1529 lines
42 KiB
JSON
1529 lines
42 KiB
JSON
|
{
|
||
|
|
"best_global_step": null,
|
||
|
|
"best_metric": null,
|
||
|
|
"best_model_checkpoint": null,
|
||
|
|
"epoch": 7.0,
|
||
|
|
"eval_steps": 500,
|
||
|
|
"global_step": 679,
|
||
|
|
"is_hyper_param_search": false,
|
||
|
|
"is_local_process_zero": true,
|
||
|
|
"is_world_process_zero": true,
|
||
|
|
"log_history": [
|
||
|
|
{
|
||
|
|
"epoch": 0.05154639175257732,
|
||
|
|
"grad_norm": 12.161818843525861,
|
||
|
|
"learning_rate": 2.3529411764705885e-06,
|
||
|
|
"loss": 0.8719,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2826420068740845,
|
||
|
|
"step": 5,
|
||
|
|
"valid_targets_mean": 7206.0,
|
||
|
|
"valid_targets_min": 2217
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.10309278350515463,
|
||
|
|
"grad_norm": 6.274544999869601,
|
||
|
|
"learning_rate": 5.294117647058824e-06,
|
||
|
|
"loss": 0.8419,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27338725328445435,
|
||
|
|
"step": 10,
|
||
|
|
"valid_targets_mean": 8297.6,
|
||
|
|
"valid_targets_min": 2914
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15463917525773196,
|
||
|
|
"grad_norm": 2.3481981912392293,
|
||
|
|
"learning_rate": 8.23529411764706e-06,
|
||
|
|
"loss": 0.7585,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23505288362503052,
|
||
|
|
"step": 15,
|
||
|
|
"valid_targets_mean": 7449.2,
|
||
|
|
"valid_targets_min": 2138
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.20618556701030927,
|
||
|
|
"grad_norm": 1.5353307761767716,
|
||
|
|
"learning_rate": 1.1176470588235295e-05,
|
||
|
|
"loss": 0.7092,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25314459204673767,
|
||
|
|
"step": 20,
|
||
|
|
"valid_targets_mean": 7483.8,
|
||
|
|
"valid_targets_min": 751
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.25773195876288657,
|
||
|
|
"grad_norm": 1.0382452883991766,
|
||
|
|
"learning_rate": 1.4117647058823532e-05,
|
||
|
|
"loss": 0.6704,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20421409606933594,
|
||
|
|
"step": 25,
|
||
|
|
"valid_targets_mean": 7920.1,
|
||
|
|
"valid_targets_min": 1516
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.30927835051546393,
|
||
|
|
"grad_norm": 0.7945196565054293,
|
||
|
|
"learning_rate": 1.7058823529411767e-05,
|
||
|
|
"loss": 0.6411,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.200191468000412,
|
||
|
|
"step": 30,
|
||
|
|
"valid_targets_mean": 7371.4,
|
||
|
|
"valid_targets_min": 3073
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.36082474226804123,
|
||
|
|
"grad_norm": 0.6179707179087484,
|
||
|
|
"learning_rate": 2e-05,
|
||
|
|
"loss": 0.602,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21440792083740234,
|
||
|
|
"step": 35,
|
||
|
|
"valid_targets_mean": 8468.3,
|
||
|
|
"valid_targets_min": 2927
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.41237113402061853,
|
||
|
|
"grad_norm": 0.48798515379944946,
|
||
|
|
"learning_rate": 2.2941176470588237e-05,
|
||
|
|
"loss": 0.584,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20928455889225006,
|
||
|
|
"step": 40,
|
||
|
|
"valid_targets_mean": 8256.4,
|
||
|
|
"valid_targets_min": 2038
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4639175257731959,
|
||
|
|
"grad_norm": 0.43415441926926357,
|
||
|
|
"learning_rate": 2.5882352941176475e-05,
|
||
|
|
"loss": 0.5619,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20116469264030457,
|
||
|
|
"step": 45,
|
||
|
|
"valid_targets_mean": 7732.8,
|
||
|
|
"valid_targets_min": 2806
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5154639175257731,
|
||
|
|
"grad_norm": 0.4287544880831311,
|
||
|
|
"learning_rate": 2.8823529411764707e-05,
|
||
|
|
"loss": 0.5199,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19549432396888733,
|
||
|
|
"step": 50,
|
||
|
|
"valid_targets_mean": 8253.3,
|
||
|
|
"valid_targets_min": 2394
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5670103092783505,
|
||
|
|
"grad_norm": 0.34966772981542454,
|
||
|
|
"learning_rate": 3.1764705882352945e-05,
|
||
|
|
"loss": 0.5072,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17478598654270172,
|
||
|
|
"step": 55,
|
||
|
|
"valid_targets_mean": 7541.5,
|
||
|
|
"valid_targets_min": 2368
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6185567010309279,
|
||
|
|
"grad_norm": 0.29814887772019083,
|
||
|
|
"learning_rate": 3.470588235294118e-05,
|
||
|
|
"loss": 0.497,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.160207599401474,
|
||
|
|
"step": 60,
|
||
|
|
"valid_targets_mean": 7963.1,
|
||
|
|
"valid_targets_min": 2595
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6701030927835051,
|
||
|
|
"grad_norm": 0.2629403316432251,
|
||
|
|
"learning_rate": 3.7647058823529415e-05,
|
||
|
|
"loss": 0.4782,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1807478666305542,
|
||
|
|
"step": 65,
|
||
|
|
"valid_targets_mean": 8748.9,
|
||
|
|
"valid_targets_min": 2994
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7216494845360825,
|
||
|
|
"grad_norm": 0.2694266615416935,
|
||
|
|
"learning_rate": 3.999973562744509e-05,
|
||
|
|
"loss": 0.4671,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16049513220787048,
|
||
|
|
"step": 70,
|
||
|
|
"valid_targets_mean": 7966.9,
|
||
|
|
"valid_targets_min": 2645
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7731958762886598,
|
||
|
|
"grad_norm": 0.2709203402662577,
|
||
|
|
"learning_rate": 3.999048332187732e-05,
|
||
|
|
"loss": 0.4527,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.141615629196167,
|
||
|
|
"step": 75,
|
||
|
|
"valid_targets_mean": 7022.3,
|
||
|
|
"valid_targets_min": 724
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8247422680412371,
|
||
|
|
"grad_norm": 0.2760272332458017,
|
||
|
|
"learning_rate": 3.996801937701812e-05,
|
||
|
|
"loss": 0.4505,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1317310482263565,
|
||
|
|
"step": 80,
|
||
|
|
"valid_targets_mean": 6580.1,
|
||
|
|
"valid_targets_min": 2996
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8762886597938144,
|
||
|
|
"grad_norm": 0.3840065178804756,
|
||
|
|
"learning_rate": 3.9932358639208713e-05,
|
||
|
|
"loss": 0.4436,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.131992369890213,
|
||
|
|
"step": 85,
|
||
|
|
"valid_targets_mean": 6960.8,
|
||
|
|
"valid_targets_min": 2550
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9278350515463918,
|
||
|
|
"grad_norm": 0.3020895947985698,
|
||
|
|
"learning_rate": 3.988352467650382e-05,
|
||
|
|
"loss": 0.4405,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14092403650283813,
|
||
|
|
"step": 90,
|
||
|
|
"valid_targets_mean": 7277.2,
|
||
|
|
"valid_targets_min": 504
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.979381443298969,
|
||
|
|
"grad_norm": 0.26537634384021036,
|
||
|
|
"learning_rate": 3.9821549763095606e-05,
|
||
|
|
"loss": 0.4406,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1356378197669983,
|
||
|
|
"step": 95,
|
||
|
|
"valid_targets_mean": 7548.2,
|
||
|
|
"valid_targets_min": 1583
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0309278350515463,
|
||
|
|
"grad_norm": 0.286922881872239,
|
||
|
|
"learning_rate": 3.9746474857983815e-05,
|
||
|
|
"loss": 0.4333,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15483158826828003,
|
||
|
|
"step": 100,
|
||
|
|
"valid_targets_mean": 8870.0,
|
||
|
|
"valid_targets_min": 2591
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0824742268041236,
|
||
|
|
"grad_norm": 0.26802122611172374,
|
||
|
|
"learning_rate": 3.965834957790608e-05,
|
||
|
|
"loss": 0.4253,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1555008590221405,
|
||
|
|
"step": 105,
|
||
|
|
"valid_targets_mean": 9067.3,
|
||
|
|
"valid_targets_min": 2556
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.134020618556701,
|
||
|
|
"grad_norm": 0.2628083647869015,
|
||
|
|
"learning_rate": 3.9557232164546405e-05,
|
||
|
|
"loss": 0.4246,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1437469869852066,
|
||
|
|
"step": 110,
|
||
|
|
"valid_targets_mean": 8403.8,
|
||
|
|
"valid_targets_min": 3217
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1855670103092784,
|
||
|
|
"grad_norm": 0.25065229791808624,
|
||
|
|
"learning_rate": 3.944318944604347e-05,
|
||
|
|
"loss": 0.4175,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13575485348701477,
|
||
|
|
"step": 115,
|
||
|
|
"valid_targets_mean": 7651.8,
|
||
|
|
"valid_targets_min": 2562
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2371134020618557,
|
||
|
|
"grad_norm": 0.27340997684040397,
|
||
|
|
"learning_rate": 3.9316296792824096e-05,
|
||
|
|
"loss": 0.4075,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11437910795211792,
|
||
|
|
"step": 120,
|
||
|
|
"valid_targets_mean": 6541.0,
|
||
|
|
"valid_targets_min": 2223
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2886597938144329,
|
||
|
|
"grad_norm": 0.2728823881222885,
|
||
|
|
"learning_rate": 3.917663806779125e-05,
|
||
|
|
"loss": 0.4208,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13078531622886658,
|
||
|
|
"step": 125,
|
||
|
|
"valid_targets_mean": 7127.0,
|
||
|
|
"valid_targets_min": 504
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3402061855670104,
|
||
|
|
"grad_norm": 0.2748224627702122,
|
||
|
|
"learning_rate": 3.9024305570899345e-05,
|
||
|
|
"loss": 0.4144,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1383627951145172,
|
||
|
|
"step": 130,
|
||
|
|
"valid_targets_mean": 7506.9,
|
||
|
|
"valid_targets_min": 3080
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3917525773195876,
|
||
|
|
"grad_norm": 0.25700551954600065,
|
||
|
|
"learning_rate": 3.885939997815349e-05,
|
||
|
|
"loss": 0.4105,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14661987125873566,
|
||
|
|
"step": 135,
|
||
|
|
"valid_targets_mean": 8519.0,
|
||
|
|
"valid_targets_min": 2974
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.443298969072165,
|
||
|
|
"grad_norm": 0.2704288869246834,
|
||
|
|
"learning_rate": 3.8682030275073125e-05,
|
||
|
|
"loss": 0.406,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1586286425590515,
|
||
|
|
"step": 140,
|
||
|
|
"valid_targets_mean": 8352.6,
|
||
|
|
"valid_targets_min": 2550
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4948453608247423,
|
||
|
|
"grad_norm": 0.28692093881388453,
|
||
|
|
"learning_rate": 3.849231368466383e-05,
|
||
|
|
"loss": 0.4086,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13457578420639038,
|
||
|
|
"step": 145,
|
||
|
|
"valid_targets_mean": 7580.2,
|
||
|
|
"valid_targets_min": 3472
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5463917525773194,
|
||
|
|
"grad_norm": 0.2465143262388595,
|
||
|
|
"learning_rate": 3.829037558994513e-05,
|
||
|
|
"loss": 0.4011,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11324875056743622,
|
||
|
|
"step": 150,
|
||
|
|
"valid_targets_mean": 6850.6,
|
||
|
|
"valid_targets_min": 1841
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.597938144329897,
|
||
|
|
"grad_norm": 0.28135471399817474,
|
||
|
|
"learning_rate": 3.807634945108521e-05,
|
||
|
|
"loss": 0.3975,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13441982865333557,
|
||
|
|
"step": 155,
|
||
|
|
"valid_targets_mean": 7777.2,
|
||
|
|
"valid_targets_min": 2662
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6494845360824741,
|
||
|
|
"grad_norm": 0.2873817941611567,
|
||
|
|
"learning_rate": 3.785037671719763e-05,
|
||
|
|
"loss": 0.3997,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13682210445404053,
|
||
|
|
"step": 160,
|
||
|
|
"valid_targets_mean": 7694.7,
|
||
|
|
"valid_targets_min": 3521
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7010309278350515,
|
||
|
|
"grad_norm": 0.2676863740524645,
|
||
|
|
"learning_rate": 3.76126067328581e-05,
|
||
|
|
"loss": 0.3994,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1269443929195404,
|
||
|
|
"step": 165,
|
||
|
|
"valid_targets_mean": 7338.2,
|
||
|
|
"valid_targets_min": 1583
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7525773195876289,
|
||
|
|
"grad_norm": 0.2829869921556731,
|
||
|
|
"learning_rate": 3.736319663940316e-05,
|
||
|
|
"loss": 0.3986,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12589965760707855,
|
||
|
|
"step": 170,
|
||
|
|
"valid_targets_mean": 6862.6,
|
||
|
|
"valid_targets_min": 2529
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8041237113402062,
|
||
|
|
"grad_norm": 0.26163280022519214,
|
||
|
|
"learning_rate": 3.710231127107606e-05,
|
||
|
|
"loss": 0.3985,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14018520712852478,
|
||
|
|
"step": 175,
|
||
|
|
"valid_targets_mean": 7897.4,
|
||
|
|
"valid_targets_min": 2563
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8556701030927836,
|
||
|
|
"grad_norm": 0.29284664172335323,
|
||
|
|
"learning_rate": 3.683012304608837e-05,
|
||
|
|
"loss": 0.401,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13740304112434387,
|
||
|
|
"step": 180,
|
||
|
|
"valid_targets_mean": 8097.0,
|
||
|
|
"valid_targets_min": 2440
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9072164948453607,
|
||
|
|
"grad_norm": 0.24879509168866065,
|
||
|
|
"learning_rate": 3.654681185266939e-05,
|
||
|
|
"loss": 0.3945,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13531193137168884,
|
||
|
|
"step": 185,
|
||
|
|
"valid_targets_mean": 7567.4,
|
||
|
|
"valid_targets_min": 2246
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9587628865979383,
|
||
|
|
"grad_norm": 0.2608310476515075,
|
||
|
|
"learning_rate": 3.6252564930178705e-05,
|
||
|
|
"loss": 0.3976,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11826080083847046,
|
||
|
|
"step": 190,
|
||
|
|
"valid_targets_mean": 7180.1,
|
||
|
|
"valid_targets_min": 811
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0103092783505154,
|
||
|
|
"grad_norm": 0.25594290047927687,
|
||
|
|
"learning_rate": 3.59475767453603e-05,
|
||
|
|
"loss": 0.392,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1198672205209732,
|
||
|
|
"step": 195,
|
||
|
|
"valid_targets_mean": 7441.7,
|
||
|
|
"valid_targets_min": 1933
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0618556701030926,
|
||
|
|
"grad_norm": 0.2923979338415982,
|
||
|
|
"learning_rate": 3.563204886382024e-05,
|
||
|
|
"loss": 0.3861,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12066236883401871,
|
||
|
|
"step": 200,
|
||
|
|
"valid_targets_mean": 7081.2,
|
||
|
|
"valid_targets_min": 350
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.11340206185567,
|
||
|
|
"grad_norm": 0.29674050150327025,
|
||
|
|
"learning_rate": 3.530618981681261e-05,
|
||
|
|
"loss": 0.3852,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12159411609172821,
|
||
|
|
"step": 205,
|
||
|
|
"valid_targets_mean": 7327.8,
|
||
|
|
"valid_targets_min": 560
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1649484536082473,
|
||
|
|
"grad_norm": 0.2685401540335565,
|
||
|
|
"learning_rate": 3.497021496342203e-05,
|
||
|
|
"loss": 0.3826,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10820683091878891,
|
||
|
|
"step": 210,
|
||
|
|
"valid_targets_mean": 6862.6,
|
||
|
|
"valid_targets_min": 2905
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.216494845360825,
|
||
|
|
"grad_norm": 0.3036957972406054,
|
||
|
|
"learning_rate": 3.4624346348233526e-05,
|
||
|
|
"loss": 0.375,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12980154156684875,
|
||
|
|
"step": 215,
|
||
|
|
"valid_targets_mean": 7444.1,
|
||
|
|
"valid_targets_min": 2623
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.268041237113402,
|
||
|
|
"grad_norm": 0.2985596980456637,
|
||
|
|
"learning_rate": 3.426881255458411e-05,
|
||
|
|
"loss": 0.3844,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10776925086975098,
|
||
|
|
"step": 220,
|
||
|
|
"valid_targets_mean": 6507.0,
|
||
|
|
"valid_targets_min": 1569
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.319587628865979,
|
||
|
|
"grad_norm": 0.2716194216592204,
|
||
|
|
"learning_rate": 3.390384855349285e-05,
|
||
|
|
"loss": 0.3832,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1313299685716629,
|
||
|
|
"step": 225,
|
||
|
|
"valid_targets_mean": 7795.2,
|
||
|
|
"valid_targets_min": 3122
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3711340206185567,
|
||
|
|
"grad_norm": 0.3358798614727285,
|
||
|
|
"learning_rate": 3.352969554836933e-05,
|
||
|
|
"loss": 0.3846,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13108502328395844,
|
||
|
|
"step": 230,
|
||
|
|
"valid_targets_mean": 7350.4,
|
||
|
|
"valid_targets_min": 2948
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.422680412371134,
|
||
|
|
"grad_norm": 0.286522549065935,
|
||
|
|
"learning_rate": 3.314660081560323e-05,
|
||
|
|
"loss": 0.3852,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1345348358154297,
|
||
|
|
"step": 235,
|
||
|
|
"valid_targets_mean": 7835.4,
|
||
|
|
"valid_targets_min": 2681
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4742268041237114,
|
||
|
|
"grad_norm": 0.2836276555771362,
|
||
|
|
"learning_rate": 3.2754817541140185e-05,
|
||
|
|
"loss": 0.3791,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.126967191696167,
|
||
|
|
"step": 240,
|
||
|
|
"valid_targets_mean": 7941.1,
|
||
|
|
"valid_targets_min": 3037
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5257731958762886,
|
||
|
|
"grad_norm": 0.2701584019036541,
|
||
|
|
"learning_rate": 3.235460465315213e-05,
|
||
|
|
"loss": 0.3866,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10711121559143066,
|
||
|
|
"step": 245,
|
||
|
|
"valid_targets_mean": 6325.0,
|
||
|
|
"valid_targets_min": 2280
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5773195876288657,
|
||
|
|
"grad_norm": 0.24537454268101747,
|
||
|
|
"learning_rate": 3.194622665091258e-05,
|
||
|
|
"loss": 0.3772,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13384535908699036,
|
||
|
|
"step": 250,
|
||
|
|
"valid_targets_mean": 8951.4,
|
||
|
|
"valid_targets_min": 3700
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6288659793814433,
|
||
|
|
"grad_norm": 0.2769985040169458,
|
||
|
|
"learning_rate": 3.152995342999002e-05,
|
||
|
|
"loss": 0.3809,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13421593606472015,
|
||
|
|
"step": 255,
|
||
|
|
"valid_targets_mean": 7729.7,
|
||
|
|
"valid_targets_min": 2910
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.680412371134021,
|
||
|
|
"grad_norm": 0.2983371759197875,
|
||
|
|
"learning_rate": 3.110606010387483e-05,
|
||
|
|
"loss": 0.3868,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13421210646629333,
|
||
|
|
"step": 260,
|
||
|
|
"valid_targets_mean": 7750.3,
|
||
|
|
"valid_targets_min": 2067
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.731958762886598,
|
||
|
|
"grad_norm": 0.3149278541436496,
|
||
|
|
"learning_rate": 3.067482682215783e-05,
|
||
|
|
"loss": 0.3782,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1094997227191925,
|
||
|
|
"step": 265,
|
||
|
|
"valid_targets_mean": 6997.8,
|
||
|
|
"valid_targets_min": 981
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.783505154639175,
|
||
|
|
"grad_norm": 0.296573764639055,
|
||
|
|
"learning_rate": 3.023653858538037e-05,
|
||
|
|
"loss": 0.3757,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14565624296665192,
|
||
|
|
"step": 270,
|
||
|
|
"valid_targets_mean": 8690.6,
|
||
|
|
"valid_targets_min": 2461
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8350515463917527,
|
||
|
|
"grad_norm": 0.28239915346948635,
|
||
|
|
"learning_rate": 2.9791485056678478e-05,
|
||
|
|
"loss": 0.3823,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13386495411396027,
|
||
|
|
"step": 275,
|
||
|
|
"valid_targets_mean": 8028.3,
|
||
|
|
"valid_targets_min": 1583
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.88659793814433,
|
||
|
|
"grad_norm": 0.2687057593736694,
|
||
|
|
"learning_rate": 2.933996037034556e-05,
|
||
|
|
"loss": 0.3747,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1334458738565445,
|
||
|
|
"step": 280,
|
||
|
|
"valid_targets_mean": 7567.1,
|
||
|
|
"valid_targets_min": 2365
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9381443298969074,
|
||
|
|
"grad_norm": 0.27538103642903583,
|
||
|
|
"learning_rate": 2.8882262937440076e-05,
|
||
|
|
"loss": 0.3789,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11708861589431763,
|
||
|
|
"step": 285,
|
||
|
|
"valid_targets_mean": 7347.5,
|
||
|
|
"valid_targets_min": 2775
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9896907216494846,
|
||
|
|
"grad_norm": 0.25420349468596576,
|
||
|
|
"learning_rate": 2.8418695248566703e-05,
|
||
|
|
"loss": 0.3714,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12286879122257233,
|
||
|
|
"step": 290,
|
||
|
|
"valid_targets_mean": 7598.0,
|
||
|
|
"valid_targets_min": 2989
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0412371134020617,
|
||
|
|
"grad_norm": 0.2825621241033727,
|
||
|
|
"learning_rate": 2.7949563673961425e-05,
|
||
|
|
"loss": 0.3692,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12423013150691986,
|
||
|
|
"step": 295,
|
||
|
|
"valid_targets_mean": 7747.8,
|
||
|
|
"valid_targets_min": 3304
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0927835051546393,
|
||
|
|
"grad_norm": 0.2928726216580958,
|
||
|
|
"learning_rate": 2.7475178261012492e-05,
|
||
|
|
"loss": 0.3701,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1302967518568039,
|
||
|
|
"step": 300,
|
||
|
|
"valid_targets_mean": 8288.3,
|
||
|
|
"valid_targets_min": 2803
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1443298969072164,
|
||
|
|
"grad_norm": 0.29199868148758423,
|
||
|
|
"learning_rate": 2.6995852529351227e-05,
|
||
|
|
"loss": 0.3644,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11168663203716278,
|
||
|
|
"step": 305,
|
||
|
|
"valid_targets_mean": 6574.4,
|
||
|
|
"valid_targets_min": 2447
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.195876288659794,
|
||
|
|
"grad_norm": 0.30611924720524425,
|
||
|
|
"learning_rate": 2.6511903263648008e-05,
|
||
|
|
"loss": 0.3665,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10399296879768372,
|
||
|
|
"step": 310,
|
||
|
|
"valid_targets_mean": 6869.6,
|
||
|
|
"valid_targets_min": 1486
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.247422680412371,
|
||
|
|
"grad_norm": 0.26364712262090423,
|
||
|
|
"learning_rate": 2.602365030425041e-05,
|
||
|
|
"loss": 0.366,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12233500182628632,
|
||
|
|
"step": 315,
|
||
|
|
"valid_targets_mean": 8237.5,
|
||
|
|
"valid_targets_min": 4039
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2989690721649483,
|
||
|
|
"grad_norm": 0.2636445174086616,
|
||
|
|
"learning_rate": 2.553141633580185e-05,
|
||
|
|
"loss": 0.3719,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11217635869979858,
|
||
|
|
"step": 320,
|
||
|
|
"valid_targets_mean": 7427.2,
|
||
|
|
"valid_targets_min": 3281
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.350515463917526,
|
||
|
|
"grad_norm": 0.28148465155423386,
|
||
|
|
"learning_rate": 2.5035526673980463e-05,
|
||
|
|
"loss": 0.3645,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11682005226612091,
|
||
|
|
"step": 325,
|
||
|
|
"valid_targets_mean": 7872.5,
|
||
|
|
"valid_targets_min": 3287
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.402061855670103,
|
||
|
|
"grad_norm": 0.29166580935856323,
|
||
|
|
"learning_rate": 2.453630905049913e-05,
|
||
|
|
"loss": 0.3742,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.132327139377594,
|
||
|
|
"step": 330,
|
||
|
|
"valid_targets_mean": 8424.6,
|
||
|
|
"valid_targets_min": 2531
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4536082474226806,
|
||
|
|
"grad_norm": 0.3153648451305088,
|
||
|
|
"learning_rate": 2.4034093396508752e-05,
|
||
|
|
"loss": 0.3687,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11530999094247818,
|
||
|
|
"step": 335,
|
||
|
|
"valid_targets_mean": 7157.7,
|
||
|
|
"valid_targets_min": 2568
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5051546391752577,
|
||
|
|
"grad_norm": 0.2455576665947615,
|
||
|
|
"learning_rate": 2.3529211624547914e-05,
|
||
|
|
"loss": 0.3689,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11953712999820709,
|
||
|
|
"step": 340,
|
||
|
|
"valid_targets_mean": 7321.0,
|
||
|
|
"valid_targets_min": 2334
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.556701030927835,
|
||
|
|
"grad_norm": 0.25354722556250475,
|
||
|
|
"learning_rate": 2.3021997409183086e-05,
|
||
|
|
"loss": 0.3664,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12187885493040085,
|
||
|
|
"step": 345,
|
||
|
|
"valid_targets_mean": 7613.2,
|
||
|
|
"valid_targets_min": 2811
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6082474226804124,
|
||
|
|
"grad_norm": 0.24825333130630592,
|
||
|
|
"learning_rate": 2.2512785966484286e-05,
|
||
|
|
"loss": 0.3688,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11898311227560043,
|
||
|
|
"step": 350,
|
||
|
|
"valid_targets_mean": 7999.2,
|
||
|
|
"valid_targets_min": 3229
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6597938144329896,
|
||
|
|
"grad_norm": 0.2445310664924181,
|
||
|
|
"learning_rate": 2.200191383248197e-05,
|
||
|
|
"loss": 0.3704,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10550983995199203,
|
||
|
|
"step": 355,
|
||
|
|
"valid_targets_mean": 6803.0,
|
||
|
|
"valid_targets_min": 2854
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.711340206185567,
|
||
|
|
"grad_norm": 0.261238526946729,
|
||
|
|
"learning_rate": 2.148971864075156e-05,
|
||
|
|
"loss": 0.3668,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1359022557735443,
|
||
|
|
"step": 360,
|
||
|
|
"valid_targets_mean": 8219.3,
|
||
|
|
"valid_targets_min": 724
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7628865979381443,
|
||
|
|
"grad_norm": 0.28139816200921625,
|
||
|
|
"learning_rate": 2.0976538899272632e-05,
|
||
|
|
"loss": 0.3645,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11895483732223511,
|
||
|
|
"step": 365,
|
||
|
|
"valid_targets_mean": 6931.2,
|
||
|
|
"valid_targets_min": 2700
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8144329896907214,
|
||
|
|
"grad_norm": 0.2490827060035925,
|
||
|
|
"learning_rate": 2.0462713766710184e-05,
|
||
|
|
"loss": 0.3662,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12993857264518738,
|
||
|
|
"step": 370,
|
||
|
|
"valid_targets_mean": 8544.9,
|
||
|
|
"valid_targets_min": 3261
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.865979381443299,
|
||
|
|
"grad_norm": 0.26122086251082494,
|
||
|
|
"learning_rate": 1.994858282826588e-05,
|
||
|
|
"loss": 0.3641,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11261369287967682,
|
||
|
|
"step": 375,
|
||
|
|
"valid_targets_mean": 7284.2,
|
||
|
|
"valid_targets_min": 1841
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.917525773195876,
|
||
|
|
"grad_norm": 0.27407300785810573,
|
||
|
|
"learning_rate": 1.9434485871247414e-05,
|
||
|
|
"loss": 0.3724,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11535194516181946,
|
||
|
|
"step": 380,
|
||
|
|
"valid_targets_mean": 7251.3,
|
||
|
|
"valid_targets_min": 385
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9690721649484537,
|
||
|
|
"grad_norm": 0.23704812602782077,
|
||
|
|
"learning_rate": 1.8920762660504254e-05,
|
||
|
|
"loss": 0.3626,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11967391520738602,
|
||
|
|
"step": 385,
|
||
|
|
"valid_targets_mean": 7830.1,
|
||
|
|
"valid_targets_min": 2526
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.020618556701031,
|
||
|
|
"grad_norm": 0.25234053146240715,
|
||
|
|
"learning_rate": 1.840775271387831e-05,
|
||
|
|
"loss": 0.3648,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1164061427116394,
|
||
|
|
"step": 390,
|
||
|
|
"valid_targets_mean": 7833.2,
|
||
|
|
"valid_targets_min": 3013
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.072164948453608,
|
||
|
|
"grad_norm": 0.25269275602633845,
|
||
|
|
"learning_rate": 1.7895795077817766e-05,
|
||
|
|
"loss": 0.3571,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.116996631026268,
|
||
|
|
"step": 395,
|
||
|
|
"valid_targets_mean": 7916.6,
|
||
|
|
"valid_targets_min": 2811
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.123711340206185,
|
||
|
|
"grad_norm": 0.24708635846902258,
|
||
|
|
"learning_rate": 1.7385228103302486e-05,
|
||
|
|
"loss": 0.363,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11654889583587646,
|
||
|
|
"step": 400,
|
||
|
|
"valid_targets_mean": 6906.6,
|
||
|
|
"valid_targets_min": 2485
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.175257731958763,
|
||
|
|
"grad_norm": 0.2743304557546906,
|
||
|
|
"learning_rate": 1.6876389222229095e-05,
|
||
|
|
"loss": 0.3653,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.114975705742836,
|
||
|
|
"step": 405,
|
||
|
|
"valid_targets_mean": 6983.0,
|
||
|
|
"valid_targets_min": 3140
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.22680412371134,
|
||
|
|
"grad_norm": 0.25944424786243014,
|
||
|
|
"learning_rate": 1.6369614724403374e-05,
|
||
|
|
"loss": 0.3585,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11670075356960297,
|
||
|
|
"step": 410,
|
||
|
|
"valid_targets_mean": 7260.9,
|
||
|
|
"valid_targets_min": 2416
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.278350515463917,
|
||
|
|
"grad_norm": 0.2503809393194736,
|
||
|
|
"learning_rate": 1.5865239535287603e-05,
|
||
|
|
"loss": 0.3584,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11413806676864624,
|
||
|
|
"step": 415,
|
||
|
|
"valid_targets_mean": 6915.8,
|
||
|
|
"valid_targets_min": 2634
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.329896907216495,
|
||
|
|
"grad_norm": 0.23553236345958803,
|
||
|
|
"learning_rate": 1.5363596994649433e-05,
|
||
|
|
"loss": 0.3616,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10793957114219666,
|
||
|
|
"step": 420,
|
||
|
|
"valid_targets_mean": 6831.2,
|
||
|
|
"valid_targets_min": 724
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.381443298969073,
|
||
|
|
"grad_norm": 0.2560177473281376,
|
||
|
|
"learning_rate": 1.4865018636258902e-05,
|
||
|
|
"loss": 0.3599,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11091448366641998,
|
||
|
|
"step": 425,
|
||
|
|
"valid_targets_mean": 6891.7,
|
||
|
|
"valid_targets_min": 2602
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.43298969072165,
|
||
|
|
"grad_norm": 0.21796050318357524,
|
||
|
|
"learning_rate": 1.4369833968778868e-05,
|
||
|
|
"loss": 0.3585,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.110523521900177,
|
||
|
|
"step": 430,
|
||
|
|
"valid_targets_mean": 7605.9,
|
||
|
|
"valid_targets_min": 1426
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.484536082474227,
|
||
|
|
"grad_norm": 0.23809540416717972,
|
||
|
|
"learning_rate": 1.3878370257993954e-05,
|
||
|
|
"loss": 0.3572,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12010036408901215,
|
||
|
|
"step": 435,
|
||
|
|
"valid_targets_mean": 7319.8,
|
||
|
|
"valid_targets_min": 2885
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.536082474226804,
|
||
|
|
"grad_norm": 0.28145142462459166,
|
||
|
|
"learning_rate": 1.3390952310521765e-05,
|
||
|
|
"loss": 0.3623,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13759326934814453,
|
||
|
|
"step": 440,
|
||
|
|
"valid_targets_mean": 8758.2,
|
||
|
|
"valid_targets_min": 2394
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.587628865979381,
|
||
|
|
"grad_norm": 0.2371669525246716,
|
||
|
|
"learning_rate": 1.2907902259149287e-05,
|
||
|
|
"loss": 0.3655,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11463523656129837,
|
||
|
|
"step": 445,
|
||
|
|
"valid_targets_mean": 7297.6,
|
||
|
|
"valid_targets_min": 2823
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.639175257731958,
|
||
|
|
"grad_norm": 0.24107613993303856,
|
||
|
|
"learning_rate": 1.2429539349936567e-05,
|
||
|
|
"loss": 0.3624,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10125042498111725,
|
||
|
|
"step": 450,
|
||
|
|
"valid_targets_mean": 6893.2,
|
||
|
|
"valid_targets_min": 2640
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.690721649484536,
|
||
|
|
"grad_norm": 0.2515233302611333,
|
||
|
|
"learning_rate": 1.1956179731228033e-05,
|
||
|
|
"loss": 0.356,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13197514414787292,
|
||
|
|
"step": 455,
|
||
|
|
"valid_targets_mean": 8226.9,
|
||
|
|
"valid_targets_min": 2295
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.742268041237113,
|
||
|
|
"grad_norm": 0.2942969848339942,
|
||
|
|
"learning_rate": 1.1488136244711254e-05,
|
||
|
|
"loss": 0.3608,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11008858680725098,
|
||
|
|
"step": 460,
|
||
|
|
"valid_targets_mean": 5676.1,
|
||
|
|
"valid_targets_min": 798
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.793814432989691,
|
||
|
|
"grad_norm": 0.23903289400004588,
|
||
|
|
"learning_rate": 1.1025718218660947e-05,
|
||
|
|
"loss": 0.3555,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11992950737476349,
|
||
|
|
"step": 465,
|
||
|
|
"valid_targets_mean": 7718.1,
|
||
|
|
"valid_targets_min": 3261
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.845360824742268,
|
||
|
|
"grad_norm": 0.2603996712375844,
|
||
|
|
"learning_rate": 1.056923126350499e-05,
|
||
|
|
"loss": 0.3538,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1116849035024643,
|
||
|
|
"step": 470,
|
||
|
|
"valid_targets_mean": 6496.6,
|
||
|
|
"valid_targets_min": 2910
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.896907216494846,
|
||
|
|
"grad_norm": 0.25231124355945356,
|
||
|
|
"learning_rate": 1.0118977069847661e-05,
|
||
|
|
"loss": 0.3563,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10944770276546478,
|
||
|
|
"step": 475,
|
||
|
|
"valid_targets_mean": 7013.9,
|
||
|
|
"valid_targets_min": 2827
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.948453608247423,
|
||
|
|
"grad_norm": 0.2594062407842193,
|
||
|
|
"learning_rate": 9.675253209083315e-06,
|
||
|
|
"loss": 0.3625,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12593770027160645,
|
||
|
|
"step": 480,
|
||
|
|
"valid_targets_mean": 8752.2,
|
||
|
|
"valid_targets_min": 2428
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.0,
|
||
|
|
"grad_norm": 0.24490141683147196,
|
||
|
|
"learning_rate": 9.238352936732549e-06,
|
||
|
|
"loss": 0.3567,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11434824764728546,
|
||
|
|
"step": 485,
|
||
|
|
"valid_targets_mean": 7529.8,
|
||
|
|
"valid_targets_min": 2913
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.051546391752577,
|
||
|
|
"grad_norm": 0.23469157694081677,
|
||
|
|
"learning_rate": 8.808564998630639e-06,
|
||
|
|
"loss": 0.3568,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11697466671466827,
|
||
|
|
"step": 490,
|
||
|
|
"valid_targets_mean": 8082.2,
|
||
|
|
"valid_targets_min": 3052
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.103092783505154,
|
||
|
|
"grad_norm": 0.2392346881419023,
|
||
|
|
"learning_rate": 8.386173440096499e-06,
|
||
|
|
"loss": 0.3537,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12970125675201416,
|
||
|
|
"step": 495,
|
||
|
|
"valid_targets_mean": 8101.6,
|
||
|
|
"valid_targets_min": 2321
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.154639175257732,
|
||
|
|
"grad_norm": 0.23536936233851433,
|
||
|
|
"learning_rate": 7.971457418208081e-06,
|
||
|
|
"loss": 0.3592,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11510544270277023,
|
||
|
|
"step": 500,
|
||
|
|
"valid_targets_mean": 7822.4,
|
||
|
|
"valid_targets_min": 675
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.206185567010309,
|
||
|
|
"grad_norm": 0.22475393543787167,
|
||
|
|
"learning_rate": 7.56469101730847e-06,
|
||
|
|
"loss": 0.355,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1176222562789917,
|
||
|
|
"step": 505,
|
||
|
|
"valid_targets_mean": 8190.4,
|
||
|
|
"valid_targets_min": 3324
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.257731958762887,
|
||
|
|
"grad_norm": 0.22302648318258872,
|
||
|
|
"learning_rate": 7.1661430678645635e-06,
|
||
|
|
"loss": 0.3535,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10517322272062302,
|
||
|
|
"step": 510,
|
||
|
|
"valid_targets_mean": 7344.0,
|
||
|
|
"valid_targets_min": 2640
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.309278350515464,
|
||
|
|
"grad_norm": 0.2541892319612184,
|
||
|
|
"learning_rate": 6.776076968797906e-06,
|
||
|
|
"loss": 0.3516,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12366795539855957,
|
||
|
|
"step": 515,
|
||
|
|
"valid_targets_mean": 7618.8,
|
||
|
|
"valid_targets_min": 3046
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.360824742268041,
|
||
|
|
"grad_norm": 0.2413520158883131,
|
||
|
|
"learning_rate": 6.394750513405336e-06,
|
||
|
|
"loss": 0.3524,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13033932447433472,
|
||
|
|
"step": 520,
|
||
|
|
"valid_targets_mean": 7899.2,
|
||
|
|
"valid_targets_min": 2831
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.412371134020619,
|
||
|
|
"grad_norm": 0.22801880397200894,
|
||
|
|
"learning_rate": 6.022415718984262e-06,
|
||
|
|
"loss": 0.3514,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11976046860218048,
|
||
|
|
"step": 525,
|
||
|
|
"valid_targets_mean": 7971.2,
|
||
|
|
"valid_targets_min": 2428
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.463917525773196,
|
||
|
|
"grad_norm": 0.23901848100735304,
|
||
|
|
"learning_rate": 5.659318660275386e-06,
|
||
|
|
"loss": 0.3573,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1043827012181282,
|
||
|
|
"step": 530,
|
||
|
|
"valid_targets_mean": 6595.1,
|
||
|
|
"valid_targets_min": 2782
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.515463917525773,
|
||
|
|
"grad_norm": 0.22350153531585404,
|
||
|
|
"learning_rate": 5.305699306832708e-06,
|
||
|
|
"loss": 0.3525,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10420400649309158,
|
||
|
|
"step": 535,
|
||
|
|
"valid_targets_mean": 6950.8,
|
||
|
|
"valid_targets_min": 2368
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.56701030927835,
|
||
|
|
"grad_norm": 0.23132694800824183,
|
||
|
|
"learning_rate": 4.961791364428572e-06,
|
||
|
|
"loss": 0.3554,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12143643945455551,
|
||
|
|
"step": 540,
|
||
|
|
"valid_targets_mean": 7798.6,
|
||
|
|
"valid_targets_min": 1826
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.618556701030927,
|
||
|
|
"grad_norm": 0.23470350350591393,
|
||
|
|
"learning_rate": 4.627822120598327e-06,
|
||
|
|
"loss": 0.3511,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11844462156295776,
|
||
|
|
"step": 545,
|
||
|
|
"valid_targets_mean": 7074.9,
|
||
|
|
"valid_targets_min": 2973
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.670103092783505,
|
||
|
|
"grad_norm": 0.2266355635638624,
|
||
|
|
"learning_rate": 4.304012294426781e-06,
|
||
|
|
"loss": 0.3533,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13807958364486694,
|
||
|
|
"step": 550,
|
||
|
|
"valid_targets_mean": 9070.3,
|
||
|
|
"valid_targets_min": 4630
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.721649484536083,
|
||
|
|
"grad_norm": 0.21107606598774695,
|
||
|
|
"learning_rate": 3.990575890675787e-06,
|
||
|
|
"loss": 0.3531,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.119442418217659,
|
||
|
|
"step": 555,
|
||
|
|
"valid_targets_mean": 8072.2,
|
||
|
|
"valid_targets_min": 2885
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.77319587628866,
|
||
|
|
"grad_norm": 0.22780036688173633,
|
||
|
|
"learning_rate": 3.6877200583492202e-06,
|
||
|
|
"loss": 0.3555,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10234396159648895,
|
||
|
|
"step": 560,
|
||
|
|
"valid_targets_mean": 6451.7,
|
||
|
|
"valid_targets_min": 2184
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.824742268041237,
|
||
|
|
"grad_norm": 0.23683633035467933,
|
||
|
|
"learning_rate": 3.3956449537889545e-06,
|
||
|
|
"loss": 0.3577,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1127454936504364,
|
||
|
|
"step": 565,
|
||
|
|
"valid_targets_mean": 7794.7,
|
||
|
|
"valid_targets_min": 2496
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.876288659793815,
|
||
|
|
"grad_norm": 0.21476488292675394,
|
||
|
|
"learning_rate": 3.114543608392242e-06,
|
||
|
|
"loss": 0.3563,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11769910156726837,
|
||
|
|
"step": 570,
|
||
|
|
"valid_targets_mean": 8090.6,
|
||
|
|
"valid_targets_min": 2287
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.927835051546392,
|
||
|
|
"grad_norm": 0.224559045966915,
|
||
|
|
"learning_rate": 2.8446018010379584e-06,
|
||
|
|
"loss": 0.3572,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11433263123035431,
|
||
|
|
"step": 575,
|
||
|
|
"valid_targets_mean": 7847.3,
|
||
|
|
"valid_targets_min": 3411
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.979381443298969,
|
||
|
|
"grad_norm": 0.20902532049371286,
|
||
|
|
"learning_rate": 2.585997935306004e-06,
|
||
|
|
"loss": 0.3611,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13010427355766296,
|
||
|
|
"step": 580,
|
||
|
|
"valid_targets_mean": 8684.2,
|
||
|
|
"valid_targets_min": 2535
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.030927835051546,
|
||
|
|
"grad_norm": 0.2177808862824972,
|
||
|
|
"learning_rate": 2.3389029215709935e-06,
|
||
|
|
"loss": 0.3577,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1019224226474762,
|
||
|
|
"step": 585,
|
||
|
|
"valid_targets_mean": 7131.1,
|
||
|
|
"valid_targets_min": 2634
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.082474226804123,
|
||
|
|
"grad_norm": 0.23233054033967332,
|
||
|
|
"learning_rate": 2.1034800640482266e-06,
|
||
|
|
"loss": 0.3508,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10640084743499756,
|
||
|
|
"step": 590,
|
||
|
|
"valid_targets_mean": 7122.7,
|
||
|
|
"valid_targets_min": 3143
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.134020618556701,
|
||
|
|
"grad_norm": 0.20422103034075942,
|
||
|
|
"learning_rate": 1.8798849528664864e-06,
|
||
|
|
"loss": 0.3502,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11052947491407394,
|
||
|
|
"step": 595,
|
||
|
|
"valid_targets_mean": 7450.5,
|
||
|
|
"valid_targets_min": 811
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.185567010309279,
|
||
|
|
"grad_norm": 0.2270530210683725,
|
||
|
|
"learning_rate": 1.668265361239092e-06,
|
||
|
|
"loss": 0.36,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1277143955230713,
|
||
|
|
"step": 600,
|
||
|
|
"valid_targets_mean": 7666.4,
|
||
|
|
"valid_targets_min": 2053
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.237113402061856,
|
||
|
|
"grad_norm": 0.20938041332255236,
|
||
|
|
"learning_rate": 1.4687611478010943e-06,
|
||
|
|
"loss": 0.358,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13399212062358856,
|
||
|
|
"step": 605,
|
||
|
|
"valid_targets_mean": 8836.5,
|
||
|
|
"valid_targets_min": 2966
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.288659793814433,
|
||
|
|
"grad_norm": 0.24121864822302883,
|
||
|
|
"learning_rate": 1.281504164177232e-06,
|
||
|
|
"loss": 0.3541,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10749071091413498,
|
||
|
|
"step": 610,
|
||
|
|
"valid_targets_mean": 7254.9,
|
||
|
|
"valid_targets_min": 1326
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.34020618556701,
|
||
|
|
"grad_norm": 0.20849782657668386,
|
||
|
|
"learning_rate": 1.1066181678416266e-06,
|
||
|
|
"loss": 0.3555,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12396785616874695,
|
||
|
|
"step": 615,
|
||
|
|
"valid_targets_mean": 7806.3,
|
||
|
|
"valid_targets_min": 3761
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.391752577319588,
|
||
|
|
"grad_norm": 0.21568836694053126,
|
||
|
|
"learning_rate": 9.442187403269187e-07,
|
||
|
|
"loss": 0.3495,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13657146692276,
|
||
|
|
"step": 620,
|
||
|
|
"valid_targets_mean": 8428.0,
|
||
|
|
"valid_targets_min": 2749
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.443298969072165,
|
||
|
|
"grad_norm": 0.21131510926443703,
|
||
|
|
"learning_rate": 7.94413210836864e-07,
|
||
|
|
"loss": 0.3638,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11734045296907425,
|
||
|
|
"step": 625,
|
||
|
|
"valid_targets_mean": 7591.2,
|
||
|
|
"valid_targets_min": 2599
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.494845360824742,
|
||
|
|
"grad_norm": 0.2070549152110076,
|
||
|
|
"learning_rate": 6.573005853128145e-07,
|
||
|
|
"loss": 0.3461,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11593832820653915,
|
||
|
|
"step": 630,
|
||
|
|
"valid_targets_mean": 7376.8,
|
||
|
|
"valid_targets_min": 3374
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.546391752577319,
|
||
|
|
"grad_norm": 0.23331512697838694,
|
||
|
|
"learning_rate": 5.3297148100107e-07,
|
||
|
|
"loss": 0.3555,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11142148077487946,
|
||
|
|
"step": 635,
|
||
|
|
"valid_targets_mean": 7318.5,
|
||
|
|
"valid_targets_min": 3467
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.597938144329897,
|
||
|
|
"grad_norm": 0.21615996095685228,
|
||
|
|
"learning_rate": 4.2150806656424014e-07,
|
||
|
|
"loss": 0.3533,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11097370088100433,
|
||
|
|
"step": 640,
|
||
|
|
"valid_targets_mean": 8043.0,
|
||
|
|
"valid_targets_min": 1933
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.649484536082475,
|
||
|
|
"grad_norm": 0.20398533955581283,
|
||
|
|
"learning_rate": 3.2298400777629027e-07,
|
||
|
|
"loss": 0.3534,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10068753361701965,
|
||
|
|
"step": 645,
|
||
|
|
"valid_targets_mean": 7075.9,
|
||
|
|
"valid_targets_min": 2521
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.701030927835052,
|
||
|
|
"grad_norm": 0.21078736226133643,
|
||
|
|
"learning_rate": 2.3746441883707006e-07,
|
||
|
|
"loss": 0.352,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10676050931215286,
|
||
|
|
"step": 650,
|
||
|
|
"valid_targets_mean": 7000.8,
|
||
|
|
"valid_targets_min": 2447
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.752577319587629,
|
||
|
|
"grad_norm": 0.20951532441960802,
|
||
|
|
"learning_rate": 1.6500581933859284e-07,
|
||
|
|
"loss": 0.3499,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12206561863422394,
|
||
|
|
"step": 655,
|
||
|
|
"valid_targets_mean": 8637.4,
|
||
|
|
"valid_targets_min": 3948
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.804123711340206,
|
||
|
|
"grad_norm": 0.22052352403089653,
|
||
|
|
"learning_rate": 1.0565609691142398e-07,
|
||
|
|
"loss": 0.3521,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11783221364021301,
|
||
|
|
"step": 660,
|
||
|
|
"valid_targets_mean": 7440.9,
|
||
|
|
"valid_targets_min": 2162
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.855670103092783,
|
||
|
|
"grad_norm": 0.23528393884116397,
|
||
|
|
"learning_rate": 5.9454475575917434e-08,
|
||
|
|
"loss": 0.3479,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1227865070104599,
|
||
|
|
"step": 665,
|
||
|
|
"valid_targets_mean": 7391.8,
|
||
|
|
"valid_targets_min": 2973
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.907216494845361,
|
||
|
|
"grad_norm": 0.2339135522469977,
|
||
|
|
"learning_rate": 2.6431489819207512e-08,
|
||
|
|
"loss": 0.3466,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11582382023334503,
|
||
|
|
"step": 670,
|
||
|
|
"valid_targets_mean": 7328.8,
|
||
|
|
"valid_targets_min": 2062
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.958762886597938,
|
||
|
|
"grad_norm": 0.21912615869468077,
|
||
|
|
"learning_rate": 6.608964415066865e-09,
|
||
|
|
"loss": 0.3576,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11818139255046844,
|
||
|
|
"step": 675,
|
||
|
|
"valid_targets_mean": 7593.4,
|
||
|
|
"valid_targets_min": 2662
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 7.0,
|
||
|
|
"step": 679,
|
||
|
|
"total_flos": 2.2525720316251996e+18,
|
||
|
|
"train_loss": 0.0,
|
||
|
|
"train_runtime": 1.2685,
|
||
|
|
"train_samples_per_second": 51247.723,
|
||
|
|
"train_steps_per_second": 535.267
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"logging_steps": 5,
|
||
|
|
"max_steps": 679,
|
||
|
|
"num_input_tokens_seen": 0,
|
||
|
|
"num_train_epochs": 7,
|
||
|
|
"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": 2.2525720316251996e+18,
|
||
|
|
"train_batch_size": 1,
|
||
|
|
"trial_name": null,
|
||
|
|
"trial_params": null
|
||
|
|
}
|