6732 lines
187 KiB
JSON
6732 lines
187 KiB
JSON
|
{
|
||
|
|
"best_global_step": null,
|
||
|
|
"best_metric": null,
|
||
|
|
"best_model_checkpoint": null,
|
||
|
|
"epoch": 5.0,
|
||
|
|
"eval_steps": 500,
|
||
|
|
"global_step": 3040,
|
||
|
|
"is_hyper_param_search": false,
|
||
|
|
"is_local_process_zero": true,
|
||
|
|
"is_world_process_zero": true,
|
||
|
|
"log_history": [
|
||
|
|
{
|
||
|
|
"epoch": 0.008228195282501372,
|
||
|
|
"grad_norm": 13.661592714819882,
|
||
|
|
"learning_rate": 5.263157894736843e-07,
|
||
|
|
"loss": 0.9672,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.33080172538757324,
|
||
|
|
"step": 5,
|
||
|
|
"valid_targets_mean": 8170.6,
|
||
|
|
"valid_targets_min": 3077
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.016456390565002744,
|
||
|
|
"grad_norm": 11.927844187875976,
|
||
|
|
"learning_rate": 1.1842105263157894e-06,
|
||
|
|
"loss": 0.9636,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.30587294697761536,
|
||
|
|
"step": 10,
|
||
|
|
"valid_targets_mean": 6980.7,
|
||
|
|
"valid_targets_min": 2298
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.024684585847504114,
|
||
|
|
"grad_norm": 6.744886263610615,
|
||
|
|
"learning_rate": 1.8421052631578948e-06,
|
||
|
|
"loss": 0.9073,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2631882429122925,
|
||
|
|
"step": 15,
|
||
|
|
"valid_targets_mean": 6148.2,
|
||
|
|
"valid_targets_min": 2167
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03291278113000549,
|
||
|
|
"grad_norm": 2.892063783530988,
|
||
|
|
"learning_rate": 2.5e-06,
|
||
|
|
"loss": 0.8329,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.280141681432724,
|
||
|
|
"step": 20,
|
||
|
|
"valid_targets_mean": 7165.8,
|
||
|
|
"valid_targets_min": 2278
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.04114097641250686,
|
||
|
|
"grad_norm": 1.7886922348255612,
|
||
|
|
"learning_rate": 3.157894736842105e-06,
|
||
|
|
"loss": 0.7725,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25607916712760925,
|
||
|
|
"step": 25,
|
||
|
|
"valid_targets_mean": 7202.3,
|
||
|
|
"valid_targets_min": 2907
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.04936917169500823,
|
||
|
|
"grad_norm": 1.4610050364093832,
|
||
|
|
"learning_rate": 3.815789473684211e-06,
|
||
|
|
"loss": 0.7477,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24936680495738983,
|
||
|
|
"step": 30,
|
||
|
|
"valid_targets_mean": 7477.5,
|
||
|
|
"valid_targets_min": 2714
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.0575973669775096,
|
||
|
|
"grad_norm": 0.9352394471095702,
|
||
|
|
"learning_rate": 4.473684210526316e-06,
|
||
|
|
"loss": 0.7078,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23676790297031403,
|
||
|
|
"step": 35,
|
||
|
|
"valid_targets_mean": 7314.0,
|
||
|
|
"valid_targets_min": 2883
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06582556226001098,
|
||
|
|
"grad_norm": 0.7295427249654216,
|
||
|
|
"learning_rate": 5.131578947368422e-06,
|
||
|
|
"loss": 0.6618,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21210411190986633,
|
||
|
|
"step": 40,
|
||
|
|
"valid_targets_mean": 7126.6,
|
||
|
|
"valid_targets_min": 2597
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07405375754251234,
|
||
|
|
"grad_norm": 0.6133963494588915,
|
||
|
|
"learning_rate": 5.789473684210527e-06,
|
||
|
|
"loss": 0.6337,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19923362135887146,
|
||
|
|
"step": 45,
|
||
|
|
"valid_targets_mean": 7369.8,
|
||
|
|
"valid_targets_min": 3375
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08228195282501372,
|
||
|
|
"grad_norm": 0.4495388936542107,
|
||
|
|
"learning_rate": 6.447368421052632e-06,
|
||
|
|
"loss": 0.5998,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19039945304393768,
|
||
|
|
"step": 50,
|
||
|
|
"valid_targets_mean": 7356.4,
|
||
|
|
"valid_targets_min": 2005
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09051014810751508,
|
||
|
|
"grad_norm": 0.3995911861766218,
|
||
|
|
"learning_rate": 7.1052631578947375e-06,
|
||
|
|
"loss": 0.5558,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17436350882053375,
|
||
|
|
"step": 55,
|
||
|
|
"valid_targets_mean": 8719.0,
|
||
|
|
"valid_targets_min": 589
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09873834339001646,
|
||
|
|
"grad_norm": 0.3499717213397111,
|
||
|
|
"learning_rate": 7.763157894736843e-06,
|
||
|
|
"loss": 0.5219,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18273428082466125,
|
||
|
|
"step": 60,
|
||
|
|
"valid_targets_mean": 9029.5,
|
||
|
|
"valid_targets_min": 4655
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.10696653867251783,
|
||
|
|
"grad_norm": 0.26109198291101654,
|
||
|
|
"learning_rate": 8.421052631578948e-06,
|
||
|
|
"loss": 0.4966,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16715273261070251,
|
||
|
|
"step": 65,
|
||
|
|
"valid_targets_mean": 10019.8,
|
||
|
|
"valid_targets_min": 4996
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1151947339550192,
|
||
|
|
"grad_norm": 0.24495980513506116,
|
||
|
|
"learning_rate": 9.078947368421054e-06,
|
||
|
|
"loss": 0.4791,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15863889455795288,
|
||
|
|
"step": 70,
|
||
|
|
"valid_targets_mean": 9413.2,
|
||
|
|
"valid_targets_min": 3337
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12342292923752057,
|
||
|
|
"grad_norm": 0.27613600523015236,
|
||
|
|
"learning_rate": 9.736842105263159e-06,
|
||
|
|
"loss": 0.4742,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16512612998485565,
|
||
|
|
"step": 75,
|
||
|
|
"valid_targets_mean": 10127.1,
|
||
|
|
"valid_targets_min": 3360
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13165112452002195,
|
||
|
|
"grad_norm": 0.22054068931113005,
|
||
|
|
"learning_rate": 1.0394736842105264e-05,
|
||
|
|
"loss": 0.4552,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16071820259094238,
|
||
|
|
"step": 80,
|
||
|
|
"valid_targets_mean": 9319.1,
|
||
|
|
"valid_targets_min": 685
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1398793198025233,
|
||
|
|
"grad_norm": 0.2616751014414881,
|
||
|
|
"learning_rate": 1.105263157894737e-05,
|
||
|
|
"loss": 0.4578,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15120165050029755,
|
||
|
|
"step": 85,
|
||
|
|
"valid_targets_mean": 9109.9,
|
||
|
|
"valid_targets_min": 4379
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.14810751508502468,
|
||
|
|
"grad_norm": 0.25389258411623156,
|
||
|
|
"learning_rate": 1.1710526315789475e-05,
|
||
|
|
"loss": 0.443,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14307349920272827,
|
||
|
|
"step": 90,
|
||
|
|
"valid_targets_mean": 9313.2,
|
||
|
|
"valid_targets_min": 3452
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15633571036752605,
|
||
|
|
"grad_norm": 0.25388019851426047,
|
||
|
|
"learning_rate": 1.236842105263158e-05,
|
||
|
|
"loss": 0.4362,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14230628311634064,
|
||
|
|
"step": 95,
|
||
|
|
"valid_targets_mean": 8640.9,
|
||
|
|
"valid_targets_min": 383
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16456390565002743,
|
||
|
|
"grad_norm": 0.2405595523038858,
|
||
|
|
"learning_rate": 1.3026315789473684e-05,
|
||
|
|
"loss": 0.4342,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14278680086135864,
|
||
|
|
"step": 100,
|
||
|
|
"valid_targets_mean": 9029.9,
|
||
|
|
"valid_targets_min": 4552
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1727921009325288,
|
||
|
|
"grad_norm": 0.22176898805086837,
|
||
|
|
"learning_rate": 1.3684210526315791e-05,
|
||
|
|
"loss": 0.4262,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14409483969211578,
|
||
|
|
"step": 105,
|
||
|
|
"valid_targets_mean": 10390.2,
|
||
|
|
"valid_targets_min": 4833
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18102029621503016,
|
||
|
|
"grad_norm": 0.243090924751181,
|
||
|
|
"learning_rate": 1.4342105263157895e-05,
|
||
|
|
"loss": 0.4172,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1436709314584732,
|
||
|
|
"step": 110,
|
||
|
|
"valid_targets_mean": 9771.6,
|
||
|
|
"valid_targets_min": 3641
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18924849149753153,
|
||
|
|
"grad_norm": 0.2625711173343529,
|
||
|
|
"learning_rate": 1.5000000000000002e-05,
|
||
|
|
"loss": 0.4153,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1286260485649109,
|
||
|
|
"step": 115,
|
||
|
|
"valid_targets_mean": 9316.0,
|
||
|
|
"valid_targets_min": 1929
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1974766867800329,
|
||
|
|
"grad_norm": 0.22939426799124796,
|
||
|
|
"learning_rate": 1.5657894736842107e-05,
|
||
|
|
"loss": 0.4154,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13896283507347107,
|
||
|
|
"step": 120,
|
||
|
|
"valid_targets_mean": 9838.7,
|
||
|
|
"valid_targets_min": 4324
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2057048820625343,
|
||
|
|
"grad_norm": 0.24963282482796506,
|
||
|
|
"learning_rate": 1.6315789473684213e-05,
|
||
|
|
"loss": 0.4208,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14296197891235352,
|
||
|
|
"step": 125,
|
||
|
|
"valid_targets_mean": 9693.1,
|
||
|
|
"valid_targets_min": 3811
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.21393307734503567,
|
||
|
|
"grad_norm": 0.30535130323003906,
|
||
|
|
"learning_rate": 1.6973684210526318e-05,
|
||
|
|
"loss": 0.4063,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12299084663391113,
|
||
|
|
"step": 130,
|
||
|
|
"valid_targets_mean": 9686.9,
|
||
|
|
"valid_targets_min": 3338
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22216127262753702,
|
||
|
|
"grad_norm": 0.284400177589226,
|
||
|
|
"learning_rate": 1.763157894736842e-05,
|
||
|
|
"loss": 0.3964,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.123782217502594,
|
||
|
|
"step": 135,
|
||
|
|
"valid_targets_mean": 9590.6,
|
||
|
|
"valid_targets_min": 4785
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2303894679100384,
|
||
|
|
"grad_norm": 0.23276724546182528,
|
||
|
|
"learning_rate": 1.828947368421053e-05,
|
||
|
|
"loss": 0.3988,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14101338386535645,
|
||
|
|
"step": 140,
|
||
|
|
"valid_targets_mean": 10718.9,
|
||
|
|
"valid_targets_min": 4892
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23861766319253977,
|
||
|
|
"grad_norm": 0.26025681943282925,
|
||
|
|
"learning_rate": 1.894736842105263e-05,
|
||
|
|
"loss": 0.4023,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1524072289466858,
|
||
|
|
"step": 145,
|
||
|
|
"valid_targets_mean": 10753.2,
|
||
|
|
"valid_targets_min": 3836
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24684585847504115,
|
||
|
|
"grad_norm": 0.2358512322697237,
|
||
|
|
"learning_rate": 1.960526315789474e-05,
|
||
|
|
"loss": 0.3961,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11710461974143982,
|
||
|
|
"step": 150,
|
||
|
|
"valid_targets_mean": 8877.9,
|
||
|
|
"valid_targets_min": 424
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2550740537575425,
|
||
|
|
"grad_norm": 0.23568354731119345,
|
||
|
|
"learning_rate": 2.0263157894736842e-05,
|
||
|
|
"loss": 0.3967,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14156289398670197,
|
||
|
|
"step": 155,
|
||
|
|
"valid_targets_mean": 11518.0,
|
||
|
|
"valid_targets_min": 4504
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2633022490400439,
|
||
|
|
"grad_norm": 0.26669374560220394,
|
||
|
|
"learning_rate": 2.0921052631578947e-05,
|
||
|
|
"loss": 0.3946,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14294372498989105,
|
||
|
|
"step": 160,
|
||
|
|
"valid_targets_mean": 10316.1,
|
||
|
|
"valid_targets_min": 4842
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.27153044432254525,
|
||
|
|
"grad_norm": 0.24067166137023635,
|
||
|
|
"learning_rate": 2.1578947368421056e-05,
|
||
|
|
"loss": 0.3925,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1319475769996643,
|
||
|
|
"step": 165,
|
||
|
|
"valid_targets_mean": 10630.5,
|
||
|
|
"valid_targets_min": 5085
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2797586396050466,
|
||
|
|
"grad_norm": 0.24772975734231953,
|
||
|
|
"learning_rate": 2.223684210526316e-05,
|
||
|
|
"loss": 0.3894,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14050891995429993,
|
||
|
|
"step": 170,
|
||
|
|
"valid_targets_mean": 10322.6,
|
||
|
|
"valid_targets_min": 4928
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.287986834887548,
|
||
|
|
"grad_norm": 0.278650053623802,
|
||
|
|
"learning_rate": 2.2894736842105263e-05,
|
||
|
|
"loss": 0.3853,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13519951701164246,
|
||
|
|
"step": 175,
|
||
|
|
"valid_targets_mean": 10297.2,
|
||
|
|
"valid_targets_min": 3888
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.29621503017004935,
|
||
|
|
"grad_norm": 0.2589128123009836,
|
||
|
|
"learning_rate": 2.355263157894737e-05,
|
||
|
|
"loss": 0.3841,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12493768334388733,
|
||
|
|
"step": 180,
|
||
|
|
"valid_targets_mean": 8726.7,
|
||
|
|
"valid_targets_min": 265
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.30444322545255076,
|
||
|
|
"grad_norm": 0.29125365436187955,
|
||
|
|
"learning_rate": 2.4210526315789474e-05,
|
||
|
|
"loss": 0.3787,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11316806077957153,
|
||
|
|
"step": 185,
|
||
|
|
"valid_targets_mean": 8473.2,
|
||
|
|
"valid_targets_min": 3703
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3126714207350521,
|
||
|
|
"grad_norm": 0.2553461910788868,
|
||
|
|
"learning_rate": 2.4868421052631583e-05,
|
||
|
|
"loss": 0.3812,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12836065888404846,
|
||
|
|
"step": 190,
|
||
|
|
"valid_targets_mean": 10562.8,
|
||
|
|
"valid_targets_min": 4059
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.32089961601755346,
|
||
|
|
"grad_norm": 0.33557598477032663,
|
||
|
|
"learning_rate": 2.5526315789473688e-05,
|
||
|
|
"loss": 0.4022,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13181576132774353,
|
||
|
|
"step": 195,
|
||
|
|
"valid_targets_mean": 5732.4,
|
||
|
|
"valid_targets_min": 996
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.32912781130005486,
|
||
|
|
"grad_norm": 0.34352411636418323,
|
||
|
|
"learning_rate": 2.618421052631579e-05,
|
||
|
|
"loss": 0.418,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13641154766082764,
|
||
|
|
"step": 200,
|
||
|
|
"valid_targets_mean": 5909.7,
|
||
|
|
"valid_targets_min": 322
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3373560065825562,
|
||
|
|
"grad_norm": 3.4954428619882982,
|
||
|
|
"learning_rate": 2.6842105263157896e-05,
|
||
|
|
"loss": 0.6654,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.34785598516464233,
|
||
|
|
"step": 205,
|
||
|
|
"valid_targets_mean": 3622.3,
|
||
|
|
"valid_targets_min": 1417
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3455842018650576,
|
||
|
|
"grad_norm": 1.077861726475296,
|
||
|
|
"learning_rate": 2.75e-05,
|
||
|
|
"loss": 0.8938,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3038213849067688,
|
||
|
|
"step": 210,
|
||
|
|
"valid_targets_mean": 4234.8,
|
||
|
|
"valid_targets_min": 1229
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.35381239714755897,
|
||
|
|
"grad_norm": 0.5677105395157203,
|
||
|
|
"learning_rate": 2.815789473684211e-05,
|
||
|
|
"loss": 0.7774,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18273773789405823,
|
||
|
|
"step": 215,
|
||
|
|
"valid_targets_mean": 3496.8,
|
||
|
|
"valid_targets_min": 792
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3620405924300603,
|
||
|
|
"grad_norm": 0.44846475537880953,
|
||
|
|
"learning_rate": 2.8815789473684215e-05,
|
||
|
|
"loss": 0.8285,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2714836001396179,
|
||
|
|
"step": 220,
|
||
|
|
"valid_targets_mean": 4790.7,
|
||
|
|
"valid_targets_min": 1423
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3702687877125617,
|
||
|
|
"grad_norm": 0.3456686765451588,
|
||
|
|
"learning_rate": 2.9473684210526317e-05,
|
||
|
|
"loss": 0.7522,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28811782598495483,
|
||
|
|
"step": 225,
|
||
|
|
"valid_targets_mean": 4529.3,
|
||
|
|
"valid_targets_min": 1236
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.37849698299506307,
|
||
|
|
"grad_norm": 0.37556290856776664,
|
||
|
|
"learning_rate": 3.0131578947368423e-05,
|
||
|
|
"loss": 0.7109,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21670036017894745,
|
||
|
|
"step": 230,
|
||
|
|
"valid_targets_mean": 3307.1,
|
||
|
|
"valid_targets_min": 1399
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3867251782775645,
|
||
|
|
"grad_norm": 0.33500865716540523,
|
||
|
|
"learning_rate": 3.078947368421053e-05,
|
||
|
|
"loss": 0.718,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23109467327594757,
|
||
|
|
"step": 235,
|
||
|
|
"valid_targets_mean": 4017.8,
|
||
|
|
"valid_targets_min": 1176
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3949533735600658,
|
||
|
|
"grad_norm": 0.31598383847429745,
|
||
|
|
"learning_rate": 3.144736842105264e-05,
|
||
|
|
"loss": 0.6864,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14081542193889618,
|
||
|
|
"step": 240,
|
||
|
|
"valid_targets_mean": 1564.6,
|
||
|
|
"valid_targets_min": 635
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4031815688425672,
|
||
|
|
"grad_norm": 0.3067298171268501,
|
||
|
|
"learning_rate": 3.210526315789474e-05,
|
||
|
|
"loss": 0.704,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23977398872375488,
|
||
|
|
"step": 245,
|
||
|
|
"valid_targets_mean": 4694.7,
|
||
|
|
"valid_targets_min": 1910
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4114097641250686,
|
||
|
|
"grad_norm": 0.34304197183917,
|
||
|
|
"learning_rate": 3.276315789473684e-05,
|
||
|
|
"loss": 0.6787,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2708185315132141,
|
||
|
|
"step": 250,
|
||
|
|
"valid_targets_mean": 4543.5,
|
||
|
|
"valid_targets_min": 712
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4196379594075699,
|
||
|
|
"grad_norm": 0.3757101087371721,
|
||
|
|
"learning_rate": 3.342105263157895e-05,
|
||
|
|
"loss": 0.6624,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23688086867332458,
|
||
|
|
"step": 255,
|
||
|
|
"valid_targets_mean": 3509.4,
|
||
|
|
"valid_targets_min": 1442
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.42786615469007133,
|
||
|
|
"grad_norm": 0.3090407246266453,
|
||
|
|
"learning_rate": 3.407894736842106e-05,
|
||
|
|
"loss": 0.6722,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1637752205133438,
|
||
|
|
"step": 260,
|
||
|
|
"valid_targets_mean": 3599.4,
|
||
|
|
"valid_targets_min": 328
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4360943499725727,
|
||
|
|
"grad_norm": 0.33923285458485086,
|
||
|
|
"learning_rate": 3.473684210526316e-05,
|
||
|
|
"loss": 0.635,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11224784702062607,
|
||
|
|
"step": 265,
|
||
|
|
"valid_targets_mean": 980.5,
|
||
|
|
"valid_targets_min": 475
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.44432254525507403,
|
||
|
|
"grad_norm": 0.3392379658417928,
|
||
|
|
"learning_rate": 3.539473684210526e-05,
|
||
|
|
"loss": 0.6713,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24215322732925415,
|
||
|
|
"step": 270,
|
||
|
|
"valid_targets_mean": 4849.9,
|
||
|
|
"valid_targets_min": 1243
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.45255074053757544,
|
||
|
|
"grad_norm": 0.3150618515035599,
|
||
|
|
"learning_rate": 3.605263157894737e-05,
|
||
|
|
"loss": 0.6483,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2192254364490509,
|
||
|
|
"step": 275,
|
||
|
|
"valid_targets_mean": 4380.0,
|
||
|
|
"valid_targets_min": 1142
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4607789358200768,
|
||
|
|
"grad_norm": 0.3472360677625879,
|
||
|
|
"learning_rate": 3.671052631578948e-05,
|
||
|
|
"loss": 0.6846,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2536545991897583,
|
||
|
|
"step": 280,
|
||
|
|
"valid_targets_mean": 3904.7,
|
||
|
|
"valid_targets_min": 1187
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4690071311025782,
|
||
|
|
"grad_norm": 0.3156614973458573,
|
||
|
|
"learning_rate": 3.736842105263158e-05,
|
||
|
|
"loss": 0.6585,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1983037143945694,
|
||
|
|
"step": 285,
|
||
|
|
"valid_targets_mean": 3962.2,
|
||
|
|
"valid_targets_min": 1010
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.47723532638507954,
|
||
|
|
"grad_norm": 0.37742389790153924,
|
||
|
|
"learning_rate": 3.802631578947369e-05,
|
||
|
|
"loss": 0.6333,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1320263296365738,
|
||
|
|
"step": 290,
|
||
|
|
"valid_targets_mean": 1099.3,
|
||
|
|
"valid_targets_min": 280
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4854635216675809,
|
||
|
|
"grad_norm": 0.2962337727307249,
|
||
|
|
"learning_rate": 3.868421052631579e-05,
|
||
|
|
"loss": 0.6569,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.246580570936203,
|
||
|
|
"step": 295,
|
||
|
|
"valid_targets_mean": 5050.1,
|
||
|
|
"valid_targets_min": 1178
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4936917169500823,
|
||
|
|
"grad_norm": 0.2856275355602754,
|
||
|
|
"learning_rate": 3.9342105263157895e-05,
|
||
|
|
"loss": 0.6432,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2303171455860138,
|
||
|
|
"step": 300,
|
||
|
|
"valid_targets_mean": 5151.1,
|
||
|
|
"valid_targets_min": 1346
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5019199122325837,
|
||
|
|
"grad_norm": 0.33300122404573673,
|
||
|
|
"learning_rate": 4e-05,
|
||
|
|
"loss": 0.4498,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08987970650196075,
|
||
|
|
"step": 305,
|
||
|
|
"valid_targets_mean": 5515.5,
|
||
|
|
"valid_targets_min": 1936
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.510148107515085,
|
||
|
|
"grad_norm": 0.29692627282197165,
|
||
|
|
"learning_rate": 3.999967038544942e-05,
|
||
|
|
"loss": 0.2477,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07394493371248245,
|
||
|
|
"step": 310,
|
||
|
|
"valid_targets_mean": 5127.6,
|
||
|
|
"valid_targets_min": 2180
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5183763027975864,
|
||
|
|
"grad_norm": 0.25936222317294383,
|
||
|
|
"learning_rate": 3.9998681552662254e-05,
|
||
|
|
"loss": 0.2281,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06937037408351898,
|
||
|
|
"step": 315,
|
||
|
|
"valid_targets_mean": 5021.8,
|
||
|
|
"valid_targets_min": 2808
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5266044980800878,
|
||
|
|
"grad_norm": 0.3327155662385847,
|
||
|
|
"learning_rate": 3.999703353423185e-05,
|
||
|
|
"loss": 0.2203,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.072842076420784,
|
||
|
|
"step": 320,
|
||
|
|
"valid_targets_mean": 5327.4,
|
||
|
|
"valid_targets_min": 2771
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5348326933625891,
|
||
|
|
"grad_norm": 0.2504539380850954,
|
||
|
|
"learning_rate": 3.999472638447933e-05,
|
||
|
|
"loss": 0.2143,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07985015213489532,
|
||
|
|
"step": 325,
|
||
|
|
"valid_targets_mean": 5568.0,
|
||
|
|
"valid_targets_min": 3219
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5430608886450905,
|
||
|
|
"grad_norm": 0.2814701429958317,
|
||
|
|
"learning_rate": 3.999176017945168e-05,
|
||
|
|
"loss": 0.2107,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06997586786746979,
|
||
|
|
"step": 330,
|
||
|
|
"valid_targets_mean": 5279.6,
|
||
|
|
"valid_targets_min": 3339
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5512890839275919,
|
||
|
|
"grad_norm": 0.23889431711323486,
|
||
|
|
"learning_rate": 3.998813501691934e-05,
|
||
|
|
"loss": 0.205,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06921573728322983,
|
||
|
|
"step": 335,
|
||
|
|
"valid_targets_mean": 5462.8,
|
||
|
|
"valid_targets_min": 1825
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5595172792100932,
|
||
|
|
"grad_norm": 0.2620420014965469,
|
||
|
|
"learning_rate": 3.9983851016372945e-05,
|
||
|
|
"loss": 0.2049,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06677880883216858,
|
||
|
|
"step": 340,
|
||
|
|
"valid_targets_mean": 5447.9,
|
||
|
|
"valid_targets_min": 2456
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5677454744925946,
|
||
|
|
"grad_norm": 0.2222928848241994,
|
||
|
|
"learning_rate": 3.997890831901938e-05,
|
||
|
|
"loss": 0.2062,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06969676911830902,
|
||
|
|
"step": 345,
|
||
|
|
"valid_targets_mean": 5365.3,
|
||
|
|
"valid_targets_min": 2902
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.575973669775096,
|
||
|
|
"grad_norm": 0.24107676881576715,
|
||
|
|
"learning_rate": 3.997330708777714e-05,
|
||
|
|
"loss": 0.1978,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06276493519544601,
|
||
|
|
"step": 350,
|
||
|
|
"valid_targets_mean": 4987.0,
|
||
|
|
"valid_targets_min": 3287
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5842018650575974,
|
||
|
|
"grad_norm": 0.21062109587656125,
|
||
|
|
"learning_rate": 3.996704750727097e-05,
|
||
|
|
"loss": 0.197,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05960414931178093,
|
||
|
|
"step": 355,
|
||
|
|
"valid_targets_mean": 5047.8,
|
||
|
|
"valid_targets_min": 2252
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5924300603400987,
|
||
|
|
"grad_norm": 0.2165529180549222,
|
||
|
|
"learning_rate": 3.9960129783825746e-05,
|
||
|
|
"loss": 0.1981,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07164271175861359,
|
||
|
|
"step": 360,
|
||
|
|
"valid_targets_mean": 5323.9,
|
||
|
|
"valid_targets_min": 2401
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6006582556226001,
|
||
|
|
"grad_norm": 0.2602414889343082,
|
||
|
|
"learning_rate": 3.995255414545969e-05,
|
||
|
|
"loss": 0.1927,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06296073645353317,
|
||
|
|
"step": 365,
|
||
|
|
"valid_targets_mean": 5421.0,
|
||
|
|
"valid_targets_min": 916
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6088864509051015,
|
||
|
|
"grad_norm": 0.22113365838524923,
|
||
|
|
"learning_rate": 3.994432084187688e-05,
|
||
|
|
"loss": 0.1932,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05878105387091637,
|
||
|
|
"step": 370,
|
||
|
|
"valid_targets_mean": 5157.3,
|
||
|
|
"valid_targets_min": 2787
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6171146461876028,
|
||
|
|
"grad_norm": 0.2726984781426435,
|
||
|
|
"learning_rate": 3.993543014445897e-05,
|
||
|
|
"loss": 0.1935,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06641367077827454,
|
||
|
|
"step": 375,
|
||
|
|
"valid_targets_mean": 5285.9,
|
||
|
|
"valid_targets_min": 2725
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6253428414701042,
|
||
|
|
"grad_norm": 0.24103613454295666,
|
||
|
|
"learning_rate": 3.992588234625629e-05,
|
||
|
|
"loss": 0.1943,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06343524158000946,
|
||
|
|
"step": 380,
|
||
|
|
"valid_targets_mean": 4471.1,
|
||
|
|
"valid_targets_min": 773
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6335710367526056,
|
||
|
|
"grad_norm": 0.23661985213609446,
|
||
|
|
"learning_rate": 3.991567776197815e-05,
|
||
|
|
"loss": 0.1924,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06470175087451935,
|
||
|
|
"step": 385,
|
||
|
|
"valid_targets_mean": 5323.2,
|
||
|
|
"valid_targets_min": 3113
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6417992320351069,
|
||
|
|
"grad_norm": 0.2365355448830919,
|
||
|
|
"learning_rate": 3.990481672798251e-05,
|
||
|
|
"loss": 0.1941,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06568753719329834,
|
||
|
|
"step": 390,
|
||
|
|
"valid_targets_mean": 5507.8,
|
||
|
|
"valid_targets_min": 3054
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6500274273176083,
|
||
|
|
"grad_norm": 0.250970116784741,
|
||
|
|
"learning_rate": 3.989329960226486e-05,
|
||
|
|
"loss": 0.1924,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06060680001974106,
|
||
|
|
"step": 395,
|
||
|
|
"valid_targets_mean": 5065.1,
|
||
|
|
"valid_targets_min": 2739
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6582556226001097,
|
||
|
|
"grad_norm": 0.24890957035716,
|
||
|
|
"learning_rate": 3.988112676444639e-05,
|
||
|
|
"loss": 0.1874,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06493669748306274,
|
||
|
|
"step": 400,
|
||
|
|
"valid_targets_mean": 5267.3,
|
||
|
|
"valid_targets_min": 3285
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6664838178826111,
|
||
|
|
"grad_norm": 0.2123920671176028,
|
||
|
|
"learning_rate": 3.9868298615761586e-05,
|
||
|
|
"loss": 0.1854,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06611112505197525,
|
||
|
|
"step": 405,
|
||
|
|
"valid_targets_mean": 5350.7,
|
||
|
|
"valid_targets_min": 2809
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6747120131651124,
|
||
|
|
"grad_norm": 0.45669439712276166,
|
||
|
|
"learning_rate": 3.9854815579044866e-05,
|
||
|
|
"loss": 0.5087,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1900215744972229,
|
||
|
|
"step": 410,
|
||
|
|
"valid_targets_mean": 4092.8,
|
||
|
|
"valid_targets_min": 1262
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6829402084476138,
|
||
|
|
"grad_norm": 0.34680859400827757,
|
||
|
|
"learning_rate": 3.984067809871675e-05,
|
||
|
|
"loss": 0.5508,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1873103380203247,
|
||
|
|
"step": 415,
|
||
|
|
"valid_targets_mean": 3959.6,
|
||
|
|
"valid_targets_min": 922
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6911684037301152,
|
||
|
|
"grad_norm": 0.33789768611131743,
|
||
|
|
"learning_rate": 3.982588664076916e-05,
|
||
|
|
"loss": 0.5498,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.172877237200737,
|
||
|
|
"step": 420,
|
||
|
|
"valid_targets_mean": 3500.9,
|
||
|
|
"valid_targets_min": 1193
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6993965990126165,
|
||
|
|
"grad_norm": 0.3422411868231782,
|
||
|
|
"learning_rate": 3.981044169275006e-05,
|
||
|
|
"loss": 0.5395,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16920597851276398,
|
||
|
|
"step": 425,
|
||
|
|
"valid_targets_mean": 3293.8,
|
||
|
|
"valid_targets_min": 1211
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7076247942951179,
|
||
|
|
"grad_norm": 0.2819763328446606,
|
||
|
|
"learning_rate": 3.979434376374744e-05,
|
||
|
|
"loss": 0.5325,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19173139333724976,
|
||
|
|
"step": 430,
|
||
|
|
"valid_targets_mean": 4374.8,
|
||
|
|
"valid_targets_min": 1271
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7158529895776193,
|
||
|
|
"grad_norm": 0.3302657868370585,
|
||
|
|
"learning_rate": 3.9777593384372436e-05,
|
||
|
|
"loss": 0.5511,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15441164374351501,
|
||
|
|
"step": 435,
|
||
|
|
"valid_targets_mean": 2742.0,
|
||
|
|
"valid_targets_min": 1163
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7240811848601206,
|
||
|
|
"grad_norm": 0.36559426634652425,
|
||
|
|
"learning_rate": 3.9760191106741935e-05,
|
||
|
|
"loss": 0.525,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17873308062553406,
|
||
|
|
"step": 440,
|
||
|
|
"valid_targets_mean": 2944.6,
|
||
|
|
"valid_targets_min": 747
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.732309380142622,
|
||
|
|
"grad_norm": 0.29776253451819223,
|
||
|
|
"learning_rate": 3.9742137504460326e-05,
|
||
|
|
"loss": 0.4966,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09976305067539215,
|
||
|
|
"step": 445,
|
||
|
|
"valid_targets_mean": 1390.7,
|
||
|
|
"valid_targets_min": 617
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7405375754251234,
|
||
|
|
"grad_norm": 0.2741654148501185,
|
||
|
|
"learning_rate": 3.972343317260061e-05,
|
||
|
|
"loss": 0.556,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20187368988990784,
|
||
|
|
"step": 450,
|
||
|
|
"valid_targets_mean": 5282.3,
|
||
|
|
"valid_targets_min": 1416
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7487657707076248,
|
||
|
|
"grad_norm": 0.29327103026837464,
|
||
|
|
"learning_rate": 3.970407872768478e-05,
|
||
|
|
"loss": 0.5153,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21182456612586975,
|
||
|
|
"step": 455,
|
||
|
|
"valid_targets_mean": 5102.7,
|
||
|
|
"valid_targets_min": 635
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7569939659901261,
|
||
|
|
"grad_norm": 0.3027137360580798,
|
||
|
|
"learning_rate": 3.968407480766352e-05,
|
||
|
|
"loss": 0.5377,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17653965950012207,
|
||
|
|
"step": 460,
|
||
|
|
"valid_targets_mean": 4471.8,
|
||
|
|
"valid_targets_min": 1166
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7652221612726275,
|
||
|
|
"grad_norm": 0.3131357916514081,
|
||
|
|
"learning_rate": 3.9663422071895103e-05,
|
||
|
|
"loss": 0.5099,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1851877123117447,
|
||
|
|
"step": 465,
|
||
|
|
"valid_targets_mean": 3015.3,
|
||
|
|
"valid_targets_min": 644
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.773450356555129,
|
||
|
|
"grad_norm": 0.30585055011385276,
|
||
|
|
"learning_rate": 3.964212120112379e-05,
|
||
|
|
"loss": 0.5066,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16202497482299805,
|
||
|
|
"step": 470,
|
||
|
|
"valid_targets_mean": 4168.5,
|
||
|
|
"valid_targets_min": 1159
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7816785518376302,
|
||
|
|
"grad_norm": 0.3185891945520371,
|
||
|
|
"learning_rate": 3.962017289745724e-05,
|
||
|
|
"loss": 0.5529,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18970099091529846,
|
||
|
|
"step": 475,
|
||
|
|
"valid_targets_mean": 4139.7,
|
||
|
|
"valid_targets_min": 1171
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7899067471201316,
|
||
|
|
"grad_norm": 0.3193022031463681,
|
||
|
|
"learning_rate": 3.959757788434351e-05,
|
||
|
|
"loss": 0.5281,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1599006950855255,
|
||
|
|
"step": 480,
|
||
|
|
"valid_targets_mean": 3524.7,
|
||
|
|
"valid_targets_min": 706
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.798134942402633,
|
||
|
|
"grad_norm": 0.47454037589519754,
|
||
|
|
"learning_rate": 3.957433690654709e-05,
|
||
|
|
"loss": 0.5262,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13566067814826965,
|
||
|
|
"step": 485,
|
||
|
|
"valid_targets_mean": 974.1,
|
||
|
|
"valid_targets_min": 373
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8063631376851343,
|
||
|
|
"grad_norm": 0.27137258308574463,
|
||
|
|
"learning_rate": 3.955045073012443e-05,
|
||
|
|
"loss": 0.5126,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1551508903503418,
|
||
|
|
"step": 490,
|
||
|
|
"valid_targets_mean": 4034.9,
|
||
|
|
"valid_targets_min": 1413
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8145913329676358,
|
||
|
|
"grad_norm": 0.2891185235274737,
|
||
|
|
"learning_rate": 3.952592014239867e-05,
|
||
|
|
"loss": 0.5201,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18325182795524597,
|
||
|
|
"step": 495,
|
||
|
|
"valid_targets_mean": 4597.8,
|
||
|
|
"valid_targets_min": 1386
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8228195282501372,
|
||
|
|
"grad_norm": 0.30627166767474445,
|
||
|
|
"learning_rate": 3.950074595193366e-05,
|
||
|
|
"loss": 0.5038,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14900541305541992,
|
||
|
|
"step": 500,
|
||
|
|
"valid_targets_mean": 3344.0,
|
||
|
|
"valid_targets_min": 1074
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8310477235326386,
|
||
|
|
"grad_norm": 0.2910945730347226,
|
||
|
|
"learning_rate": 3.947492898850736e-05,
|
||
|
|
"loss": 0.5297,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2500567138195038,
|
||
|
|
"step": 505,
|
||
|
|
"valid_targets_mean": 4781.9,
|
||
|
|
"valid_targets_min": 1038
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8392759188151399,
|
||
|
|
"grad_norm": 0.28859181669258777,
|
||
|
|
"learning_rate": 3.9448470103084436e-05,
|
||
|
|
"loss": 0.5061,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14351779222488403,
|
||
|
|
"step": 510,
|
||
|
|
"valid_targets_mean": 4052.2,
|
||
|
|
"valid_targets_min": 1148
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8475041140976413,
|
||
|
|
"grad_norm": 0.2849202691210404,
|
||
|
|
"learning_rate": 3.942137016778826e-05,
|
||
|
|
"loss": 0.4405,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09605582058429718,
|
||
|
|
"step": 515,
|
||
|
|
"valid_targets_mean": 5513.2,
|
||
|
|
"valid_targets_min": 260
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8557323093801427,
|
||
|
|
"grad_norm": 0.24873959226778644,
|
||
|
|
"learning_rate": 3.939363007587213e-05,
|
||
|
|
"loss": 0.3552,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1327248215675354,
|
||
|
|
"step": 520,
|
||
|
|
"valid_targets_mean": 6674.3,
|
||
|
|
"valid_targets_min": 667
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.863960504662644,
|
||
|
|
"grad_norm": 0.2519692215563831,
|
||
|
|
"learning_rate": 3.9365250741689835e-05,
|
||
|
|
"loss": 0.3647,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11264800280332565,
|
||
|
|
"step": 525,
|
||
|
|
"valid_targets_mean": 5004.6,
|
||
|
|
"valid_targets_min": 184
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8721886999451454,
|
||
|
|
"grad_norm": 0.2342060171912288,
|
||
|
|
"learning_rate": 3.933623310066554e-05,
|
||
|
|
"loss": 0.3511,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11143757402896881,
|
||
|
|
"step": 530,
|
||
|
|
"valid_targets_mean": 6059.3,
|
||
|
|
"valid_targets_min": 496
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8804168952276468,
|
||
|
|
"grad_norm": 0.271362287684106,
|
||
|
|
"learning_rate": 3.9306578109262894e-05,
|
||
|
|
"loss": 0.347,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05567864701151848,
|
||
|
|
"step": 535,
|
||
|
|
"valid_targets_mean": 1604.8,
|
||
|
|
"valid_targets_min": 520
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8886450905101481,
|
||
|
|
"grad_norm": 0.2597578802375837,
|
||
|
|
"learning_rate": 3.927628674495357e-05,
|
||
|
|
"loss": 0.3431,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09169554710388184,
|
||
|
|
"step": 540,
|
||
|
|
"valid_targets_mean": 6672.8,
|
||
|
|
"valid_targets_min": 750
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8968732857926495,
|
||
|
|
"grad_norm": 0.21749824898637793,
|
||
|
|
"learning_rate": 3.924536000618501e-05,
|
||
|
|
"loss": 0.3249,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10265297442674637,
|
||
|
|
"step": 545,
|
||
|
|
"valid_targets_mean": 5849.8,
|
||
|
|
"valid_targets_min": 225
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9051014810751509,
|
||
|
|
"grad_norm": 0.2241552515651752,
|
||
|
|
"learning_rate": 3.921379891234753e-05,
|
||
|
|
"loss": 0.3254,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11423005163669586,
|
||
|
|
"step": 550,
|
||
|
|
"valid_targets_mean": 6783.3,
|
||
|
|
"valid_targets_min": 643
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9133296763576523,
|
||
|
|
"grad_norm": 0.2158340366537389,
|
||
|
|
"learning_rate": 3.9181604503740714e-05,
|
||
|
|
"loss": 0.3199,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09704571962356567,
|
||
|
|
"step": 555,
|
||
|
|
"valid_targets_mean": 6440.8,
|
||
|
|
"valid_targets_min": 392
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9215578716401536,
|
||
|
|
"grad_norm": 0.26183611514857996,
|
||
|
|
"learning_rate": 3.914877784153909e-05,
|
||
|
|
"loss": 0.3367,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17041084170341492,
|
||
|
|
"step": 560,
|
||
|
|
"valid_targets_mean": 8714.9,
|
||
|
|
"valid_targets_min": 1646
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.929786066922655,
|
||
|
|
"grad_norm": 0.20432961383680928,
|
||
|
|
"learning_rate": 3.9115320007757225e-05,
|
||
|
|
"loss": 0.3168,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0803799033164978,
|
||
|
|
"step": 565,
|
||
|
|
"valid_targets_mean": 6242.6,
|
||
|
|
"valid_targets_min": 1338
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9380142622051564,
|
||
|
|
"grad_norm": 0.2092825817868237,
|
||
|
|
"learning_rate": 3.9081232105214e-05,
|
||
|
|
"loss": 0.3229,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13302992284297943,
|
||
|
|
"step": 570,
|
||
|
|
"valid_targets_mean": 7600.2,
|
||
|
|
"valid_targets_min": 399
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9462424574876577,
|
||
|
|
"grad_norm": 0.19714866749443535,
|
||
|
|
"learning_rate": 3.9046515257496295e-05,
|
||
|
|
"loss": 0.3387,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0796457827091217,
|
||
|
|
"step": 575,
|
||
|
|
"valid_targets_mean": 6031.7,
|
||
|
|
"valid_targets_min": 993
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9544706527701591,
|
||
|
|
"grad_norm": 0.20932629070373632,
|
||
|
|
"learning_rate": 3.9011170608921904e-05,
|
||
|
|
"loss": 0.295,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09574976563453674,
|
||
|
|
"step": 580,
|
||
|
|
"valid_targets_mean": 6480.3,
|
||
|
|
"valid_targets_min": 279
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9626988480526605,
|
||
|
|
"grad_norm": 0.19884348922834869,
|
||
|
|
"learning_rate": 3.897519932450189e-05,
|
||
|
|
"loss": 0.3276,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08668152987957001,
|
||
|
|
"step": 585,
|
||
|
|
"valid_targets_mean": 6095.1,
|
||
|
|
"valid_targets_min": 1420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9709270433351618,
|
||
|
|
"grad_norm": 0.1916490209833201,
|
||
|
|
"learning_rate": 3.893860258990212e-05,
|
||
|
|
"loss": 0.3026,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09876567125320435,
|
||
|
|
"step": 590,
|
||
|
|
"valid_targets_mean": 6487.8,
|
||
|
|
"valid_targets_min": 491
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9791552386176632,
|
||
|
|
"grad_norm": 0.24087156746724506,
|
||
|
|
"learning_rate": 3.890138161140421e-05,
|
||
|
|
"loss": 0.2973,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08016130328178406,
|
||
|
|
"step": 595,
|
||
|
|
"valid_targets_mean": 1157.9,
|
||
|
|
"valid_targets_min": 493
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9873834339001646,
|
||
|
|
"grad_norm": 0.21112403994094667,
|
||
|
|
"learning_rate": 3.886353761586579e-05,
|
||
|
|
"loss": 0.3127,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10186732560396194,
|
||
|
|
"step": 600,
|
||
|
|
"valid_targets_mean": 6671.6,
|
||
|
|
"valid_targets_min": 354
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9956116291826659,
|
||
|
|
"grad_norm": 0.20983694665510674,
|
||
|
|
"learning_rate": 3.8825071850679996e-05,
|
||
|
|
"loss": 0.3009,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12147838622331619,
|
||
|
|
"step": 605,
|
||
|
|
"valid_targets_mean": 7020.9,
|
||
|
|
"valid_targets_min": 848
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0032912781130006,
|
||
|
|
"grad_norm": 1.659397044858526,
|
||
|
|
"learning_rate": 3.878598558373443e-05,
|
||
|
|
"loss": 0.5467,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18808016180992126,
|
||
|
|
"step": 610,
|
||
|
|
"valid_targets_mean": 6370.5,
|
||
|
|
"valid_targets_min": 2089
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.011519473395502,
|
||
|
|
"grad_norm": 0.7209332976643361,
|
||
|
|
"learning_rate": 3.874628010336932e-05,
|
||
|
|
"loss": 0.5739,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20240893959999084,
|
||
|
|
"step": 615,
|
||
|
|
"valid_targets_mean": 7651.2,
|
||
|
|
"valid_targets_min": 2581
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0197476686780034,
|
||
|
|
"grad_norm": 0.4882521280358526,
|
||
|
|
"learning_rate": 3.870595671833508e-05,
|
||
|
|
"loss": 0.5011,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16959220170974731,
|
||
|
|
"step": 620,
|
||
|
|
"valid_targets_mean": 7917.3,
|
||
|
|
"valid_targets_min": 2931
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0279758639605046,
|
||
|
|
"grad_norm": 0.3524983441886274,
|
||
|
|
"learning_rate": 3.866501675774914e-05,
|
||
|
|
"loss": 0.4729,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16359078884124756,
|
||
|
|
"step": 625,
|
||
|
|
"valid_targets_mean": 7465.7,
|
||
|
|
"valid_targets_min": 1655
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.036204059243006,
|
||
|
|
"grad_norm": 0.285845283172002,
|
||
|
|
"learning_rate": 3.862346157105219e-05,
|
||
|
|
"loss": 0.444,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13884960114955902,
|
||
|
|
"step": 630,
|
||
|
|
"valid_targets_mean": 6986.5,
|
||
|
|
"valid_targets_min": 2745
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0444322545255074,
|
||
|
|
"grad_norm": 0.2637082715943371,
|
||
|
|
"learning_rate": 3.858129252796363e-05,
|
||
|
|
"loss": 0.4377,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12356004118919373,
|
||
|
|
"step": 635,
|
||
|
|
"valid_targets_mean": 6226.8,
|
||
|
|
"valid_targets_min": 2078
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0526604498080088,
|
||
|
|
"grad_norm": 0.28887382496755204,
|
||
|
|
"learning_rate": 3.853851101843649e-05,
|
||
|
|
"loss": 0.4291,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12742659449577332,
|
||
|
|
"step": 640,
|
||
|
|
"valid_targets_mean": 6504.0,
|
||
|
|
"valid_targets_min": 3466
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0608886450905102,
|
||
|
|
"grad_norm": 0.24476017136685832,
|
||
|
|
"learning_rate": 3.8495118452611574e-05,
|
||
|
|
"loss": 0.4201,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14294564723968506,
|
||
|
|
"step": 645,
|
||
|
|
"valid_targets_mean": 7776.8,
|
||
|
|
"valid_targets_min": 2749
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0691168403730116,
|
||
|
|
"grad_norm": 0.2726802244958609,
|
||
|
|
"learning_rate": 3.845111626077097e-05,
|
||
|
|
"loss": 0.4119,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12050601094961166,
|
||
|
|
"step": 650,
|
||
|
|
"valid_targets_mean": 6354.8,
|
||
|
|
"valid_targets_min": 1941
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.077345035655513,
|
||
|
|
"grad_norm": 0.2773725725381812,
|
||
|
|
"learning_rate": 3.840650589329098e-05,
|
||
|
|
"loss": 0.4111,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1138051226735115,
|
||
|
|
"step": 655,
|
||
|
|
"valid_targets_mean": 6384.4,
|
||
|
|
"valid_targets_min": 2441
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0855732309380142,
|
||
|
|
"grad_norm": 0.26559994898000056,
|
||
|
|
"learning_rate": 3.83612888205942e-05,
|
||
|
|
"loss": 0.4077,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13822872936725616,
|
||
|
|
"step": 660,
|
||
|
|
"valid_targets_mean": 7465.9,
|
||
|
|
"valid_targets_min": 2228
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0938014262205156,
|
||
|
|
"grad_norm": 0.22208110131792275,
|
||
|
|
"learning_rate": 3.8315466533101154e-05,
|
||
|
|
"loss": 0.3852,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11882581561803818,
|
||
|
|
"step": 665,
|
||
|
|
"valid_targets_mean": 8703.7,
|
||
|
|
"valid_targets_min": 4392
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.102029621503017,
|
||
|
|
"grad_norm": 0.21922616622876517,
|
||
|
|
"learning_rate": 3.82690405411811e-05,
|
||
|
|
"loss": 0.3758,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11796137690544128,
|
||
|
|
"step": 670,
|
||
|
|
"valid_targets_mean": 9525.9,
|
||
|
|
"valid_targets_min": 3807
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1102578167855184,
|
||
|
|
"grad_norm": 0.2335012875150806,
|
||
|
|
"learning_rate": 3.82220123751023e-05,
|
||
|
|
"loss": 0.3727,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12342832237482071,
|
||
|
|
"step": 675,
|
||
|
|
"valid_targets_mean": 8976.7,
|
||
|
|
"valid_targets_min": 2713
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1184860120680198,
|
||
|
|
"grad_norm": 0.20615998005456446,
|
||
|
|
"learning_rate": 3.8174383584981525e-05,
|
||
|
|
"loss": 0.3683,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12510226666927338,
|
||
|
|
"step": 680,
|
||
|
|
"valid_targets_mean": 9808.0,
|
||
|
|
"valid_targets_min": 5333
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1267142073505212,
|
||
|
|
"grad_norm": 0.20549746954269069,
|
||
|
|
"learning_rate": 3.812615574073301e-05,
|
||
|
|
"loss": 0.3675,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11540381610393524,
|
||
|
|
"step": 685,
|
||
|
|
"valid_targets_mean": 8583.8,
|
||
|
|
"valid_targets_min": 2662
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1349424026330226,
|
||
|
|
"grad_norm": 0.354746331129916,
|
||
|
|
"learning_rate": 3.807733043201666e-05,
|
||
|
|
"loss": 0.3624,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11764580011367798,
|
||
|
|
"step": 690,
|
||
|
|
"valid_targets_mean": 8976.7,
|
||
|
|
"valid_targets_min": 899
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1431705979155238,
|
||
|
|
"grad_norm": 0.22026383511742792,
|
||
|
|
"learning_rate": 3.8027909268185695e-05,
|
||
|
|
"loss": 0.3701,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1217566654086113,
|
||
|
|
"step": 695,
|
||
|
|
"valid_targets_mean": 8862.6,
|
||
|
|
"valid_targets_min": 363
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1513987931980252,
|
||
|
|
"grad_norm": 0.24232587865327862,
|
||
|
|
"learning_rate": 3.7977893878233604e-05,
|
||
|
|
"loss": 0.3592,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12140684574842453,
|
||
|
|
"step": 700,
|
||
|
|
"valid_targets_mean": 9068.4,
|
||
|
|
"valid_targets_min": 532
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1596269884805266,
|
||
|
|
"grad_norm": 0.21509299959913547,
|
||
|
|
"learning_rate": 3.792728591074041e-05,
|
||
|
|
"loss": 0.3619,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12871021032333374,
|
||
|
|
"step": 705,
|
||
|
|
"valid_targets_mean": 10089.5,
|
||
|
|
"valid_targets_min": 5419
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.167855183763028,
|
||
|
|
"grad_norm": 0.231174997822983,
|
||
|
|
"learning_rate": 3.7876087033818345e-05,
|
||
|
|
"loss": 0.3584,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11500117927789688,
|
||
|
|
"step": 710,
|
||
|
|
"valid_targets_mean": 9003.8,
|
||
|
|
"valid_targets_min": 387
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1760833790455294,
|
||
|
|
"grad_norm": 0.22555157163732995,
|
||
|
|
"learning_rate": 3.78242989350569e-05,
|
||
|
|
"loss": 0.3579,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12305817008018494,
|
||
|
|
"step": 715,
|
||
|
|
"valid_targets_mean": 10012.9,
|
||
|
|
"valid_targets_min": 5197
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1843115743280308,
|
||
|
|
"grad_norm": 0.23487162908677478,
|
||
|
|
"learning_rate": 3.7771923321467163e-05,
|
||
|
|
"loss": 0.3555,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12105908989906311,
|
||
|
|
"step": 720,
|
||
|
|
"valid_targets_mean": 9826.7,
|
||
|
|
"valid_targets_min": 464
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1925397696105322,
|
||
|
|
"grad_norm": 0.2517013646791283,
|
||
|
|
"learning_rate": 3.771896191942556e-05,
|
||
|
|
"loss": 0.3526,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10659809410572052,
|
||
|
|
"step": 725,
|
||
|
|
"valid_targets_mean": 8842.1,
|
||
|
|
"valid_targets_min": 4239
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2007679648930334,
|
||
|
|
"grad_norm": 0.2156644331644718,
|
||
|
|
"learning_rate": 3.7665416474616986e-05,
|
||
|
|
"loss": 0.3569,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10782045125961304,
|
||
|
|
"step": 730,
|
||
|
|
"valid_targets_mean": 9048.3,
|
||
|
|
"valid_targets_min": 5085
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2089961601755348,
|
||
|
|
"grad_norm": 0.26346428818115725,
|
||
|
|
"learning_rate": 3.761128875197719e-05,
|
||
|
|
"loss": 0.362,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11525176465511322,
|
||
|
|
"step": 735,
|
||
|
|
"valid_targets_mean": 9484.0,
|
||
|
|
"valid_targets_min": 4512
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2172243554580362,
|
||
|
|
"grad_norm": 0.22755907847000245,
|
||
|
|
"learning_rate": 3.7556580535634685e-05,
|
||
|
|
"loss": 0.3485,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11937953531742096,
|
||
|
|
"step": 740,
|
||
|
|
"valid_targets_mean": 9844.3,
|
||
|
|
"valid_targets_min": 3863
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2254525507405376,
|
||
|
|
"grad_norm": 0.22714801320069883,
|
||
|
|
"learning_rate": 3.750129362885188e-05,
|
||
|
|
"loss": 0.3469,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10537885129451752,
|
||
|
|
"step": 745,
|
||
|
|
"valid_targets_mean": 9963.1,
|
||
|
|
"valid_targets_min": 5225
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.233680746023039,
|
||
|
|
"grad_norm": 0.2147068269485451,
|
||
|
|
"learning_rate": 3.744542985396566e-05,
|
||
|
|
"loss": 0.3486,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10538201034069061,
|
||
|
|
"step": 750,
|
||
|
|
"valid_targets_mean": 9133.7,
|
||
|
|
"valid_targets_min": 2782
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2419089413055404,
|
||
|
|
"grad_norm": 0.22999817220649277,
|
||
|
|
"learning_rate": 3.738899105232734e-05,
|
||
|
|
"loss": 0.3524,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12429474294185638,
|
||
|
|
"step": 755,
|
||
|
|
"valid_targets_mean": 9841.2,
|
||
|
|
"valid_targets_min": 885
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2501371365880418,
|
||
|
|
"grad_norm": 0.2182279465433469,
|
||
|
|
"learning_rate": 3.733197908424194e-05,
|
||
|
|
"loss": 0.3437,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10498598963022232,
|
||
|
|
"step": 760,
|
||
|
|
"valid_targets_mean": 9260.1,
|
||
|
|
"valid_targets_min": 378
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.258365331870543,
|
||
|
|
"grad_norm": 0.2176991000254134,
|
||
|
|
"learning_rate": 3.727439582890689e-05,
|
||
|
|
"loss": 0.3521,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11775383353233337,
|
||
|
|
"step": 765,
|
||
|
|
"valid_targets_mean": 9212.2,
|
||
|
|
"valid_targets_min": 3467
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2665935271530444,
|
||
|
|
"grad_norm": 0.22474069393548862,
|
||
|
|
"learning_rate": 3.721624318435006e-05,
|
||
|
|
"loss": 0.3442,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11866054683923721,
|
||
|
|
"step": 770,
|
||
|
|
"valid_targets_mean": 10565.9,
|
||
|
|
"valid_targets_min": 723
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2748217224355458,
|
||
|
|
"grad_norm": 0.3553450372737768,
|
||
|
|
"learning_rate": 3.715752306736724e-05,
|
||
|
|
"loss": 0.3491,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11934887617826462,
|
||
|
|
"step": 775,
|
||
|
|
"valid_targets_mean": 10235.8,
|
||
|
|
"valid_targets_min": 5141
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2830499177180472,
|
||
|
|
"grad_norm": 0.24117058830790514,
|
||
|
|
"learning_rate": 3.709823741345894e-05,
|
||
|
|
"loss": 0.3408,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12090543657541275,
|
||
|
|
"step": 780,
|
||
|
|
"valid_targets_mean": 10594.5,
|
||
|
|
"valid_targets_min": 4701
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2912781130005486,
|
||
|
|
"grad_norm": 0.27672367825987276,
|
||
|
|
"learning_rate": 3.703838817676654e-05,
|
||
|
|
"loss": 0.3399,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1089012622833252,
|
||
|
|
"step": 785,
|
||
|
|
"valid_targets_mean": 9161.7,
|
||
|
|
"valid_targets_min": 3022
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2995063082830498,
|
||
|
|
"grad_norm": 0.21099639116289254,
|
||
|
|
"learning_rate": 3.6977977330008e-05,
|
||
|
|
"loss": 0.3432,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1100674569606781,
|
||
|
|
"step": 790,
|
||
|
|
"valid_targets_mean": 9593.7,
|
||
|
|
"valid_targets_min": 5120
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3077345035655512,
|
||
|
|
"grad_norm": 0.27202881396330664,
|
||
|
|
"learning_rate": 3.691700686441272e-05,
|
||
|
|
"loss": 0.3365,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10544559359550476,
|
||
|
|
"step": 795,
|
||
|
|
"valid_targets_mean": 9896.2,
|
||
|
|
"valid_targets_min": 3759
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3159626988480526,
|
||
|
|
"grad_norm": 0.29008227740554365,
|
||
|
|
"learning_rate": 3.685547878965595e-05,
|
||
|
|
"loss": 0.3418,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11258372664451599,
|
||
|
|
"step": 800,
|
||
|
|
"valid_targets_mean": 9676.5,
|
||
|
|
"valid_targets_min": 4171
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.324190894130554,
|
||
|
|
"grad_norm": 0.2648068230573467,
|
||
|
|
"learning_rate": 3.679339513379257e-05,
|
||
|
|
"loss": 0.3654,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11481667309999466,
|
||
|
|
"step": 805,
|
||
|
|
"valid_targets_mean": 6269.3,
|
||
|
|
"valid_targets_min": 470
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3324190894130554,
|
||
|
|
"grad_norm": 0.3302755564975962,
|
||
|
|
"learning_rate": 3.673075794319022e-05,
|
||
|
|
"loss": 0.3703,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1273985505104065,
|
||
|
|
"step": 810,
|
||
|
|
"valid_targets_mean": 6197.9,
|
||
|
|
"valid_targets_min": 350
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3406472846955568,
|
||
|
|
"grad_norm": 0.9946048795797314,
|
||
|
|
"learning_rate": 3.6667569282461835e-05,
|
||
|
|
"loss": 0.7179,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.29462897777557373,
|
||
|
|
"step": 815,
|
||
|
|
"valid_targets_mean": 5028.6,
|
||
|
|
"valid_targets_min": 935
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3488754799780582,
|
||
|
|
"grad_norm": 0.5672650836141364,
|
||
|
|
"learning_rate": 3.660383123439761e-05,
|
||
|
|
"loss": 0.6726,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1995016634464264,
|
||
|
|
"step": 820,
|
||
|
|
"valid_targets_mean": 3933.0,
|
||
|
|
"valid_targets_min": 797
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3571036752605594,
|
||
|
|
"grad_norm": 0.42579660991027274,
|
||
|
|
"learning_rate": 3.653954589989637e-05,
|
||
|
|
"loss": 0.6596,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18716245889663696,
|
||
|
|
"step": 825,
|
||
|
|
"valid_targets_mean": 3912.4,
|
||
|
|
"valid_targets_min": 1137
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3653318705430608,
|
||
|
|
"grad_norm": 0.30342608710540986,
|
||
|
|
"learning_rate": 3.647471539789626e-05,
|
||
|
|
"loss": 0.6193,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2176496684551239,
|
||
|
|
"step": 830,
|
||
|
|
"valid_targets_mean": 3427.5,
|
||
|
|
"valid_targets_min": 570
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3735600658255622,
|
||
|
|
"grad_norm": 0.2692438365384602,
|
||
|
|
"learning_rate": 3.640934186530496e-05,
|
||
|
|
"loss": 0.612,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17222759127616882,
|
||
|
|
"step": 835,
|
||
|
|
"valid_targets_mean": 4285.9,
|
||
|
|
"valid_targets_min": 1385
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3817882611080636,
|
||
|
|
"grad_norm": 0.3252340214730497,
|
||
|
|
"learning_rate": 3.634342745692924e-05,
|
||
|
|
"loss": 0.6096,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23037424683570862,
|
||
|
|
"step": 840,
|
||
|
|
"valid_targets_mean": 4571.2,
|
||
|
|
"valid_targets_min": 1195
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.390016456390565,
|
||
|
|
"grad_norm": 0.283339399288758,
|
||
|
|
"learning_rate": 3.62769743454039e-05,
|
||
|
|
"loss": 0.5952,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17118248343467712,
|
||
|
|
"step": 845,
|
||
|
|
"valid_targets_mean": 3360.0,
|
||
|
|
"valid_targets_min": 791
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3982446516730664,
|
||
|
|
"grad_norm": 0.2935408435366708,
|
||
|
|
"learning_rate": 3.6209984721120195e-05,
|
||
|
|
"loss": 0.6073,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18600332736968994,
|
||
|
|
"step": 850,
|
||
|
|
"valid_targets_mean": 3748.8,
|
||
|
|
"valid_targets_min": 1312
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4064728469555678,
|
||
|
|
"grad_norm": 0.2785551163427506,
|
||
|
|
"learning_rate": 3.614246079215361e-05,
|
||
|
|
"loss": 0.5841,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2570624053478241,
|
||
|
|
"step": 855,
|
||
|
|
"valid_targets_mean": 4315.9,
|
||
|
|
"valid_targets_min": 858
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.414701042238069,
|
||
|
|
"grad_norm": 0.2571094312626162,
|
||
|
|
"learning_rate": 3.6074404784191084e-05,
|
||
|
|
"loss": 0.5723,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2021176815032959,
|
||
|
|
"step": 860,
|
||
|
|
"valid_targets_mean": 5165.7,
|
||
|
|
"valid_targets_min": 1332
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4229292375205704,
|
||
|
|
"grad_norm": 0.29270135824966964,
|
||
|
|
"learning_rate": 3.600581894045768e-05,
|
||
|
|
"loss": 0.5871,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17256325483322144,
|
||
|
|
"step": 865,
|
||
|
|
"valid_targets_mean": 4124.3,
|
||
|
|
"valid_targets_min": 1668
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4311574328030718,
|
||
|
|
"grad_norm": 0.2943443689190457,
|
||
|
|
"learning_rate": 3.593670552164261e-05,
|
||
|
|
"loss": 0.5814,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19908928871154785,
|
||
|
|
"step": 870,
|
||
|
|
"valid_targets_mean": 4181.1,
|
||
|
|
"valid_targets_min": 897
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4393856280855732,
|
||
|
|
"grad_norm": 0.30602158217934583,
|
||
|
|
"learning_rate": 3.586706680582471e-05,
|
||
|
|
"loss": 0.5616,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17146259546279907,
|
||
|
|
"step": 875,
|
||
|
|
"valid_targets_mean": 3683.8,
|
||
|
|
"valid_targets_min": 1300
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4476138233680746,
|
||
|
|
"grad_norm": 0.2818001970674873,
|
||
|
|
"learning_rate": 3.579690508839738e-05,
|
||
|
|
"loss": 0.5933,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2810094952583313,
|
||
|
|
"step": 880,
|
||
|
|
"valid_targets_mean": 4964.7,
|
||
|
|
"valid_targets_min": 906
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.455842018650576,
|
||
|
|
"grad_norm": 0.26338390068846385,
|
||
|
|
"learning_rate": 3.572622268199292e-05,
|
||
|
|
"loss": 0.5654,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1930113285779953,
|
||
|
|
"step": 885,
|
||
|
|
"valid_targets_mean": 4663.9,
|
||
|
|
"valid_targets_min": 1014
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4640702139330775,
|
||
|
|
"grad_norm": 0.2852703233091329,
|
||
|
|
"learning_rate": 3.5655021916406295e-05,
|
||
|
|
"loss": 0.6099,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20813152194023132,
|
||
|
|
"step": 890,
|
||
|
|
"valid_targets_mean": 5404.8,
|
||
|
|
"valid_targets_min": 1229
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4722984092155786,
|
||
|
|
"grad_norm": 0.28634164602408574,
|
||
|
|
"learning_rate": 3.558330513851833e-05,
|
||
|
|
"loss": 0.5761,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18061956763267517,
|
||
|
|
"step": 895,
|
||
|
|
"valid_targets_mean": 4281.5,
|
||
|
|
"valid_targets_min": 1057
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.48052660449808,
|
||
|
|
"grad_norm": 0.3363155802523227,
|
||
|
|
"learning_rate": 3.55110747122184e-05,
|
||
|
|
"loss": 0.5784,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21961042284965515,
|
||
|
|
"step": 900,
|
||
|
|
"valid_targets_mean": 4570.4,
|
||
|
|
"valid_targets_min": 1376
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4887547997805815,
|
||
|
|
"grad_norm": 0.3152139088458827,
|
||
|
|
"learning_rate": 3.543833301832642e-05,
|
||
|
|
"loss": 0.591,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21911880373954773,
|
||
|
|
"step": 905,
|
||
|
|
"valid_targets_mean": 4505.6,
|
||
|
|
"valid_targets_min": 1222
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4969829950630829,
|
||
|
|
"grad_norm": 0.2725299656670677,
|
||
|
|
"learning_rate": 3.5365082454514493e-05,
|
||
|
|
"loss": 0.5635,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.155883327126503,
|
||
|
|
"step": 910,
|
||
|
|
"valid_targets_mean": 4379.0,
|
||
|
|
"valid_targets_min": 558
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5052111903455843,
|
||
|
|
"grad_norm": 0.3099063769029409,
|
||
|
|
"learning_rate": 3.529132543522777e-05,
|
||
|
|
"loss": 0.2335,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.062484048306941986,
|
||
|
|
"step": 915,
|
||
|
|
"valid_targets_mean": 5018.2,
|
||
|
|
"valid_targets_min": 2523
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5134393856280854,
|
||
|
|
"grad_norm": 0.24229893596824933,
|
||
|
|
"learning_rate": 3.521706439160494e-05,
|
||
|
|
"loss": 0.1825,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05739575996994972,
|
||
|
|
"step": 920,
|
||
|
|
"valid_targets_mean": 5262.3,
|
||
|
|
"valid_targets_min": 2967
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.521667580910587,
|
||
|
|
"grad_norm": 0.22759412709293747,
|
||
|
|
"learning_rate": 3.514230177139805e-05,
|
||
|
|
"loss": 0.1755,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05191062390804291,
|
||
|
|
"step": 925,
|
||
|
|
"valid_targets_mean": 5112.8,
|
||
|
|
"valid_targets_min": 3095
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5298957761930883,
|
||
|
|
"grad_norm": 0.2099880413854461,
|
||
|
|
"learning_rate": 3.5067040038891834e-05,
|
||
|
|
"loss": 0.174,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.054002054035663605,
|
||
|
|
"step": 930,
|
||
|
|
"valid_targets_mean": 5352.1,
|
||
|
|
"valid_targets_min": 2034
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5381239714755897,
|
||
|
|
"grad_norm": 0.211986257010731,
|
||
|
|
"learning_rate": 3.499128167482253e-05,
|
||
|
|
"loss": 0.1723,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.059163033962249756,
|
||
|
|
"step": 935,
|
||
|
|
"valid_targets_mean": 5610.1,
|
||
|
|
"valid_targets_min": 3069
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.546352166758091,
|
||
|
|
"grad_norm": 0.20019735272394734,
|
||
|
|
"learning_rate": 3.491502917629602e-05,
|
||
|
|
"loss": 0.1712,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05470460653305054,
|
||
|
|
"step": 940,
|
||
|
|
"valid_targets_mean": 5206.3,
|
||
|
|
"valid_targets_min": 3294
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5545803620405925,
|
||
|
|
"grad_norm": 0.22719278970124465,
|
||
|
|
"learning_rate": 3.483828505670563e-05,
|
||
|
|
"loss": 0.1687,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05662545561790466,
|
||
|
|
"step": 945,
|
||
|
|
"valid_targets_mean": 5573.9,
|
||
|
|
"valid_targets_min": 1997
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5628085573230939,
|
||
|
|
"grad_norm": 0.20264846243090592,
|
||
|
|
"learning_rate": 3.476105184564921e-05,
|
||
|
|
"loss": 0.1731,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05749613046646118,
|
||
|
|
"step": 950,
|
||
|
|
"valid_targets_mean": 5469.3,
|
||
|
|
"valid_targets_min": 3506
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.571036752605595,
|
||
|
|
"grad_norm": 0.19906120158231985,
|
||
|
|
"learning_rate": 3.468333208884576e-05,
|
||
|
|
"loss": 0.1673,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.056417644023895264,
|
||
|
|
"step": 955,
|
||
|
|
"valid_targets_mean": 5436.2,
|
||
|
|
"valid_targets_min": 2886
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5792649478880967,
|
||
|
|
"grad_norm": 0.21488228662363817,
|
||
|
|
"learning_rate": 3.4605128348051566e-05,
|
||
|
|
"loss": 0.165,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06093878298997879,
|
||
|
|
"step": 960,
|
||
|
|
"valid_targets_mean": 5540.5,
|
||
|
|
"valid_targets_min": 3229
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5874931431705979,
|
||
|
|
"grad_norm": 0.20141714461992172,
|
||
|
|
"learning_rate": 3.4526443200975704e-05,
|
||
|
|
"loss": 0.1674,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.057458825409412384,
|
||
|
|
"step": 965,
|
||
|
|
"valid_targets_mean": 5831.1,
|
||
|
|
"valid_targets_min": 2745
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5957213384530993,
|
||
|
|
"grad_norm": 0.19477058216940663,
|
||
|
|
"learning_rate": 3.444727924119511e-05,
|
||
|
|
"loss": 0.1654,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05204876884818077,
|
||
|
|
"step": 970,
|
||
|
|
"valid_targets_mean": 5347.5,
|
||
|
|
"valid_targets_min": 2417
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6039495337356007,
|
||
|
|
"grad_norm": 0.22865468044098994,
|
||
|
|
"learning_rate": 3.436763907806911e-05,
|
||
|
|
"loss": 0.1664,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.053488366305828094,
|
||
|
|
"step": 975,
|
||
|
|
"valid_targets_mean": 5258.5,
|
||
|
|
"valid_targets_min": 720
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.612177729018102,
|
||
|
|
"grad_norm": 0.22074279987171683,
|
||
|
|
"learning_rate": 3.4287525336653335e-05,
|
||
|
|
"loss": 0.1626,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.052560918033123016,
|
||
|
|
"step": 980,
|
||
|
|
"valid_targets_mean": 5099.8,
|
||
|
|
"valid_targets_min": 1583
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6204059243006035,
|
||
|
|
"grad_norm": 0.17850347287453902,
|
||
|
|
"learning_rate": 3.420694065761328e-05,
|
||
|
|
"loss": 0.1632,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05743696168065071,
|
||
|
|
"step": 985,
|
||
|
|
"valid_targets_mean": 5578.4,
|
||
|
|
"valid_targets_min": 2798
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6286341195831047,
|
||
|
|
"grad_norm": 0.19214912575505724,
|
||
|
|
"learning_rate": 3.412588769713723e-05,
|
||
|
|
"loss": 0.1675,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.050386179238557816,
|
||
|
|
"step": 990,
|
||
|
|
"valid_targets_mean": 5036.6,
|
||
|
|
"valid_targets_min": 2310
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6368623148656063,
|
||
|
|
"grad_norm": 0.19389336692541487,
|
||
|
|
"learning_rate": 3.40443691268487e-05,
|
||
|
|
"loss": 0.1641,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.053170040249824524,
|
||
|
|
"step": 995,
|
||
|
|
"valid_targets_mean": 5270.8,
|
||
|
|
"valid_targets_min": 2978
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6450905101481075,
|
||
|
|
"grad_norm": 0.21103448932645819,
|
||
|
|
"learning_rate": 3.396238763371837e-05,
|
||
|
|
"loss": 0.1656,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05385826528072357,
|
||
|
|
"step": 1000,
|
||
|
|
"valid_targets_mean": 5462.3,
|
||
|
|
"valid_targets_min": 2571
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6533187054306089,
|
||
|
|
"grad_norm": 0.20655305153916342,
|
||
|
|
"learning_rate": 3.387994591997554e-05,
|
||
|
|
"loss": 0.1658,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.050711266696453094,
|
||
|
|
"step": 1005,
|
||
|
|
"valid_targets_mean": 5422.2,
|
||
|
|
"valid_targets_min": 2689
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6615469007131103,
|
||
|
|
"grad_norm": 0.20405599334414792,
|
||
|
|
"learning_rate": 3.379704670301906e-05,
|
||
|
|
"loss": 0.1614,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05435357242822647,
|
||
|
|
"step": 1010,
|
||
|
|
"valid_targets_mean": 5195.3,
|
||
|
|
"valid_targets_min": 3110
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6697750959956115,
|
||
|
|
"grad_norm": 0.8352584687550971,
|
||
|
|
"learning_rate": 3.371369271532775e-05,
|
||
|
|
"loss": 0.2498,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19201651215553284,
|
||
|
|
"step": 1015,
|
||
|
|
"valid_targets_mean": 3959.7,
|
||
|
|
"valid_targets_min": 1023
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.678003291278113,
|
||
|
|
"grad_norm": 0.5025769864208273,
|
||
|
|
"learning_rate": 3.362988670437031e-05,
|
||
|
|
"loss": 0.4966,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14757052063941956,
|
||
|
|
"step": 1020,
|
||
|
|
"valid_targets_mean": 3057.9,
|
||
|
|
"valid_targets_min": 1066
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6862314865606143,
|
||
|
|
"grad_norm": 0.3026403609768399,
|
||
|
|
"learning_rate": 3.354563143251483e-05,
|
||
|
|
"loss": 0.4843,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1374441534280777,
|
||
|
|
"step": 1025,
|
||
|
|
"valid_targets_mean": 4226.2,
|
||
|
|
"valid_targets_min": 902
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.694459681843116,
|
||
|
|
"grad_norm": 0.3072152435136929,
|
||
|
|
"learning_rate": 3.346092967693764e-05,
|
||
|
|
"loss": 0.4969,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14073555171489716,
|
||
|
|
"step": 1030,
|
||
|
|
"valid_targets_mean": 3537.5,
|
||
|
|
"valid_targets_min": 1090
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.702687877125617,
|
||
|
|
"grad_norm": 0.31412887585449273,
|
||
|
|
"learning_rate": 3.3375784229531864e-05,
|
||
|
|
"loss": 0.4868,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13491514325141907,
|
||
|
|
"step": 1035,
|
||
|
|
"valid_targets_mean": 3058.2,
|
||
|
|
"valid_targets_min": 771
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7109160724081185,
|
||
|
|
"grad_norm": 0.36543147309442925,
|
||
|
|
"learning_rate": 3.3290197896815344e-05,
|
||
|
|
"loss": 0.4708,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15971092879772186,
|
||
|
|
"step": 1040,
|
||
|
|
"valid_targets_mean": 1120.0,
|
||
|
|
"valid_targets_min": 521
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.71914426769062,
|
||
|
|
"grad_norm": 0.26724180927424696,
|
||
|
|
"learning_rate": 3.320417349983813e-05,
|
||
|
|
"loss": 0.4831,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16630011796951294,
|
||
|
|
"step": 1045,
|
||
|
|
"valid_targets_mean": 4731.8,
|
||
|
|
"valid_targets_min": 1389
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.727372462973121,
|
||
|
|
"grad_norm": 0.24730868974709785,
|
||
|
|
"learning_rate": 3.3117713874089516e-05,
|
||
|
|
"loss": 0.455,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1551167070865631,
|
||
|
|
"step": 1050,
|
||
|
|
"valid_targets_mean": 4816.6,
|
||
|
|
"valid_targets_min": 1341
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7356006582556227,
|
||
|
|
"grad_norm": 0.28605184203121603,
|
||
|
|
"learning_rate": 3.303082186940458e-05,
|
||
|
|
"loss": 0.4822,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14953917264938354,
|
||
|
|
"step": 1055,
|
||
|
|
"valid_targets_mean": 3374.7,
|
||
|
|
"valid_targets_min": 1239
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.743828853538124,
|
||
|
|
"grad_norm": 0.2673582874164761,
|
||
|
|
"learning_rate": 3.294350034987022e-05,
|
||
|
|
"loss": 0.4692,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16546636819839478,
|
||
|
|
"step": 1060,
|
||
|
|
"valid_targets_mean": 3214.4,
|
||
|
|
"valid_targets_min": 621
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7520570488206253,
|
||
|
|
"grad_norm": 0.2633944966457584,
|
||
|
|
"learning_rate": 3.285575219373079e-05,
|
||
|
|
"loss": 0.4608,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17115873098373413,
|
||
|
|
"step": 1065,
|
||
|
|
"valid_targets_mean": 5321.0,
|
||
|
|
"valid_targets_min": 1407
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7602852441031267,
|
||
|
|
"grad_norm": 0.28138662960460287,
|
||
|
|
"learning_rate": 3.276758029329318e-05,
|
||
|
|
"loss": 0.482,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1477651447057724,
|
||
|
|
"step": 1070,
|
||
|
|
"valid_targets_mean": 3536.7,
|
||
|
|
"valid_targets_min": 915
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.768513439385628,
|
||
|
|
"grad_norm": 0.30575007256496134,
|
||
|
|
"learning_rate": 3.267898755483153e-05,
|
||
|
|
"loss": 0.4575,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12381979823112488,
|
||
|
|
"step": 1075,
|
||
|
|
"valid_targets_mean": 2772.1,
|
||
|
|
"valid_targets_min": 1106
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7767416346681295,
|
||
|
|
"grad_norm": 0.4606810569188011,
|
||
|
|
"learning_rate": 3.258997689849142e-05,
|
||
|
|
"loss": 0.4827,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.29282310605049133,
|
||
|
|
"step": 1080,
|
||
|
|
"valid_targets_mean": 3487.8,
|
||
|
|
"valid_targets_min": 523
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7849698299506307,
|
||
|
|
"grad_norm": 0.27883070137062055,
|
||
|
|
"learning_rate": 3.250055125819358e-05,
|
||
|
|
"loss": 0.4813,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14349916577339172,
|
||
|
|
"step": 1085,
|
||
|
|
"valid_targets_mean": 4307.9,
|
||
|
|
"valid_targets_min": 1207
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7931980252331323,
|
||
|
|
"grad_norm": 0.2848599685713568,
|
||
|
|
"learning_rate": 3.241071358153723e-05,
|
||
|
|
"loss": 0.4659,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17098680138587952,
|
||
|
|
"step": 1090,
|
||
|
|
"valid_targets_mean": 4449.8,
|
||
|
|
"valid_targets_min": 1036
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8014262205156335,
|
||
|
|
"grad_norm": 0.30529896453924493,
|
||
|
|
"learning_rate": 3.232046682970293e-05,
|
||
|
|
"loss": 0.4793,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13532951474189758,
|
||
|
|
"step": 1095,
|
||
|
|
"valid_targets_mean": 3510.9,
|
||
|
|
"valid_targets_min": 1014
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.809654415798135,
|
||
|
|
"grad_norm": 0.28169285171294794,
|
||
|
|
"learning_rate": 3.2229813977354926e-05,
|
||
|
|
"loss": 0.4687,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16980434954166412,
|
||
|
|
"step": 1100,
|
||
|
|
"valid_targets_mean": 4310.5,
|
||
|
|
"valid_targets_min": 1193
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8178826110806363,
|
||
|
|
"grad_norm": 0.26956538256356954,
|
||
|
|
"learning_rate": 3.213875801254314e-05,
|
||
|
|
"loss": 0.4568,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13664710521697998,
|
||
|
|
"step": 1105,
|
||
|
|
"valid_targets_mean": 4566.1,
|
||
|
|
"valid_targets_min": 1394
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8261108063631377,
|
||
|
|
"grad_norm": 0.32339359248442834,
|
||
|
|
"learning_rate": 3.204730193660466e-05,
|
||
|
|
"loss": 0.4743,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1973041146993637,
|
||
|
|
"step": 1110,
|
||
|
|
"valid_targets_mean": 3325.2,
|
||
|
|
"valid_targets_min": 883
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8343390016456391,
|
||
|
|
"grad_norm": 0.3143153411480853,
|
||
|
|
"learning_rate": 3.195544876406482e-05,
|
||
|
|
"loss": 0.4771,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16181105375289917,
|
||
|
|
"step": 1115,
|
||
|
|
"valid_targets_mean": 3990.3,
|
||
|
|
"valid_targets_min": 1374
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8425671969281403,
|
||
|
|
"grad_norm": 0.4224508251606666,
|
||
|
|
"learning_rate": 3.1863201522537843e-05,
|
||
|
|
"loss": 0.4355,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.143344908952713,
|
||
|
|
"step": 1120,
|
||
|
|
"valid_targets_mean": 5897.4,
|
||
|
|
"valid_targets_min": 347
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.850795392210642,
|
||
|
|
"grad_norm": 0.278768311774877,
|
||
|
|
"learning_rate": 3.177056325262704e-05,
|
||
|
|
"loss": 0.3117,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09472917020320892,
|
||
|
|
"step": 1125,
|
||
|
|
"valid_targets_mean": 6455.6,
|
||
|
|
"valid_targets_min": 1359
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8590235874931431,
|
||
|
|
"grad_norm": 0.22708249957012502,
|
||
|
|
"learning_rate": 3.167753700782457e-05,
|
||
|
|
"loss": 0.281,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.097655288875103,
|
||
|
|
"step": 1130,
|
||
|
|
"valid_targets_mean": 6676.2,
|
||
|
|
"valid_targets_min": 1143
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8672517827756445,
|
||
|
|
"grad_norm": 0.22955456223919352,
|
||
|
|
"learning_rate": 3.1584125854410824e-05,
|
||
|
|
"loss": 0.2943,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11995258927345276,
|
||
|
|
"step": 1135,
|
||
|
|
"valid_targets_mean": 6722.4,
|
||
|
|
"valid_targets_min": 505
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.875479978058146,
|
||
|
|
"grad_norm": 0.18939517331373842,
|
||
|
|
"learning_rate": 3.149033287135335e-05,
|
||
|
|
"loss": 0.2902,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10605036467313766,
|
||
|
|
"step": 1140,
|
||
|
|
"valid_targets_mean": 7329.8,
|
||
|
|
"valid_targets_min": 1300
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8837081733406473,
|
||
|
|
"grad_norm": 0.21555429419147118,
|
||
|
|
"learning_rate": 3.1396161150205324e-05,
|
||
|
|
"loss": 0.3081,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07604040205478668,
|
||
|
|
"step": 1145,
|
||
|
|
"valid_targets_mean": 6100.7,
|
||
|
|
"valid_targets_min": 1328
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8919363686231487,
|
||
|
|
"grad_norm": 0.18721522620193756,
|
||
|
|
"learning_rate": 3.130161379500371e-05,
|
||
|
|
"loss": 0.2738,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06492835283279419,
|
||
|
|
"step": 1150,
|
||
|
|
"valid_targets_mean": 5562.0,
|
||
|
|
"valid_targets_min": 246
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.90016456390565,
|
||
|
|
"grad_norm": 0.6559450509008863,
|
||
|
|
"learning_rate": 3.120669392216692e-05,
|
||
|
|
"loss": 0.2918,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06112710386514664,
|
||
|
|
"step": 1155,
|
||
|
|
"valid_targets_mean": 1929.8,
|
||
|
|
"valid_targets_min": 503
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9083927591881515,
|
||
|
|
"grad_norm": 0.19521939318451423,
|
||
|
|
"learning_rate": 3.111140466039205e-05,
|
||
|
|
"loss": 0.2809,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0632120668888092,
|
||
|
|
"step": 1160,
|
||
|
|
"valid_targets_mean": 5253.4,
|
||
|
|
"valid_targets_min": 595
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9166209544706527,
|
||
|
|
"grad_norm": 0.19764156857068718,
|
||
|
|
"learning_rate": 3.1015749150551835e-05,
|
||
|
|
"loss": 0.274,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07447328418493271,
|
||
|
|
"step": 1165,
|
||
|
|
"valid_targets_mean": 5247.3,
|
||
|
|
"valid_targets_min": 257
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9248491497531541,
|
||
|
|
"grad_norm": 0.19862374493029875,
|
||
|
|
"learning_rate": 3.091973054559106e-05,
|
||
|
|
"loss": 0.2975,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07149604707956314,
|
||
|
|
"step": 1170,
|
||
|
|
"valid_targets_mean": 5607.3,
|
||
|
|
"valid_targets_min": 452
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9330773450356555,
|
||
|
|
"grad_norm": 0.2011133107853708,
|
||
|
|
"learning_rate": 3.082335201042266e-05,
|
||
|
|
"loss": 0.291,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10824010521173477,
|
||
|
|
"step": 1175,
|
||
|
|
"valid_targets_mean": 7863.1,
|
||
|
|
"valid_targets_min": 1340
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9413055403181567,
|
||
|
|
"grad_norm": 0.23295729957170552,
|
||
|
|
"learning_rate": 3.0726616721823394e-05,
|
||
|
|
"loss": 0.3024,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15341416001319885,
|
||
|
|
"step": 1180,
|
||
|
|
"valid_targets_mean": 7620.9,
|
||
|
|
"valid_targets_min": 1084
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9495337356006583,
|
||
|
|
"grad_norm": 0.19668525599135733,
|
||
|
|
"learning_rate": 3.062952786832912e-05,
|
||
|
|
"loss": 0.2652,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08159739524126053,
|
||
|
|
"step": 1185,
|
||
|
|
"valid_targets_mean": 5906.8,
|
||
|
|
"valid_targets_min": 1288
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9577619308831595,
|
||
|
|
"grad_norm": 0.21883749091726576,
|
||
|
|
"learning_rate": 3.053208865012973e-05,
|
||
|
|
"loss": 0.2714,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0845392718911171,
|
||
|
|
"step": 1190,
|
||
|
|
"valid_targets_mean": 5153.4,
|
||
|
|
"valid_targets_min": 221
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9659901261656612,
|
||
|
|
"grad_norm": 0.21677746139321524,
|
||
|
|
"learning_rate": 3.0434302278963623e-05,
|
||
|
|
"loss": 0.2811,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10049721598625183,
|
||
|
|
"step": 1195,
|
||
|
|
"valid_targets_mean": 7219.4,
|
||
|
|
"valid_targets_min": 219
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9742183214481623,
|
||
|
|
"grad_norm": 0.1734827884546814,
|
||
|
|
"learning_rate": 3.0336171978011885e-05,
|
||
|
|
"loss": 0.2645,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07478731125593185,
|
||
|
|
"step": 1200,
|
||
|
|
"valid_targets_mean": 5985.7,
|
||
|
|
"valid_targets_min": 292
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9824465167306637,
|
||
|
|
"grad_norm": 0.19391076236488966,
|
||
|
|
"learning_rate": 3.0237700981792023e-05,
|
||
|
|
"loss": 0.2844,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06688568741083145,
|
||
|
|
"step": 1205,
|
||
|
|
"valid_targets_mean": 5758.3,
|
||
|
|
"valid_targets_min": 310
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9906747120131651,
|
||
|
|
"grad_norm": 0.1785729712568323,
|
||
|
|
"learning_rate": 3.013889253605135e-05,
|
||
|
|
"loss": 0.269,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05671679973602295,
|
||
|
|
"step": 1210,
|
||
|
|
"valid_targets_mean": 5599.4,
|
||
|
|
"valid_targets_min": 982
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9989029072956663,
|
||
|
|
"grad_norm": 0.20992664106258474,
|
||
|
|
"learning_rate": 3.0039749897660005e-05,
|
||
|
|
"loss": 0.2789,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05317822843790054,
|
||
|
|
"step": 1215,
|
||
|
|
"valid_targets_mean": 2148.5,
|
||
|
|
"valid_targets_min": 457
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.006582556226001,
|
||
|
|
"grad_norm": 0.9638756157661459,
|
||
|
|
"learning_rate": 2.9940276334503617e-05,
|
||
|
|
"loss": 0.6179,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1680884063243866,
|
||
|
|
"step": 1220,
|
||
|
|
"valid_targets_mean": 6954.8,
|
||
|
|
"valid_targets_min": 1761
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0148107515085023,
|
||
|
|
"grad_norm": 0.6978276908779179,
|
||
|
|
"learning_rate": 2.984047512537557e-05,
|
||
|
|
"loss": 0.4752,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17175118625164032,
|
||
|
|
"step": 1225,
|
||
|
|
"valid_targets_mean": 8220.0,
|
||
|
|
"valid_targets_min": 2842
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.023038946791004,
|
||
|
|
"grad_norm": 0.4316421614618443,
|
||
|
|
"learning_rate": 2.9740349559868918e-05,
|
||
|
|
"loss": 0.4364,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13174858689308167,
|
||
|
|
"step": 1230,
|
||
|
|
"valid_targets_mean": 7166.5,
|
||
|
|
"valid_targets_min": 2501
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.031267142073505,
|
||
|
|
"grad_norm": 0.3683054870009327,
|
||
|
|
"learning_rate": 2.9639902938267994e-05,
|
||
|
|
"loss": 0.4043,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13023295998573303,
|
||
|
|
"step": 1235,
|
||
|
|
"valid_targets_mean": 6987.4,
|
||
|
|
"valid_targets_min": 1709
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0394953373560067,
|
||
|
|
"grad_norm": 0.27694022329011136,
|
||
|
|
"learning_rate": 2.9539138571439614e-05,
|
||
|
|
"loss": 0.3873,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1342620551586151,
|
||
|
|
"step": 1240,
|
||
|
|
"valid_targets_mean": 7609.8,
|
||
|
|
"valid_targets_min": 2002
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.047723532638508,
|
||
|
|
"grad_norm": 0.2579222382415089,
|
||
|
|
"learning_rate": 2.943805978072391e-05,
|
||
|
|
"loss": 0.3884,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12417769432067871,
|
||
|
|
"step": 1245,
|
||
|
|
"valid_targets_mean": 7067.3,
|
||
|
|
"valid_targets_min": 2449
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.055951727921009,
|
||
|
|
"grad_norm": 0.25837969440910324,
|
||
|
|
"learning_rate": 2.933666989782491e-05,
|
||
|
|
"loss": 0.3801,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11597777903079987,
|
||
|
|
"step": 1250,
|
||
|
|
"valid_targets_mean": 7091.4,
|
||
|
|
"valid_targets_min": 2861
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0641799232035107,
|
||
|
|
"grad_norm": 0.2491458862466655,
|
||
|
|
"learning_rate": 2.9234972264700687e-05,
|
||
|
|
"loss": 0.3696,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1361021101474762,
|
||
|
|
"step": 1255,
|
||
|
|
"valid_targets_mean": 7742.7,
|
||
|
|
"valid_targets_min": 341
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.072408118486012,
|
||
|
|
"grad_norm": 0.21808961060612198,
|
||
|
|
"learning_rate": 2.913297023345319e-05,
|
||
|
|
"loss": 0.3725,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14099043607711792,
|
||
|
|
"step": 1260,
|
||
|
|
"valid_targets_mean": 8509.0,
|
||
|
|
"valid_targets_min": 3043
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0806363137685135,
|
||
|
|
"grad_norm": 0.22565312170460802,
|
||
|
|
"learning_rate": 2.903066716621779e-05,
|
||
|
|
"loss": 0.3689,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11987649649381638,
|
||
|
|
"step": 1265,
|
||
|
|
"valid_targets_mean": 6931.7,
|
||
|
|
"valid_targets_min": 2360
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0888645090510147,
|
||
|
|
"grad_norm": 0.20315983397248213,
|
||
|
|
"learning_rate": 2.892806643505245e-05,
|
||
|
|
"loss": 0.3582,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1136256754398346,
|
||
|
|
"step": 1270,
|
||
|
|
"valid_targets_mean": 9712.0,
|
||
|
|
"valid_targets_min": 2685
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0970927043335164,
|
||
|
|
"grad_norm": 0.20097162887853337,
|
||
|
|
"learning_rate": 2.8825171421826555e-05,
|
||
|
|
"loss": 0.3366,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11705143749713898,
|
||
|
|
"step": 1275,
|
||
|
|
"valid_targets_mean": 9876.8,
|
||
|
|
"valid_targets_min": 387
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1053208996160175,
|
||
|
|
"grad_norm": 0.20588709003765224,
|
||
|
|
"learning_rate": 2.8721985518109457e-05,
|
||
|
|
"loss": 0.3355,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10917450487613678,
|
||
|
|
"step": 1280,
|
||
|
|
"valid_targets_mean": 9530.9,
|
||
|
|
"valid_targets_min": 5556
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1135490948985187,
|
||
|
|
"grad_norm": 0.2078925688921669,
|
||
|
|
"learning_rate": 2.861851212505869e-05,
|
||
|
|
"loss": 0.325,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11689627170562744,
|
||
|
|
"step": 1285,
|
||
|
|
"valid_targets_mean": 9594.7,
|
||
|
|
"valid_targets_min": 4603
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1217772901810203,
|
||
|
|
"grad_norm": 0.20054485627408972,
|
||
|
|
"learning_rate": 2.8514754653307836e-05,
|
||
|
|
"loss": 0.3343,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11187094449996948,
|
||
|
|
"step": 1290,
|
||
|
|
"valid_targets_mean": 9466.0,
|
||
|
|
"valid_targets_min": 6389
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1300054854635215,
|
||
|
|
"grad_norm": 0.22534885286504078,
|
||
|
|
"learning_rate": 2.8410716522854152e-05,
|
||
|
|
"loss": 0.3245,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10926699638366699,
|
||
|
|
"step": 1295,
|
||
|
|
"valid_targets_mean": 9956.4,
|
||
|
|
"valid_targets_min": 4744
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.138233680746023,
|
||
|
|
"grad_norm": 0.22280970786481757,
|
||
|
|
"learning_rate": 2.8306401162945795e-05,
|
||
|
|
"loss": 0.3325,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11518752574920654,
|
||
|
|
"step": 1300,
|
||
|
|
"valid_targets_mean": 9960.5,
|
||
|
|
"valid_targets_min": 6172
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1464618760285243,
|
||
|
|
"grad_norm": 0.21839411779582002,
|
||
|
|
"learning_rate": 2.8201812011968807e-05,
|
||
|
|
"loss": 0.3282,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10992267727851868,
|
||
|
|
"step": 1305,
|
||
|
|
"valid_targets_mean": 9848.8,
|
||
|
|
"valid_targets_min": 751
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.154690071311026,
|
||
|
|
"grad_norm": 0.21794272450141897,
|
||
|
|
"learning_rate": 2.809695251733379e-05,
|
||
|
|
"loss": 0.3265,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1086210310459137,
|
||
|
|
"step": 1310,
|
||
|
|
"valid_targets_mean": 9407.0,
|
||
|
|
"valid_targets_min": 2003
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.162918266593527,
|
||
|
|
"grad_norm": 0.2210245319595092,
|
||
|
|
"learning_rate": 2.799182613536226e-05,
|
||
|
|
"loss": 0.3264,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10626842081546783,
|
||
|
|
"step": 1315,
|
||
|
|
"valid_targets_mean": 9471.5,
|
||
|
|
"valid_targets_min": 4297
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1711464618760283,
|
||
|
|
"grad_norm": 0.22990804540391607,
|
||
|
|
"learning_rate": 2.7886436331172745e-05,
|
||
|
|
"loss": 0.3257,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10737085342407227,
|
||
|
|
"step": 1320,
|
||
|
|
"valid_targets_mean": 8577.6,
|
||
|
|
"valid_targets_min": 3933
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.17937465715853,
|
||
|
|
"grad_norm": 0.205922355272621,
|
||
|
|
"learning_rate": 2.7780786578566524e-05,
|
||
|
|
"loss": 0.3213,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1152525395154953,
|
||
|
|
"step": 1325,
|
||
|
|
"valid_targets_mean": 9960.7,
|
||
|
|
"valid_targets_min": 1490
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.187602852441031,
|
||
|
|
"grad_norm": 0.1876585353072035,
|
||
|
|
"learning_rate": 2.7674880359913183e-05,
|
||
|
|
"loss": 0.3235,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10439243167638779,
|
||
|
|
"step": 1330,
|
||
|
|
"valid_targets_mean": 9948.3,
|
||
|
|
"valid_targets_min": 3344
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1958310477235328,
|
||
|
|
"grad_norm": 0.20859391800680568,
|
||
|
|
"learning_rate": 2.7568721166035778e-05,
|
||
|
|
"loss": 0.3254,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09820768982172012,
|
||
|
|
"step": 1335,
|
||
|
|
"valid_targets_mean": 9342.2,
|
||
|
|
"valid_targets_min": 3715
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.204059243006034,
|
||
|
|
"grad_norm": 0.22628943045344366,
|
||
|
|
"learning_rate": 2.7462312496095805e-05,
|
||
|
|
"loss": 0.3337,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11237113922834396,
|
||
|
|
"step": 1340,
|
||
|
|
"valid_targets_mean": 9508.4,
|
||
|
|
"valid_targets_min": 434
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2122874382885356,
|
||
|
|
"grad_norm": 0.20135082037861282,
|
||
|
|
"learning_rate": 2.735565785747787e-05,
|
||
|
|
"loss": 0.3271,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10496258735656738,
|
||
|
|
"step": 1345,
|
||
|
|
"valid_targets_mean": 9982.4,
|
||
|
|
"valid_targets_min": 3744
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2205156335710368,
|
||
|
|
"grad_norm": 0.22074725520877453,
|
||
|
|
"learning_rate": 2.7248760765674033e-05,
|
||
|
|
"loss": 0.3156,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10917022824287415,
|
||
|
|
"step": 1350,
|
||
|
|
"valid_targets_mean": 9832.3,
|
||
|
|
"valid_targets_min": 5100
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.228743828853538,
|
||
|
|
"grad_norm": 0.2212465257169665,
|
||
|
|
"learning_rate": 2.7141624744168e-05,
|
||
|
|
"loss": 0.3177,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1113644689321518,
|
||
|
|
"step": 1355,
|
||
|
|
"valid_targets_mean": 10835.8,
|
||
|
|
"valid_targets_min": 2829
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2369720241360396,
|
||
|
|
"grad_norm": 0.19921061855711575,
|
||
|
|
"learning_rate": 2.703425332431891e-05,
|
||
|
|
"loss": 0.3222,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10436619818210602,
|
||
|
|
"step": 1360,
|
||
|
|
"valid_targets_mean": 10256.2,
|
||
|
|
"valid_targets_min": 5379
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2452002194185408,
|
||
|
|
"grad_norm": 0.23316937250636952,
|
||
|
|
"learning_rate": 2.6926650045245014e-05,
|
||
|
|
"loss": 0.3216,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09406248480081558,
|
||
|
|
"step": 1365,
|
||
|
|
"valid_targets_mean": 9251.9,
|
||
|
|
"valid_targets_min": 648
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2534284147010424,
|
||
|
|
"grad_norm": 0.1967403350878879,
|
||
|
|
"learning_rate": 2.6818818453706944e-05,
|
||
|
|
"loss": 0.3191,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11360487341880798,
|
||
|
|
"step": 1370,
|
||
|
|
"valid_targets_mean": 11007.1,
|
||
|
|
"valid_targets_min": 1456
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2616566099835436,
|
||
|
|
"grad_norm": 0.22084889534546884,
|
||
|
|
"learning_rate": 2.6710762103990856e-05,
|
||
|
|
"loss": 0.323,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10803453624248505,
|
||
|
|
"step": 1375,
|
||
|
|
"valid_targets_mean": 9657.8,
|
||
|
|
"valid_targets_min": 2733
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.269884805266045,
|
||
|
|
"grad_norm": 0.20093362465257963,
|
||
|
|
"learning_rate": 2.660248455779128e-05,
|
||
|
|
"loss": 0.318,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12841132283210754,
|
||
|
|
"step": 1380,
|
||
|
|
"valid_targets_mean": 10753.2,
|
||
|
|
"valid_targets_min": 5581
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2781130005485464,
|
||
|
|
"grad_norm": 0.20269530317142032,
|
||
|
|
"learning_rate": 2.6493989384093674e-05,
|
||
|
|
"loss": 0.321,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10728751868009567,
|
||
|
|
"step": 1385,
|
||
|
|
"valid_targets_mean": 9586.8,
|
||
|
|
"valid_targets_min": 3841
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2863411958310476,
|
||
|
|
"grad_norm": 0.2946661577910316,
|
||
|
|
"learning_rate": 2.6385280159056838e-05,
|
||
|
|
"loss": 0.3144,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09770511090755463,
|
||
|
|
"step": 1390,
|
||
|
|
"valid_targets_mean": 9004.5,
|
||
|
|
"valid_targets_min": 434
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.294569391113549,
|
||
|
|
"grad_norm": 0.23257347797723024,
|
||
|
|
"learning_rate": 2.6276360465895004e-05,
|
||
|
|
"loss": 0.3172,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09528092294931412,
|
||
|
|
"step": 1395,
|
||
|
|
"valid_targets_mean": 9294.9,
|
||
|
|
"valid_targets_min": 4130
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3027975863960504,
|
||
|
|
"grad_norm": 0.22704053553458559,
|
||
|
|
"learning_rate": 2.6167233894759743e-05,
|
||
|
|
"loss": 0.3109,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10657741874456406,
|
||
|
|
"step": 1400,
|
||
|
|
"valid_targets_mean": 10179.0,
|
||
|
|
"valid_targets_min": 4267
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.311025781678552,
|
||
|
|
"grad_norm": 0.21046171710437353,
|
||
|
|
"learning_rate": 2.6057904042621625e-05,
|
||
|
|
"loss": 0.3157,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09410691261291504,
|
||
|
|
"step": 1405,
|
||
|
|
"valid_targets_mean": 8532.7,
|
||
|
|
"valid_targets_min": 417
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.319253976961053,
|
||
|
|
"grad_norm": 0.24250405396092567,
|
||
|
|
"learning_rate": 2.5948374513151668e-05,
|
||
|
|
"loss": 0.3236,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12210428714752197,
|
||
|
|
"step": 1410,
|
||
|
|
"valid_targets_mean": 6747.5,
|
||
|
|
"valid_targets_min": 401
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3274821722435544,
|
||
|
|
"grad_norm": 0.2738588406052067,
|
||
|
|
"learning_rate": 2.583864891660252e-05,
|
||
|
|
"loss": 0.3357,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10755414515733719,
|
||
|
|
"step": 1415,
|
||
|
|
"valid_targets_mean": 5423.5,
|
||
|
|
"valid_targets_min": 369
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.335710367526056,
|
||
|
|
"grad_norm": 2.4982151917439483,
|
||
|
|
"learning_rate": 2.5728730869689505e-05,
|
||
|
|
"loss": 0.4375,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2848479747772217,
|
||
|
|
"step": 1420,
|
||
|
|
"valid_targets_mean": 3791.8,
|
||
|
|
"valid_targets_min": 1231
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.343938562808557,
|
||
|
|
"grad_norm": 0.6190427970922461,
|
||
|
|
"learning_rate": 2.5618623995471394e-05,
|
||
|
|
"loss": 0.6956,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20855411887168884,
|
||
|
|
"step": 1425,
|
||
|
|
"valid_targets_mean": 4159.7,
|
||
|
|
"valid_targets_min": 1267
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.352166758091059,
|
||
|
|
"grad_norm": 0.5754941700932448,
|
||
|
|
"learning_rate": 2.5508331923230963e-05,
|
||
|
|
"loss": 0.6226,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1878442019224167,
|
||
|
|
"step": 1430,
|
||
|
|
"valid_targets_mean": 3654.0,
|
||
|
|
"valid_targets_min": 1421
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.36039495337356,
|
||
|
|
"grad_norm": 0.3930300076882507,
|
||
|
|
"learning_rate": 2.5397858288355397e-05,
|
||
|
|
"loss": 0.6106,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17351096868515015,
|
||
|
|
"step": 1435,
|
||
|
|
"valid_targets_mean": 2945.1,
|
||
|
|
"valid_targets_min": 996
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3686231486560616,
|
||
|
|
"grad_norm": 0.32216825686035716,
|
||
|
|
"learning_rate": 2.5287206732216453e-05,
|
||
|
|
"loss": 0.5819,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20648203790187836,
|
||
|
|
"step": 1440,
|
||
|
|
"valid_targets_mean": 4179.1,
|
||
|
|
"valid_targets_min": 1183
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.376851343938563,
|
||
|
|
"grad_norm": 0.2956785702719909,
|
||
|
|
"learning_rate": 2.5176380902050418e-05,
|
||
|
|
"loss": 0.5457,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17796429991722107,
|
||
|
|
"step": 1445,
|
||
|
|
"valid_targets_mean": 4048.5,
|
||
|
|
"valid_targets_min": 1297
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3850795392210644,
|
||
|
|
"grad_norm": 0.28597121366642053,
|
||
|
|
"learning_rate": 2.5065384450837916e-05,
|
||
|
|
"loss": 0.5422,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17518360912799835,
|
||
|
|
"step": 1450,
|
||
|
|
"valid_targets_mean": 4272.8,
|
||
|
|
"valid_targets_min": 1187
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3933077345035656,
|
||
|
|
"grad_norm": 0.2921318011955971,
|
||
|
|
"learning_rate": 2.495422103718349e-05,
|
||
|
|
"loss": 0.5542,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2636011242866516,
|
||
|
|
"step": 1455,
|
||
|
|
"valid_targets_mean": 4842.6,
|
||
|
|
"valid_targets_min": 1298
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.401535929786067,
|
||
|
|
"grad_norm": 0.28775359794170824,
|
||
|
|
"learning_rate": 2.4842894325194996e-05,
|
||
|
|
"loss": 0.5454,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21379245817661285,
|
||
|
|
"step": 1460,
|
||
|
|
"valid_targets_mean": 4255.8,
|
||
|
|
"valid_targets_min": 1314
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4097641250685684,
|
||
|
|
"grad_norm": 0.27746274584622366,
|
||
|
|
"learning_rate": 2.473140798436285e-05,
|
||
|
|
"loss": 0.5415,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16948026418685913,
|
||
|
|
"step": 1465,
|
||
|
|
"valid_targets_mean": 4050.4,
|
||
|
|
"valid_targets_min": 1624
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4179923203510696,
|
||
|
|
"grad_norm": 0.2809157030625528,
|
||
|
|
"learning_rate": 2.4619765689439064e-05,
|
||
|
|
"loss": 0.5188,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16582998633384705,
|
||
|
|
"step": 1470,
|
||
|
|
"valid_targets_mean": 3251.1,
|
||
|
|
"valid_targets_min": 1018
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.426220515633571,
|
||
|
|
"grad_norm": 0.27242860017545745,
|
||
|
|
"learning_rate": 2.4507971120316128e-05,
|
||
|
|
"loss": 0.5321,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18139344453811646,
|
||
|
|
"step": 1475,
|
||
|
|
"valid_targets_mean": 3961.1,
|
||
|
|
"valid_targets_min": 1033
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4344487109160724,
|
||
|
|
"grad_norm": 0.2572967478845345,
|
||
|
|
"learning_rate": 2.4396027961905704e-05,
|
||
|
|
"loss": 0.5298,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19407576322555542,
|
||
|
|
"step": 1480,
|
||
|
|
"valid_targets_mean": 5018.0,
|
||
|
|
"valid_targets_min": 1101
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4426769061985736,
|
||
|
|
"grad_norm": 0.2729508689206608,
|
||
|
|
"learning_rate": 2.4283939904017183e-05,
|
||
|
|
"loss": 0.5344,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19475191831588745,
|
||
|
|
"step": 1485,
|
||
|
|
"valid_targets_mean": 4231.1,
|
||
|
|
"valid_targets_min": 1310
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.450905101481075,
|
||
|
|
"grad_norm": 0.2922531287196005,
|
||
|
|
"learning_rate": 2.4171710641236045e-05,
|
||
|
|
"loss": 0.5288,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16344180703163147,
|
||
|
|
"step": 1490,
|
||
|
|
"valid_targets_mean": 3704.2,
|
||
|
|
"valid_targets_min": 1093
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4591332967635764,
|
||
|
|
"grad_norm": 0.29625471879334625,
|
||
|
|
"learning_rate": 2.4059343872802084e-05,
|
||
|
|
"loss": 0.5464,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17003777623176575,
|
||
|
|
"step": 1495,
|
||
|
|
"valid_targets_mean": 3797.5,
|
||
|
|
"valid_targets_min": 1323
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.467361492046078,
|
||
|
|
"grad_norm": 0.28096087867759284,
|
||
|
|
"learning_rate": 2.3946843302487497e-05,
|
||
|
|
"loss": 0.5458,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17858433723449707,
|
||
|
|
"step": 1500,
|
||
|
|
"valid_targets_mean": 3499.0,
|
||
|
|
"valid_targets_min": 665
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.475589687328579,
|
||
|
|
"grad_norm": 0.261118921385376,
|
||
|
|
"learning_rate": 2.3834212638474773e-05,
|
||
|
|
"loss": 0.5307,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18844759464263916,
|
||
|
|
"step": 1505,
|
||
|
|
"valid_targets_mean": 5092.9,
|
||
|
|
"valid_targets_min": 1149
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.483817882611081,
|
||
|
|
"grad_norm": 0.2831486710543651,
|
||
|
|
"learning_rate": 2.372145559323448e-05,
|
||
|
|
"loss": 0.5428,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2189253866672516,
|
||
|
|
"step": 1510,
|
||
|
|
"valid_targets_mean": 4786.8,
|
||
|
|
"valid_targets_min": 1492
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.492046077893582,
|
||
|
|
"grad_norm": 0.27484534472121397,
|
||
|
|
"learning_rate": 2.3608575883402903e-05,
|
||
|
|
"loss": 0.5438,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18232977390289307,
|
||
|
|
"step": 1515,
|
||
|
|
"valid_targets_mean": 4219.8,
|
||
|
|
"valid_targets_min": 776
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5002742731760836,
|
||
|
|
"grad_norm": 0.4465063057356539,
|
||
|
|
"learning_rate": 2.3495577229659515e-05,
|
||
|
|
"loss": 0.4163,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06493039429187775,
|
||
|
|
"step": 1520,
|
||
|
|
"valid_targets_mean": 5702.6,
|
||
|
|
"valid_targets_min": 2936
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.508502468458585,
|
||
|
|
"grad_norm": 0.340790384582099,
|
||
|
|
"learning_rate": 2.3382463356604378e-05,
|
||
|
|
"loss": 0.1673,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05560822784900665,
|
||
|
|
"step": 1525,
|
||
|
|
"valid_targets_mean": 5383.9,
|
||
|
|
"valid_targets_min": 3193
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.516730663741086,
|
||
|
|
"grad_norm": 0.23433249286165134,
|
||
|
|
"learning_rate": 2.3269237992635318e-05,
|
||
|
|
"loss": 0.1557,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05000593140721321,
|
||
|
|
"step": 1530,
|
||
|
|
"valid_targets_mean": 5404.5,
|
||
|
|
"valid_targets_min": 3254
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5249588590235876,
|
||
|
|
"grad_norm": 0.21914046561271092,
|
||
|
|
"learning_rate": 2.31559048698251e-05,
|
||
|
|
"loss": 0.1538,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.051547035574913025,
|
||
|
|
"step": 1535,
|
||
|
|
"valid_targets_mean": 5669.6,
|
||
|
|
"valid_targets_min": 2339
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.533187054306089,
|
||
|
|
"grad_norm": 0.2022698368996799,
|
||
|
|
"learning_rate": 2.3042467723798335e-05,
|
||
|
|
"loss": 0.1473,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05184007063508034,
|
||
|
|
"step": 1540,
|
||
|
|
"valid_targets_mean": 5712.7,
|
||
|
|
"valid_targets_min": 2963
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5414152495885904,
|
||
|
|
"grad_norm": 0.1880847339367831,
|
||
|
|
"learning_rate": 2.2928930293608435e-05,
|
||
|
|
"loss": 0.1494,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0460038036108017,
|
||
|
|
"step": 1545,
|
||
|
|
"valid_targets_mean": 5044.9,
|
||
|
|
"valid_targets_min": 2759
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5496434448710916,
|
||
|
|
"grad_norm": 0.18085215046098602,
|
||
|
|
"learning_rate": 2.281529632161429e-05,
|
||
|
|
"loss": 0.1451,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0443473756313324,
|
||
|
|
"step": 1550,
|
||
|
|
"valid_targets_mean": 5219.8,
|
||
|
|
"valid_targets_min": 2654
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.557871640153593,
|
||
|
|
"grad_norm": 0.19614621783834488,
|
||
|
|
"learning_rate": 2.2701569553356963e-05,
|
||
|
|
"loss": 0.1497,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.047740936279296875,
|
||
|
|
"step": 1555,
|
||
|
|
"valid_targets_mean": 5413.7,
|
||
|
|
"valid_targets_min": 3427
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5660998354360944,
|
||
|
|
"grad_norm": 0.18143070301123154,
|
||
|
|
"learning_rate": 2.2587753737436217e-05,
|
||
|
|
"loss": 0.1477,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04606207087635994,
|
||
|
|
"step": 1560,
|
||
|
|
"valid_targets_mean": 5462.3,
|
||
|
|
"valid_targets_min": 3458
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5743280307185956,
|
||
|
|
"grad_norm": 0.21959475347948587,
|
||
|
|
"learning_rate": 2.247385262538696e-05,
|
||
|
|
"loss": 0.1437,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05131494253873825,
|
||
|
|
"step": 1565,
|
||
|
|
"valid_targets_mean": 5651.2,
|
||
|
|
"valid_targets_min": 2078
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5825562260010972,
|
||
|
|
"grad_norm": 0.18550933231717945,
|
||
|
|
"learning_rate": 2.235986997155556e-05,
|
||
|
|
"loss": 0.1423,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04456018656492233,
|
||
|
|
"step": 1570,
|
||
|
|
"valid_targets_mean": 5431.2,
|
||
|
|
"valid_targets_min": 1278
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5907844212835984,
|
||
|
|
"grad_norm": 0.25278930145039497,
|
||
|
|
"learning_rate": 2.2245809532976157e-05,
|
||
|
|
"loss": 0.1468,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04588646814227104,
|
||
|
|
"step": 1575,
|
||
|
|
"valid_targets_mean": 5013.1,
|
||
|
|
"valid_targets_min": 2521
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5990126165660996,
|
||
|
|
"grad_norm": 0.21124774470898763,
|
||
|
|
"learning_rate": 2.2131675069246758e-05,
|
||
|
|
"loss": 0.1418,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.042148131877183914,
|
||
|
|
"step": 1580,
|
||
|
|
"valid_targets_mean": 5014.3,
|
||
|
|
"valid_targets_min": 2703
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6072408118486012,
|
||
|
|
"grad_norm": 0.20744467482599036,
|
||
|
|
"learning_rate": 2.201747034240537e-05,
|
||
|
|
"loss": 0.1434,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.044687531888484955,
|
||
|
|
"step": 1585,
|
||
|
|
"valid_targets_mean": 5006.2,
|
||
|
|
"valid_targets_min": 2670
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6154690071311024,
|
||
|
|
"grad_norm": 0.18358585696946206,
|
||
|
|
"learning_rate": 2.1903199116805953e-05,
|
||
|
|
"loss": 0.1442,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05012306571006775,
|
||
|
|
"step": 1590,
|
||
|
|
"valid_targets_mean": 6038.5,
|
||
|
|
"valid_targets_min": 3631
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.623697202413604,
|
||
|
|
"grad_norm": 0.2389717332734137,
|
||
|
|
"learning_rate": 2.1788865158994384e-05,
|
||
|
|
"loss": 0.1424,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05136081576347351,
|
||
|
|
"step": 1595,
|
||
|
|
"valid_targets_mean": 4937.4,
|
||
|
|
"valid_targets_min": 2598
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6319253976961052,
|
||
|
|
"grad_norm": 0.1946054551208735,
|
||
|
|
"learning_rate": 2.1674472237584272e-05,
|
||
|
|
"loss": 0.1414,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.046021468937397,
|
||
|
|
"step": 1600,
|
||
|
|
"valid_targets_mean": 5467.6,
|
||
|
|
"valid_targets_min": 2990
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.640153592978607,
|
||
|
|
"grad_norm": 0.2157540275683584,
|
||
|
|
"learning_rate": 2.1560024123132755e-05,
|
||
|
|
"loss": 0.144,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04845328629016876,
|
||
|
|
"step": 1605,
|
||
|
|
"valid_targets_mean": 5111.8,
|
||
|
|
"valid_targets_min": 2044
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.648381788261108,
|
||
|
|
"grad_norm": 0.20199076365851348,
|
||
|
|
"learning_rate": 2.1445524588016214e-05,
|
||
|
|
"loss": 0.1454,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0499688945710659,
|
||
|
|
"step": 1610,
|
||
|
|
"valid_targets_mean": 5763.9,
|
||
|
|
"valid_targets_min": 2904
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6566099835436097,
|
||
|
|
"grad_norm": 0.19383269744904036,
|
||
|
|
"learning_rate": 2.1330977406305933e-05,
|
||
|
|
"loss": 0.1419,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.044633202254772186,
|
||
|
|
"step": 1615,
|
||
|
|
"valid_targets_mean": 4922.0,
|
||
|
|
"valid_targets_min": 2500
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.664838178826111,
|
||
|
|
"grad_norm": 0.20125926925518162,
|
||
|
|
"learning_rate": 2.1216386353643686e-05,
|
||
|
|
"loss": 0.1418,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04471147805452347,
|
||
|
|
"step": 1620,
|
||
|
|
"valid_targets_mean": 5118.9,
|
||
|
|
"valid_targets_min": 2923
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.673066374108612,
|
||
|
|
"grad_norm": 0.3738083632546402,
|
||
|
|
"learning_rate": 2.110175520711731e-05,
|
||
|
|
"loss": 0.3568,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14241358637809753,
|
||
|
|
"step": 1625,
|
||
|
|
"valid_targets_mean": 3042.4,
|
||
|
|
"valid_targets_min": 1100
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6812945693911137,
|
||
|
|
"grad_norm": 0.3399624573266519,
|
||
|
|
"learning_rate": 2.098708774513619e-05,
|
||
|
|
"loss": 0.4743,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18090471625328064,
|
||
|
|
"step": 1630,
|
||
|
|
"valid_targets_mean": 4385.5,
|
||
|
|
"valid_targets_min": 891
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.689522764673615,
|
||
|
|
"grad_norm": 0.38175053443981743,
|
||
|
|
"learning_rate": 2.0872387747306725e-05,
|
||
|
|
"loss": 0.4392,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1515863686800003,
|
||
|
|
"step": 1635,
|
||
|
|
"valid_targets_mean": 1813.8,
|
||
|
|
"valid_targets_min": 507
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6977509599561165,
|
||
|
|
"grad_norm": 0.2944089441499447,
|
||
|
|
"learning_rate": 2.075765899430773e-05,
|
||
|
|
"loss": 0.4485,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13034245371818542,
|
||
|
|
"step": 1640,
|
||
|
|
"valid_targets_mean": 3846.0,
|
||
|
|
"valid_targets_min": 1237
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7059791552386177,
|
||
|
|
"grad_norm": 0.26016259566125427,
|
||
|
|
"learning_rate": 2.0642905267765846e-05,
|
||
|
|
"loss": 0.4374,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1433674395084381,
|
||
|
|
"step": 1645,
|
||
|
|
"valid_targets_mean": 4735.8,
|
||
|
|
"valid_targets_min": 910
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.714207350521119,
|
||
|
|
"grad_norm": 0.29076860147451133,
|
||
|
|
"learning_rate": 2.0528130350130867e-05,
|
||
|
|
"loss": 0.4346,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1388690173625946,
|
||
|
|
"step": 1650,
|
||
|
|
"valid_targets_mean": 3882.9,
|
||
|
|
"valid_targets_min": 1176
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7224355458036205,
|
||
|
|
"grad_norm": 0.27279308035940975,
|
||
|
|
"learning_rate": 2.041333802455109e-05,
|
||
|
|
"loss": 0.4311,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1723819077014923,
|
||
|
|
"step": 1655,
|
||
|
|
"valid_targets_mean": 4241.8,
|
||
|
|
"valid_targets_min": 1013
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7306637410861216,
|
||
|
|
"grad_norm": 0.22293614986974064,
|
||
|
|
"learning_rate": 2.0298532074748594e-05,
|
||
|
|
"loss": 0.4087,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11925294995307922,
|
||
|
|
"step": 1660,
|
||
|
|
"valid_targets_mean": 4934.8,
|
||
|
|
"valid_targets_min": 1378
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7388919363686233,
|
||
|
|
"grad_norm": 0.2864792947502029,
|
||
|
|
"learning_rate": 2.0183716284894533e-05,
|
||
|
|
"loss": 0.451,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1562422215938568,
|
||
|
|
"step": 1665,
|
||
|
|
"valid_targets_mean": 3635.7,
|
||
|
|
"valid_targets_min": 1162
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7471201316511245,
|
||
|
|
"grad_norm": 0.28246275393534515,
|
||
|
|
"learning_rate": 2.00688944394844e-05,
|
||
|
|
"loss": 0.4236,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12074942141771317,
|
||
|
|
"step": 1670,
|
||
|
|
"valid_targets_mean": 3513.8,
|
||
|
|
"valid_targets_min": 1441
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7553483269336256,
|
||
|
|
"grad_norm": 0.6175672954446245,
|
||
|
|
"learning_rate": 1.9954070323213296e-05,
|
||
|
|
"loss": 0.4337,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2847403883934021,
|
||
|
|
"step": 1675,
|
||
|
|
"valid_targets_mean": 4413.0,
|
||
|
|
"valid_targets_min": 1233
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7635765222161273,
|
||
|
|
"grad_norm": 0.23873518139179745,
|
||
|
|
"learning_rate": 1.9839247720851178e-05,
|
||
|
|
"loss": 0.4245,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10815485566854477,
|
||
|
|
"step": 1680,
|
||
|
|
"valid_targets_mean": 3737.7,
|
||
|
|
"valid_targets_min": 1650
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.771804717498629,
|
||
|
|
"grad_norm": 0.27032318641137143,
|
||
|
|
"learning_rate": 1.9724430417118074e-05,
|
||
|
|
"loss": 0.4213,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11345283687114716,
|
||
|
|
"step": 1685,
|
||
|
|
"valid_targets_mean": 3766.6,
|
||
|
|
"valid_targets_min": 1250
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.78003291278113,
|
||
|
|
"grad_norm": 0.2881813172917136,
|
||
|
|
"learning_rate": 1.9609622196559402e-05,
|
||
|
|
"loss": 0.4447,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13241040706634521,
|
||
|
|
"step": 1690,
|
||
|
|
"valid_targets_mean": 3043.5,
|
||
|
|
"valid_targets_min": 1208
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7882611080636313,
|
||
|
|
"grad_norm": 0.25597327371881795,
|
||
|
|
"learning_rate": 1.9494826843421147e-05,
|
||
|
|
"loss": 0.4473,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15462671220302582,
|
||
|
|
"step": 1695,
|
||
|
|
"valid_targets_mean": 4615.1,
|
||
|
|
"valid_targets_min": 269
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.796489303346133,
|
||
|
|
"grad_norm": 0.2656100599955537,
|
||
|
|
"learning_rate": 1.9380048141525194e-05,
|
||
|
|
"loss": 0.4193,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13796931505203247,
|
||
|
|
"step": 1700,
|
||
|
|
"valid_targets_mean": 5172.2,
|
||
|
|
"valid_targets_min": 1589
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.804717498628634,
|
||
|
|
"grad_norm": 0.28598719744173756,
|
||
|
|
"learning_rate": 1.9265289874144554e-05,
|
||
|
|
"loss": 0.446,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1630500853061676,
|
||
|
|
"step": 1705,
|
||
|
|
"valid_targets_mean": 4159.3,
|
||
|
|
"valid_targets_min": 1283
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8129456939111357,
|
||
|
|
"grad_norm": 0.27527634681188917,
|
||
|
|
"learning_rate": 1.9150555823878708e-05,
|
||
|
|
"loss": 0.4301,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.155914306640625,
|
||
|
|
"step": 1710,
|
||
|
|
"valid_targets_mean": 4058.6,
|
||
|
|
"valid_targets_min": 876
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.821173889193637,
|
||
|
|
"grad_norm": 0.2853993592486656,
|
||
|
|
"learning_rate": 1.9035849772528907e-05,
|
||
|
|
"loss": 0.4168,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1859429031610489,
|
||
|
|
"step": 1715,
|
||
|
|
"valid_targets_mean": 4401.0,
|
||
|
|
"valid_targets_min": 1349
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.829402084476138,
|
||
|
|
"grad_norm": 0.2519814408323229,
|
||
|
|
"learning_rate": 1.8921175500973496e-05,
|
||
|
|
"loss": 0.4388,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12586364150047302,
|
||
|
|
"step": 1720,
|
||
|
|
"valid_targets_mean": 3505.7,
|
||
|
|
"valid_targets_min": 809
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8376302797586397,
|
||
|
|
"grad_norm": 0.25743712856714657,
|
||
|
|
"learning_rate": 1.8806536789043322e-05,
|
||
|
|
"loss": 0.4373,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11665894091129303,
|
||
|
|
"step": 1725,
|
||
|
|
"valid_targets_mean": 3860.3,
|
||
|
|
"valid_targets_min": 602
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.845858475041141,
|
||
|
|
"grad_norm": 0.3117452905892116,
|
||
|
|
"learning_rate": 1.869193741539714e-05,
|
||
|
|
"loss": 0.3654,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11202247440814972,
|
||
|
|
"step": 1730,
|
||
|
|
"valid_targets_mean": 8001.6,
|
||
|
|
"valid_targets_min": 1252
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8540866703236425,
|
||
|
|
"grad_norm": 0.21933985341011952,
|
||
|
|
"learning_rate": 1.8577381157397056e-05,
|
||
|
|
"loss": 0.2606,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08016058057546616,
|
||
|
|
"step": 1735,
|
||
|
|
"valid_targets_mean": 6021.0,
|
||
|
|
"valid_targets_min": 389
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8623148656061437,
|
||
|
|
"grad_norm": 0.22902224638423593,
|
||
|
|
"learning_rate": 1.8462871790984015e-05,
|
||
|
|
"loss": 0.2631,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12745118141174316,
|
||
|
|
"step": 1740,
|
||
|
|
"valid_targets_mean": 6741.6,
|
||
|
|
"valid_targets_min": 473
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.870543060888645,
|
||
|
|
"grad_norm": 0.21401347256438896,
|
||
|
|
"learning_rate": 1.8348413090553356e-05,
|
||
|
|
"loss": 0.2638,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07021795213222504,
|
||
|
|
"step": 1745,
|
||
|
|
"valid_targets_mean": 5457.3,
|
||
|
|
"valid_targets_min": 253
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8787712561711465,
|
||
|
|
"grad_norm": 0.19462902063881504,
|
||
|
|
"learning_rate": 1.8234008828830386e-05,
|
||
|
|
"loss": 0.2677,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08573737740516663,
|
||
|
|
"step": 1750,
|
||
|
|
"valid_targets_mean": 6283.3,
|
||
|
|
"valid_targets_min": 146
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8869994514536477,
|
||
|
|
"grad_norm": 0.202942917615318,
|
||
|
|
"learning_rate": 1.8119662776746043e-05,
|
||
|
|
"loss": 0.2703,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08039772510528564,
|
||
|
|
"step": 1755,
|
||
|
|
"valid_targets_mean": 6143.3,
|
||
|
|
"valid_targets_min": 233
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8952276467361493,
|
||
|
|
"grad_norm": 0.17992757298307224,
|
||
|
|
"learning_rate": 1.800537870331257e-05,
|
||
|
|
"loss": 0.2531,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06841991096735,
|
||
|
|
"step": 1760,
|
||
|
|
"valid_targets_mean": 6087.1,
|
||
|
|
"valid_targets_min": 1118
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9034558420186505,
|
||
|
|
"grad_norm": 0.17660821788526068,
|
||
|
|
"learning_rate": 1.789116037549933e-05,
|
||
|
|
"loss": 0.2646,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08433274179697037,
|
||
|
|
"step": 1765,
|
||
|
|
"valid_targets_mean": 6520.0,
|
||
|
|
"valid_targets_min": 178
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.911684037301152,
|
||
|
|
"grad_norm": 0.188845019321141,
|
||
|
|
"learning_rate": 1.77770115581086e-05,
|
||
|
|
"loss": 0.2545,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08890101313591003,
|
||
|
|
"step": 1770,
|
||
|
|
"valid_targets_mean": 6715.5,
|
||
|
|
"valid_targets_min": 744
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9199122325836533,
|
||
|
|
"grad_norm": 0.2250077260143483,
|
||
|
|
"learning_rate": 1.7662936013651493e-05,
|
||
|
|
"loss": 0.2663,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.053828299045562744,
|
||
|
|
"step": 1775,
|
||
|
|
"valid_targets_mean": 1099.6,
|
||
|
|
"valid_targets_min": 316
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.928140427866155,
|
||
|
|
"grad_norm": 0.18320311040212517,
|
||
|
|
"learning_rate": 1.7548937502223932e-05,
|
||
|
|
"loss": 0.2602,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.061102431267499924,
|
||
|
|
"step": 1780,
|
||
|
|
"valid_targets_mean": 5019.2,
|
||
|
|
"valid_targets_min": 247
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.936368623148656,
|
||
|
|
"grad_norm": 0.17206280683714112,
|
||
|
|
"learning_rate": 1.7435019781382737e-05,
|
||
|
|
"loss": 0.2659,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09793031215667725,
|
||
|
|
"step": 1785,
|
||
|
|
"valid_targets_mean": 7409.7,
|
||
|
|
"valid_targets_min": 1034
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9445968184311573,
|
||
|
|
"grad_norm": 0.19580444227484753,
|
||
|
|
"learning_rate": 1.732118660602175e-05,
|
||
|
|
"loss": 0.2784,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.054457198828458786,
|
||
|
|
"step": 1790,
|
||
|
|
"valid_targets_mean": 4764.1,
|
||
|
|
"valid_targets_min": 243
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.952825013713659,
|
||
|
|
"grad_norm": 0.19600050749352133,
|
||
|
|
"learning_rate": 1.7207441728248055e-05,
|
||
|
|
"loss": 0.2427,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08765925467014313,
|
||
|
|
"step": 1795,
|
||
|
|
"valid_targets_mean": 6958.2,
|
||
|
|
"valid_targets_min": 1080
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.96105320899616,
|
||
|
|
"grad_norm": 0.20856937565072015,
|
||
|
|
"learning_rate": 1.7093788897258338e-05,
|
||
|
|
"loss": 0.261,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13853465020656586,
|
||
|
|
"step": 1800,
|
||
|
|
"valid_targets_mean": 7221.4,
|
||
|
|
"valid_targets_min": 405
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9692814042786617,
|
||
|
|
"grad_norm": 0.1961668365424937,
|
||
|
|
"learning_rate": 1.698023185921526e-05,
|
||
|
|
"loss": 0.2532,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10164403915405273,
|
||
|
|
"step": 1805,
|
||
|
|
"valid_targets_mean": 7585.1,
|
||
|
|
"valid_targets_min": 1495
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.977509599561163,
|
||
|
|
"grad_norm": 0.19497785486683333,
|
||
|
|
"learning_rate": 1.6866774357124054e-05,
|
||
|
|
"loss": 0.2356,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1096169725060463,
|
||
|
|
"step": 1810,
|
||
|
|
"valid_targets_mean": 6407.4,
|
||
|
|
"valid_targets_min": 201
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.985737794843664,
|
||
|
|
"grad_norm": 0.17898913986619544,
|
||
|
|
"learning_rate": 1.675342013070905e-05,
|
||
|
|
"loss": 0.2582,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06940888613462448,
|
||
|
|
"step": 1815,
|
||
|
|
"valid_targets_mean": 6268.8,
|
||
|
|
"valid_targets_min": 536
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9939659901261657,
|
||
|
|
"grad_norm": 0.18783301380836664,
|
||
|
|
"learning_rate": 1.6640172916290515e-05,
|
||
|
|
"loss": 0.2508,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11000367999076843,
|
||
|
|
"step": 1820,
|
||
|
|
"valid_targets_mean": 6778.3,
|
||
|
|
"valid_targets_min": 674
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0016456390565,
|
||
|
|
"grad_norm": 2.617429539265205,
|
||
|
|
"learning_rate": 1.6527036446661396e-05,
|
||
|
|
"loss": 0.4103,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22323724627494812,
|
||
|
|
"step": 1825,
|
||
|
|
"valid_targets_mean": 7264.4,
|
||
|
|
"valid_targets_min": 1973
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0098738343390017,
|
||
|
|
"grad_norm": 0.8824033570259873,
|
||
|
|
"learning_rate": 1.641401445096436e-05,
|
||
|
|
"loss": 0.5455,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1985154151916504,
|
||
|
|
"step": 1830,
|
||
|
|
"valid_targets_mean": 9100.2,
|
||
|
|
"valid_targets_min": 3178
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.018102029621503,
|
||
|
|
"grad_norm": 0.7118784346337131,
|
||
|
|
"learning_rate": 1.6301110654568833e-05,
|
||
|
|
"loss": 0.4481,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13510319590568542,
|
||
|
|
"step": 1835,
|
||
|
|
"valid_targets_mean": 6959.6,
|
||
|
|
"valid_targets_min": 2166
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0263302249040045,
|
||
|
|
"grad_norm": 0.48035427814678705,
|
||
|
|
"learning_rate": 1.6188328778948238e-05,
|
||
|
|
"loss": 0.4066,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1353878676891327,
|
||
|
|
"step": 1840,
|
||
|
|
"valid_targets_mean": 7188.0,
|
||
|
|
"valid_targets_min": 2166
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0345584201865057,
|
||
|
|
"grad_norm": 0.3563748368494854,
|
||
|
|
"learning_rate": 1.6075672541557287e-05,
|
||
|
|
"loss": 0.3745,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13746076822280884,
|
||
|
|
"step": 1845,
|
||
|
|
"valid_targets_mean": 6959.9,
|
||
|
|
"valid_targets_min": 2083
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0427866154690073,
|
||
|
|
"grad_norm": 0.2528267340757668,
|
||
|
|
"learning_rate": 1.5963145655709495e-05,
|
||
|
|
"loss": 0.366,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13522320985794067,
|
||
|
|
"step": 1850,
|
||
|
|
"valid_targets_mean": 7818.4,
|
||
|
|
"valid_targets_min": 2042
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0510148107515085,
|
||
|
|
"grad_norm": 0.235073091120978,
|
||
|
|
"learning_rate": 1.5850751830454747e-05,
|
||
|
|
"loss": 0.3588,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12187724560499191,
|
||
|
|
"step": 1855,
|
||
|
|
"valid_targets_mean": 7926.4,
|
||
|
|
"valid_targets_min": 3527
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0592430060340097,
|
||
|
|
"grad_norm": 0.22130659813335654,
|
||
|
|
"learning_rate": 1.573849477045706e-05,
|
||
|
|
"loss": 0.3523,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10538642853498459,
|
||
|
|
"step": 1860,
|
||
|
|
"valid_targets_mean": 6605.4,
|
||
|
|
"valid_targets_min": 2848
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0674712013165113,
|
||
|
|
"grad_norm": 0.22662966396011508,
|
||
|
|
"learning_rate": 1.5626378175872486e-05,
|
||
|
|
"loss": 0.3435,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11400727927684784,
|
||
|
|
"step": 1865,
|
||
|
|
"valid_targets_mean": 6947.1,
|
||
|
|
"valid_targets_min": 2244
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0756993965990125,
|
||
|
|
"grad_norm": 0.23238049650837628,
|
||
|
|
"learning_rate": 1.5514405742227103e-05,
|
||
|
|
"loss": 0.3478,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11836406588554382,
|
||
|
|
"step": 1870,
|
||
|
|
"valid_targets_mean": 7027.9,
|
||
|
|
"valid_targets_min": 340
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.083927591881514,
|
||
|
|
"grad_norm": 0.20760299511274713,
|
||
|
|
"learning_rate": 1.5402581160295265e-05,
|
||
|
|
"loss": 0.3424,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13461612164974213,
|
||
|
|
"step": 1875,
|
||
|
|
"valid_targets_mean": 8959.9,
|
||
|
|
"valid_targets_min": 3446
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0921557871640153,
|
||
|
|
"grad_norm": 0.20502093564997662,
|
||
|
|
"learning_rate": 1.5290908115977884e-05,
|
||
|
|
"loss": 0.3233,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10294201970100403,
|
||
|
|
"step": 1880,
|
||
|
|
"valid_targets_mean": 9482.4,
|
||
|
|
"valid_targets_min": 5095
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.100383982446517,
|
||
|
|
"grad_norm": 0.19747411277171628,
|
||
|
|
"learning_rate": 1.5179390290181013e-05,
|
||
|
|
"loss": 0.3113,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09898518025875092,
|
||
|
|
"step": 1885,
|
||
|
|
"valid_targets_mean": 8913.7,
|
||
|
|
"valid_targets_min": 251
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.108612177729018,
|
||
|
|
"grad_norm": 0.19042094134064866,
|
||
|
|
"learning_rate": 1.5068031358694437e-05,
|
||
|
|
"loss": 0.3056,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10452181100845337,
|
||
|
|
"step": 1890,
|
||
|
|
"valid_targets_mean": 8780.3,
|
||
|
|
"valid_targets_min": 349
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1168403730115193,
|
||
|
|
"grad_norm": 0.18674667249593555,
|
||
|
|
"learning_rate": 1.4956834992070589e-05,
|
||
|
|
"loss": 0.3058,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10294011235237122,
|
||
|
|
"step": 1895,
|
||
|
|
"valid_targets_mean": 9303.8,
|
||
|
|
"valid_targets_min": 773
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.125068568294021,
|
||
|
|
"grad_norm": 0.18439896531534866,
|
||
|
|
"learning_rate": 1.4845804855503494e-05,
|
||
|
|
"loss": 0.3039,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1009887233376503,
|
||
|
|
"step": 1900,
|
||
|
|
"valid_targets_mean": 9819.0,
|
||
|
|
"valid_targets_min": 5584
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.133296763576522,
|
||
|
|
"grad_norm": 0.18693371137153128,
|
||
|
|
"learning_rate": 1.4734944608708022e-05,
|
||
|
|
"loss": 0.3029,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10432776808738708,
|
||
|
|
"step": 1905,
|
||
|
|
"valid_targets_mean": 9100.7,
|
||
|
|
"valid_targets_min": 306
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1415249588590237,
|
||
|
|
"grad_norm": 0.1968021432855259,
|
||
|
|
"learning_rate": 1.46242579057992e-05,
|
||
|
|
"loss": 0.3124,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10337679833173752,
|
||
|
|
"step": 1910,
|
||
|
|
"valid_targets_mean": 9212.3,
|
||
|
|
"valid_targets_min": 378
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.149753154141525,
|
||
|
|
"grad_norm": 0.19727420663031295,
|
||
|
|
"learning_rate": 1.451374839517183e-05,
|
||
|
|
"loss": 0.3035,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09127824008464813,
|
||
|
|
"step": 1915,
|
||
|
|
"valid_targets_mean": 8648.9,
|
||
|
|
"valid_targets_min": 1557
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1579813494240265,
|
||
|
|
"grad_norm": 0.2133142568969134,
|
||
|
|
"learning_rate": 1.4403419719380161e-05,
|
||
|
|
"loss": 0.306,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11271902918815613,
|
||
|
|
"step": 1920,
|
||
|
|
"valid_targets_mean": 10183.4,
|
||
|
|
"valid_targets_min": 3886
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1662095447065277,
|
||
|
|
"grad_norm": 0.19051586500080203,
|
||
|
|
"learning_rate": 1.42932755150179e-05,
|
||
|
|
"loss": 0.3011,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09688104689121246,
|
||
|
|
"step": 1925,
|
||
|
|
"valid_targets_mean": 9594.6,
|
||
|
|
"valid_targets_min": 344
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.174437739989029,
|
||
|
|
"grad_norm": 0.1872830976899931,
|
||
|
|
"learning_rate": 1.4183319412598274e-05,
|
||
|
|
"loss": 0.3048,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09433311969041824,
|
||
|
|
"step": 1930,
|
||
|
|
"valid_targets_mean": 8928.9,
|
||
|
|
"valid_targets_min": 897
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1826659352715305,
|
||
|
|
"grad_norm": 0.1999039787769261,
|
||
|
|
"learning_rate": 1.4073555036434423e-05,
|
||
|
|
"loss": 0.3028,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10328617691993713,
|
||
|
|
"step": 1935,
|
||
|
|
"valid_targets_mean": 9772.7,
|
||
|
|
"valid_targets_min": 3574
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1908941305540317,
|
||
|
|
"grad_norm": 0.19542875185819147,
|
||
|
|
"learning_rate": 1.3963986004519885e-05,
|
||
|
|
"loss": 0.3015,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10275737941265106,
|
||
|
|
"step": 1940,
|
||
|
|
"valid_targets_mean": 10406.6,
|
||
|
|
"valid_targets_min": 5877
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1991223258365333,
|
||
|
|
"grad_norm": 0.1732017772846815,
|
||
|
|
"learning_rate": 1.385461592840939e-05,
|
||
|
|
"loss": 0.3036,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1027611643075943,
|
||
|
|
"step": 1945,
|
||
|
|
"valid_targets_mean": 9535.4,
|
||
|
|
"valid_targets_min": 3557
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2073505211190345,
|
||
|
|
"grad_norm": 0.1882390427186111,
|
||
|
|
"learning_rate": 1.3745448413099795e-05,
|
||
|
|
"loss": 0.3113,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09596308320760727,
|
||
|
|
"step": 1950,
|
||
|
|
"valid_targets_mean": 9136.9,
|
||
|
|
"valid_targets_min": 1343
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.215578716401536,
|
||
|
|
"grad_norm": 0.1810142486999098,
|
||
|
|
"learning_rate": 1.3636487056911236e-05,
|
||
|
|
"loss": 0.3013,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08757059276103973,
|
||
|
|
"step": 1955,
|
||
|
|
"valid_targets_mean": 9388.7,
|
||
|
|
"valid_targets_min": 3282
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2238069116840373,
|
||
|
|
"grad_norm": 0.18209264751554483,
|
||
|
|
"learning_rate": 1.3527735451368567e-05,
|
||
|
|
"loss": 0.2971,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0974527895450592,
|
||
|
|
"step": 1960,
|
||
|
|
"valid_targets_mean": 9729.2,
|
||
|
|
"valid_targets_min": 3231
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2320351069665385,
|
||
|
|
"grad_norm": 0.18591292208455748,
|
||
|
|
"learning_rate": 1.3419197181082937e-05,
|
||
|
|
"loss": 0.3001,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11141443252563477,
|
||
|
|
"step": 1965,
|
||
|
|
"valid_targets_mean": 10332.3,
|
||
|
|
"valid_targets_min": 4344
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.24026330224904,
|
||
|
|
"grad_norm": 0.18085390289564288,
|
||
|
|
"learning_rate": 1.3310875823633675e-05,
|
||
|
|
"loss": 0.3029,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10907731205224991,
|
||
|
|
"step": 1970,
|
||
|
|
"valid_targets_mean": 11408.0,
|
||
|
|
"valid_targets_min": 5726
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2484914975315413,
|
||
|
|
"grad_norm": 0.1893653762416235,
|
||
|
|
"learning_rate": 1.3202774949450326e-05,
|
||
|
|
"loss": 0.3018,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10792022943496704,
|
||
|
|
"step": 1975,
|
||
|
|
"valid_targets_mean": 10250.5,
|
||
|
|
"valid_targets_min": 2607
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.256719692814043,
|
||
|
|
"grad_norm": 0.2029953340139833,
|
||
|
|
"learning_rate": 1.3094898121695008e-05,
|
||
|
|
"loss": 0.3008,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09804176539182663,
|
||
|
|
"step": 1980,
|
||
|
|
"valid_targets_mean": 9131.8,
|
||
|
|
"valid_targets_min": 761
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.264947888096544,
|
||
|
|
"grad_norm": 0.1990065296907557,
|
||
|
|
"learning_rate": 1.2987248896144915e-05,
|
||
|
|
"loss": 0.299,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10255561023950577,
|
||
|
|
"step": 1985,
|
||
|
|
"valid_targets_mean": 10329.2,
|
||
|
|
"valid_targets_min": 5291
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2731760833790453,
|
||
|
|
"grad_norm": 0.19438225738776518,
|
||
|
|
"learning_rate": 1.2879830821075174e-05,
|
||
|
|
"loss": 0.3009,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09681476652622223,
|
||
|
|
"step": 1990,
|
||
|
|
"valid_targets_mean": 8999.2,
|
||
|
|
"valid_targets_min": 2381
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.281404278661547,
|
||
|
|
"grad_norm": 0.18865449062923906,
|
||
|
|
"learning_rate": 1.277264743714182e-05,
|
||
|
|
"loss": 0.2964,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10528258979320526,
|
||
|
|
"step": 1995,
|
||
|
|
"valid_targets_mean": 9227.1,
|
||
|
|
"valid_targets_min": 437
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.289632473944048,
|
||
|
|
"grad_norm": 0.20620698231027396,
|
||
|
|
"learning_rate": 1.2665702277265168e-05,
|
||
|
|
"loss": 0.2996,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09422826766967773,
|
||
|
|
"step": 2000,
|
||
|
|
"valid_targets_mean": 9084.0,
|
||
|
|
"valid_targets_min": 1451
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2978606692265497,
|
||
|
|
"grad_norm": 0.18470245913703312,
|
||
|
|
"learning_rate": 1.2558998866513283e-05,
|
||
|
|
"loss": 0.2958,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08594414591789246,
|
||
|
|
"step": 2005,
|
||
|
|
"valid_targets_mean": 9274.7,
|
||
|
|
"valid_targets_min": 4259
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.306088864509051,
|
||
|
|
"grad_norm": 0.18549116063553767,
|
||
|
|
"learning_rate": 1.245254072198585e-05,
|
||
|
|
"loss": 0.2922,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09195162355899811,
|
||
|
|
"step": 2010,
|
||
|
|
"valid_targets_mean": 8931.9,
|
||
|
|
"valid_targets_min": 410
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3143170597915526,
|
||
|
|
"grad_norm": 0.1963520402334605,
|
||
|
|
"learning_rate": 1.2346331352698206e-05,
|
||
|
|
"loss": 0.3002,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12253104895353317,
|
||
|
|
"step": 2015,
|
||
|
|
"valid_targets_mean": 11522.8,
|
||
|
|
"valid_targets_min": 620
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3225452550740537,
|
||
|
|
"grad_norm": 0.2583659342136034,
|
||
|
|
"learning_rate": 1.224037425946571e-05,
|
||
|
|
"loss": 0.3058,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10416905581951141,
|
||
|
|
"step": 2020,
|
||
|
|
"valid_targets_mean": 6154.2,
|
||
|
|
"valid_targets_min": 386
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3307734503565554,
|
||
|
|
"grad_norm": 0.24178543153686508,
|
||
|
|
"learning_rate": 1.2134672934788338e-05,
|
||
|
|
"loss": 0.3072,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1027805358171463,
|
||
|
|
"step": 2025,
|
||
|
|
"valid_targets_mean": 6070.9,
|
||
|
|
"valid_targets_min": 2967
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3390016456390565,
|
||
|
|
"grad_norm": 1.895090237361272,
|
||
|
|
"learning_rate": 1.202923086273554e-05,
|
||
|
|
"loss": 0.5715,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20451006293296814,
|
||
|
|
"step": 2030,
|
||
|
|
"valid_targets_mean": 3383.0,
|
||
|
|
"valid_targets_min": 1270
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3472298409215577,
|
||
|
|
"grad_norm": 0.7358808746545971,
|
||
|
|
"learning_rate": 1.1924051518831444e-05,
|
||
|
|
"loss": 0.6695,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16195747256278992,
|
||
|
|
"step": 2035,
|
||
|
|
"valid_targets_mean": 2937.2,
|
||
|
|
"valid_targets_min": 1026
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3554580362040594,
|
||
|
|
"grad_norm": 0.6745202805527456,
|
||
|
|
"learning_rate": 1.1819138369940251e-05,
|
||
|
|
"loss": 0.5868,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2747642397880554,
|
||
|
|
"step": 2040,
|
||
|
|
"valid_targets_mean": 4071.6,
|
||
|
|
"valid_targets_min": 1539
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3636862314865605,
|
||
|
|
"grad_norm": 0.3884525181714535,
|
||
|
|
"learning_rate": 1.1714494874152025e-05,
|
||
|
|
"loss": 0.5628,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18681445717811584,
|
||
|
|
"step": 2045,
|
||
|
|
"valid_targets_mean": 5239.0,
|
||
|
|
"valid_targets_min": 1739
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.371914426769062,
|
||
|
|
"grad_norm": 0.3025604115847852,
|
||
|
|
"learning_rate": 1.1610124480668636e-05,
|
||
|
|
"loss": 0.5438,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1633882224559784,
|
||
|
|
"step": 2050,
|
||
|
|
"valid_targets_mean": 4039.2,
|
||
|
|
"valid_targets_min": 1204
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3801426220515634,
|
||
|
|
"grad_norm": 0.29935962744985817,
|
||
|
|
"learning_rate": 1.1506030629690124e-05,
|
||
|
|
"loss": 0.513,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1476593315601349,
|
||
|
|
"step": 2055,
|
||
|
|
"valid_targets_mean": 3473.9,
|
||
|
|
"valid_targets_min": 1130
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3883708173340645,
|
||
|
|
"grad_norm": 0.2901294118474583,
|
||
|
|
"learning_rate": 1.140221675230127e-05,
|
||
|
|
"loss": 0.5226,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.155238538980484,
|
||
|
|
"step": 2060,
|
||
|
|
"valid_targets_mean": 3420.7,
|
||
|
|
"valid_targets_min": 891
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.396599012616566,
|
||
|
|
"grad_norm": 0.31111984150155464,
|
||
|
|
"learning_rate": 1.1298686270358542e-05,
|
||
|
|
"loss": 0.5053,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22876670956611633,
|
||
|
|
"step": 2065,
|
||
|
|
"valid_targets_mean": 3772.6,
|
||
|
|
"valid_targets_min": 1497
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4048272078990673,
|
||
|
|
"grad_norm": 0.25109657866341767,
|
||
|
|
"learning_rate": 1.1195442596377253e-05,
|
||
|
|
"loss": 0.4982,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15915924310684204,
|
||
|
|
"step": 2070,
|
||
|
|
"valid_targets_mean": 4168.5,
|
||
|
|
"valid_targets_min": 972
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.413055403181569,
|
||
|
|
"grad_norm": 0.2583604576475023,
|
||
|
|
"learning_rate": 1.1092489133419137e-05,
|
||
|
|
"loss": 0.4989,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11599528789520264,
|
||
|
|
"step": 2075,
|
||
|
|
"valid_targets_mean": 3252.4,
|
||
|
|
"valid_targets_min": 636
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.42128359846407,
|
||
|
|
"grad_norm": 0.27203191468375026,
|
||
|
|
"learning_rate": 1.0989829274980126e-05,
|
||
|
|
"loss": 0.4889,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19358757138252258,
|
||
|
|
"step": 2080,
|
||
|
|
"valid_targets_mean": 4318.8,
|
||
|
|
"valid_targets_min": 1103
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.429511793746572,
|
||
|
|
"grad_norm": 0.29509970677561226,
|
||
|
|
"learning_rate": 1.088746640487854e-05,
|
||
|
|
"loss": 0.505,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1461564600467682,
|
||
|
|
"step": 2085,
|
||
|
|
"valid_targets_mean": 3735.0,
|
||
|
|
"valid_targets_min": 1048
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.437739989029073,
|
||
|
|
"grad_norm": 0.30173358477297363,
|
||
|
|
"learning_rate": 1.078540389714351e-05,
|
||
|
|
"loss": 0.4832,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21707960963249207,
|
||
|
|
"step": 2090,
|
||
|
|
"valid_targets_mean": 4098.8,
|
||
|
|
"valid_targets_min": 1319
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.445968184311574,
|
||
|
|
"grad_norm": 0.24353067059478417,
|
||
|
|
"learning_rate": 1.0683645115903811e-05,
|
||
|
|
"loss": 0.4899,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1496349275112152,
|
||
|
|
"step": 2095,
|
||
|
|
"valid_targets_mean": 4343.0,
|
||
|
|
"valid_targets_min": 1454
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4541963795940758,
|
||
|
|
"grad_norm": 0.2728265370581586,
|
||
|
|
"learning_rate": 1.0582193415276931e-05,
|
||
|
|
"loss": 0.5025,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14801722764968872,
|
||
|
|
"step": 2100,
|
||
|
|
"valid_targets_mean": 3758.9,
|
||
|
|
"valid_targets_min": 1151
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.462424574876577,
|
||
|
|
"grad_norm": 0.27268401905899425,
|
||
|
|
"learning_rate": 1.048105213925853e-05,
|
||
|
|
"loss": 0.5072,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14311590790748596,
|
||
|
|
"step": 2105,
|
||
|
|
"valid_targets_mean": 3194.6,
|
||
|
|
"valid_targets_min": 994
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4706527701590786,
|
||
|
|
"grad_norm": 0.27509131095220457,
|
||
|
|
"learning_rate": 1.0380224621612252e-05,
|
||
|
|
"loss": 0.5125,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15198372304439545,
|
||
|
|
"step": 2110,
|
||
|
|
"valid_targets_mean": 3790.3,
|
||
|
|
"valid_targets_min": 1391
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4788809654415798,
|
||
|
|
"grad_norm": 0.2806449483852979,
|
||
|
|
"learning_rate": 1.0279714185759771e-05,
|
||
|
|
"loss": 0.4921,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2070062756538391,
|
||
|
|
"step": 2115,
|
||
|
|
"valid_targets_mean": 4685.3,
|
||
|
|
"valid_targets_min": 822
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4871091607240814,
|
||
|
|
"grad_norm": 0.2501045793083265,
|
||
|
|
"learning_rate": 1.0179524144671315e-05,
|
||
|
|
"loss": 0.4981,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13317403197288513,
|
||
|
|
"step": 2120,
|
||
|
|
"valid_targets_mean": 3750.0,
|
||
|
|
"valid_targets_min": 1314
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4953373560065826,
|
||
|
|
"grad_norm": 0.2653477626320092,
|
||
|
|
"learning_rate": 1.0079657800756409e-05,
|
||
|
|
"loss": 0.5039,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15906351804733276,
|
||
|
|
"step": 2125,
|
||
|
|
"valid_targets_mean": 4435.8,
|
||
|
|
"valid_targets_min": 901
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5035655512890838,
|
||
|
|
"grad_norm": 0.27270201439588876,
|
||
|
|
"learning_rate": 9.980118445755072e-06,
|
||
|
|
"loss": 0.2588,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05085363984107971,
|
||
|
|
"step": 2130,
|
||
|
|
"valid_targets_mean": 5312.0,
|
||
|
|
"valid_targets_min": 2490
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5117937465715854,
|
||
|
|
"grad_norm": 0.25850926643571376,
|
||
|
|
"learning_rate": 9.880909360629265e-06,
|
||
|
|
"loss": 0.1504,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.052780479192733765,
|
||
|
|
"step": 2135,
|
||
|
|
"valid_targets_mean": 5549.4,
|
||
|
|
"valid_targets_min": 3000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5200219418540866,
|
||
|
|
"grad_norm": 0.21165170628572016,
|
||
|
|
"learning_rate": 9.782033815454806e-06,
|
||
|
|
"loss": 0.1382,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04313386231660843,
|
||
|
|
"step": 2140,
|
||
|
|
"valid_targets_mean": 5357.1,
|
||
|
|
"valid_targets_min": 2787
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.528250137136588,
|
||
|
|
"grad_norm": 0.181714134126696,
|
||
|
|
"learning_rate": 9.683495069313527e-06,
|
||
|
|
"loss": 0.1371,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05347670614719391,
|
||
|
|
"step": 2145,
|
||
|
|
"valid_targets_mean": 5853.2,
|
||
|
|
"valid_targets_min": 1305
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5364783324190894,
|
||
|
|
"grad_norm": 0.19681955685226582,
|
||
|
|
"learning_rate": 9.585296370185875e-06,
|
||
|
|
"loss": 0.1309,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04260285198688507,
|
||
|
|
"step": 2150,
|
||
|
|
"valid_targets_mean": 5323.5,
|
||
|
|
"valid_targets_min": 2609
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5447065277015906,
|
||
|
|
"grad_norm": 0.18774837820824186,
|
||
|
|
"learning_rate": 9.487440954843856e-06,
|
||
|
|
"loss": 0.1299,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04370295628905296,
|
||
|
|
"step": 2155,
|
||
|
|
"valid_targets_mean": 5320.9,
|
||
|
|
"valid_targets_min": 3076
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.552934722984092,
|
||
|
|
"grad_norm": 0.17697030654769377,
|
||
|
|
"learning_rate": 9.38993204874436e-06,
|
||
|
|
"loss": 0.1282,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.03998667374253273,
|
||
|
|
"step": 2160,
|
||
|
|
"valid_targets_mean": 5294.3,
|
||
|
|
"valid_targets_min": 1502
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5611629182665934,
|
||
|
|
"grad_norm": 0.17674626038431754,
|
||
|
|
"learning_rate": 9.292772865922792e-06,
|
||
|
|
"loss": 0.1314,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04547334834933281,
|
||
|
|
"step": 2165,
|
||
|
|
"valid_targets_mean": 5738.7,
|
||
|
|
"valid_targets_min": 3508
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.569391113549095,
|
||
|
|
"grad_norm": 0.18303379919738216,
|
||
|
|
"learning_rate": 9.195966608887212e-06,
|
||
|
|
"loss": 0.1306,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.040250808000564575,
|
||
|
|
"step": 2170,
|
||
|
|
"valid_targets_mean": 5052.5,
|
||
|
|
"valid_targets_min": 2487
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.577619308831596,
|
||
|
|
"grad_norm": 0.18797337361410427,
|
||
|
|
"learning_rate": 9.099516468512692e-06,
|
||
|
|
"loss": 0.1241,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04151705279946327,
|
||
|
|
"step": 2175,
|
||
|
|
"valid_targets_mean": 5108.5,
|
||
|
|
"valid_targets_min": 3236
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.585847504114098,
|
||
|
|
"grad_norm": 0.17156816751481446,
|
||
|
|
"learning_rate": 9.003425623936208e-06,
|
||
|
|
"loss": 0.1262,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.040634635835886,
|
||
|
|
"step": 2180,
|
||
|
|
"valid_targets_mean": 5397.8,
|
||
|
|
"valid_targets_min": 2803
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.594075699396599,
|
||
|
|
"grad_norm": 0.18585953552751092,
|
||
|
|
"learning_rate": 8.907697242451825e-06,
|
||
|
|
"loss": 0.1277,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.03988367319107056,
|
||
|
|
"step": 2185,
|
||
|
|
"valid_targets_mean": 5513.7,
|
||
|
|
"valid_targets_min": 2867
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6023038946791006,
|
||
|
|
"grad_norm": 0.18577716318480716,
|
||
|
|
"learning_rate": 8.812334479406266e-06,
|
||
|
|
"loss": 0.128,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0442671924829483,
|
||
|
|
"step": 2190,
|
||
|
|
"valid_targets_mean": 5217.7,
|
||
|
|
"valid_targets_min": 2433
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.610532089961602,
|
||
|
|
"grad_norm": 0.19363655736757662,
|
||
|
|
"learning_rate": 8.71734047809498e-06,
|
||
|
|
"loss": 0.1242,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.040493179112672806,
|
||
|
|
"step": 2195,
|
||
|
|
"valid_targets_mean": 5169.1,
|
||
|
|
"valid_targets_min": 2789
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.618760285244103,
|
||
|
|
"grad_norm": 0.16716286573436492,
|
||
|
|
"learning_rate": 8.62271836965846e-06,
|
||
|
|
"loss": 0.1258,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04132536053657532,
|
||
|
|
"step": 2200,
|
||
|
|
"valid_targets_mean": 5809.1,
|
||
|
|
"valid_targets_min": 2841
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6269884805266046,
|
||
|
|
"grad_norm": 0.18637501105182114,
|
||
|
|
"learning_rate": 8.528471272979083e-06,
|
||
|
|
"loss": 0.128,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04350794851779938,
|
||
|
|
"step": 2205,
|
||
|
|
"valid_targets_mean": 5502.1,
|
||
|
|
"valid_targets_min": 2911
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.635216675809106,
|
||
|
|
"grad_norm": 0.19484985192773233,
|
||
|
|
"learning_rate": 8.434602294578285e-06,
|
||
|
|
"loss": 0.123,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04264716058969498,
|
||
|
|
"step": 2210,
|
||
|
|
"valid_targets_mean": 5274.9,
|
||
|
|
"valid_targets_min": 2672
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6434448710916074,
|
||
|
|
"grad_norm": 0.18319091175169402,
|
||
|
|
"learning_rate": 8.341114528514192e-06,
|
||
|
|
"loss": 0.1262,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04070258140563965,
|
||
|
|
"step": 2215,
|
||
|
|
"valid_targets_mean": 4966.8,
|
||
|
|
"valid_targets_min": 2931
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6516730663741086,
|
||
|
|
"grad_norm": 0.18406967880449163,
|
||
|
|
"learning_rate": 8.248011056279588e-06,
|
||
|
|
"loss": 0.1269,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04556259140372276,
|
||
|
|
"step": 2220,
|
||
|
|
"valid_targets_mean": 5898.0,
|
||
|
|
"valid_targets_min": 3337
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.65990126165661,
|
||
|
|
"grad_norm": 0.19470458062154897,
|
||
|
|
"learning_rate": 8.155294946700402e-06,
|
||
|
|
"loss": 0.123,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.043193720281124115,
|
||
|
|
"step": 2225,
|
||
|
|
"valid_targets_mean": 5214.2,
|
||
|
|
"valid_targets_min": 2978
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6681294569391114,
|
||
|
|
"grad_norm": 0.25333892509494665,
|
||
|
|
"learning_rate": 8.062969255834505e-06,
|
||
|
|
"loss": 0.1339,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09117273986339569,
|
||
|
|
"step": 2230,
|
||
|
|
"valid_targets_mean": 3851.8,
|
||
|
|
"valid_targets_min": 887
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6763576522216126,
|
||
|
|
"grad_norm": 0.3855404264715868,
|
||
|
|
"learning_rate": 7.971037026871016e-06,
|
||
|
|
"loss": 0.4519,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1366964727640152,
|
||
|
|
"step": 2235,
|
||
|
|
"valid_targets_mean": 4596.2,
|
||
|
|
"valid_targets_min": 1143
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.684585847504114,
|
||
|
|
"grad_norm": 0.2895125132387707,
|
||
|
|
"learning_rate": 7.879501290029954e-06,
|
||
|
|
"loss": 0.4387,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.136347234249115,
|
||
|
|
"step": 2240,
|
||
|
|
"valid_targets_mean": 4087.0,
|
||
|
|
"valid_targets_min": 1058
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6928140427866154,
|
||
|
|
"grad_norm": 0.32375236588358164,
|
||
|
|
"learning_rate": 7.788365062462411e-06,
|
||
|
|
"loss": 0.4199,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1273779571056366,
|
||
|
|
"step": 2245,
|
||
|
|
"valid_targets_mean": 3146.8,
|
||
|
|
"valid_targets_min": 960
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7010422380691166,
|
||
|
|
"grad_norm": 0.2765774016675542,
|
||
|
|
"learning_rate": 7.697631348151048e-06,
|
||
|
|
"loss": 0.419,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14948008954524994,
|
||
|
|
"step": 2250,
|
||
|
|
"valid_targets_mean": 4243.4,
|
||
|
|
"valid_targets_min": 890
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.709270433351618,
|
||
|
|
"grad_norm": 0.2405231497548994,
|
||
|
|
"learning_rate": 7.607303137811108e-06,
|
||
|
|
"loss": 0.3877,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10435382276773453,
|
||
|
|
"step": 2255,
|
||
|
|
"valid_targets_mean": 4787.6,
|
||
|
|
"valid_targets_min": 989
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.71749862863412,
|
||
|
|
"grad_norm": 0.28959377843811446,
|
||
|
|
"learning_rate": 7.517383408791847e-06,
|
||
|
|
"loss": 0.4175,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1839148998260498,
|
||
|
|
"step": 2260,
|
||
|
|
"valid_targets_mean": 4654.7,
|
||
|
|
"valid_targets_min": 385
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.725726823916621,
|
||
|
|
"grad_norm": 0.27304162379886376,
|
||
|
|
"learning_rate": 7.427875124978359e-06,
|
||
|
|
"loss": 0.3918,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13247811794281006,
|
||
|
|
"step": 2265,
|
||
|
|
"valid_targets_mean": 3702.7,
|
||
|
|
"valid_targets_min": 1391
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.733955019199122,
|
||
|
|
"grad_norm": 0.30525674061003377,
|
||
|
|
"learning_rate": 7.33878123669393e-06,
|
||
|
|
"loss": 0.3865,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13155576586723328,
|
||
|
|
"step": 2270,
|
||
|
|
"valid_targets_mean": 3575.6,
|
||
|
|
"valid_targets_min": 1115
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.742183214481624,
|
||
|
|
"grad_norm": 0.23940187715657524,
|
||
|
|
"learning_rate": 7.2501046806027456e-06,
|
||
|
|
"loss": 0.41,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11278781294822693,
|
||
|
|
"step": 2275,
|
||
|
|
"valid_targets_mean": 3883.8,
|
||
|
|
"valid_targets_min": 1256
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.750411409764125,
|
||
|
|
"grad_norm": 0.26157914151130646,
|
||
|
|
"learning_rate": 7.161848379613134e-06,
|
||
|
|
"loss": 0.394,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11556685715913773,
|
||
|
|
"step": 2280,
|
||
|
|
"valid_targets_mean": 4064.0,
|
||
|
|
"valid_targets_min": 1346
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7586396050466266,
|
||
|
|
"grad_norm": 0.2754096668324058,
|
||
|
|
"learning_rate": 7.074015242781181e-06,
|
||
|
|
"loss": 0.4103,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.136755108833313,
|
||
|
|
"step": 2285,
|
||
|
|
"valid_targets_mean": 3096.2,
|
||
|
|
"valid_targets_min": 695
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.766867800329128,
|
||
|
|
"grad_norm": 0.2673295797880238,
|
||
|
|
"learning_rate": 6.986608165214892e-06,
|
||
|
|
"loss": 0.3874,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13044080138206482,
|
||
|
|
"step": 2290,
|
||
|
|
"valid_targets_mean": 3573.1,
|
||
|
|
"valid_targets_min": 1272
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.775095995611629,
|
||
|
|
"grad_norm": 0.24565165248996393,
|
||
|
|
"learning_rate": 6.899630027978717e-06,
|
||
|
|
"loss": 0.3862,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11896165460348129,
|
||
|
|
"step": 2295,
|
||
|
|
"valid_targets_mean": 4696.2,
|
||
|
|
"valid_targets_min": 939
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7833241908941306,
|
||
|
|
"grad_norm": 0.26076112407188706,
|
||
|
|
"learning_rate": 6.8130836979986236e-06,
|
||
|
|
"loss": 0.4288,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1774311065673828,
|
||
|
|
"step": 2300,
|
||
|
|
"valid_targets_mean": 4314.5,
|
||
|
|
"valid_targets_min": 1135
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.791552386176632,
|
||
|
|
"grad_norm": 0.26624764344340895,
|
||
|
|
"learning_rate": 6.7269720279675755e-06,
|
||
|
|
"loss": 0.3908,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1311570405960083,
|
||
|
|
"step": 2305,
|
||
|
|
"valid_targets_mean": 3551.0,
|
||
|
|
"valid_targets_min": 909
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7997805814591334,
|
||
|
|
"grad_norm": 0.27025205393210866,
|
||
|
|
"learning_rate": 6.641297856251514e-06,
|
||
|
|
"loss": 0.4191,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.134062260389328,
|
||
|
|
"step": 2310,
|
||
|
|
"valid_targets_mean": 3728.1,
|
||
|
|
"valid_targets_min": 970
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8080087767416346,
|
||
|
|
"grad_norm": 0.24904092712840964,
|
||
|
|
"learning_rate": 6.556064006795795e-06,
|
||
|
|
"loss": 0.3918,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1097637414932251,
|
||
|
|
"step": 2315,
|
||
|
|
"valid_targets_mean": 3462.3,
|
||
|
|
"valid_targets_min": 1420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.816236972024136,
|
||
|
|
"grad_norm": 0.24275402888828918,
|
||
|
|
"learning_rate": 6.471273289032125e-06,
|
||
|
|
"loss": 0.4029,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1297762244939804,
|
||
|
|
"step": 2320,
|
||
|
|
"valid_targets_mean": 4587.7,
|
||
|
|
"valid_targets_min": 1133
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8244651673066374,
|
||
|
|
"grad_norm": 0.2654856935959466,
|
||
|
|
"learning_rate": 6.386928497785929e-06,
|
||
|
|
"loss": 0.3892,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1179320216178894,
|
||
|
|
"step": 2325,
|
||
|
|
"valid_targets_mean": 3638.8,
|
||
|
|
"valid_targets_min": 1166
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8326933625891386,
|
||
|
|
"grad_norm": 0.2648669254710519,
|
||
|
|
"learning_rate": 6.303032413184256e-06,
|
||
|
|
"loss": 0.4142,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13473305106163025,
|
||
|
|
"step": 2330,
|
||
|
|
"valid_targets_mean": 4151.9,
|
||
|
|
"valid_targets_min": 1404
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8409215578716402,
|
||
|
|
"grad_norm": 0.25274953101806685,
|
||
|
|
"learning_rate": 6.219587800564135e-06,
|
||
|
|
"loss": 0.3908,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07814621180295944,
|
||
|
|
"step": 2335,
|
||
|
|
"valid_targets_mean": 1979.0,
|
||
|
|
"valid_targets_min": 506
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8491497531541414,
|
||
|
|
"grad_norm": 0.2799408422446661,
|
||
|
|
"learning_rate": 6.136597410381404e-06,
|
||
|
|
"loss": 0.2928,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0938805416226387,
|
||
|
|
"step": 2340,
|
||
|
|
"valid_targets_mean": 6913.1,
|
||
|
|
"valid_targets_min": 817
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.857377948436643,
|
||
|
|
"grad_norm": 0.2128799029136465,
|
||
|
|
"learning_rate": 6.054063978120093e-06,
|
||
|
|
"loss": 0.2429,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08730052411556244,
|
||
|
|
"step": 2345,
|
||
|
|
"valid_targets_mean": 6177.3,
|
||
|
|
"valid_targets_min": 509
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8656061437191442,
|
||
|
|
"grad_norm": 0.20296058207505324,
|
||
|
|
"learning_rate": 5.971990224202209e-06,
|
||
|
|
"loss": 0.2501,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08442901074886322,
|
||
|
|
"step": 2350,
|
||
|
|
"valid_targets_mean": 6737.9,
|
||
|
|
"valid_targets_min": 1241
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.873834339001646,
|
||
|
|
"grad_norm": 0.17908986617598158,
|
||
|
|
"learning_rate": 5.890378853898106e-06,
|
||
|
|
"loss": 0.2559,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09330446273088455,
|
||
|
|
"step": 2355,
|
||
|
|
"valid_targets_mean": 7625.8,
|
||
|
|
"valid_targets_min": 597
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.882062534284147,
|
||
|
|
"grad_norm": 0.24834625855840978,
|
||
|
|
"learning_rate": 5.809232557237292e-06,
|
||
|
|
"loss": 0.2583,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1216353327035904,
|
||
|
|
"step": 2360,
|
||
|
|
"valid_targets_mean": 6605.6,
|
||
|
|
"valid_targets_min": 335
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8902907295666482,
|
||
|
|
"grad_norm": 0.17101902840906216,
|
||
|
|
"learning_rate": 5.728554008919794e-06,
|
||
|
|
"loss": 0.2444,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08080480992794037,
|
||
|
|
"step": 2365,
|
||
|
|
"valid_targets_mean": 6378.9,
|
||
|
|
"valid_targets_min": 281
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.89851892484915,
|
||
|
|
"grad_norm": 0.17341671996138142,
|
||
|
|
"learning_rate": 5.6483458682279354e-06,
|
||
|
|
"loss": 0.2398,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06936435401439667,
|
||
|
|
"step": 2370,
|
||
|
|
"valid_targets_mean": 5468.6,
|
||
|
|
"valid_targets_min": 543
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.906747120131651,
|
||
|
|
"grad_norm": 0.16456626810742647,
|
||
|
|
"learning_rate": 5.568610778938761e-06,
|
||
|
|
"loss": 0.2386,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07711958140134811,
|
||
|
|
"step": 2375,
|
||
|
|
"valid_targets_mean": 6290.5,
|
||
|
|
"valid_targets_min": 301
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9149753154141527,
|
||
|
|
"grad_norm": 0.1695584593994994,
|
||
|
|
"learning_rate": 5.489351369236817e-06,
|
||
|
|
"loss": 0.241,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05445744842290878,
|
||
|
|
"step": 2380,
|
||
|
|
"valid_targets_mean": 4845.6,
|
||
|
|
"valid_targets_min": 401
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.923203510696654,
|
||
|
|
"grad_norm": 0.1800936264070191,
|
||
|
|
"learning_rate": 5.410570251627587e-06,
|
||
|
|
"loss": 0.255,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08194078505039215,
|
||
|
|
"step": 2385,
|
||
|
|
"valid_targets_mean": 6736.4,
|
||
|
|
"valid_targets_min": 134
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.931431705979155,
|
||
|
|
"grad_norm": 0.174046744890005,
|
||
|
|
"learning_rate": 5.332270022851327e-06,
|
||
|
|
"loss": 0.2462,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09117108583450317,
|
||
|
|
"step": 2390,
|
||
|
|
"valid_targets_mean": 7333.2,
|
||
|
|
"valid_targets_min": 1427
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9396599012616567,
|
||
|
|
"grad_norm": 0.2458671161035671,
|
||
|
|
"learning_rate": 5.254453263797521e-06,
|
||
|
|
"loss": 0.2531,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04853862524032593,
|
||
|
|
"step": 2395,
|
||
|
|
"valid_targets_mean": 1272.8,
|
||
|
|
"valid_targets_min": 374
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.947888096544158,
|
||
|
|
"grad_norm": 0.26767848712622627,
|
||
|
|
"learning_rate": 5.177122539419763e-06,
|
||
|
|
"loss": 0.2445,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0782180204987526,
|
||
|
|
"step": 2400,
|
||
|
|
"valid_targets_mean": 6596.2,
|
||
|
|
"valid_targets_min": 491
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9561162918266595,
|
||
|
|
"grad_norm": 0.16636456794257964,
|
||
|
|
"learning_rate": 5.10028039865126e-06,
|
||
|
|
"loss": 0.2243,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05681359022855759,
|
||
|
|
"step": 2405,
|
||
|
|
"valid_targets_mean": 5984.0,
|
||
|
|
"valid_targets_min": 350
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9643444871091607,
|
||
|
|
"grad_norm": 0.1766915803923583,
|
||
|
|
"learning_rate": 5.023929374320779e-06,
|
||
|
|
"loss": 0.2485,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08463890850543976,
|
||
|
|
"step": 2410,
|
||
|
|
"valid_targets_mean": 6808.1,
|
||
|
|
"valid_targets_min": 367
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.972572682391662,
|
||
|
|
"grad_norm": 0.18083090564134296,
|
||
|
|
"learning_rate": 4.948071983069167e-06,
|
||
|
|
"loss": 0.2359,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08655136823654175,
|
||
|
|
"step": 2415,
|
||
|
|
"valid_targets_mean": 6776.0,
|
||
|
|
"valid_targets_min": 475
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9808008776741635,
|
||
|
|
"grad_norm": 0.2150095285524866,
|
||
|
|
"learning_rate": 4.8727107252664315e-06,
|
||
|
|
"loss": 0.2271,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10201592743396759,
|
||
|
|
"step": 2420,
|
||
|
|
"valid_targets_mean": 7100.5,
|
||
|
|
"valid_targets_min": 370
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.989029072956665,
|
||
|
|
"grad_norm": 0.1714470668143074,
|
||
|
|
"learning_rate": 4.797848084929271e-06,
|
||
|
|
"loss": 0.2384,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06286249309778214,
|
||
|
|
"step": 2425,
|
||
|
|
"valid_targets_mean": 5628.6,
|
||
|
|
"valid_targets_min": 503
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9972572682391663,
|
||
|
|
"grad_norm": 0.16916735134740704,
|
||
|
|
"learning_rate": 4.723486529639252e-06,
|
||
|
|
"loss": 0.2387,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08586841821670532,
|
||
|
|
"step": 2430,
|
||
|
|
"valid_targets_mean": 6859.2,
|
||
|
|
"valid_targets_min": 578
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.004936917169501,
|
||
|
|
"grad_norm": 2.336956202835139,
|
||
|
|
"learning_rate": 4.649628510461428e-06,
|
||
|
|
"loss": 0.514,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21151629090309143,
|
||
|
|
"step": 2435,
|
||
|
|
"valid_targets_mean": 7854.0,
|
||
|
|
"valid_targets_min": 2676
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.013165112452002,
|
||
|
|
"grad_norm": 1.41923931878634,
|
||
|
|
"learning_rate": 4.576276461863589e-06,
|
||
|
|
"loss": 0.5277,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15032610297203064,
|
||
|
|
"step": 2440,
|
||
|
|
"valid_targets_mean": 6740.2,
|
||
|
|
"valid_targets_min": 3017
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.021393307734503,
|
||
|
|
"grad_norm": 0.8500945954778326,
|
||
|
|
"learning_rate": 4.503432801635976e-06,
|
||
|
|
"loss": 0.46,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13372503221035004,
|
||
|
|
"step": 2445,
|
||
|
|
"valid_targets_mean": 7294.2,
|
||
|
|
"valid_targets_min": 3200
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.029621503017005,
|
||
|
|
"grad_norm": 0.6004641788267427,
|
||
|
|
"learning_rate": 4.431099930811633e-06,
|
||
|
|
"loss": 0.4076,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14208507537841797,
|
||
|
|
"step": 2450,
|
||
|
|
"valid_targets_mean": 7212.2,
|
||
|
|
"valid_targets_min": 2554
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.037849698299507,
|
||
|
|
"grad_norm": 0.46434648019740743,
|
||
|
|
"learning_rate": 4.359280233587229e-06,
|
||
|
|
"loss": 0.3746,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12539725005626678,
|
||
|
|
"step": 2455,
|
||
|
|
"valid_targets_mean": 7016.0,
|
||
|
|
"valid_targets_min": 2429
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.046077893582008,
|
||
|
|
"grad_norm": 0.3135597964334871,
|
||
|
|
"learning_rate": 4.28797607724448e-06,
|
||
|
|
"loss": 0.3607,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11195516586303711,
|
||
|
|
"step": 2460,
|
||
|
|
"valid_targets_mean": 7897.7,
|
||
|
|
"valid_targets_min": 2091
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.054306088864509,
|
||
|
|
"grad_norm": 0.26939488289825964,
|
||
|
|
"learning_rate": 4.217189812072131e-06,
|
||
|
|
"loss": 0.3497,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12193445861339569,
|
||
|
|
"step": 2465,
|
||
|
|
"valid_targets_mean": 7362.4,
|
||
|
|
"valid_targets_min": 1674
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.06253428414701,
|
||
|
|
"grad_norm": 0.26655830700455374,
|
||
|
|
"learning_rate": 4.146923771288489e-06,
|
||
|
|
"loss": 0.336,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11060521006584167,
|
||
|
|
"step": 2470,
|
||
|
|
"valid_targets_mean": 6896.6,
|
||
|
|
"valid_targets_min": 2522
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.070762479429511,
|
||
|
|
"grad_norm": 0.21602585425107565,
|
||
|
|
"learning_rate": 4.077180270964487e-06,
|
||
|
|
"loss": 0.3354,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10641120374202728,
|
||
|
|
"step": 2475,
|
||
|
|
"valid_targets_mean": 7470.7,
|
||
|
|
"valid_targets_min": 2790
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.0789906747120135,
|
||
|
|
"grad_norm": 0.19078412445817078,
|
||
|
|
"learning_rate": 4.007961609947391e-06,
|
||
|
|
"loss": 0.3356,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11443803459405899,
|
||
|
|
"step": 2480,
|
||
|
|
"valid_targets_mean": 8266.8,
|
||
|
|
"valid_targets_min": 3307
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.087218869994515,
|
||
|
|
"grad_norm": 0.2034314805463107,
|
||
|
|
"learning_rate": 3.93927006978497e-06,
|
||
|
|
"loss": 0.3246,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09000586718320847,
|
||
|
|
"step": 2485,
|
||
|
|
"valid_targets_mean": 8286.0,
|
||
|
|
"valid_targets_min": 325
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.095447065277016,
|
||
|
|
"grad_norm": 0.1845196550420189,
|
||
|
|
"learning_rate": 3.8711079146503474e-06,
|
||
|
|
"loss": 0.3023,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10727182030677795,
|
||
|
|
"step": 2490,
|
||
|
|
"valid_targets_mean": 9267.8,
|
||
|
|
"valid_targets_min": 5215
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.103675260559517,
|
||
|
|
"grad_norm": 0.17034619427285855,
|
||
|
|
"learning_rate": 3.8034773912673383e-06,
|
||
|
|
"loss": 0.2955,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10079951584339142,
|
||
|
|
"step": 2495,
|
||
|
|
"valid_targets_mean": 9206.4,
|
||
|
|
"valid_targets_min": 3552
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.111903455842018,
|
||
|
|
"grad_norm": 0.16544617657311927,
|
||
|
|
"learning_rate": 3.736380728836393e-06,
|
||
|
|
"loss": 0.2899,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1036929041147232,
|
||
|
|
"step": 2500,
|
||
|
|
"valid_targets_mean": 9805.0,
|
||
|
|
"valid_targets_min": 435
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.12013165112452,
|
||
|
|
"grad_norm": 0.1675109415038507,
|
||
|
|
"learning_rate": 3.6698201389611423e-06,
|
||
|
|
"loss": 0.2945,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10505664348602295,
|
||
|
|
"step": 2505,
|
||
|
|
"valid_targets_mean": 9971.8,
|
||
|
|
"valid_targets_min": 3896
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.1283598464070215,
|
||
|
|
"grad_norm": 0.17732216459505157,
|
||
|
|
"learning_rate": 3.6037978155754737e-06,
|
||
|
|
"loss": 0.2894,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09879560768604279,
|
||
|
|
"step": 2510,
|
||
|
|
"valid_targets_mean": 9100.1,
|
||
|
|
"valid_targets_min": 3598
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.136588041689523,
|
||
|
|
"grad_norm": 0.17652461132533978,
|
||
|
|
"learning_rate": 3.53831593487123e-06,
|
||
|
|
"loss": 0.2923,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09788352251052856,
|
||
|
|
"step": 2515,
|
||
|
|
"valid_targets_mean": 9095.8,
|
||
|
|
"valid_targets_min": 3870
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.144816236972024,
|
||
|
|
"grad_norm": 0.1751708078026008,
|
||
|
|
"learning_rate": 3.473376655226479e-06,
|
||
|
|
"loss": 0.2952,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09423353523015976,
|
||
|
|
"step": 2520,
|
||
|
|
"valid_targets_mean": 8719.1,
|
||
|
|
"valid_targets_min": 3803
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.153044432254526,
|
||
|
|
"grad_norm": 0.1617553991074427,
|
||
|
|
"learning_rate": 3.408982117134374e-06,
|
||
|
|
"loss": 0.2892,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10206412523984909,
|
||
|
|
"step": 2525,
|
||
|
|
"valid_targets_mean": 9952.8,
|
||
|
|
"valid_targets_min": 703
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.161272627537027,
|
||
|
|
"grad_norm": 0.17010544092490168,
|
||
|
|
"learning_rate": 3.3451344431325806e-06,
|
||
|
|
"loss": 0.292,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09758783876895905,
|
||
|
|
"step": 2530,
|
||
|
|
"valid_targets_mean": 9790.2,
|
||
|
|
"valid_targets_min": 1990
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.169500822819528,
|
||
|
|
"grad_norm": 0.1759410398694626,
|
||
|
|
"learning_rate": 3.2818357377333455e-06,
|
||
|
|
"loss": 0.291,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09575256705284119,
|
||
|
|
"step": 2535,
|
||
|
|
"valid_targets_mean": 9145.0,
|
||
|
|
"valid_targets_min": 3036
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.1777290181020295,
|
||
|
|
"grad_norm": 0.16216951102404267,
|
||
|
|
"learning_rate": 3.219088087354092e-06,
|
||
|
|
"loss": 0.2892,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08874919265508652,
|
||
|
|
"step": 2540,
|
||
|
|
"valid_targets_mean": 9393.3,
|
||
|
|
"valid_targets_min": 749
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.185957213384531,
|
||
|
|
"grad_norm": 0.1633207914795439,
|
||
|
|
"learning_rate": 3.156893560248688e-06,
|
||
|
|
"loss": 0.2885,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08750354498624802,
|
||
|
|
"step": 2545,
|
||
|
|
"valid_targets_mean": 8753.6,
|
||
|
|
"valid_targets_min": 721
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.194185408667033,
|
||
|
|
"grad_norm": 0.17268830072169394,
|
||
|
|
"learning_rate": 3.095254206439233e-06,
|
||
|
|
"loss": 0.2904,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09792269021272659,
|
||
|
|
"step": 2550,
|
||
|
|
"valid_targets_mean": 9578.5,
|
||
|
|
"valid_targets_min": 2306
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.202413603949534,
|
||
|
|
"grad_norm": 0.1631186432521016,
|
||
|
|
"learning_rate": 3.0341720576485277e-06,
|
||
|
|
"loss": 0.2959,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09698028117418289,
|
||
|
|
"step": 2555,
|
||
|
|
"valid_targets_mean": 9520.8,
|
||
|
|
"valid_targets_min": 4270
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.210641799232035,
|
||
|
|
"grad_norm": 0.1635172649108156,
|
||
|
|
"learning_rate": 2.9736491272330694e-06,
|
||
|
|
"loss": 0.2955,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10328608751296997,
|
||
|
|
"step": 2560,
|
||
|
|
"valid_targets_mean": 10044.1,
|
||
|
|
"valid_targets_min": 5057
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.218869994514536,
|
||
|
|
"grad_norm": 0.1588271104676105,
|
||
|
|
"learning_rate": 2.9136874101167034e-06,
|
||
|
|
"loss": 0.2842,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09239079058170319,
|
||
|
|
"step": 2565,
|
||
|
|
"valid_targets_mean": 9430.1,
|
||
|
|
"valid_targets_min": 4478
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.227098189797037,
|
||
|
|
"grad_norm": 0.15675985150107885,
|
||
|
|
"learning_rate": 2.854288882724885e-06,
|
||
|
|
"loss": 0.287,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09478726983070374,
|
||
|
|
"step": 2570,
|
||
|
|
"valid_targets_mean": 9906.8,
|
||
|
|
"valid_targets_min": 4602
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.2353263850795395,
|
||
|
|
"grad_norm": 0.15893873290014987,
|
||
|
|
"learning_rate": 2.795455502919493e-06,
|
||
|
|
"loss": 0.2877,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09879006445407867,
|
||
|
|
"step": 2575,
|
||
|
|
"valid_targets_mean": 9816.9,
|
||
|
|
"valid_targets_min": 600
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.243554580362041,
|
||
|
|
"grad_norm": 0.16008381725374082,
|
||
|
|
"learning_rate": 2.7371892099343455e-06,
|
||
|
|
"loss": 0.2942,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09585876762866974,
|
||
|
|
"step": 2580,
|
||
|
|
"valid_targets_mean": 9615.7,
|
||
|
|
"valid_targets_min": 321
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.251782775644542,
|
||
|
|
"grad_norm": 0.16240540291320715,
|
||
|
|
"learning_rate": 2.679491924311226e-06,
|
||
|
|
"loss": 0.2837,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09075018763542175,
|
||
|
|
"step": 2585,
|
||
|
|
"valid_targets_mean": 9669.4,
|
||
|
|
"valid_targets_min": 4373
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.260010970927043,
|
||
|
|
"grad_norm": 0.16123315756596274,
|
||
|
|
"learning_rate": 2.622365547836636e-06,
|
||
|
|
"loss": 0.291,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10122531652450562,
|
||
|
|
"step": 2590,
|
||
|
|
"valid_targets_mean": 10030.3,
|
||
|
|
"valid_targets_min": 4760
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.268239166209545,
|
||
|
|
"grad_norm": 0.1614050818709832,
|
||
|
|
"learning_rate": 2.5658119634790526e-06,
|
||
|
|
"loss": 0.285,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09613090753555298,
|
||
|
|
"step": 2595,
|
||
|
|
"valid_targets_mean": 9881.2,
|
||
|
|
"valid_targets_min": 4785
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.276467361492046,
|
||
|
|
"grad_norm": 0.16136146793987857,
|
||
|
|
"learning_rate": 2.5098330353269164e-06,
|
||
|
|
"loss": 0.2898,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09504969418048859,
|
||
|
|
"step": 2600,
|
||
|
|
"valid_targets_mean": 9669.7,
|
||
|
|
"valid_targets_min": 3932
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.2846955567745475,
|
||
|
|
"grad_norm": 0.16203287980747294,
|
||
|
|
"learning_rate": 2.4544306085271406e-06,
|
||
|
|
"loss": 0.2843,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09451472014188766,
|
||
|
|
"step": 2605,
|
||
|
|
"valid_targets_mean": 9425.6,
|
||
|
|
"valid_targets_min": 3063
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.292923752057049,
|
||
|
|
"grad_norm": 0.1664226870010622,
|
||
|
|
"learning_rate": 2.399606509224337e-06,
|
||
|
|
"loss": 0.2875,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10325511544942856,
|
||
|
|
"step": 2610,
|
||
|
|
"valid_targets_mean": 9942.2,
|
||
|
|
"valid_targets_min": 3468
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.30115194733955,
|
||
|
|
"grad_norm": 0.16319711424290925,
|
||
|
|
"learning_rate": 2.345362544500589e-06,
|
||
|
|
"loss": 0.2802,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09199801087379456,
|
||
|
|
"step": 2615,
|
||
|
|
"valid_targets_mean": 9693.8,
|
||
|
|
"valid_targets_min": 562
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.309380142622052,
|
||
|
|
"grad_norm": 0.16661531597560844,
|
||
|
|
"learning_rate": 2.2917005023158966e-06,
|
||
|
|
"loss": 0.285,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10052244365215302,
|
||
|
|
"step": 2620,
|
||
|
|
"valid_targets_mean": 10011.0,
|
||
|
|
"valid_targets_min": 3774
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.317608337904553,
|
||
|
|
"grad_norm": 0.18510374807764027,
|
||
|
|
"learning_rate": 2.2386221514492502e-06,
|
||
|
|
"loss": 0.2862,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06400661915540695,
|
||
|
|
"step": 2625,
|
||
|
|
"valid_targets_mean": 5209.4,
|
||
|
|
"valid_targets_min": 484
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.325836533187054,
|
||
|
|
"grad_norm": 0.22137356711405826,
|
||
|
|
"learning_rate": 2.186129241440336e-06,
|
||
|
|
"loss": 0.2915,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09740819036960602,
|
||
|
|
"step": 2630,
|
||
|
|
"valid_targets_mean": 6386.3,
|
||
|
|
"valid_targets_min": 3796
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.3340647284695555,
|
||
|
|
"grad_norm": 0.23070165597543357,
|
||
|
|
"learning_rate": 2.134223502531838e-06,
|
||
|
|
"loss": 0.2902,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10796046257019043,
|
||
|
|
"step": 2635,
|
||
|
|
"valid_targets_mean": 6407.3,
|
||
|
|
"valid_targets_min": 2561
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.342292923752057,
|
||
|
|
"grad_norm": 1.634978861513987,
|
||
|
|
"learning_rate": 2.0829066456124415e-06,
|
||
|
|
"loss": 0.6808,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21645402908325195,
|
||
|
|
"step": 2640,
|
||
|
|
"valid_targets_mean": 4497.1,
|
||
|
|
"valid_targets_min": 1348
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.350521119034559,
|
||
|
|
"grad_norm": 1.2882498753330385,
|
||
|
|
"learning_rate": 2.032180362160423e-06,
|
||
|
|
"loss": 0.6324,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17659324407577515,
|
||
|
|
"step": 2645,
|
||
|
|
"valid_targets_mean": 4428.4,
|
||
|
|
"valid_targets_min": 1031
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.35874931431706,
|
||
|
|
"grad_norm": 1.043040981513531,
|
||
|
|
"learning_rate": 1.9820463241878873e-06,
|
||
|
|
"loss": 0.62,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22026868164539337,
|
||
|
|
"step": 2650,
|
||
|
|
"valid_targets_mean": 3730.1,
|
||
|
|
"valid_targets_min": 827
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.366977509599561,
|
||
|
|
"grad_norm": 0.6949809377584997,
|
||
|
|
"learning_rate": 1.9325061841856808e-06,
|
||
|
|
"loss": 0.5603,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21149209141731262,
|
||
|
|
"step": 2655,
|
||
|
|
"valid_targets_mean": 4497.8,
|
||
|
|
"valid_targets_min": 1422
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.375205704882062,
|
||
|
|
"grad_norm": 0.5381576597915756,
|
||
|
|
"learning_rate": 1.8835615750688997e-06,
|
||
|
|
"loss": 0.5239,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14553728699684143,
|
||
|
|
"step": 2660,
|
||
|
|
"valid_targets_mean": 2116.2,
|
||
|
|
"valid_targets_min": 785
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.3834339001645635,
|
||
|
|
"grad_norm": 0.3521266371539034,
|
||
|
|
"learning_rate": 1.8352141101230758e-06,
|
||
|
|
"loss": 0.5108,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15841825306415558,
|
||
|
|
"step": 2665,
|
||
|
|
"valid_targets_mean": 4756.1,
|
||
|
|
"valid_targets_min": 1301
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.3916620954470655,
|
||
|
|
"grad_norm": 0.2892377482447778,
|
||
|
|
"learning_rate": 1.787465382950999e-06,
|
||
|
|
"loss": 0.4972,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15879112482070923,
|
||
|
|
"step": 2670,
|
||
|
|
"valid_targets_mean": 4760.8,
|
||
|
|
"valid_targets_min": 758
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.399890290729567,
|
||
|
|
"grad_norm": 0.31603860554853436,
|
||
|
|
"learning_rate": 1.7403169674202036e-06,
|
||
|
|
"loss": 0.5039,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16973236203193665,
|
||
|
|
"step": 2675,
|
||
|
|
"valid_targets_mean": 3936.6,
|
||
|
|
"valid_targets_min": 1412
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.408118486012068,
|
||
|
|
"grad_norm": 0.31041708575716764,
|
||
|
|
"learning_rate": 1.6937704176110582e-06,
|
||
|
|
"loss": 0.4889,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.175504669547081,
|
||
|
|
"step": 2680,
|
||
|
|
"valid_targets_mean": 4049.9,
|
||
|
|
"valid_targets_min": 928
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.416346681294569,
|
||
|
|
"grad_norm": 0.31096923174048335,
|
||
|
|
"learning_rate": 1.6478272677655804e-06,
|
||
|
|
"loss": 0.4648,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1734139621257782,
|
||
|
|
"step": 2685,
|
||
|
|
"valid_targets_mean": 3266.9,
|
||
|
|
"valid_targets_min": 782
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.424574876577071,
|
||
|
|
"grad_norm": 0.24393298018131654,
|
||
|
|
"learning_rate": 1.6024890322368358e-06,
|
||
|
|
"loss": 0.4778,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16197149455547333,
|
||
|
|
"step": 2690,
|
||
|
|
"valid_targets_mean": 4822.6,
|
||
|
|
"valid_targets_min": 1093
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.432803071859572,
|
||
|
|
"grad_norm": 0.2514174477374428,
|
||
|
|
"learning_rate": 1.5577572054390388e-06,
|
||
|
|
"loss": 0.4799,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14338165521621704,
|
||
|
|
"step": 2695,
|
||
|
|
"valid_targets_mean": 3932.0,
|
||
|
|
"valid_targets_min": 1179
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.4410312671420735,
|
||
|
|
"grad_norm": 0.26955755821506955,
|
||
|
|
"learning_rate": 1.5136332617982863e-06,
|
||
|
|
"loss": 0.4633,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16022084653377533,
|
||
|
|
"step": 2700,
|
||
|
|
"valid_targets_mean": 3117.2,
|
||
|
|
"valid_targets_min": 1152
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.449259462424575,
|
||
|
|
"grad_norm": 0.2573136431453555,
|
||
|
|
"learning_rate": 1.4701186557039648e-06,
|
||
|
|
"loss": 0.4941,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13580411672592163,
|
||
|
|
"step": 2705,
|
||
|
|
"valid_targets_mean": 3402.1,
|
||
|
|
"valid_targets_min": 1357
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.457487657707076,
|
||
|
|
"grad_norm": 0.38826071621804453,
|
||
|
|
"learning_rate": 1.4272148214608073e-06,
|
||
|
|
"loss": 0.4766,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2044774889945984,
|
||
|
|
"step": 2710,
|
||
|
|
"valid_targets_mean": 3427.9,
|
||
|
|
"valid_targets_min": 1115
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.465715852989578,
|
||
|
|
"grad_norm": 0.227099580695401,
|
||
|
|
"learning_rate": 1.384923173241619e-06,
|
||
|
|
"loss": 0.4828,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14953580498695374,
|
||
|
|
"step": 2715,
|
||
|
|
"valid_targets_mean": 4640.0,
|
||
|
|
"valid_targets_min": 1563
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.473944048272079,
|
||
|
|
"grad_norm": 0.22830741711269528,
|
||
|
|
"learning_rate": 1.3432451050406603e-06,
|
||
|
|
"loss": 0.4872,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15632256865501404,
|
||
|
|
"step": 2720,
|
||
|
|
"valid_targets_mean": 3996.0,
|
||
|
|
"valid_targets_min": 1296
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.48217224355458,
|
||
|
|
"grad_norm": 0.2507077891470695,
|
||
|
|
"learning_rate": 1.3021819906277021e-06,
|
||
|
|
"loss": 0.4717,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.148903489112854,
|
||
|
|
"step": 2725,
|
||
|
|
"valid_targets_mean": 3636.5,
|
||
|
|
"valid_targets_min": 1273
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.4904004388370815,
|
||
|
|
"grad_norm": 0.2604213246216982,
|
||
|
|
"learning_rate": 1.2617351835027481e-06,
|
||
|
|
"loss": 0.5015,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16615310311317444,
|
||
|
|
"step": 2730,
|
||
|
|
"valid_targets_mean": 3391.8,
|
||
|
|
"valid_targets_min": 1536
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.498628634119583,
|
||
|
|
"grad_norm": 0.42603564994275517,
|
||
|
|
"learning_rate": 1.2219060168514086e-06,
|
||
|
|
"loss": 0.4304,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09591139853000641,
|
||
|
|
"step": 2735,
|
||
|
|
"valid_targets_mean": 5537.1,
|
||
|
|
"valid_targets_min": 2564
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.506856829402085,
|
||
|
|
"grad_norm": 0.44103615162619425,
|
||
|
|
"learning_rate": 1.1826958035009773e-06,
|
||
|
|
"loss": 0.1567,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.054220978170633316,
|
||
|
|
"step": 2740,
|
||
|
|
"valid_targets_mean": 5478.3,
|
||
|
|
"valid_targets_min": 2226
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.515085024684586,
|
||
|
|
"grad_norm": 0.28894591073573966,
|
||
|
|
"learning_rate": 1.1441058358771317e-06,
|
||
|
|
"loss": 0.1415,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04055090993642807,
|
||
|
|
"step": 2745,
|
||
|
|
"valid_targets_mean": 4935.5,
|
||
|
|
"valid_targets_min": 2518
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.523313219967087,
|
||
|
|
"grad_norm": 0.23637425110289348,
|
||
|
|
"learning_rate": 1.1061373859613634e-06,
|
||
|
|
"loss": 0.1359,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.045938000082969666,
|
||
|
|
"step": 2750,
|
||
|
|
"valid_targets_mean": 5529.9,
|
||
|
|
"valid_targets_min": 3140
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.531541415249588,
|
||
|
|
"grad_norm": 0.21800711396947803,
|
||
|
|
"learning_rate": 1.0687917052490193e-06,
|
||
|
|
"loss": 0.1283,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.039560433477163315,
|
||
|
|
"step": 2755,
|
||
|
|
"valid_targets_mean": 4935.2,
|
||
|
|
"valid_targets_min": 2984
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.53976961053209,
|
||
|
|
"grad_norm": 0.197365764296847,
|
||
|
|
"learning_rate": 1.032070024708085e-06,
|
||
|
|
"loss": 0.1269,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.039238572120666504,
|
||
|
|
"step": 2760,
|
||
|
|
"valid_targets_mean": 4934.8,
|
||
|
|
"valid_targets_min": 2648
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.547997805814592,
|
||
|
|
"grad_norm": 0.18608496032386362,
|
||
|
|
"learning_rate": 9.959735547385762e-07,
|
||
|
|
"loss": 0.1223,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04007216915488243,
|
||
|
|
"step": 2765,
|
||
|
|
"valid_targets_mean": 5136.8,
|
||
|
|
"valid_targets_min": 2159
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.556226001097093,
|
||
|
|
"grad_norm": 0.1728807254072536,
|
||
|
|
"learning_rate": 9.605034851326644e-07,
|
||
|
|
"loss": 0.124,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.038183607161045074,
|
||
|
|
"step": 2770,
|
||
|
|
"valid_targets_mean": 5029.4,
|
||
|
|
"valid_targets_min": 2963
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.564454196379594,
|
||
|
|
"grad_norm": 0.19435780088328164,
|
||
|
|
"learning_rate": 9.256609850354636e-07,
|
||
|
|
"loss": 0.1264,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04276277869939804,
|
||
|
|
"step": 2775,
|
||
|
|
"valid_targets_mean": 5930.7,
|
||
|
|
"valid_targets_min": 2714
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.572682391662095,
|
||
|
|
"grad_norm": 0.17792201797210133,
|
||
|
|
"learning_rate": 8.91447202906468e-07,
|
||
|
|
"loss": 0.1185,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.03755832463502884,
|
||
|
|
"step": 2780,
|
||
|
|
"valid_targets_mean": 5439.3,
|
||
|
|
"valid_targets_min": 2668
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.580910586944597,
|
||
|
|
"grad_norm": 0.1662125580893836,
|
||
|
|
"learning_rate": 8.578632664817177e-07,
|
||
|
|
"loss": 0.1181,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.03744211420416832,
|
||
|
|
"step": 2785,
|
||
|
|
"valid_targets_mean": 5153.3,
|
||
|
|
"valid_targets_min": 2730
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.589138782227098,
|
||
|
|
"grad_norm": 0.18003515710383633,
|
||
|
|
"learning_rate": 8.249102827366306e-07,
|
||
|
|
"loss": 0.1215,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.045447975397109985,
|
||
|
|
"step": 2790,
|
||
|
|
"valid_targets_mean": 5870.0,
|
||
|
|
"valid_targets_min": 3065
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.5973669775095996,
|
||
|
|
"grad_norm": 0.16711301896155625,
|
||
|
|
"learning_rate": 7.925893378494942e-07,
|
||
|
|
"loss": 0.1181,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.041192252188920975,
|
||
|
|
"step": 2795,
|
||
|
|
"valid_targets_mean": 5658.6,
|
||
|
|
"valid_targets_min": 2803
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.605595172792101,
|
||
|
|
"grad_norm": 0.1852534235518299,
|
||
|
|
"learning_rate": 7.609014971656803e-07,
|
||
|
|
"loss": 0.1187,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.03670789301395416,
|
||
|
|
"step": 2800,
|
||
|
|
"valid_targets_mean": 5224.7,
|
||
|
|
"valid_targets_min": 2628
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.613823368074602,
|
||
|
|
"grad_norm": 0.16950321374451233,
|
||
|
|
"learning_rate": 7.298478051625335e-07,
|
||
|
|
"loss": 0.1187,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04173382371664047,
|
||
|
|
"step": 2805,
|
||
|
|
"valid_targets_mean": 5735.1,
|
||
|
|
"valid_targets_min": 3533
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.622051563357104,
|
||
|
|
"grad_norm": 0.17829538445378323,
|
||
|
|
"learning_rate": 6.994292854149165e-07,
|
||
|
|
"loss": 0.117,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.038754336535930634,
|
||
|
|
"step": 2810,
|
||
|
|
"valid_targets_mean": 5265.8,
|
||
|
|
"valid_targets_min": 2050
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.630279758639605,
|
||
|
|
"grad_norm": 0.17636268345082917,
|
||
|
|
"learning_rate": 6.696469405615102e-07,
|
||
|
|
"loss": 0.1159,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04200731962919235,
|
||
|
|
"step": 2815,
|
||
|
|
"valid_targets_mean": 5609.0,
|
||
|
|
"valid_targets_min": 3609
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.638507953922106,
|
||
|
|
"grad_norm": 0.1829464386645835,
|
||
|
|
"learning_rate": 6.405017522717316e-07,
|
||
|
|
"loss": 0.1168,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.037692733108997345,
|
||
|
|
"step": 2820,
|
||
|
|
"valid_targets_mean": 5324.1,
|
||
|
|
"valid_targets_min": 1713
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.6467361492046075,
|
||
|
|
"grad_norm": 0.18881807104757656,
|
||
|
|
"learning_rate": 6.119946812133926e-07,
|
||
|
|
"loss": 0.1176,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.03966955095529556,
|
||
|
|
"step": 2825,
|
||
|
|
"valid_targets_mean": 4990.3,
|
||
|
|
"valid_targets_min": 665
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.654964344487109,
|
||
|
|
"grad_norm": 0.18184605173434396,
|
||
|
|
"learning_rate": 5.841266670210366e-07,
|
||
|
|
"loss": 0.1163,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.03901213780045509,
|
||
|
|
"step": 2830,
|
||
|
|
"valid_targets_mean": 5599.9,
|
||
|
|
"valid_targets_min": 2657
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.663192539769611,
|
||
|
|
"grad_norm": 0.18166398336136586,
|
||
|
|
"learning_rate": 5.568986282649636e-07,
|
||
|
|
"loss": 0.1144,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04104125499725342,
|
||
|
|
"step": 2835,
|
||
|
|
"valid_targets_mean": 5342.1,
|
||
|
|
"valid_targets_min": 2880
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.671420735052112,
|
||
|
|
"grad_norm": 0.6466547891111,
|
||
|
|
"learning_rate": 5.303114624209449e-07,
|
||
|
|
"loss": 0.261,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1742110550403595,
|
||
|
|
"step": 2840,
|
||
|
|
"valid_targets_mean": 3803.8,
|
||
|
|
"valid_targets_min": 1361
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.679648930334613,
|
||
|
|
"grad_norm": 0.5463616335992676,
|
||
|
|
"learning_rate": 5.043660458406563e-07,
|
||
|
|
"loss": 0.4345,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14823535084724426,
|
||
|
|
"step": 2845,
|
||
|
|
"valid_targets_mean": 4280.1,
|
||
|
|
"valid_targets_min": 341
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.687877125617114,
|
||
|
|
"grad_norm": 0.42219045429720925,
|
||
|
|
"learning_rate": 4.790632337227785e-07,
|
||
|
|
"loss": 0.4006,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11169308423995972,
|
||
|
|
"step": 2850,
|
||
|
|
"valid_targets_mean": 4561.8,
|
||
|
|
"valid_targets_min": 1051
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.6961053208996155,
|
||
|
|
"grad_norm": 0.36411612300260526,
|
||
|
|
"learning_rate": 4.544038600848155e-07,
|
||
|
|
"loss": 0.4252,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14665555953979492,
|
||
|
|
"step": 2855,
|
||
|
|
"valid_targets_mean": 3689.7,
|
||
|
|
"valid_targets_min": 1235
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.704333516182118,
|
||
|
|
"grad_norm": 0.30679805398991083,
|
||
|
|
"learning_rate": 4.303887377356053e-07,
|
||
|
|
"loss": 0.3976,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11447066068649292,
|
||
|
|
"step": 2860,
|
||
|
|
"valid_targets_mean": 3426.4,
|
||
|
|
"valid_targets_min": 974
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.712561711464619,
|
||
|
|
"grad_norm": 0.3123137521770701,
|
||
|
|
"learning_rate": 4.070186582485214e-07,
|
||
|
|
"loss": 0.396,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1737116575241089,
|
||
|
|
"step": 2865,
|
||
|
|
"valid_targets_mean": 3986.3,
|
||
|
|
"valid_targets_min": 1079
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.72078990674712,
|
||
|
|
"grad_norm": 0.26330053899557904,
|
||
|
|
"learning_rate": 3.842943919353914e-07,
|
||
|
|
"loss": 0.3833,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11449083685874939,
|
||
|
|
"step": 2870,
|
||
|
|
"valid_targets_mean": 3593.2,
|
||
|
|
"valid_targets_min": 926
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.729018102029621,
|
||
|
|
"grad_norm": 0.2556769576018494,
|
||
|
|
"learning_rate": 3.6221668782109534e-07,
|
||
|
|
"loss": 0.375,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11261357367038727,
|
||
|
|
"step": 2875,
|
||
|
|
"valid_targets_mean": 3480.5,
|
||
|
|
"valid_targets_min": 1151
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.737246297312123,
|
||
|
|
"grad_norm": 0.2669807142158802,
|
||
|
|
"learning_rate": 3.4078627361888717e-07,
|
||
|
|
"loss": 0.3942,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14366257190704346,
|
||
|
|
"step": 2880,
|
||
|
|
"valid_targets_mean": 3933.4,
|
||
|
|
"valid_targets_min": 1036
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.745474492594624,
|
||
|
|
"grad_norm": 0.2664773916564826,
|
||
|
|
"learning_rate": 3.2000385570640114e-07,
|
||
|
|
"loss": 0.3869,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12738540768623352,
|
||
|
|
"step": 2885,
|
||
|
|
"valid_targets_mean": 3254.5,
|
||
|
|
"valid_targets_min": 1190
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.753702687877126,
|
||
|
|
"grad_norm": 0.23815293184151845,
|
||
|
|
"learning_rate": 2.998701191023701e-07,
|
||
|
|
"loss": 0.3684,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09321914613246918,
|
||
|
|
"step": 2890,
|
||
|
|
"valid_targets_mean": 3184.5,
|
||
|
|
"valid_targets_min": 627
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.761930883159627,
|
||
|
|
"grad_norm": 0.2341598205900795,
|
||
|
|
"learning_rate": 2.80385727444048e-07,
|
||
|
|
"loss": 0.3995,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12059900164604187,
|
||
|
|
"step": 2895,
|
||
|
|
"valid_targets_mean": 4686.7,
|
||
|
|
"valid_targets_min": 1218
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.770159078442129,
|
||
|
|
"grad_norm": 0.24528309024526104,
|
||
|
|
"learning_rate": 2.615513229653366e-07,
|
||
|
|
"loss": 0.3732,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14497415721416473,
|
||
|
|
"step": 2900,
|
||
|
|
"valid_targets_mean": 4532.6,
|
||
|
|
"valid_targets_min": 613
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.77838727372463,
|
||
|
|
"grad_norm": 0.2573155357244181,
|
||
|
|
"learning_rate": 2.4336752647561304e-07,
|
||
|
|
"loss": 0.397,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12354221940040588,
|
||
|
|
"step": 2905,
|
||
|
|
"valid_targets_mean": 3717.2,
|
||
|
|
"valid_targets_min": 1137
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.786615469007131,
|
||
|
|
"grad_norm": 0.24346065047412174,
|
||
|
|
"learning_rate": 2.2583493733926655e-07,
|
||
|
|
"loss": 0.3876,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10926102101802826,
|
||
|
|
"step": 2910,
|
||
|
|
"valid_targets_mean": 3432.2,
|
||
|
|
"valid_targets_min": 766
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.794843664289632,
|
||
|
|
"grad_norm": 0.24025285632400223,
|
||
|
|
"learning_rate": 2.0895413345594527e-07,
|
||
|
|
"loss": 0.3908,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14541694521903992,
|
||
|
|
"step": 2915,
|
||
|
|
"valid_targets_mean": 4030.4,
|
||
|
|
"valid_targets_min": 1035
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.803071859572134,
|
||
|
|
"grad_norm": 0.2502583573082134,
|
||
|
|
"learning_rate": 1.9272567124150932e-07,
|
||
|
|
"loss": 0.3916,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11226493120193481,
|
||
|
|
"step": 2920,
|
||
|
|
"valid_targets_mean": 3379.8,
|
||
|
|
"valid_targets_min": 1265
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.811300054854636,
|
||
|
|
"grad_norm": 0.27135058705651266,
|
||
|
|
"learning_rate": 1.771500856096875e-07,
|
||
|
|
"loss": 0.3932,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11371229588985443,
|
||
|
|
"step": 2925,
|
||
|
|
"valid_targets_mean": 2594.7,
|
||
|
|
"valid_targets_min": 1135
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.819528250137137,
|
||
|
|
"grad_norm": 0.28328121691854996,
|
||
|
|
"learning_rate": 1.6222788995444272e-07,
|
||
|
|
"loss": 0.369,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06900590658187866,
|
||
|
|
"step": 2930,
|
||
|
|
"valid_targets_mean": 972.6,
|
||
|
|
"valid_targets_min": 546
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.827756445419638,
|
||
|
|
"grad_norm": 0.2318305134373849,
|
||
|
|
"learning_rate": 1.4795957613305877e-07,
|
||
|
|
"loss": 0.3937,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16254112124443054,
|
||
|
|
"step": 2935,
|
||
|
|
"valid_targets_mean": 5099.9,
|
||
|
|
"valid_targets_min": 1465
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.835984640702139,
|
||
|
|
"grad_norm": 0.24068535986602976,
|
||
|
|
"learning_rate": 1.3434561444992e-07,
|
||
|
|
"loss": 0.3936,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15346771478652954,
|
||
|
|
"step": 2940,
|
||
|
|
"valid_targets_mean": 5051.9,
|
||
|
|
"valid_targets_min": 986
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.84421283598464,
|
||
|
|
"grad_norm": 0.39007386893679075,
|
||
|
|
"learning_rate": 1.2138645364101032e-07,
|
||
|
|
"loss": 0.3477,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1031792089343071,
|
||
|
|
"step": 2945,
|
||
|
|
"valid_targets_mean": 6354.4,
|
||
|
|
"valid_targets_min": 348
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.852441031267142,
|
||
|
|
"grad_norm": 0.3829605064351524,
|
||
|
|
"learning_rate": 1.0908252085912952e-07,
|
||
|
|
"loss": 0.2616,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07880100607872009,
|
||
|
|
"step": 2950,
|
||
|
|
"valid_targets_mean": 6446.1,
|
||
|
|
"valid_targets_min": 139
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.860669226549644,
|
||
|
|
"grad_norm": 0.3816630053349074,
|
||
|
|
"learning_rate": 9.743422165980454e-08,
|
||
|
|
"loss": 0.2401,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06825752556324005,
|
||
|
|
"step": 2955,
|
||
|
|
"valid_targets_mean": 3949.5,
|
||
|
|
"valid_targets_min": 306
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.868897421832145,
|
||
|
|
"grad_norm": 0.3686134242305812,
|
||
|
|
"learning_rate": 8.64419399879246e-08,
|
||
|
|
"loss": 0.2575,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09232122451066971,
|
||
|
|
"step": 2960,
|
||
|
|
"valid_targets_mean": 6445.5,
|
||
|
|
"valid_targets_min": 501
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.877125617114646,
|
||
|
|
"grad_norm": 0.35617768490637586,
|
||
|
|
"learning_rate": 7.61060381650891e-08,
|
||
|
|
"loss": 0.2476,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08181267231702805,
|
||
|
|
"step": 2965,
|
||
|
|
"valid_targets_mean": 6484.1,
|
||
|
|
"valid_targets_min": 702
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.885353812397147,
|
||
|
|
"grad_norm": 0.3539720544440604,
|
||
|
|
"learning_rate": 6.642685687766159e-08,
|
||
|
|
"loss": 0.264,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08130834996700287,
|
||
|
|
"step": 2970,
|
||
|
|
"valid_targets_mean": 7440.5,
|
||
|
|
"valid_targets_min": 914
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.893582007679649,
|
||
|
|
"grad_norm": 0.34241063452198034,
|
||
|
|
"learning_rate": 5.740471516553881e-08,
|
||
|
|
"loss": 0.2368,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09114379435777664,
|
||
|
|
"step": 2975,
|
||
|
|
"valid_targets_mean": 7078.0,
|
||
|
|
"valid_targets_min": 280
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.90181020296215,
|
||
|
|
"grad_norm": 0.3171560903782373,
|
||
|
|
"learning_rate": 4.9039910411643466e-08,
|
||
|
|
"loss": 0.2492,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08341857045888901,
|
||
|
|
"step": 2980,
|
||
|
|
"valid_targets_mean": 6463.1,
|
||
|
|
"valid_targets_min": 1221
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.910038398244652,
|
||
|
|
"grad_norm": 0.3168640296013643,
|
||
|
|
"learning_rate": 4.133271833210772e-08,
|
||
|
|
"loss": 0.234,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06438249349594116,
|
||
|
|
"step": 2985,
|
||
|
|
"valid_targets_mean": 6061.8,
|
||
|
|
"valid_targets_min": 1412
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.918266593527153,
|
||
|
|
"grad_norm": 0.33532587380496615,
|
||
|
|
"learning_rate": 3.428339296719596e-08,
|
||
|
|
"loss": 0.2502,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10326523333787918,
|
||
|
|
"step": 2990,
|
||
|
|
"valid_targets_mean": 6472.4,
|
||
|
|
"valid_targets_min": 277
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.926494788809654,
|
||
|
|
"grad_norm": 0.2979188578639744,
|
||
|
|
"learning_rate": 2.789216667293593e-08,
|
||
|
|
"loss": 0.2527,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06032399833202362,
|
||
|
|
"step": 2995,
|
||
|
|
"valid_targets_mean": 7596.1,
|
||
|
|
"valid_targets_min": 2298
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.934722984092156,
|
||
|
|
"grad_norm": 0.3044209930110469,
|
||
|
|
"learning_rate": 2.2159250113438223e-08,
|
||
|
|
"loss": 0.2546,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07601773738861084,
|
||
|
|
"step": 3000,
|
||
|
|
"valid_targets_mean": 7278.7,
|
||
|
|
"valid_targets_min": 299
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.942951179374657,
|
||
|
|
"grad_norm": 0.3175588292516497,
|
||
|
|
"learning_rate": 1.708483225397961e-08,
|
||
|
|
"loss": 0.2639,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09058444201946259,
|
||
|
|
"step": 3005,
|
||
|
|
"valid_targets_mean": 6472.9,
|
||
|
|
"valid_targets_min": 578
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.951179374657158,
|
||
|
|
"grad_norm": 0.32066451016675285,
|
||
|
|
"learning_rate": 1.266908035475467e-08,
|
||
|
|
"loss": 0.2289,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08107155561447144,
|
||
|
|
"step": 3010,
|
||
|
|
"valid_targets_mean": 6411.4,
|
||
|
|
"valid_targets_min": 176
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.95940756993966,
|
||
|
|
"grad_norm": 0.31698950477993576,
|
||
|
|
"learning_rate": 8.912139965369105e-09,
|
||
|
|
"loss": 0.2386,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09014058858156204,
|
||
|
|
"step": 3015,
|
||
|
|
"valid_targets_mean": 5064.2,
|
||
|
|
"valid_targets_min": 746
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.967635765222162,
|
||
|
|
"grad_norm": 0.316410358732829,
|
||
|
|
"learning_rate": 5.814134920048009e-09,
|
||
|
|
"loss": 0.2425,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10155069828033447,
|
||
|
|
"step": 3020,
|
||
|
|
"valid_targets_mean": 6593.2,
|
||
|
|
"valid_targets_min": 1045
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.975863960504663,
|
||
|
|
"grad_norm": 0.3192028950014772,
|
||
|
|
"learning_rate": 3.3751673335458147e-09,
|
||
|
|
"loss": 0.2281,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06748147308826447,
|
||
|
|
"step": 3025,
|
||
|
|
"valid_targets_mean": 5822.4,
|
||
|
|
"valid_targets_min": 952
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.984092155787164,
|
||
|
|
"grad_norm": 0.31201193824927753,
|
||
|
|
"learning_rate": 1.5953175977778679e-09,
|
||
|
|
"loss": 0.2468,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1104644238948822,
|
||
|
|
"step": 3030,
|
||
|
|
"valid_targets_mean": 8131.2,
|
||
|
|
"valid_targets_min": 270
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.992320351069665,
|
||
|
|
"grad_norm": 0.3171727370891135,
|
||
|
|
"learning_rate": 4.746443791869837e-10,
|
||
|
|
"loss": 0.2333,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07531018555164337,
|
||
|
|
"step": 3035,
|
||
|
|
"valid_targets_mean": 6484.8,
|
||
|
|
"valid_targets_min": 1247
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.0,
|
||
|
|
"grad_norm": 1.355991457881435,
|
||
|
|
"learning_rate": 1.3184616789452264e-11,
|
||
|
|
"loss": 0.2833,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.36725589632987976,
|
||
|
|
"step": 3040,
|
||
|
|
"valid_targets_mean": 6065.3,
|
||
|
|
"valid_targets_min": 1413
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.0,
|
||
|
|
"step": 3040,
|
||
|
|
"total_flos": 9.271417366388933e+18,
|
||
|
|
"train_loss": 0.0,
|
||
|
|
"train_runtime": 4.0738,
|
||
|
|
"train_samples_per_second": 71564.652,
|
||
|
|
"train_steps_per_second": 746.232
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"logging_steps": 5,
|
||
|
|
"max_steps": 3040,
|
||
|
|
"num_input_tokens_seen": 0,
|
||
|
|
"num_train_epochs": 5,
|
||
|
|
"save_steps": 300,
|
||
|
|
"stateful_callbacks": {
|
||
|
|
"TrainerControl": {
|
||
|
|
"args": {
|
||
|
|
"should_epoch_stop": false,
|
||
|
|
"should_evaluate": false,
|
||
|
|
"should_log": false,
|
||
|
|
"should_save": true,
|
||
|
|
"should_training_stop": true
|
||
|
|
},
|
||
|
|
"attributes": {}
|
||
|
|
}
|
||
|
|
},
|
||
|
|
"total_flos": 9.271417366388933e+18,
|
||
|
|
"train_batch_size": 1,
|
||
|
|
"trial_name": null,
|
||
|
|
"trial_params": null
|
||
|
|
}
|