
"step": 1529,
|
|
"token_acc": 0.6829411764705883
|
|
},
|
|
{
|
|
"epoch": 0.8968347010550997,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 2.8841310123253865e-07,
|
|
"loss": 1.1995351314544678,
|
|
"step": 1530,
|
|
"token_acc": 0.6864620030196276
|
|
},
|
|
{
|
|
"epoch": 0.8974208675263775,
|
|
"grad_norm": 3.921875,
|
|
"learning_rate": 2.8517641909562075e-07,
|
|
"loss": 1.1603044271469116,
|
|
"step": 1531,
|
|
"token_acc": 0.6879394267171444
|
|
},
|
|
{
|
|
"epoch": 0.8980070339976554,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 2.819574680320825e-07,
|
|
"loss": 1.2841908931732178,
|
|
"step": 1532,
|
|
"token_acc": 0.6796141152576796
|
|
},
|
|
{
|
|
"epoch": 0.8985932004689332,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 2.7875626014746026e-07,
|
|
"loss": 1.1420249938964844,
|
|
"step": 1533,
|
|
"token_acc": 0.6994480682388359
|
|
},
|
|
{
|
|
"epoch": 0.8991793669402111,
|
|
"grad_norm": 3.609375,
|
|
"learning_rate": 2.755728074805597e-07,
|
|
"loss": 1.1422362327575684,
|
|
"step": 1534,
|
|
"token_acc": 0.6968586387434555
|
|
},
|
|
{
|
|
"epoch": 0.8997655334114889,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 2.724071220034158e-07,
|
|
"loss": 1.1294820308685303,
|
|
"step": 1535,
|
|
"token_acc": 0.6994551190134787
|
|
},
|
|
{
|
|
"epoch": 0.9003516998827668,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 2.6925921562124867e-07,
|
|
"loss": 1.127701997756958,
|
|
"step": 1536,
|
|
"token_acc": 0.6934434434434434
|
|
},
|
|
{
|
|
"epoch": 0.9009378663540446,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 2.6612910017241036e-07,
|
|
"loss": 1.161512851715088,
|
|
"step": 1537,
|
|
"token_acc": 0.6968677193890759
|
|
},
|
|
{
|
|
"epoch": 0.9015240328253223,
|
|
"grad_norm": 3.921875,
|
|
"learning_rate": 2.63016787428354e-07,
|
|
"loss": 1.1306143999099731,
|
|
"step": 1538,
|
|
"token_acc": 0.6954695469546954
|
|
},
|
|
{
|
|
"epoch": 0.9021101992966002,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 2.5992228909357486e-07,
|
|
"loss": 1.1870734691619873,
|
|
"step": 1539,
|
|
"token_acc": 0.6758504357604723
|
|
},
|
|
{
|
|
"epoch": 0.902696365767878,
|
|
"grad_norm": 3.75,
|
|
"learning_rate": 2.5684561680557995e-07,
|
|
"loss": 1.1518449783325195,
|
|
"step": 1540,
|
|
"token_acc": 0.6941333333333334
|
|
},
|
|
{
|
|
"epoch": 0.9032825322391559,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 2.5378678213483057e-07,
|
|
"loss": 1.1549478769302368,
|
|
"step": 1541,
|
|
"token_acc": 0.6893840104849279
|
|
},
|
|
{
|
|
"epoch": 0.9038686987104337,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 2.5074579658471266e-07,
|
|
"loss": 1.2101107835769653,
|
|
"step": 1542,
|
|
"token_acc": 0.6870988446726572
|
|
},
|
|
{
|
|
"epoch": 0.9044548651817116,
|
|
"grad_norm": 3.53125,
|
|
"learning_rate": 2.477226715914838e-07,
|
|
"loss": 1.161818027496338,
|
|
"step": 1543,
|
|
"token_acc": 0.6903127383676583
|
|
},
|
|
{
|
|
"epoch": 0.9050410316529894,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 2.447174185242324e-07,
|
|
"loss": 1.182915449142456,
|
|
"step": 1544,
|
|
"token_acc": 0.6934523809523809
|
|
},
|
|
{
|
|
"epoch": 0.9056271981242673,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 2.417300486848373e-07,
|
|
"loss": 1.2088913917541504,
|
|
"step": 1545,
|
|
"token_acc": 0.6869493392070485
|
|
},
|
|
{
|
|
"epoch": 0.9062133645955451,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 2.3876057330792344e-07,
|
|
"loss": 1.1839567422866821,
|
|
"step": 1546,
|
|
"token_acc": 0.6894595401612421
|
|
},
|
|
{
|
|
"epoch": 0.906799531066823,
|
|
"grad_norm": 3.375,
|
|
"learning_rate": 2.3580900356081903e-07,
|
|
"loss": 1.1491632461547852,
|
|
"step": 1547,
|
|
"token_acc": 0.7023887793118563
|
|
},
|
|
{
|
|
"epoch": 0.9073856975381008,
|
|
"grad_norm": 3.484375,
|
|
"learning_rate": 2.3287535054351716e-07,
|
|
"loss": 1.1867594718933105,
|
|
"step": 1548,
|
|
"token_acc": 0.6950207468879668
|
|
},
|
|
{
|
|
"epoch": 0.9079718640093787,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 2.2995962528862657e-07,
|
|
"loss": 1.2417415380477905,
|
|
"step": 1549,
|
|
"token_acc": 0.6812053925455988
|
|
},
|
|
{
|
|
"epoch": 0.9085580304806565,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 2.2706183876134047e-07,
|
|
"loss": 1.2784538269042969,
|
|
"step": 1550,
|
|
"token_acc": 0.6782945736434108
|
|
},
|
|
{
|
|
"epoch": 0.9091441969519344,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 2.2418200185938488e-07,
|
|
"loss": 1.219226360321045,
|
|
"step": 1551,
|
|
"token_acc": 0.6919104991394148
|
|
},
|
|
{
|
|
"epoch": 0.9097303634232122,
|
|
"grad_norm": 3.46875,
|
|
"learning_rate": 2.2132012541298542e-07,
|
|
"loss": 1.1442694664001465,
|
|
"step": 1552,
|
|
"token_acc": 0.6978812450954748
|
|
},
|
|
{
|
|
"epoch": 0.9103165298944901,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 2.1847622018482283e-07,
|
|
"loss": 1.219637393951416,
|
|
"step": 1553,
|
|
"token_acc": 0.6771653543307087
|
|
},
|
|
{
|
|
"epoch": 0.9109026963657679,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 2.1565029686999306e-07,
|
|
"loss": 1.2168278694152832,
|
|
"step": 1554,
|
|
"token_acc": 0.6794451450189155
|
|
},
|
|
{
|
|
"epoch": 0.9114888628370457,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 2.1284236609596886e-07,
|
|
"loss": 1.1365243196487427,
|
|
"step": 1555,
|
|
"token_acc": 0.6988178398710371
|
|
},
|
|
{
|
|
"epoch": 0.9120750293083235,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 2.1005243842255552e-07,
|
|
"loss": 1.187851905822754,
|
|
"step": 1556,
|
|
"token_acc": 0.6882657228423632
|
|
},
|
|
{
|
|
"epoch": 0.9126611957796014,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 2.0728052434185742e-07,
|
|
"loss": 1.1412928104400635,
|
|
"step": 1557,
|
|
"token_acc": 0.7049861495844876
|
|
},
|
|
{
|
|
"epoch": 0.9132473622508792,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 2.0452663427823093e-07,
|
|
"loss": 1.1649284362792969,
|
|
"step": 1558,
|
|
"token_acc": 0.6867947544983226
|
|
},
|
|
{
|
|
"epoch": 0.9138335287221571,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 2.0179077858825445e-07,
|
|
"loss": 1.2690011262893677,
|
|
"step": 1559,
|
|
"token_acc": 0.6764232008592911
|
|
},
|
|
{
|
|
"epoch": 0.9144196951934349,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 1.990729675606784e-07,
|
|
"loss": 1.1789381504058838,
|
|
"step": 1560,
|
|
"token_acc": 0.683622463981182
|
|
},
|
|
{
|
|
"epoch": 0.9150058616647128,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 1.9637321141639743e-07,
|
|
"loss": 1.1797910928726196,
|
|
"step": 1561,
|
|
"token_acc": 0.6888756692444973
|
|
},
|
|
{
|
|
"epoch": 0.9155920281359906,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 1.9369152030840553e-07,
|
|
"loss": 1.1718943119049072,
|
|
"step": 1562,
|
|
"token_acc": 0.6837315130830489
|
|
},
|
|
{
|
|
"epoch": 0.9161781946072685,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 1.910279043217589e-07,
|
|
"loss": 1.1090738773345947,
|
|
"step": 1563,
|
|
"token_acc": 0.7017687074829932
|
|
},
|
|
{
|
|
"epoch": 0.9167643610785463,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 1.8838237347353848e-07,
|
|
"loss": 1.1922394037246704,
|
|
"step": 1564,
|
|
"token_acc": 0.6865152919369787
|
|
},
|
|
{
|
|
"epoch": 0.9173505275498242,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 1.8575493771281205e-07,
|
|
"loss": 1.2071478366851807,
|
|
"step": 1565,
|
|
"token_acc": 0.6917019987886129
|
|
},
|
|
{
|
|
"epoch": 0.917936694021102,
|
|
"grad_norm": 3.609375,
|
|
"learning_rate": 1.8314560692059836e-07,
|
|
"loss": 1.1875050067901611,
|
|
"step": 1566,
|
|
"token_acc": 0.6833779508396204
|
|
},
|
|
{
|
|
"epoch": 0.9185228604923799,
|
|
"grad_norm": 4.3125,
|
|
"learning_rate": 1.8055439090982784e-07,
|
|
"loss": 1.217982530593872,
|
|
"step": 1567,
|
|
"token_acc": 0.6775396085740913
|
|
},
|
|
{
|
|
"epoch": 0.9191090269636577,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 1.779812994253055e-07,
|
|
"loss": 1.1643027067184448,
|
|
"step": 1568,
|
|
"token_acc": 0.6920289855072463
|
|
},
|
|
{
|
|
"epoch": 0.9196951934349356,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 1.7542634214367748e-07,
|
|
"loss": 1.2324070930480957,
|
|
"step": 1569,
|
|
"token_acc": 0.6843894899536321
|
|
},
|
|
{
|
|
"epoch": 0.9202813599062134,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 1.728895286733906e-07,
|
|
"loss": 1.209564208984375,
|
|
"step": 1570,
|
|
"token_acc": 0.6872220575155648
|
|
},
|
|
{
|
|
"epoch": 0.9208675263774913,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 1.7037086855465902e-07,
|
|
"loss": 1.0810272693634033,
|
|
"step": 1571,
|
|
"token_acc": 0.697507681802663
|
|
},
|
|
{
|
|
"epoch": 0.9214536928487691,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 1.6787037125942706e-07,
|
|
"loss": 1.1677509546279907,
|
|
"step": 1572,
|
|
"token_acc": 0.6898608349900597
|
|
},
|
|
{
|
|
"epoch": 0.9220398593200468,
|
|
"grad_norm": 3.546875,
|
|
"learning_rate": 1.6538804619133485e-07,
|
|
"loss": 1.1684002876281738,
|
|
"step": 1573,
|
|
"token_acc": 0.704076497232008
|
|
},
|
|
{
|
|
"epoch": 0.9226260257913247,
|
|
"grad_norm": 3.46875,
|
|
"learning_rate": 1.6292390268568103e-07,
|
|
"loss": 1.1727193593978882,
|
|
"step": 1574,
|
|
"token_acc": 0.6918507959135186
|
|
},
|
|
{
|
|
"epoch": 0.9232121922626025,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 1.6047795000938782e-07,
|
|
"loss": 1.2292934656143188,
|
|
"step": 1575,
|
|
"token_acc": 0.6795929592959296
|
|
},
|
|
{
|
|
"epoch": 0.9237983587338804,
|
|
"grad_norm": 3.59375,
|
|
"learning_rate": 1.5805019736097105e-07,
|
|
"loss": 1.1350016593933105,
|
|
"step": 1576,
|
|
"token_acc": 0.6938775510204082
|
|
},
|
|
{
|
|
"epoch": 0.9243845252051582,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 1.556406538704963e-07,
|
|
"loss": 1.2458807229995728,
|
|
"step": 1577,
|
|
"token_acc": 0.6793960923623446
|
|
},
|
|
{
|
|
"epoch": 0.9249706916764361,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 1.53249328599554e-07,
|
|
"loss": 1.1816558837890625,
|
|
"step": 1578,
|
|
"token_acc": 0.68830242510699
|
|
},
|
|
{
|
|
"epoch": 0.9255568581477139,
|
|
"grad_norm": 4.34375,
|
|
"learning_rate": 1.5087623054121813e-07,
|
|
"loss": 1.1485333442687988,
|
|
"step": 1579,
|
|
"token_acc": 0.699897575964493
|
|
},
|
|
{
|
|
"epoch": 0.9261430246189918,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 1.4852136862001766e-07,
|
|
"loss": 1.194901704788208,
|
|
"step": 1580,
|
|
"token_acc": 0.6960503720663995
|
|
},
|
|
{
|
|
"epoch": 0.9267291910902696,
|
|
"grad_norm": 3.53125,
|
|
"learning_rate": 1.4618475169190017e-07,
|
|
"loss": 1.1235371828079224,
|
|
"step": 1581,
|
|
"token_acc": 0.6970059880239521
|
|
},
|
|
{
|
|
"epoch": 0.9273153575615475,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 1.438663885441982e-07,
|
|
"loss": 1.2141356468200684,
|
|
"step": 1582,
|
|
"token_acc": 0.6798418972332015
|
|
},
|
|
{
|
|
"epoch": 0.9279015240328253,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 1.4156628789559924e-07,
|
|
"loss": 1.1616910696029663,
|
|
"step": 1583,
|
|
"token_acc": 0.6981475857880353
|
|
},
|
|
{
|
|
"epoch": 0.9284876905041032,
|
|
"grad_norm": 4.3125,
|
|
"learning_rate": 1.3928445839610782e-07,
|
|
"loss": 1.0604296922683716,
|
|
"step": 1584,
|
|
"token_acc": 0.7068713450292398
|
|
},
|
|
{
|
|
"epoch": 0.929073856975381,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 1.3702090862701855e-07,
|
|
"loss": 1.2138644456863403,
|
|
"step": 1585,
|
|
"token_acc": 0.6818851251840943
|
|
},
|
|
{
|
|
"epoch": 0.9296600234466589,
|
|
"grad_norm": 3.546875,
|
|
"learning_rate": 1.3477564710088097e-07,
|
|
"loss": 1.1064062118530273,
|
|
"step": 1586,
|
|
"token_acc": 0.7064700285442436
|
|
},
|
|
{
|
|
"epoch": 0.9302461899179367,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 1.3254868226146688e-07,
|
|
"loss": 1.1603219509124756,
|
|
"step": 1587,
|
|
"token_acc": 0.684559520050521
|
|
},
|
|
{
|
|
"epoch": 0.9308323563892146,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 1.303400224837398e-07,
|
|
"loss": 1.2612087726593018,
|
|
"step": 1588,
|
|
"token_acc": 0.6731443134756292
|
|
},
|
|
{
|
|
"epoch": 0.9314185228604924,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 1.2814967607382433e-07,
|
|
"loss": 1.148130178451538,
|
|
"step": 1589,
|
|
"token_acc": 0.696105702364395
|
|
},
|
|
{
|
|
"epoch": 0.9320046893317703,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 1.25977651268972e-07,
|
|
"loss": 1.2368993759155273,
|
|
"step": 1590,
|
|
"token_acc": 0.6825880758807588
|
|
},
|
|
{
|
|
"epoch": 0.932590855803048,
|
|
"grad_norm": 4.3125,
|
|
"learning_rate": 1.2382395623753484e-07,
|
|
"loss": 1.240231990814209,
|
|
"step": 1591,
|
|
"token_acc": 0.6814304461942258
|
|
},
|
|
{
|
|
"epoch": 0.9331770222743259,
|
|
"grad_norm": 4.1875,
|
|
"learning_rate": 1.2168859907892904e-07,
|
|
"loss": 1.1808828115463257,
|
|
"step": 1592,
|
|
"token_acc": 0.6819901076520221
|
|
},
|
|
{
|
|
"epoch": 0.9337631887456037,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 1.1957158782360922e-07,
|
|
"loss": 1.083528757095337,
|
|
"step": 1593,
|
|
"token_acc": 0.7040097949188858
|
|
},
|
|
{
|
|
"epoch": 0.9343493552168816,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 1.174729304330352e-07,
|
|
"loss": 1.1720170974731445,
|
|
"step": 1594,
|
|
"token_acc": 0.6903474903474903
|
|
},
|
|
{
|
|
"epoch": 0.9349355216881594,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 1.1539263479964535e-07,
|
|
"loss": 1.1909143924713135,
|
|
"step": 1595,
|
|
"token_acc": 0.6940527283874923
|
|
},
|
|
{
|
|
"epoch": 0.9355216881594373,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 1.1333070874682217e-07,
|
|
"loss": 1.178584337234497,
|
|
"step": 1596,
|
|
"token_acc": 0.6914441747572816
|
|
},
|
|
{
|
|
"epoch": 0.9361078546307151,
|
|
"grad_norm": 3.921875,
|
|
"learning_rate": 1.1128716002886841e-07,
|
|
"loss": 1.1065185070037842,
|
|
"step": 1597,
|
|
"token_acc": 0.7037784828381887
|
|
},
|
|
{
|
|
"epoch": 0.936694021101993,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 1.0926199633097156e-07,
|
|
"loss": 1.2129840850830078,
|
|
"step": 1598,
|
|
"token_acc": 0.6952993844432009
|
|
},
|
|
{
|
|
"epoch": 0.9372801875732708,
|
|
"grad_norm": 3.546875,
|
|
"learning_rate": 1.0725522526918275e-07,
|
|
"loss": 1.1688727140426636,
|
|
"step": 1599,
|
|
"token_acc": 0.6874843240531728
|
|
},
|
|
{
|
|
"epoch": 0.9378663540445487,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 1.0526685439037843e-07,
|
|
"loss": 1.1666312217712402,
|
|
"step": 1600,
|
|
"token_acc": 0.6857749469214437
|
|
},
|
|
{
|
|
"epoch": 0.9384525205158265,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 1.0329689117224262e-07,
|
|
"loss": 1.198572039604187,
|
|
"step": 1601,
|
|
"token_acc": 0.6917315757938886
|
|
},
|
|
{
|
|
"epoch": 0.9390386869871044,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 1.0134534302323029e-07,
|
|
"loss": 1.1933400630950928,
|
|
"step": 1602,
|
|
"token_acc": 0.6892638036809816
|
|
},
|
|
{
|
|
"epoch": 0.9396248534583822,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 9.941221728254347e-08,
|
|
"loss": 1.2221391201019287,
|
|
"step": 1603,
|
|
"token_acc": 0.6794213719085395
|
|
},
|
|
{
|
|
"epoch": 0.94021101992966,
|
|
"grad_norm": 3.96875,
|
|
"learning_rate": 9.749752122010347e-08,
|
|
"loss": 1.2146235704421997,
|
|
"step": 1604,
|
|
"token_acc": 0.6727011494252874
|
|
},
|
|
{
|
|
"epoch": 0.9407971864009379,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 9.560126203652265e-08,
|
|
"loss": 1.2025160789489746,
|
|
"step": 1605,
|
|
"token_acc": 0.6857070961248113
|
|
},
|
|
{
|
|
"epoch": 0.9413833528722158,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 9.372344686307655e-08,
|
|
"loss": 1.1532485485076904,
|
|
"step": 1606,
|
|
"token_acc": 0.6911037891268533
|
|
},
|
|
{
|
|
"epoch": 0.9419695193434936,
|
|
"grad_norm": 3.4375,
|
|
"learning_rate": 9.186408276168012e-08,
|
|
"loss": 1.1789733171463013,
|
|
"step": 1607,
|
|
"token_acc": 0.6950320893748514
|
|
},
|
|
{
|
|
"epoch": 0.9425556858147714,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 9.002317672485828e-08,
|
|
"loss": 1.2205586433410645,
|
|
"step": 1608,
|
|
"token_acc": 0.6819541375872383
|
|
},
|
|
{
|
|
"epoch": 0.9431418522860492,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 8.82007356757203e-08,
|
|
"loss": 1.1377350091934204,
|
|
"step": 1609,
|
|
"token_acc": 0.6983842010771992
|
|
},
|
|
{
|
|
"epoch": 0.943728018757327,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 8.639676646793382e-08,
|
|
"loss": 1.1446236371994019,
|
|
"step": 1610,
|
|
"token_acc": 0.6957726957726957
|
|
},
|
|
{
|
|
"epoch": 0.9443141852286049,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 8.461127588570039e-08,
|
|
"loss": 1.0831918716430664,
|
|
"step": 1611,
|
|
"token_acc": 0.7034185401909455
|
|
},
|
|
{
|
|
"epoch": 0.9449003516998827,
|
|
"grad_norm": 3.734375,
|
|
"learning_rate": 8.284427064372769e-08,
|
|
"loss": 1.1800715923309326,
|
|
"step": 1612,
|
|
"token_acc": 0.6883720930232559
|
|
},
|
|
{
|
|
"epoch": 0.9454865181711606,
|
|
"grad_norm": 3.953125,
|
|
"learning_rate": 8.109575738720621e-08,
|
|
"loss": 1.1328818798065186,
|
|
"step": 1613,
|
|
"token_acc": 0.7031613976705491
|
|
},
|
|
{
|
|
"epoch": 0.9460726846424384,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 7.936574269178376e-08,
|
|
"loss": 1.2273576259613037,
|
|
"step": 1614,
|
|
"token_acc": 0.6885482938114517
|
|
},
|
|
{
|
|
"epoch": 0.9466588511137163,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 7.765423306353881e-08,
|
|
"loss": 1.1761486530303955,
|
|
"step": 1615,
|
|
"token_acc": 0.6854485776805251
|
|
},
|
|
{
|
|
"epoch": 0.9472450175849941,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 7.59612349389599e-08,
|
|
"loss": 1.1841951608657837,
|
|
"step": 1616,
|
|
"token_acc": 0.6964980544747081
|
|
},
|
|
{
|
|
"epoch": 0.947831184056272,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 7.428675468491742e-08,
|
|
"loss": 1.1557681560516357,
|
|
"step": 1617,
|
|
"token_acc": 0.6985119047619047
|
|
},
|
|
{
|
|
"epoch": 0.9484173505275498,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 7.263079859864298e-08,
|
|
"loss": 1.18850576877594,
|
|
"step": 1618,
|
|
"token_acc": 0.6867408906882592
|
|
},
|
|
{
|
|
"epoch": 0.9490035169988277,
|
|
"grad_norm": 3.625,
|
|
"learning_rate": 7.09933729077017e-08,
|
|
"loss": 1.1463152170181274,
|
|
"step": 1619,
|
|
"token_acc": 0.6869448183041723
|
|
},
|
|
{
|
|
"epoch": 0.9495896834701055,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 6.937448376997503e-08,
|
|
"loss": 1.1688690185546875,
|
|
"step": 1620,
|
|
"token_acc": 0.6927576601671309
|
|
},
|
|
{
|
|
"epoch": 0.9501758499413834,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 6.777413727363069e-08,
|
|
"loss": 1.1861896514892578,
|
|
"step": 1621,
|
|
"token_acc": 0.6879232505643341
|
|
},
|
|
{
|
|
"epoch": 0.9507620164126612,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 6.61923394371039e-08,
|
|
"loss": 1.146449327468872,
|
|
"step": 1622,
|
|
"token_acc": 0.6913987359164606
|
|
},
|
|
{
|
|
"epoch": 0.9513481828839391,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 6.46290962090751e-08,
|
|
"loss": 1.2255604267120361,
|
|
"step": 1623,
|
|
"token_acc": 0.6738596012018574
|
|
},
|
|
{
|
|
"epoch": 0.9519343493552169,
|
|
"grad_norm": 3.359375,
|
|
"learning_rate": 6.308441346844386e-08,
|
|
"loss": 1.1414072513580322,
|
|
"step": 1624,
|
|
"token_acc": 0.6983333333333334
|
|
},
|
|
{
|
|
"epoch": 0.9525205158264948,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 6.15582970243117e-08,
|
|
"loss": 1.1892120838165283,
|
|
"step": 1625,
|
|
"token_acc": 0.6921653300431833
|
|
},
|
|
{
|
|
"epoch": 0.9531066822977726,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 6.005075261595495e-08,
|
|
"loss": 1.2350949048995972,
|
|
"step": 1626,
|
|
"token_acc": 0.6690874644774234
|
|
},
|
|
{
|
|
"epoch": 0.9536928487690504,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 5.856178591280903e-08,
|
|
"loss": 1.113286018371582,
|
|
"step": 1627,
|
|
"token_acc": 0.7033232628398791
|
|
},
|
|
{
|
|
"epoch": 0.9542790152403282,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 5.709140251444201e-08,
|
|
"loss": 1.1649744510650635,
|
|
"step": 1628,
|
|
"token_acc": 0.6854779858915105
|
|
},
|
|
{
|
|
"epoch": 0.9548651817116061,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 5.563960795053558e-08,
|
|
"loss": 1.183915376663208,
|
|
"step": 1629,
|
|
"token_acc": 0.6901150627615062
|
|
},
|
|
{
|
|
"epoch": 0.9554513481828839,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 5.42064076808646e-08,
|
|
"loss": 1.1431870460510254,
|
|
"step": 1630,
|
|
"token_acc": 0.6927138331573389
|
|
},
|
|
{
|
|
"epoch": 0.9560375146541618,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 5.279180709527765e-08,
|
|
"loss": 1.1914196014404297,
|
|
"step": 1631,
|
|
"token_acc": 0.6818181818181818
|
|
},
|
|
{
|
|
"epoch": 0.9566236811254396,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 5.139581151367312e-08,
|
|
"loss": 1.1334843635559082,
|
|
"step": 1632,
|
|
"token_acc": 0.6989873417721519
|
|
},
|
|
{
|
|
"epoch": 0.9572098475967175,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 5.001842618598207e-08,
|
|
"loss": 1.1719768047332764,
|
|
"step": 1633,
|
|
"token_acc": 0.6919416243654822
|
|
},
|
|
{
|
|
"epoch": 0.9577960140679953,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 4.865965629214819e-08,
|
|
"loss": 1.0931975841522217,
|
|
"step": 1634,
|
|
"token_acc": 0.7094811018577835
|
|
},
|
|
{
|
|
"epoch": 0.9583821805392732,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 4.731950694210896e-08,
|
|
"loss": 1.1181995868682861,
|
|
"step": 1635,
|
|
"token_acc": 0.7001160092807425
|
|
},
|
|
{
|
|
"epoch": 0.958968347010551,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 4.599798317577342e-08,
|
|
"loss": 1.2620211839675903,
|
|
"step": 1636,
|
|
"token_acc": 0.6742051786299574
|
|
},
|
|
{
|
|
"epoch": 0.9595545134818289,
|
|
"grad_norm": 4.125,
|
|
"learning_rate": 4.469508996300664e-08,
|
|
"loss": 1.1763691902160645,
|
|
"step": 1637,
|
|
"token_acc": 0.6881387426447817
|
|
},
|
|
{
|
|
"epoch": 0.9601406799531067,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 4.3410832203608645e-08,
|
|
"loss": 1.1446363925933838,
|
|
"step": 1638,
|
|
"token_acc": 0.7026963103122044
|
|
},
|
|
{
|
|
"epoch": 0.9607268464243846,
|
|
"grad_norm": 3.5,
|
|
"learning_rate": 4.214521472729827e-08,
|
|
"loss": 1.1900640726089478,
|
|
"step": 1639,
|
|
"token_acc": 0.6868272680676576
|
|
},
|
|
{
|
|
"epoch": 0.9613130128956624,
|
|
"grad_norm": 5.46875,
|
|
"learning_rate": 4.0898242293691546e-08,
|
|
"loss": 1.2667232751846313,
|
|
"step": 1640,
|
|
"token_acc": 0.673224043715847
|
|
},
|
|
{
|
|
"epoch": 0.9618991793669402,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 3.9669919592288385e-08,
|
|
"loss": 1.1754297018051147,
|
|
"step": 1641,
|
|
"token_acc": 0.6844005777563794
|
|
},
|
|
{
|
|
"epoch": 0.9624853458382181,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 3.8460251242451454e-08,
|
|
"loss": 1.1270718574523926,
|
|
"step": 1642,
|
|
"token_acc": 0.7043558850787767
|
|
},
|
|
{
|
|
"epoch": 0.963071512309496,
|
|
"grad_norm": 3.984375,
|
|
"learning_rate": 3.726924179339009e-08,
|
|
"loss": 1.1872979402542114,
|
|
"step": 1643,
|
|
"token_acc": 0.6913690476190476
|
|
},
|
|
{
|
|
"epoch": 0.9636576787807737,
|
|
"grad_norm": 3.421875,
|
|
"learning_rate": 3.6096895724141435e-08,
|
|
"loss": 1.1511950492858887,
|
|
"step": 1644,
|
|
"token_acc": 0.6906544312967882
|
|
},
|
|
{
|
|
"epoch": 0.9642438452520515,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 3.494321744355767e-08,
|
|
"loss": 1.1618025302886963,
|
|
"step": 1645,
|
|
"token_acc": 0.6998719590268886
|
|
},
|
|
{
|
|
"epoch": 0.9648300117233294,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 3.3808211290284886e-08,
|
|
"loss": 1.2162041664123535,
|
|
"step": 1646,
|
|
"token_acc": 0.6846175089754212
|
|
},
|
|
{
|
|
"epoch": 0.9654161781946072,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 3.269188153275038e-08,
|
|
"loss": 1.1487592458724976,
|
|
"step": 1647,
|
|
"token_acc": 0.694665487768936
|
|
},
|
|
{
|
|
"epoch": 0.9660023446658851,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 3.159423236914261e-08,
|
|
"loss": 1.1773320436477661,
|
|
"step": 1648,
|
|
"token_acc": 0.6868244981986619
|
|
},
|
|
{
|
|
"epoch": 0.9665885111371629,
|
|
"grad_norm": 3.859375,
|
|
"learning_rate": 3.0515267927400116e-08,
|
|
"loss": 1.1662803888320923,
|
|
"step": 1649,
|
|
"token_acc": 0.6967808930425753
|
|
},
|
|
{
|
|
"epoch": 0.9671746776084408,
|
|
"grad_norm": 3.84375,
|
|
"learning_rate": 2.9454992265193216e-08,
|
|
"loss": 1.167703628540039,
|
|
"step": 1650,
|
|
"token_acc": 0.697907949790795
|
|
},
|
|
{
|
|
"epoch": 0.9677608440797186,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 2.8413409369907887e-08,
|
|
"loss": 1.2405099868774414,
|
|
"step": 1651,
|
|
"token_acc": 0.67423297580444
|
|
},
|
|
{
|
|
"epoch": 0.9683470105509965,
|
|
"grad_norm": 3.890625,
|
|
"learning_rate": 2.7390523158633552e-08,
|
|
"loss": 1.1604894399642944,
|
|
"step": 1652,
|
|
"token_acc": 0.6860259433962265
|
|
},
|
|
{
|
|
"epoch": 0.9689331770222743,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 2.638633747814534e-08,
|
|
"loss": 1.17080819606781,
|
|
"step": 1653,
|
|
"token_acc": 0.6891694997571637
|
|
},
|
|
{
|
|
"epoch": 0.9695193434935522,
|
|
"grad_norm": 4.25,
|
|
"learning_rate": 2.5400856104894066e-08,
|
|
"loss": 1.2003419399261475,
|
|
"step": 1654,
|
|
"token_acc": 0.6870034953924372
|
|
},
|
|
{
|
|
"epoch": 0.97010550996483,
|
|
"grad_norm": 3.921875,
|
|
"learning_rate": 2.44340827449846e-08,
|
|
"loss": 1.1847243309020996,
|
|
"step": 1655,
|
|
"token_acc": 0.681753079346892
|
|
},
|
|
{
|
|
"epoch": 0.9706916764361079,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 2.3486021034170857e-08,
|
|
"loss": 1.178354024887085,
|
|
"step": 1656,
|
|
"token_acc": 0.6950762309422645
|
|
},
|
|
{
|
|
"epoch": 0.9712778429073857,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 2.2556674537835276e-08,
|
|
"loss": 1.2082114219665527,
|
|
"step": 1657,
|
|
"token_acc": 0.6881665820213306
|
|
},
|
|
{
|
|
"epoch": 0.9718640093786636,
|
|
"grad_norm": 3.90625,
|
|
"learning_rate": 2.1646046750978255e-08,
|
|
"loss": 1.1701436042785645,
|
|
"step": 1658,
|
|
"token_acc": 0.6973386183465459
|
|
},
|
|
{
|
|
"epoch": 0.9724501758499414,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 2.0754141098204282e-08,
|
|
"loss": 1.2170751094818115,
|
|
"step": 1659,
|
|
"token_acc": 0.6908658922914467
|
|
},
|
|
{
|
|
"epoch": 0.9730363423212193,
|
|
"grad_norm": 4.3125,
|
|
"learning_rate": 1.9880960933710836e-08,
|
|
"loss": 1.1572237014770508,
|
|
"step": 1660,
|
|
"token_acc": 0.6933162775823473
|
|
},
|
|
{
|
|
"epoch": 0.9736225087924971,
|
|
"grad_norm": 3.453125,
|
|
"learning_rate": 1.9026509541272276e-08,
|
|
"loss": 1.1160807609558105,
|
|
"step": 1661,
|
|
"token_acc": 0.7033481043820778
|
|
},
|
|
{
|
|
"epoch": 0.9742086752637749,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 1.8190790134231528e-08,
|
|
"loss": 1.1553945541381836,
|
|
"step": 1662,
|
|
"token_acc": 0.6953302961275627
|
|
},
|
|
{
|
|
"epoch": 0.9747948417350527,
|
|
"grad_norm": 3.53125,
|
|
"learning_rate": 1.737380585548398e-08,
|
|
"loss": 1.1398444175720215,
|
|
"step": 1663,
|
|
"token_acc": 0.7015965819653699
|
|
},
|
|
{
|
|
"epoch": 0.9753810082063306,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 1.657555977746972e-08,
|
|
"loss": 1.1714553833007812,
|
|
"step": 1664,
|
|
"token_acc": 0.6912004345464422
|
|
},
|
|
{
|
|
"epoch": 0.9759671746776084,
|
|
"grad_norm": 3.484375,
|
|
"learning_rate": 1.5796054902157966e-08,
|
|
"loss": 1.1380133628845215,
|
|
"step": 1665,
|
|
"token_acc": 0.6851560745564788
|
|
},
|
|
{
|
|
"epoch": 0.9765533411488863,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 1.5035294161039882e-08,
|
|
"loss": 1.0883269309997559,
|
|
"step": 1666,
|
|
"token_acc": 0.7039170506912442
|
|
},
|
|
{
|
|
"epoch": 0.9771395076201641,
|
|
"grad_norm": 3.265625,
|
|
"learning_rate": 1.429328041511302e-08,
|
|
"loss": 1.154322624206543,
|
|
"step": 1667,
|
|
"token_acc": 0.6952380952380952
|
|
},
|
|
{
|
|
"epoch": 0.977725674091442,
|
|
"grad_norm": 3.703125,
|
|
"learning_rate": 1.3570016454874658e-08,
|
|
"loss": 1.181182622909546,
|
|
"step": 1668,
|
|
"token_acc": 0.6907114624505929
|
|
},
|
|
{
|
|
"epoch": 0.9783118405627198,
|
|
"grad_norm": 4.28125,
|
|
"learning_rate": 1.2865505000309586e-08,
|
|
"loss": 1.236847162246704,
|
|
"step": 1669,
|
|
"token_acc": 0.6688135593220339
|
|
},
|
|
{
|
|
"epoch": 0.9788980070339977,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 1.2179748700879013e-08,
|
|
"loss": 1.1684527397155762,
|
|
"step": 1670,
|
|
"token_acc": 0.6931216931216931
|
|
},
|
|
{
|
|
"epoch": 0.9794841735052755,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 1.1512750135511674e-08,
|
|
"loss": 1.176369547843933,
|
|
"step": 1671,
|
|
"token_acc": 0.6915676287492926
|
|
},
|
|
{
|
|
"epoch": 0.9800703399765534,
|
|
"grad_norm": 4.28125,
|
|
"learning_rate": 1.0864511812594958e-08,
|
|
"loss": 1.234694480895996,
|
|
"step": 1672,
|
|
"token_acc": 0.6773391358408455
|
|
},
|
|
{
|
|
"epoch": 0.9806565064478312,
|
|
"grad_norm": 3.8125,
|
|
"learning_rate": 1.0235036169963241e-08,
|
|
"loss": 1.167799472808838,
|
|
"step": 1673,
|
|
"token_acc": 0.6940344403444034
|
|
},
|
|
{
|
|
"epoch": 0.981242672919109,
|
|
"grad_norm": 4.28125,
|
|
"learning_rate": 9.624325574890125e-09,
|
|
"loss": 1.204697608947754,
|
|
"step": 1674,
|
|
"token_acc": 0.6845878136200717
|
|
},
|
|
{
|
|
"epoch": 0.9818288393903869,
|
|
"grad_norm": 4.34375,
|
|
"learning_rate": 9.032382324080102e-09,
|
|
"loss": 1.2196338176727295,
|
|
"step": 1675,
|
|
"token_acc": 0.6861547411285631
|
|
},
|
|
{
|
|
"epoch": 0.9824150058616647,
|
|
"grad_norm": 4.4375,
|
|
"learning_rate": 8.459208643659122e-09,
|
|
"loss": 1.199216365814209,
|
|
"step": 1676,
|
|
"token_acc": 0.6877667140825036
|
|
},
|
|
{
|
|
"epoch": 0.9830011723329426,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 7.904806689165156e-09,
|
|
"loss": 1.225955843925476,
|
|
"step": 1677,
|
|
"token_acc": 0.6870571563533302
|
|
},
|
|
{
|
|
"epoch": 0.9835873388042204,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 7.369178545542088e-09,
|
|
"loss": 1.1981971263885498,
|
|
"step": 1678,
|
|
"token_acc": 0.6907273736812882
|
|
},
|
|
{
|
|
"epoch": 0.9841735052754983,
|
|
"grad_norm": 4.03125,
|
|
"learning_rate": 6.852326227130835e-09,
|
|
"loss": 1.1561996936798096,
|
|
"step": 1679,
|
|
"token_acc": 0.7007162877608222
|
|
},
|
|
{
|
|
"epoch": 0.984759671746776,
|
|
"grad_norm": 3.875,
|
|
"learning_rate": 6.354251677661572e-09,
|
|
"loss": 1.1559357643127441,
|
|
"step": 1680,
|
|
"token_acc": 0.6980786825251601
|
|
},
|
|
{
|
|
"epoch": 0.9853458382180539,
|
|
"grad_norm": 4.0625,
|
|
"learning_rate": 5.874956770248186e-09,
|
|
"loss": 1.1268209218978882,
|
|
"step": 1681,
|
|
"token_acc": 0.6927424152290304
|
|
},
|
|
{
|
|
"epoch": 0.9859320046893317,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 5.414443307377171e-09,
|
|
"loss": 1.177412509918213,
|
|
"step": 1682,
|
|
"token_acc": 0.6946386946386947
|
|
},
|
|
{
|
|
"epoch": 0.9865181711606096,
|
|
"grad_norm": 3.71875,
|
|
"learning_rate": 4.9727130209065164e-09,
|
|
"loss": 1.2031774520874023,
|
|
"step": 1683,
|
|
"token_acc": 0.6860780065005417
|
|
},
|
|
{
|
|
"epoch": 0.9871043376318874,
|
|
"grad_norm": 3.484375,
|
|
"learning_rate": 4.5497675720540535e-09,
|
|
"loss": 1.1564052104949951,
|
|
"step": 1684,
|
|
"token_acc": 0.6950695322376739
|
|
},
|
|
{
|
|
"epoch": 0.9876905041031653,
|
|
"grad_norm": 3.34375,
|
|
"learning_rate": 4.145608551393565e-09,
|
|
"loss": 1.15203058719635,
|
|
"step": 1685,
|
|
"token_acc": 0.6945200155460551
|
|
},
|
|
{
|
|
"epoch": 0.9882766705744431,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 3.760237478849793e-09,
|
|
"loss": 1.204986572265625,
|
|
"step": 1686,
|
|
"token_acc": 0.6883313180169287
|
|
},
|
|
{
|
|
"epoch": 0.988862837045721,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 3.39365580369011e-09,
|
|
"loss": 1.1802268028259277,
|
|
"step": 1687,
|
|
"token_acc": 0.691230850501849
|
|
},
|
|
{
|
|
"epoch": 0.9894490035169988,
|
|
"grad_norm": 4.375,
|
|
"learning_rate": 3.0458649045211897e-09,
|
|
"loss": 1.2190027236938477,
|
|
"step": 1688,
|
|
"token_acc": 0.6822398939695162
|
|
},
|
|
{
|
|
"epoch": 0.9900351699882767,
|
|
"grad_norm": 3.78125,
|
|
"learning_rate": 2.716866089283454e-09,
|
|
"loss": 1.1407601833343506,
|
|
"step": 1689,
|
|
"token_acc": 0.6846089150546678
|
|
},
|
|
{
|
|
"epoch": 0.9906213364595545,
|
|
"grad_norm": 3.53125,
|
|
"learning_rate": 2.4066605952444144e-09,
|
|
"loss": 1.23799467086792,
|
|
"step": 1690,
|
|
"token_acc": 0.6828009828009828
|
|
},
|
|
{
|
|
"epoch": 0.9912075029308324,
|
|
"grad_norm": 3.671875,
|
|
"learning_rate": 2.1152495889970035e-09,
|
|
"loss": 1.2383923530578613,
|
|
"step": 1691,
|
|
"token_acc": 0.6736247544204322
|
|
},
|
|
{
|
|
"epoch": 0.9917936694021102,
|
|
"grad_norm": 3.6875,
|
|
"learning_rate": 1.8426341664529168e-09,
|
|
"loss": 1.1362744569778442,
|
|
"step": 1692,
|
|
"token_acc": 0.6990911323999017
|
|
},
|
|
{
|
|
"epoch": 0.9923798358733881,
|
|
"grad_norm": 3.453125,
|
|
"learning_rate": 1.5888153528381688e-09,
|
|
"loss": 1.1022942066192627,
|
|
"step": 1693,
|
|
"token_acc": 0.7018774033024203
|
|
},
|
|
{
|
|
"epoch": 0.9929660023446659,
|
|
"grad_norm": 4.0,
|
|
"learning_rate": 1.3537941026914302e-09,
|
|
"loss": 1.072890281677246,
|
|
"step": 1694,
|
|
"token_acc": 0.705423940149626
|
|
},
|
|
{
|
|
"epoch": 0.9935521688159438,
|
|
"grad_norm": 4.28125,
|
|
"learning_rate": 1.1375712998595855e-09,
|
|
"loss": 1.156285047531128,
|
|
"step": 1695,
|
|
"token_acc": 0.6974195828914811
|
|
},
|
|
{
|
|
"epoch": 0.9941383352872216,
|
|
"grad_norm": 3.453125,
|
|
"learning_rate": 9.401477574932927e-10,
|
|
"loss": 1.1880497932434082,
|
|
"step": 1696,
|
|
"token_acc": 0.6874725274725275
|
|
},
|
|
{
|
|
"epoch": 0.9947245017584995,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 7.615242180436521e-10,
|
|
"loss": 1.1535248756408691,
|
|
"step": 1697,
|
|
"token_acc": 0.6925795053003534
|
|
},
|
|
{
|
|
"epoch": 0.9953106682297772,
|
|
"grad_norm": 4.28125,
|
|
"learning_rate": 6.017013532627625e-10,
|
|
"loss": 1.2054600715637207,
|
|
"step": 1698,
|
|
"token_acc": 0.6987138263665594
|
|
},
|
|
{
|
|
"epoch": 0.9958968347010551,
|
|
"grad_norm": 3.921875,
|
|
"learning_rate": 4.606797641976135e-10,
|
|
"loss": 1.1637084484100342,
|
|
"step": 1699,
|
|
"token_acc": 0.7000307976593779
|
|
},
|
|
{
|
|
"epoch": 0.9964830011723329,
|
|
"grad_norm": 3.515625,
|
|
"learning_rate": 3.384599811889766e-10,
|
|
"loss": 1.124220371246338,
|
|
"step": 1700,
|
|
"token_acc": 0.7031635629347095
|
|
},
|
|
{
|
|
"epoch": 0.9970691676436108,
|
|
"grad_norm": 3.421875,
|
|
"learning_rate": 2.3504246386918394e-10,
|
|
"loss": 1.1244007349014282,
|
|
"step": 1701,
|
|
"token_acc": 0.701660042085574
|
|
},
|
|
{
|
|
"epoch": 0.9976553341148886,
|
|
"grad_norm": 4.40625,
|
|
"learning_rate": 1.504276011621286e-10,
|
|
"loss": 1.3062704801559448,
|
|
"step": 1702,
|
|
"token_acc": 0.6760957413739509
|
|
},
|
|
{
|
|
"epoch": 0.9982415005861665,
|
|
"grad_norm": 3.609375,
|
|
"learning_rate": 8.461571127882373e-11,
|
|
"loss": 1.190398931503296,
|
|
"step": 1703,
|
|
"token_acc": 0.6872118417859743
|
|
},
|
|
{
|
|
"epoch": 0.9988276670574443,
|
|
"grad_norm": 4.46875,
|
|
"learning_rate": 3.760704171962282e-11,
|
|
"loss": 1.2514241933822632,
|
|
"step": 1704,
|
|
"token_acc": 0.682760501909438
|
|
},
|
|
{
|
|
"epoch": 0.9994138335287222,
|
|
"grad_norm": 4.5625,
|
|
"learning_rate": 9.401769269223871e-12,
|
|
"loss": 1.1998621225357056,
|
|
"step": 1705,
|
|
"token_acc": 0.6889679715302491
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"grad_norm": 4.96875,
|
|
"learning_rate": 0.0,
|
|
"loss": 1.183607578277588,
|
|
"step": 1706,
|
|
"token_acc": 0.6978911230995586
|
|
}
|
|
],
|
|
"logging_steps": 1,
|
|
"max_steps": 1706,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 1,
|
|
"save_steps": 100,
|
|
"stateful_callbacks": {
|
|
"TrainerControl": {
|
|
"args": {
|
|
"should_epoch_stop": false,
|
|
"should_evaluate": false,
|
|
"should_log": false,
|
|
"should_save": true,
|
|
"should_training_stop": true
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 9.767433214686659e+17,
|
|
"train_batch_size": 4,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|