Files
nemotron-31600-opt100k__Qwe…/trainer_state.json
ModelHub XC b271f3bf83 初始化项目,由ModelHub XC社区提供模型
Model: laion/nemotron-31600-opt100k__Qwen3-8B
Source: Original Platform
2026-05-09 17:15:10 +08:00

3678 lines
102 KiB
JSON

{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 5.0,
"eval_steps": 500,
"global_step": 1650,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.015182186234817813,
"grad_norm": 9.767456630329582,
"learning_rate": 9.696969696969698e-07,
"loss": 0.9495,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.30272722244262695,
"step": 5,
"valid_targets_mean": 10378.4,
"valid_targets_min": 1358
},
{
"epoch": 0.030364372469635626,
"grad_norm": 6.271278770967426,
"learning_rate": 2.181818181818182e-06,
"loss": 0.9318,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.31232553720474243,
"step": 10,
"valid_targets_mean": 10926.1,
"valid_targets_min": 1793
},
{
"epoch": 0.04554655870445344,
"grad_norm": 2.8647197149421197,
"learning_rate": 3.3939393939393946e-06,
"loss": 0.8663,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.30162012577056885,
"step": 15,
"valid_targets_mean": 11718.9,
"valid_targets_min": 2968
},
{
"epoch": 0.06072874493927125,
"grad_norm": 1.5728241227206552,
"learning_rate": 4.606060606060606e-06,
"loss": 0.8034,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.23369714617729187,
"step": 20,
"valid_targets_mean": 9150.8,
"valid_targets_min": 2866
},
{
"epoch": 0.07591093117408906,
"grad_norm": 1.298780515091113,
"learning_rate": 5.8181818181818185e-06,
"loss": 0.7551,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.25340479612350464,
"step": 25,
"valid_targets_mean": 9598.5,
"valid_targets_min": 1022
},
{
"epoch": 0.09109311740890688,
"grad_norm": 0.753479537539852,
"learning_rate": 7.030303030303031e-06,
"loss": 0.7108,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.24694345891475677,
"step": 30,
"valid_targets_mean": 11339.5,
"valid_targets_min": 2455
},
{
"epoch": 0.1062753036437247,
"grad_norm": 0.5456863914213468,
"learning_rate": 8.242424242424243e-06,
"loss": 0.6935,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.23709480464458466,
"step": 35,
"valid_targets_mean": 10767.9,
"valid_targets_min": 2190
},
{
"epoch": 0.1214574898785425,
"grad_norm": 0.47461380782907614,
"learning_rate": 9.454545454545456e-06,
"loss": 0.6711,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.22755438089370728,
"step": 40,
"valid_targets_mean": 11884.0,
"valid_targets_min": 1925
},
{
"epoch": 0.13663967611336034,
"grad_norm": 0.4106448655482926,
"learning_rate": 1.0666666666666667e-05,
"loss": 0.6459,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.21281057596206665,
"step": 45,
"valid_targets_mean": 9801.6,
"valid_targets_min": 1654
},
{
"epoch": 0.15182186234817813,
"grad_norm": 0.35120072155847837,
"learning_rate": 1.187878787878788e-05,
"loss": 0.6144,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.21171334385871887,
"step": 50,
"valid_targets_mean": 11687.2,
"valid_targets_min": 2243
},
{
"epoch": 0.16700404858299595,
"grad_norm": 0.2896997146592483,
"learning_rate": 1.3090909090909092e-05,
"loss": 0.5917,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.18080435693264008,
"step": 55,
"valid_targets_mean": 10539.2,
"valid_targets_min": 1901
},
{
"epoch": 0.18218623481781376,
"grad_norm": 0.23549511279092397,
"learning_rate": 1.4303030303030305e-05,
"loss": 0.5778,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.18052560091018677,
"step": 60,
"valid_targets_mean": 10891.6,
"valid_targets_min": 3467
},
{
"epoch": 0.19736842105263158,
"grad_norm": 0.22554924567011164,
"learning_rate": 1.5515151515151516e-05,
"loss": 0.5724,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.20011213421821594,
"step": 65,
"valid_targets_mean": 12352.1,
"valid_targets_min": 2287
},
{
"epoch": 0.2125506072874494,
"grad_norm": 0.23069171787024742,
"learning_rate": 1.672727272727273e-05,
"loss": 0.5582,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.17292290925979614,
"step": 70,
"valid_targets_mean": 8332.5,
"valid_targets_min": 2508
},
{
"epoch": 0.22773279352226722,
"grad_norm": 0.24300716935495,
"learning_rate": 1.7939393939393942e-05,
"loss": 0.5431,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.18352295458316803,
"step": 75,
"valid_targets_mean": 10469.7,
"valid_targets_min": 3220
},
{
"epoch": 0.242914979757085,
"grad_norm": 0.2539389077011328,
"learning_rate": 1.9151515151515152e-05,
"loss": 0.5426,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.17171943187713623,
"step": 80,
"valid_targets_mean": 11575.8,
"valid_targets_min": 3367
},
{
"epoch": 0.25809716599190285,
"grad_norm": 0.26124706332109904,
"learning_rate": 2.0363636363636365e-05,
"loss": 0.5298,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1690063625574112,
"step": 85,
"valid_targets_mean": 9496.4,
"valid_targets_min": 1415
},
{
"epoch": 0.2732793522267207,
"grad_norm": 0.2365981496426835,
"learning_rate": 2.1575757575757578e-05,
"loss": 0.5298,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1702321618795395,
"step": 90,
"valid_targets_mean": 11065.8,
"valid_targets_min": 766
},
{
"epoch": 0.28846153846153844,
"grad_norm": 0.26451757214209165,
"learning_rate": 2.278787878787879e-05,
"loss": 0.5302,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.16847875714302063,
"step": 95,
"valid_targets_mean": 9680.0,
"valid_targets_min": 290
},
{
"epoch": 0.30364372469635625,
"grad_norm": 0.2741299353525031,
"learning_rate": 2.4e-05,
"loss": 0.521,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1615229696035385,
"step": 100,
"valid_targets_mean": 9314.8,
"valid_targets_min": 975
},
{
"epoch": 0.3188259109311741,
"grad_norm": 0.2992218104861107,
"learning_rate": 2.5212121212121214e-05,
"loss": 0.5119,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.17057791352272034,
"step": 105,
"valid_targets_mean": 10264.6,
"valid_targets_min": 777
},
{
"epoch": 0.3340080971659919,
"grad_norm": 0.32299047345991855,
"learning_rate": 2.6424242424242427e-05,
"loss": 0.5086,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.19047002494335175,
"step": 110,
"valid_targets_mean": 10825.8,
"valid_targets_min": 4362
},
{
"epoch": 0.3491902834008097,
"grad_norm": 0.29045987481699725,
"learning_rate": 2.763636363636364e-05,
"loss": 0.5206,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.17592790722846985,
"step": 115,
"valid_targets_mean": 10548.6,
"valid_targets_min": 2197
},
{
"epoch": 0.3643724696356275,
"grad_norm": 0.26637442919498494,
"learning_rate": 2.884848484848485e-05,
"loss": 0.5004,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13782358169555664,
"step": 120,
"valid_targets_mean": 9174.8,
"valid_targets_min": 1757
},
{
"epoch": 0.37955465587044535,
"grad_norm": 0.27717254269356884,
"learning_rate": 3.0060606060606062e-05,
"loss": 0.5053,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.17009928822517395,
"step": 125,
"valid_targets_mean": 10493.4,
"valid_targets_min": 2896
},
{
"epoch": 0.39473684210526316,
"grad_norm": 0.26050759497187553,
"learning_rate": 3.127272727272728e-05,
"loss": 0.5047,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1754511296749115,
"step": 130,
"valid_targets_mean": 11562.1,
"valid_targets_min": 2774
},
{
"epoch": 0.409919028340081,
"grad_norm": 0.2765139416546145,
"learning_rate": 3.2484848484848485e-05,
"loss": 0.4994,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.15204985439777374,
"step": 135,
"valid_targets_mean": 9299.8,
"valid_targets_min": 2076
},
{
"epoch": 0.4251012145748988,
"grad_norm": 0.27349224278807954,
"learning_rate": 3.36969696969697e-05,
"loss": 0.4968,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.16537371277809143,
"step": 140,
"valid_targets_mean": 10670.5,
"valid_targets_min": 343
},
{
"epoch": 0.4402834008097166,
"grad_norm": 0.2891276908365685,
"learning_rate": 3.490909090909091e-05,
"loss": 0.4941,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.17477017641067505,
"step": 145,
"valid_targets_mean": 11143.9,
"valid_targets_min": 1940
},
{
"epoch": 0.45546558704453444,
"grad_norm": 0.2879865399652702,
"learning_rate": 3.6121212121212124e-05,
"loss": 0.4921,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.176496684551239,
"step": 150,
"valid_targets_mean": 10958.7,
"valid_targets_min": 3124
},
{
"epoch": 0.4706477732793522,
"grad_norm": 0.287556416985574,
"learning_rate": 3.733333333333334e-05,
"loss": 0.4877,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1479555368423462,
"step": 155,
"valid_targets_mean": 9051.1,
"valid_targets_min": 1508
},
{
"epoch": 0.48582995951417,
"grad_norm": 0.2864787839954415,
"learning_rate": 3.854545454545455e-05,
"loss": 0.4942,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1827171891927719,
"step": 160,
"valid_targets_mean": 10364.4,
"valid_targets_min": 164
},
{
"epoch": 0.5010121457489879,
"grad_norm": 0.25566057907575657,
"learning_rate": 3.9757575757575757e-05,
"loss": 0.4833,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.16354942321777344,
"step": 165,
"valid_targets_mean": 11585.2,
"valid_targets_min": 3505
},
{
"epoch": 0.5161943319838057,
"grad_norm": 0.3454913577704521,
"learning_rate": 3.999928391557286e-05,
"loss": 0.4816,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12809130549430847,
"step": 170,
"valid_targets_mean": 8255.8,
"valid_targets_min": 996
},
{
"epoch": 0.5313765182186235,
"grad_norm": 0.25232423038040575,
"learning_rate": 3.999637491047052e-05,
"loss": 0.4898,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14153242111206055,
"step": 175,
"valid_targets_mean": 10202.8,
"valid_targets_min": 3853
},
{
"epoch": 0.5465587044534413,
"grad_norm": 0.38999173724416125,
"learning_rate": 3.999122855464813e-05,
"loss": 0.4903,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.19011259078979492,
"step": 180,
"valid_targets_mean": 12055.8,
"valid_targets_min": 1761
},
{
"epoch": 0.5617408906882592,
"grad_norm": 0.3358629796944954,
"learning_rate": 3.998384542392021e-05,
"loss": 0.4868,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.16506445407867432,
"step": 185,
"valid_targets_mean": 10798.4,
"valid_targets_min": 1793
},
{
"epoch": 0.5769230769230769,
"grad_norm": 0.2600759867395087,
"learning_rate": 3.9974226344369124e-05,
"loss": 0.4869,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.15860313177108765,
"step": 190,
"valid_targets_mean": 9988.2,
"valid_targets_min": 1149
},
{
"epoch": 0.5921052631578947,
"grad_norm": 0.260677411587491,
"learning_rate": 3.996237239225268e-05,
"loss": 0.4863,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.16243574023246765,
"step": 195,
"valid_targets_mean": 10358.2,
"valid_targets_min": 3235
},
{
"epoch": 0.6072874493927125,
"grad_norm": 0.2645245975670984,
"learning_rate": 3.994828489388371e-05,
"loss": 0.4822,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1531163454055786,
"step": 200,
"valid_targets_mean": 10201.0,
"valid_targets_min": 1772
},
{
"epoch": 0.6224696356275303,
"grad_norm": 0.25227333524594137,
"learning_rate": 3.993196542548162e-05,
"loss": 0.4818,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14635561406612396,
"step": 205,
"valid_targets_mean": 9432.2,
"valid_targets_min": 1884
},
{
"epoch": 0.6376518218623481,
"grad_norm": 0.28903743175057034,
"learning_rate": 3.991341581299609e-05,
"loss": 0.4835,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.17242132127285004,
"step": 210,
"valid_targets_mean": 10711.5,
"valid_targets_min": 1318
},
{
"epoch": 0.652834008097166,
"grad_norm": 0.2638604812189626,
"learning_rate": 3.9892638131902765e-05,
"loss": 0.4796,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.16491654515266418,
"step": 215,
"valid_targets_mean": 11328.2,
"valid_targets_min": 1957
},
{
"epoch": 0.6680161943319838,
"grad_norm": 0.27154894774630783,
"learning_rate": 3.9869634706971e-05,
"loss": 0.4751,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14194658398628235,
"step": 220,
"valid_targets_mean": 9687.9,
"valid_targets_min": 3468
},
{
"epoch": 0.6831983805668016,
"grad_norm": 0.25557034456032535,
"learning_rate": 3.984440811200379e-05,
"loss": 0.4661,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1502843201160431,
"step": 225,
"valid_targets_mean": 9392.2,
"valid_targets_min": 2945
},
{
"epoch": 0.6983805668016194,
"grad_norm": 0.31261870349322307,
"learning_rate": 3.981696116954973e-05,
"loss": 0.4849,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1625722348690033,
"step": 230,
"valid_targets_mean": 11870.0,
"valid_targets_min": 3248
},
{
"epoch": 0.7135627530364372,
"grad_norm": 0.2785550873091526,
"learning_rate": 3.978729695058729e-05,
"loss": 0.4818,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.17710517346858978,
"step": 235,
"valid_targets_mean": 10057.5,
"valid_targets_min": 755
},
{
"epoch": 0.728744939271255,
"grad_norm": 0.3037308974545316,
"learning_rate": 3.9755418774181146e-05,
"loss": 0.4724,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1543278694152832,
"step": 240,
"valid_targets_mean": 10055.2,
"valid_targets_min": 1140
},
{
"epoch": 0.7439271255060729,
"grad_norm": 0.2814092333238354,
"learning_rate": 3.9721330207110835e-05,
"loss": 0.4683,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13317811489105225,
"step": 245,
"valid_targets_mean": 8725.9,
"valid_targets_min": 1518
},
{
"epoch": 0.7591093117408907,
"grad_norm": 0.2937104349412505,
"learning_rate": 3.9685035063471675e-05,
"loss": 0.4747,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.17623022198677063,
"step": 250,
"valid_targets_mean": 11871.8,
"valid_targets_min": 1440
},
{
"epoch": 0.7742914979757085,
"grad_norm": 0.30478426754913396,
"learning_rate": 3.964653740424804e-05,
"loss": 0.4715,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.16203227639198303,
"step": 255,
"valid_targets_mean": 9864.8,
"valid_targets_min": 1364
},
{
"epoch": 0.7894736842105263,
"grad_norm": 0.31658164073789935,
"learning_rate": 3.960584153685895e-05,
"loss": 0.4636,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1387915015220642,
"step": 260,
"valid_targets_mean": 10663.3,
"valid_targets_min": 1778
},
{
"epoch": 0.8046558704453441,
"grad_norm": 0.38421252432540776,
"learning_rate": 3.9562952014676116e-05,
"loss": 0.4716,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13185453414916992,
"step": 265,
"valid_targets_mean": 10560.4,
"valid_targets_min": 3376
},
{
"epoch": 0.819838056680162,
"grad_norm": 0.31563470556487117,
"learning_rate": 3.9517873636514525e-05,
"loss": 0.464,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1441793292760849,
"step": 270,
"valid_targets_mean": 10806.7,
"valid_targets_min": 2834
},
{
"epoch": 0.8350202429149798,
"grad_norm": 0.24166282014582607,
"learning_rate": 3.947061144609546e-05,
"loss": 0.4638,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1455232948064804,
"step": 275,
"valid_targets_mean": 9412.5,
"valid_targets_min": 1636
},
{
"epoch": 0.8502024291497976,
"grad_norm": 0.28019156605196904,
"learning_rate": 3.942117073148221e-05,
"loss": 0.4735,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.17822107672691345,
"step": 280,
"valid_targets_mean": 11960.5,
"valid_targets_min": 3625
},
{
"epoch": 0.8653846153846154,
"grad_norm": 0.29544332084190916,
"learning_rate": 3.9369557024488345e-05,
"loss": 0.466,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13360615074634552,
"step": 285,
"valid_targets_mean": 8446.1,
"valid_targets_min": 2648
},
{
"epoch": 0.8805668016194332,
"grad_norm": 0.2521821962950855,
"learning_rate": 3.931577610005883e-05,
"loss": 0.4677,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1583632528781891,
"step": 290,
"valid_targets_mean": 10544.0,
"valid_targets_min": 2552
},
{
"epoch": 0.895748987854251,
"grad_norm": 0.2649498200492295,
"learning_rate": 3.925983397562385e-05,
"loss": 0.465,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.15108221769332886,
"step": 295,
"valid_targets_mean": 9534.8,
"valid_targets_min": 322
},
{
"epoch": 0.9109311740890689,
"grad_norm": 0.2705724371949053,
"learning_rate": 3.920173691042554e-05,
"loss": 0.4652,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1575058102607727,
"step": 300,
"valid_targets_mean": 9826.8,
"valid_targets_min": 2599
},
{
"epoch": 0.9261133603238867,
"grad_norm": 0.25599146225296465,
"learning_rate": 3.914149140481766e-05,
"loss": 0.4665,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.17281994223594666,
"step": 305,
"valid_targets_mean": 11182.3,
"valid_targets_min": 1557
},
{
"epoch": 0.9412955465587044,
"grad_norm": 0.2622184905952597,
"learning_rate": 3.9079104199538256e-05,
"loss": 0.4737,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1571318507194519,
"step": 310,
"valid_targets_mean": 9966.4,
"valid_targets_min": 2376
},
{
"epoch": 0.9564777327935222,
"grad_norm": 0.24759698838011684,
"learning_rate": 3.901458227495549e-05,
"loss": 0.4538,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.15467405319213867,
"step": 315,
"valid_targets_mean": 10853.2,
"valid_targets_min": 4115
},
{
"epoch": 0.97165991902834,
"grad_norm": 0.23434699678802617,
"learning_rate": 3.8947932850286585e-05,
"loss": 0.4666,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.15081250667572021,
"step": 320,
"valid_targets_mean": 10789.0,
"valid_targets_min": 1678
},
{
"epoch": 0.9868421052631579,
"grad_norm": 0.27532576364760486,
"learning_rate": 3.887916338279014e-05,
"loss": 0.4708,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.18342334032058716,
"step": 325,
"valid_targets_mean": 10936.7,
"valid_targets_min": 1649
},
{
"epoch": 1.0,
"grad_norm": 0.35899681080322277,
"learning_rate": 3.8808281566931675e-05,
"loss": 0.4468,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.40738117694854736,
"step": 330,
"valid_targets_mean": 10492.2,
"valid_targets_min": 2847
},
{
"epoch": 1.0151821862348178,
"grad_norm": 0.28458334885625075,
"learning_rate": 3.873529533352277e-05,
"loss": 0.4469,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13354119658470154,
"step": 335,
"valid_targets_mean": 9683.1,
"valid_targets_min": 2160
},
{
"epoch": 1.0303643724696356,
"grad_norm": 0.354345323388759,
"learning_rate": 3.8660212848833705e-05,
"loss": 0.4414,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14219801127910614,
"step": 340,
"valid_targets_mean": 9913.6,
"valid_targets_min": 2191
},
{
"epoch": 1.0455465587044535,
"grad_norm": 0.3350719189242995,
"learning_rate": 3.858304251367972e-05,
"loss": 0.4407,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.15460778772830963,
"step": 345,
"valid_targets_mean": 10729.1,
"valid_targets_min": 1941
},
{
"epoch": 1.0607287449392713,
"grad_norm": 0.2729944048209113,
"learning_rate": 3.850379296248107e-05,
"loss": 0.4479,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1340738832950592,
"step": 350,
"valid_targets_mean": 9968.5,
"valid_targets_min": 2422
},
{
"epoch": 1.075910931174089,
"grad_norm": 0.22978531889990117,
"learning_rate": 3.8422473062297e-05,
"loss": 0.4402,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.17449653148651123,
"step": 355,
"valid_targets_mean": 12469.6,
"valid_targets_min": 755
},
{
"epoch": 1.091093117408907,
"grad_norm": 0.23887951948467376,
"learning_rate": 3.8339091911833545e-05,
"loss": 0.4404,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.16491574048995972,
"step": 360,
"valid_targets_mean": 11062.2,
"valid_targets_min": 3580
},
{
"epoch": 1.1062753036437247,
"grad_norm": 0.26815554514697304,
"learning_rate": 3.825365884042553e-05,
"loss": 0.4461,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12322846055030823,
"step": 365,
"valid_targets_mean": 8607.6,
"valid_targets_min": 2790
},
{
"epoch": 1.1214574898785425,
"grad_norm": 0.25346781569245475,
"learning_rate": 3.8166183406992745e-05,
"loss": 0.4519,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.15480399131774902,
"step": 370,
"valid_targets_mean": 10520.0,
"valid_targets_min": 1761
},
{
"epoch": 1.1366396761133604,
"grad_norm": 0.26272434617647855,
"learning_rate": 3.807667539897041e-05,
"loss": 0.4349,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1547757238149643,
"step": 375,
"valid_targets_mean": 11900.1,
"valid_targets_min": 2634
},
{
"epoch": 1.1518218623481782,
"grad_norm": 0.29964984986988785,
"learning_rate": 3.798514483121408e-05,
"loss": 0.4396,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13471867144107819,
"step": 380,
"valid_targets_mean": 9117.8,
"valid_targets_min": 1867
},
{
"epoch": 1.167004048582996,
"grad_norm": 0.2533460035751316,
"learning_rate": 3.789160194487908e-05,
"loss": 0.4383,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1802617311477661,
"step": 385,
"valid_targets_mean": 12489.5,
"valid_targets_min": 2196
},
{
"epoch": 1.1821862348178138,
"grad_norm": 0.28288452724215873,
"learning_rate": 3.7796057206274686e-05,
"loss": 0.4438,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1617857664823532,
"step": 390,
"valid_targets_mean": 11982.3,
"valid_targets_min": 2783
},
{
"epoch": 1.1973684210526316,
"grad_norm": 0.2526195634079488,
"learning_rate": 3.769852130569304e-05,
"loss": 0.4429,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13690868020057678,
"step": 395,
"valid_targets_mean": 9777.0,
"valid_targets_min": 2093
},
{
"epoch": 1.2125506072874495,
"grad_norm": 0.22129078972673513,
"learning_rate": 3.7599005156213066e-05,
"loss": 0.4366,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14769168198108673,
"step": 400,
"valid_targets_mean": 11107.2,
"valid_targets_min": 3098
},
{
"epoch": 1.2277327935222673,
"grad_norm": 0.25167748549310603,
"learning_rate": 3.74975198924794e-05,
"loss": 0.4389,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13846036791801453,
"step": 405,
"valid_targets_mean": 9990.3,
"valid_targets_min": 1426
},
{
"epoch": 1.242914979757085,
"grad_norm": 0.242028968841572,
"learning_rate": 3.739407686945658e-05,
"loss": 0.4401,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12689189612865448,
"step": 410,
"valid_targets_mean": 9368.6,
"valid_targets_min": 2146
},
{
"epoch": 1.258097165991903,
"grad_norm": 0.2475327665205262,
"learning_rate": 3.728868766115854e-05,
"loss": 0.4437,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.15651461482048035,
"step": 415,
"valid_targets_mean": 9742.8,
"valid_targets_min": 1110
},
{
"epoch": 1.2732793522267207,
"grad_norm": 0.24920519930764157,
"learning_rate": 3.718136405935365e-05,
"loss": 0.4432,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13602977991104126,
"step": 420,
"valid_targets_mean": 10792.3,
"valid_targets_min": 3021
},
{
"epoch": 1.2884615384615383,
"grad_norm": 0.2722218179265993,
"learning_rate": 3.707211807224534e-05,
"loss": 0.4336,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.16734597086906433,
"step": 425,
"valid_targets_mean": 11353.7,
"valid_targets_min": 1823
},
{
"epoch": 1.3036437246963564,
"grad_norm": 0.24861353772242864,
"learning_rate": 3.696096192312852e-05,
"loss": 0.4311,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14648747444152832,
"step": 430,
"valid_targets_mean": 10162.9,
"valid_targets_min": 1861
},
{
"epoch": 1.318825910931174,
"grad_norm": 0.24353293989235789,
"learning_rate": 3.684790804902199e-05,
"loss": 0.4349,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13794153928756714,
"step": 435,
"valid_targets_mean": 9891.2,
"valid_targets_min": 1204
},
{
"epoch": 1.334008097165992,
"grad_norm": 0.2538455688242792,
"learning_rate": 3.673296909927682e-05,
"loss": 0.4463,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13822206854820251,
"step": 440,
"valid_targets_mean": 9649.3,
"valid_targets_min": 1728
},
{
"epoch": 1.3491902834008096,
"grad_norm": 0.2195458402548249,
"learning_rate": 3.661615793416109e-05,
"loss": 0.4329,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1351986676454544,
"step": 445,
"valid_targets_mean": 9482.1,
"valid_targets_min": 1607
},
{
"epoch": 1.3643724696356276,
"grad_norm": 0.262149947406993,
"learning_rate": 3.649748762342098e-05,
"loss": 0.4393,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.15256932377815247,
"step": 450,
"valid_targets_mean": 10223.6,
"valid_targets_min": 1663
},
{
"epoch": 1.3795546558704452,
"grad_norm": 0.24170742573441606,
"learning_rate": 3.637697144481839e-05,
"loss": 0.4348,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14698657393455505,
"step": 455,
"valid_targets_mean": 10525.9,
"valid_targets_min": 2458
},
{
"epoch": 1.3947368421052633,
"grad_norm": 0.2831147102052386,
"learning_rate": 3.625462288264536e-05,
"loss": 0.4401,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1650058627128601,
"step": 460,
"valid_targets_mean": 11955.9,
"valid_targets_min": 2357
},
{
"epoch": 1.4099190283400809,
"grad_norm": 0.25507823085926157,
"learning_rate": 3.613045562621533e-05,
"loss": 0.4426,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.16729950904846191,
"step": 465,
"valid_targets_mean": 11234.8,
"valid_targets_min": 1488
},
{
"epoch": 1.425101214574899,
"grad_norm": 0.21517023605472518,
"learning_rate": 3.600448356833146e-05,
"loss": 0.4435,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14533230662345886,
"step": 470,
"valid_targets_mean": 10856.3,
"valid_targets_min": 1958
},
{
"epoch": 1.4402834008097165,
"grad_norm": 0.27274183444021904,
"learning_rate": 3.587672080373219e-05,
"loss": 0.4351,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.15404292941093445,
"step": 475,
"valid_targets_mean": 11907.4,
"valid_targets_min": 2037
},
{
"epoch": 1.4554655870445345,
"grad_norm": 0.27757044107139334,
"learning_rate": 3.574718162751426e-05,
"loss": 0.4362,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1617705523967743,
"step": 480,
"valid_targets_mean": 10082.9,
"valid_targets_min": 1439
},
{
"epoch": 1.4706477732793521,
"grad_norm": 0.22467367774993238,
"learning_rate": 3.561588053353319e-05,
"loss": 0.4448,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13222670555114746,
"step": 485,
"valid_targets_mean": 9991.5,
"valid_targets_min": 1576
},
{
"epoch": 1.48582995951417,
"grad_norm": 0.23355252705926008,
"learning_rate": 3.5482832212781655e-05,
"loss": 0.4333,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.18576061725616455,
"step": 490,
"valid_targets_mean": 12468.6,
"valid_targets_min": 1985
},
{
"epoch": 1.5010121457489878,
"grad_norm": 0.2181912791479774,
"learning_rate": 3.53480515517457e-05,
"loss": 0.4354,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1390918791294098,
"step": 495,
"valid_targets_mean": 9409.0,
"valid_targets_min": 1772
},
{
"epoch": 1.5161943319838058,
"grad_norm": 0.2604310615555627,
"learning_rate": 3.5211553630739166e-05,
"loss": 0.4423,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12898939847946167,
"step": 500,
"valid_targets_mean": 9808.9,
"valid_targets_min": 2193
},
{
"epoch": 1.5313765182186234,
"grad_norm": 0.26061930166812924,
"learning_rate": 3.5073353722216334e-05,
"loss": 0.4407,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.15980827808380127,
"step": 505,
"valid_targets_mean": 11388.4,
"valid_targets_min": 1640
},
{
"epoch": 1.5465587044534415,
"grad_norm": 0.25366664801826005,
"learning_rate": 3.4933467289063156e-05,
"loss": 0.4259,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13890908658504486,
"step": 510,
"valid_targets_mean": 9539.9,
"valid_targets_min": 2659
},
{
"epoch": 1.561740890688259,
"grad_norm": 0.22395372976390027,
"learning_rate": 3.4791909982867175e-05,
"loss": 0.4339,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1427781581878662,
"step": 515,
"valid_targets_mean": 10788.7,
"valid_targets_min": 2272
},
{
"epoch": 1.5769230769230769,
"grad_norm": 0.2509901501713803,
"learning_rate": 3.464869764216622e-05,
"loss": 0.4362,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1664656698703766,
"step": 520,
"valid_targets_mean": 10282.4,
"valid_targets_min": 1719
},
{
"epoch": 1.5921052631578947,
"grad_norm": 0.23050758734822147,
"learning_rate": 3.450384629067635e-05,
"loss": 0.4386,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14430655539035797,
"step": 525,
"valid_targets_mean": 11192.8,
"valid_targets_min": 1942
},
{
"epoch": 1.6072874493927125,
"grad_norm": 0.24432443613636357,
"learning_rate": 3.435737213549896e-05,
"loss": 0.425,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.15063516795635223,
"step": 530,
"valid_targets_mean": 11381.7,
"valid_targets_min": 1608
},
{
"epoch": 1.6224696356275303,
"grad_norm": 0.22577785062585196,
"learning_rate": 3.420929156530738e-05,
"loss": 0.4337,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1626434326171875,
"step": 535,
"valid_targets_mean": 11110.8,
"valid_targets_min": 2254
},
{
"epoch": 1.6376518218623481,
"grad_norm": 0.21783409731034523,
"learning_rate": 3.405962114851324e-05,
"loss": 0.4372,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1312275528907776,
"step": 540,
"valid_targets_mean": 9964.3,
"valid_targets_min": 891
},
{
"epoch": 1.652834008097166,
"grad_norm": 0.22363734173869207,
"learning_rate": 3.390837763141261e-05,
"loss": 0.4295,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14635321497917175,
"step": 545,
"valid_targets_mean": 10282.8,
"valid_targets_min": 442
},
{
"epoch": 1.6680161943319838,
"grad_norm": 0.28509501532395254,
"learning_rate": 3.3755577936312344e-05,
"loss": 0.4235,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11927291005849838,
"step": 550,
"valid_targets_mean": 8001.4,
"valid_targets_min": 1991
},
{
"epoch": 1.6831983805668016,
"grad_norm": 0.273518896683822,
"learning_rate": 3.360123915963662e-05,
"loss": 0.4325,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14016571640968323,
"step": 555,
"valid_targets_mean": 10077.3,
"valid_targets_min": 2843
},
{
"epoch": 1.6983805668016194,
"grad_norm": 0.21366230789574794,
"learning_rate": 3.3445378570014125e-05,
"loss": 0.4364,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11847387999296188,
"step": 560,
"valid_targets_mean": 8844.1,
"valid_targets_min": 1250
},
{
"epoch": 1.7135627530364372,
"grad_norm": 0.22680113952919098,
"learning_rate": 3.328801360634585e-05,
"loss": 0.4396,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1664324402809143,
"step": 565,
"valid_targets_mean": 12107.6,
"valid_targets_min": 2496
},
{
"epoch": 1.728744939271255,
"grad_norm": 0.22907607154865234,
"learning_rate": 3.312916187585392e-05,
"loss": 0.428,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1518704891204834,
"step": 570,
"valid_targets_mean": 11503.6,
"valid_targets_min": 2129
},
{
"epoch": 1.7439271255060729,
"grad_norm": 0.22554888976194104,
"learning_rate": 3.296884115211157e-05,
"loss": 0.4224,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1115855723619461,
"step": 575,
"valid_targets_mean": 9672.9,
"valid_targets_min": 2263
},
{
"epoch": 1.7591093117408907,
"grad_norm": 0.21649372224245605,
"learning_rate": 3.280706937305445e-05,
"loss": 0.437,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.16382913291454315,
"step": 580,
"valid_targets_mean": 11463.7,
"valid_targets_min": 2365
},
{
"epoch": 1.7742914979757085,
"grad_norm": 0.2536358792335515,
"learning_rate": 3.2643864638973645e-05,
"loss": 0.4336,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.15374597907066345,
"step": 585,
"valid_targets_mean": 10600.5,
"valid_targets_min": 2442
},
{
"epoch": 1.7894736842105263,
"grad_norm": 0.2447232506114259,
"learning_rate": 3.2479245210490434e-05,
"loss": 0.4363,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14378559589385986,
"step": 590,
"valid_targets_mean": 9931.5,
"valid_targets_min": 2331
},
{
"epoch": 1.8046558704453441,
"grad_norm": 0.22625023350242304,
"learning_rate": 3.2313229506513167e-05,
"loss": 0.4251,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14341717958450317,
"step": 595,
"valid_targets_mean": 11203.3,
"valid_targets_min": 3362
},
{
"epoch": 1.819838056680162,
"grad_norm": 0.22747844454240196,
"learning_rate": 3.2145836102176424e-05,
"loss": 0.4349,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12817899882793427,
"step": 600,
"valid_targets_mean": 9786.9,
"valid_targets_min": 2551
},
{
"epoch": 1.8350202429149798,
"grad_norm": 0.21461340841905202,
"learning_rate": 3.197708372676265e-05,
"loss": 0.4328,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1507049798965454,
"step": 605,
"valid_targets_mean": 11408.8,
"valid_targets_min": 3494
},
{
"epoch": 1.8502024291497976,
"grad_norm": 0.23227807829347027,
"learning_rate": 3.1806991261606604e-05,
"loss": 0.4275,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1595226526260376,
"step": 610,
"valid_targets_mean": 10987.9,
"valid_targets_min": 477
},
{
"epoch": 1.8653846153846154,
"grad_norm": 0.22934685685191428,
"learning_rate": 3.163557773798276e-05,
"loss": 0.441,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.18910819292068481,
"step": 615,
"valid_targets_mean": 12061.2,
"valid_targets_min": 2666
},
{
"epoch": 1.8805668016194332,
"grad_norm": 0.24021212052305302,
"learning_rate": 3.146286233497593e-05,
"loss": 0.4309,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.16946552693843842,
"step": 620,
"valid_targets_mean": 11908.8,
"valid_targets_min": 1784
},
{
"epoch": 1.895748987854251,
"grad_norm": 0.23243414958739897,
"learning_rate": 3.128886437733539e-05,
"loss": 0.4274,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1634024977684021,
"step": 625,
"valid_targets_mean": 11875.7,
"valid_targets_min": 2266
},
{
"epoch": 1.9109311740890689,
"grad_norm": 0.2321897218435078,
"learning_rate": 3.111360333331263e-05,
"loss": 0.4358,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12072163820266724,
"step": 630,
"valid_targets_mean": 9380.6,
"valid_targets_min": 2435
},
{
"epoch": 1.9261133603238867,
"grad_norm": 0.2343398914268616,
"learning_rate": 3.093709881248312e-05,
"loss": 0.4244,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14065805077552795,
"step": 635,
"valid_targets_mean": 10873.3,
"valid_targets_min": 2493
},
{
"epoch": 1.9412955465587043,
"grad_norm": 0.21403036899310532,
"learning_rate": 3.075937056355225e-05,
"loss": 0.4398,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.16944406926631927,
"step": 640,
"valid_targets_mean": 12638.7,
"valid_targets_min": 766
},
{
"epoch": 1.9564777327935223,
"grad_norm": 0.2475176696543535,
"learning_rate": 3.0580438472145665e-05,
"loss": 0.4231,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1551147699356079,
"step": 645,
"valid_targets_mean": 10001.7,
"valid_targets_min": 1887
},
{
"epoch": 1.97165991902834,
"grad_norm": 0.22067069810740458,
"learning_rate": 3.0400322558584308e-05,
"loss": 0.4306,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14765214920043945,
"step": 650,
"valid_targets_mean": 10529.8,
"valid_targets_min": 1695
},
{
"epoch": 1.986842105263158,
"grad_norm": 0.2200440889393565,
"learning_rate": 3.0219042975644415e-05,
"loss": 0.4224,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14973564445972443,
"step": 655,
"valid_targets_mean": 9583.8,
"valid_targets_min": 3079
},
{
"epoch": 2.0,
"grad_norm": 0.2978629103491527,
"learning_rate": 3.0036620006302624e-05,
"loss": 0.4244,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.4037216901779175,
"step": 660,
"valid_targets_mean": 9887.7,
"valid_targets_min": 2034
},
{
"epoch": 2.0151821862348176,
"grad_norm": 0.24355665535718823,
"learning_rate": 2.9853074061466602e-05,
"loss": 0.3965,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11437727510929108,
"step": 665,
"valid_targets_mean": 9561.8,
"valid_targets_min": 2799
},
{
"epoch": 2.0303643724696356,
"grad_norm": 0.24869442243804363,
"learning_rate": 2.9668425677691278e-05,
"loss": 0.4074,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.15065011382102966,
"step": 670,
"valid_targets_mean": 10834.3,
"valid_targets_min": 1423
},
{
"epoch": 2.0455465587044532,
"grad_norm": 0.2297659593459137,
"learning_rate": 2.948269551488108e-05,
"loss": 0.4011,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14018970727920532,
"step": 675,
"valid_targets_mean": 11208.8,
"valid_targets_min": 2197
},
{
"epoch": 2.0607287449392713,
"grad_norm": 0.21760065911439302,
"learning_rate": 2.929590435397832e-05,
"loss": 0.4157,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.15401902794837952,
"step": 680,
"valid_targets_mean": 12252.5,
"valid_targets_min": 2775
},
{
"epoch": 2.075910931174089,
"grad_norm": 0.21929998454079455,
"learning_rate": 2.9108073094638066e-05,
"loss": 0.4061,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1402328610420227,
"step": 685,
"valid_targets_mean": 10520.2,
"valid_targets_min": 810
},
{
"epoch": 2.091093117408907,
"grad_norm": 0.25762854084256154,
"learning_rate": 2.8919222752889727e-05,
"loss": 0.4042,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12665215134620667,
"step": 690,
"valid_targets_mean": 8784.8,
"valid_targets_min": 2482
},
{
"epoch": 2.1062753036437245,
"grad_norm": 0.20883110039175287,
"learning_rate": 2.8729374458785647e-05,
"loss": 0.4033,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11460208892822266,
"step": 695,
"valid_targets_mean": 9300.2,
"valid_targets_min": 1087
},
{
"epoch": 2.1214574898785425,
"grad_norm": 0.2280468030852318,
"learning_rate": 2.8538549454036838e-05,
"loss": 0.3992,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12859494984149933,
"step": 700,
"valid_targets_mean": 9836.1,
"valid_targets_min": 3216
},
{
"epoch": 2.13663967611336,
"grad_norm": 0.2305434920238047,
"learning_rate": 2.834676908963636e-05,
"loss": 0.4062,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12350933253765106,
"step": 705,
"valid_targets_mean": 9741.5,
"valid_targets_min": 2195
},
{
"epoch": 2.151821862348178,
"grad_norm": 0.2377162666149193,
"learning_rate": 2.815405482347037e-05,
"loss": 0.4015,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1677446961402893,
"step": 710,
"valid_targets_mean": 12769.0,
"valid_targets_min": 2859
},
{
"epoch": 2.167004048582996,
"grad_norm": 0.21204075312597667,
"learning_rate": 2.796042821791725e-05,
"loss": 0.403,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1556544154882431,
"step": 715,
"valid_targets_mean": 12234.6,
"valid_targets_min": 2507
},
{
"epoch": 2.182186234817814,
"grad_norm": 0.2039415062906729,
"learning_rate": 2.776591093743505e-05,
"loss": 0.4106,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11131976544857025,
"step": 720,
"valid_targets_mean": 9257.6,
"valid_targets_min": 1640
},
{
"epoch": 2.1973684210526314,
"grad_norm": 0.22685508299446067,
"learning_rate": 2.7570524746137485e-05,
"loss": 0.4014,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11084121465682983,
"step": 725,
"valid_targets_mean": 8457.0,
"valid_targets_min": 1852
},
{
"epoch": 2.2125506072874495,
"grad_norm": 0.20939970979319947,
"learning_rate": 2.7374291505358818e-05,
"loss": 0.4035,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13435867428779602,
"step": 730,
"valid_targets_mean": 10587.6,
"valid_targets_min": 3988
},
{
"epoch": 2.227732793522267,
"grad_norm": 0.21631166810829094,
"learning_rate": 2.7177233171207817e-05,
"loss": 0.4058,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.145869642496109,
"step": 735,
"valid_targets_mean": 12106.4,
"valid_targets_min": 2588
},
{
"epoch": 2.242914979757085,
"grad_norm": 0.22219638160833832,
"learning_rate": 2.6979371792111147e-05,
"loss": 0.41,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12326790392398834,
"step": 740,
"valid_targets_mean": 9268.1,
"valid_targets_min": 2090
},
{
"epoch": 2.2580971659919027,
"grad_norm": 0.21644687932298712,
"learning_rate": 2.678072950634641e-05,
"loss": 0.409,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14557188749313354,
"step": 745,
"valid_targets_mean": 11407.6,
"valid_targets_min": 1829
},
{
"epoch": 2.2732793522267207,
"grad_norm": 0.2086747965666265,
"learning_rate": 2.6581328539565184e-05,
"loss": 0.4125,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1296234130859375,
"step": 750,
"valid_targets_mean": 11610.8,
"valid_targets_min": 1192
},
{
"epoch": 2.2884615384615383,
"grad_norm": 0.21997306254416282,
"learning_rate": 2.638119120230616e-05,
"loss": 0.4007,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11541731655597687,
"step": 755,
"valid_targets_mean": 9031.7,
"valid_targets_min": 1477
},
{
"epoch": 2.3036437246963564,
"grad_norm": 0.22670698334583528,
"learning_rate": 2.618033988749895e-05,
"loss": 0.409,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1477312594652176,
"step": 760,
"valid_targets_mean": 11624.3,
"valid_targets_min": 1991
},
{
"epoch": 2.318825910931174,
"grad_norm": 0.226842879488616,
"learning_rate": 2.5978797067958542e-05,
"loss": 0.402,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.16331815719604492,
"step": 765,
"valid_targets_mean": 12411.1,
"valid_targets_min": 2223
},
{
"epoch": 2.334008097165992,
"grad_norm": 0.24365624193190316,
"learning_rate": 2.5776585293870877e-05,
"loss": 0.4029,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1397906094789505,
"step": 770,
"valid_targets_mean": 10511.0,
"valid_targets_min": 3395
},
{
"epoch": 2.3491902834008096,
"grad_norm": 0.22210597485491806,
"learning_rate": 2.557372719026976e-05,
"loss": 0.4003,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14930933713912964,
"step": 775,
"valid_targets_mean": 11476.6,
"valid_targets_min": 2831
},
{
"epoch": 2.3643724696356276,
"grad_norm": 0.24020132746247502,
"learning_rate": 2.537024545450539e-05,
"loss": 0.3997,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1250806450843811,
"step": 780,
"valid_targets_mean": 10802.0,
"valid_targets_min": 1656
},
{
"epoch": 2.3795546558704452,
"grad_norm": 0.21970939450799473,
"learning_rate": 2.5166162853704825e-05,
"loss": 0.412,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1378406286239624,
"step": 785,
"valid_targets_mean": 9006.6,
"valid_targets_min": 1612
},
{
"epoch": 2.3947368421052633,
"grad_norm": 0.20791735340812462,
"learning_rate": 2.496150222222458e-05,
"loss": 0.4062,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14031660556793213,
"step": 790,
"valid_targets_mean": 11368.5,
"valid_targets_min": 2404
},
{
"epoch": 2.409919028340081,
"grad_norm": 0.26153369844841917,
"learning_rate": 2.475628645909576e-05,
"loss": 0.4036,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13041003048419952,
"step": 795,
"valid_targets_mean": 10017.9,
"valid_targets_min": 2411
},
{
"epoch": 2.425101214574899,
"grad_norm": 0.22027316195260252,
"learning_rate": 2.4550538525461963e-05,
"loss": 0.4058,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13168390095233917,
"step": 800,
"valid_targets_mean": 10116.5,
"valid_targets_min": 2301
},
{
"epoch": 2.4402834008097165,
"grad_norm": 0.20735932325707584,
"learning_rate": 2.434428144201016e-05,
"loss": 0.4031,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.141376793384552,
"step": 805,
"valid_targets_mean": 11278.5,
"valid_targets_min": 2941
},
{
"epoch": 2.4554655870445345,
"grad_norm": 0.20965832979327945,
"learning_rate": 2.4137538286394976e-05,
"loss": 0.397,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1296098828315735,
"step": 810,
"valid_targets_mean": 9996.4,
"valid_targets_min": 2669
},
{
"epoch": 2.470647773279352,
"grad_norm": 0.20323563119959276,
"learning_rate": 2.3930332190656604e-05,
"loss": 0.4064,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12276121973991394,
"step": 815,
"valid_targets_mean": 10276.2,
"valid_targets_min": 3924
},
{
"epoch": 2.48582995951417,
"grad_norm": 0.2222712599782443,
"learning_rate": 2.3722686338632602e-05,
"loss": 0.4048,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.15194165706634521,
"step": 820,
"valid_targets_mean": 11763.2,
"valid_targets_min": 1754
},
{
"epoch": 2.501012145748988,
"grad_norm": 0.21676101726038569,
"learning_rate": 2.3514623963363886e-05,
"loss": 0.399,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1177615374326706,
"step": 825,
"valid_targets_mean": 9788.6,
"valid_targets_min": 1647
},
{
"epoch": 2.516194331983806,
"grad_norm": 0.24747798535074458,
"learning_rate": 2.330616834449525e-05,
"loss": 0.3976,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.10470348596572876,
"step": 830,
"valid_targets_mean": 8228.6,
"valid_targets_min": 1640
},
{
"epoch": 2.5313765182186234,
"grad_norm": 0.23642757448228505,
"learning_rate": 2.309734280567065e-05,
"loss": 0.4008,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1282092034816742,
"step": 835,
"valid_targets_mean": 9554.1,
"valid_targets_min": 1823
},
{
"epoch": 2.5465587044534415,
"grad_norm": 0.21463605718507556,
"learning_rate": 2.28881707119236e-05,
"loss": 0.4069,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14472943544387817,
"step": 840,
"valid_targets_mean": 11038.6,
"valid_targets_min": 1316
},
{
"epoch": 2.561740890688259,
"grad_norm": 0.19830704786987963,
"learning_rate": 2.267867546706287e-05,
"loss": 0.3993,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1337677240371704,
"step": 845,
"valid_targets_mean": 11115.8,
"valid_targets_min": 1658
},
{
"epoch": 2.5769230769230766,
"grad_norm": 0.22111503225469503,
"learning_rate": 2.2468880511053896e-05,
"loss": 0.4073,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13332180678844452,
"step": 850,
"valid_targets_mean": 9738.6,
"valid_targets_min": 3298
},
{
"epoch": 2.5921052631578947,
"grad_norm": 0.22257999514713242,
"learning_rate": 2.2258809317396163e-05,
"loss": 0.4052,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1417657732963562,
"step": 855,
"valid_targets_mean": 10401.3,
"valid_targets_min": 406
},
{
"epoch": 2.6072874493927127,
"grad_norm": 0.21253273139588202,
"learning_rate": 2.2048485390496757e-05,
"loss": 0.3977,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.09866519272327423,
"step": 860,
"valid_targets_mean": 7923.8,
"valid_targets_min": 2214
},
{
"epoch": 2.6224696356275303,
"grad_norm": 0.22870814783465315,
"learning_rate": 2.1837932263040553e-05,
"loss": 0.4027,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1660136878490448,
"step": 865,
"valid_targets_mean": 11078.2,
"valid_targets_min": 2212
},
{
"epoch": 2.637651821862348,
"grad_norm": 0.29931964038640724,
"learning_rate": 2.1627173493357167e-05,
"loss": 0.4004,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14233574271202087,
"step": 870,
"valid_targets_mean": 10852.8,
"valid_targets_min": 2910
},
{
"epoch": 2.652834008097166,
"grad_norm": 0.25534440527100677,
"learning_rate": 2.1416232662785084e-05,
"loss": 0.4037,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13631418347358704,
"step": 875,
"valid_targets_mean": 10067.3,
"valid_targets_min": 1426
},
{
"epoch": 2.668016194331984,
"grad_norm": 0.2371733144530005,
"learning_rate": 2.1205133373033173e-05,
"loss": 0.404,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1404227614402771,
"step": 880,
"valid_targets_mean": 10728.2,
"valid_targets_min": 1271
},
{
"epoch": 2.6831983805668016,
"grad_norm": 0.2078109950480519,
"learning_rate": 2.0993899243539953e-05,
"loss": 0.3994,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1199427992105484,
"step": 885,
"valid_targets_mean": 8727.6,
"valid_targets_min": 1812
},
{
"epoch": 2.698380566801619,
"grad_norm": 0.2159301658980046,
"learning_rate": 2.0782553908830887e-05,
"loss": 0.4064,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13723723590373993,
"step": 890,
"valid_targets_mean": 11083.8,
"valid_targets_min": 2927
},
{
"epoch": 2.7135627530364372,
"grad_norm": 0.19636767747538167,
"learning_rate": 2.0571121015873924e-05,
"loss": 0.4057,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.16593371331691742,
"step": 895,
"valid_targets_mean": 13503.1,
"valid_targets_min": 1671
},
{
"epoch": 2.7287449392712553,
"grad_norm": 0.22327456821224023,
"learning_rate": 2.0359624221433728e-05,
"loss": 0.4058,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.127730131149292,
"step": 900,
"valid_targets_mean": 10031.6,
"valid_targets_min": 2126
},
{
"epoch": 2.743927125506073,
"grad_norm": 0.20041291146103954,
"learning_rate": 2.014808718942476e-05,
"loss": 0.3999,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13702960312366486,
"step": 905,
"valid_targets_mean": 11859.9,
"valid_targets_min": 1344
},
{
"epoch": 2.7591093117408905,
"grad_norm": 0.1959510117062367,
"learning_rate": 1.9936533588263557e-05,
"loss": 0.399,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13599321246147156,
"step": 910,
"valid_targets_mean": 11546.5,
"valid_targets_min": 721
},
{
"epoch": 2.7742914979757085,
"grad_norm": 0.2150020133924093,
"learning_rate": 1.9724987088220565e-05,
"loss": 0.4121,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14389005303382874,
"step": 915,
"valid_targets_mean": 11402.6,
"valid_targets_min": 2827
},
{
"epoch": 2.7894736842105265,
"grad_norm": 0.21215898602744604,
"learning_rate": 1.951347135877169e-05,
"loss": 0.4062,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1301729381084442,
"step": 920,
"valid_targets_mean": 9943.8,
"valid_targets_min": 2395
},
{
"epoch": 2.804655870445344,
"grad_norm": 0.20967773612549848,
"learning_rate": 1.930201006594999e-05,
"loss": 0.3988,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1343080699443817,
"step": 925,
"valid_targets_mean": 12364.9,
"valid_targets_min": 3036
},
{
"epoch": 2.8198380566801617,
"grad_norm": 0.21630796564169155,
"learning_rate": 1.9090626869697714e-05,
"loss": 0.3978,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12192901223897934,
"step": 930,
"valid_targets_mean": 9816.8,
"valid_targets_min": 2402
},
{
"epoch": 2.83502024291498,
"grad_norm": 0.20100577630192085,
"learning_rate": 1.8879345421219063e-05,
"loss": 0.4065,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1359075903892517,
"step": 935,
"valid_targets_mean": 10343.0,
"valid_targets_min": 442
},
{
"epoch": 2.850202429149798,
"grad_norm": 0.23046010924694144,
"learning_rate": 1.8668189360333923e-05,
"loss": 0.397,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11910334974527359,
"step": 940,
"valid_targets_mean": 9519.0,
"valid_targets_min": 978
},
{
"epoch": 2.8653846153846154,
"grad_norm": 0.20448111009525943,
"learning_rate": 1.845718231283281e-05,
"loss": 0.4007,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11319918185472488,
"step": 945,
"valid_targets_mean": 8395.6,
"valid_targets_min": 1575
},
{
"epoch": 2.880566801619433,
"grad_norm": 0.20645165771896762,
"learning_rate": 1.8246347887833457e-05,
"loss": 0.402,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11795958131551743,
"step": 950,
"valid_targets_mean": 9538.2,
"valid_targets_min": 343
},
{
"epoch": 2.895748987854251,
"grad_norm": 0.21282459320317798,
"learning_rate": 1.8035709675139258e-05,
"loss": 0.4056,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11727368831634521,
"step": 955,
"valid_targets_mean": 8698.4,
"valid_targets_min": 1446
},
{
"epoch": 2.910931174089069,
"grad_norm": 0.19781964939751115,
"learning_rate": 1.7825291242599837e-05,
"loss": 0.3959,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11419452726840973,
"step": 960,
"valid_targets_mean": 9466.3,
"valid_targets_min": 2153
},
{
"epoch": 2.9261133603238867,
"grad_norm": 0.21002201019480785,
"learning_rate": 1.7615116133474084e-05,
"loss": 0.4064,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13672292232513428,
"step": 965,
"valid_targets_mean": 10546.8,
"valid_targets_min": 1985
},
{
"epoch": 2.9412955465587043,
"grad_norm": 0.21275931479304183,
"learning_rate": 1.7405207863795966e-05,
"loss": 0.3967,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13758745789527893,
"step": 970,
"valid_targets_mean": 9654.3,
"valid_targets_min": 1915
},
{
"epoch": 2.9564777327935223,
"grad_norm": 0.20100360416696722,
"learning_rate": 1.719558991974339e-05,
"loss": 0.3973,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1079438179731369,
"step": 975,
"valid_targets_mean": 9043.8,
"valid_targets_min": 2879
},
{
"epoch": 2.97165991902834,
"grad_norm": 0.2269975836389704,
"learning_rate": 1.698628575501034e-05,
"loss": 0.3999,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14483851194381714,
"step": 980,
"valid_targets_mean": 11436.7,
"valid_targets_min": 1211
},
{
"epoch": 2.986842105263158,
"grad_norm": 0.1879476526932179,
"learning_rate": 1.6777318788182723e-05,
"loss": 0.4035,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14617758989334106,
"step": 985,
"valid_targets_mean": 11645.2,
"valid_targets_min": 3089
},
{
"epoch": 3.0,
"grad_norm": 0.3010199328168687,
"learning_rate": 1.6568712400118102e-05,
"loss": 0.3969,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.39078396558761597,
"step": 990,
"valid_targets_mean": 10663.5,
"valid_targets_min": 2457
},
{
"epoch": 3.0151821862348176,
"grad_norm": 0.21781751875463537,
"learning_rate": 1.636048993132969e-05,
"loss": 0.3834,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13729703426361084,
"step": 995,
"valid_targets_mean": 10773.5,
"valid_targets_min": 1279
},
{
"epoch": 3.0303643724696356,
"grad_norm": 0.2230417656760043,
"learning_rate": 1.615267467937479e-05,
"loss": 0.3792,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13346058130264282,
"step": 1000,
"valid_targets_mean": 11306.1,
"valid_targets_min": 2234
},
{
"epoch": 3.0455465587044532,
"grad_norm": 0.21733111151874662,
"learning_rate": 1.59452898962481e-05,
"loss": 0.3901,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13238683342933655,
"step": 1005,
"valid_targets_mean": 11750.0,
"valid_targets_min": 3215
},
{
"epoch": 3.0607287449392713,
"grad_norm": 0.22625370843565767,
"learning_rate": 1.573835878578013e-05,
"loss": 0.3762,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1422734260559082,
"step": 1010,
"valid_targets_mean": 11209.9,
"valid_targets_min": 2079
},
{
"epoch": 3.075910931174089,
"grad_norm": 0.19666515336266407,
"learning_rate": 1.5531904501040917e-05,
"loss": 0.3807,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1307479739189148,
"step": 1015,
"valid_targets_mean": 11612.6,
"valid_targets_min": 2971
},
{
"epoch": 3.091093117408907,
"grad_norm": 0.1928886855734126,
"learning_rate": 1.5325950141749522e-05,
"loss": 0.3826,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13941217958927155,
"step": 1020,
"valid_targets_mean": 11068.5,
"valid_targets_min": 909
},
{
"epoch": 3.1062753036437245,
"grad_norm": 0.21072607913357874,
"learning_rate": 1.5120518751689438e-05,
"loss": 0.3788,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.106941819190979,
"step": 1025,
"valid_targets_mean": 9145.7,
"valid_targets_min": 2303
},
{
"epoch": 3.1214574898785425,
"grad_norm": 0.20354552500648707,
"learning_rate": 1.4915633316130267e-05,
"loss": 0.3784,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12127313017845154,
"step": 1030,
"valid_targets_mean": 9670.7,
"valid_targets_min": 2686
},
{
"epoch": 3.13663967611336,
"grad_norm": 0.1914779822177274,
"learning_rate": 1.4711316759255963e-05,
"loss": 0.3878,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11697717756032944,
"step": 1035,
"valid_targets_mean": 9475.3,
"valid_targets_min": 1230
},
{
"epoch": 3.151821862348178,
"grad_norm": 0.20902997896333064,
"learning_rate": 1.450759194159987e-05,
"loss": 0.381,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12032449245452881,
"step": 1040,
"valid_targets_mean": 9425.8,
"valid_targets_min": 2479
},
{
"epoch": 3.167004048582996,
"grad_norm": 0.19696063514807588,
"learning_rate": 1.4304481657486955e-05,
"loss": 0.3778,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.08829522132873535,
"step": 1045,
"valid_targets_mean": 8185.7,
"valid_targets_min": 2185
},
{
"epoch": 3.182186234817814,
"grad_norm": 0.20177029184014197,
"learning_rate": 1.4102008632483344e-05,
"loss": 0.3873,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11741188913583755,
"step": 1050,
"valid_targets_mean": 8940.3,
"valid_targets_min": 2097
},
{
"epoch": 3.1973684210526314,
"grad_norm": 0.2136067023802031,
"learning_rate": 1.3900195520853628e-05,
"loss": 0.3795,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13736692070960999,
"step": 1055,
"valid_targets_mean": 12065.6,
"valid_targets_min": 2194
},
{
"epoch": 3.2125506072874495,
"grad_norm": 0.2012264827781041,
"learning_rate": 1.3699064903026149e-05,
"loss": 0.3758,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13378620147705078,
"step": 1060,
"valid_targets_mean": 11753.4,
"valid_targets_min": 4457
},
{
"epoch": 3.227732793522267,
"grad_norm": 0.19533737596737558,
"learning_rate": 1.34986392830665e-05,
"loss": 0.3834,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12444774061441422,
"step": 1065,
"valid_targets_mean": 10883.3,
"valid_targets_min": 1829
},
{
"epoch": 3.242914979757085,
"grad_norm": 0.1913483548977371,
"learning_rate": 1.3298941086159598e-05,
"loss": 0.3797,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1327345073223114,
"step": 1070,
"valid_targets_mean": 11969.5,
"valid_targets_min": 2300
},
{
"epoch": 3.2580971659919027,
"grad_norm": 0.20277132333096498,
"learning_rate": 1.3099992656100592e-05,
"loss": 0.3867,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.10380689799785614,
"step": 1075,
"valid_targets_mean": 8426.2,
"valid_targets_min": 1565
},
{
"epoch": 3.2732793522267207,
"grad_norm": 0.19529197941773585,
"learning_rate": 1.2901816252794848e-05,
"loss": 0.3804,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12796002626419067,
"step": 1080,
"valid_targets_mean": 10604.7,
"valid_targets_min": 1972
},
{
"epoch": 3.2884615384615383,
"grad_norm": 0.1842338834508977,
"learning_rate": 1.2704434049767356e-05,
"loss": 0.3928,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.15373364090919495,
"step": 1085,
"valid_targets_mean": 13930.4,
"valid_targets_min": 2820
},
{
"epoch": 3.3036437246963564,
"grad_norm": 0.1963156313504413,
"learning_rate": 1.250786813168176e-05,
"loss": 0.3814,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12112146615982056,
"step": 1090,
"valid_targets_mean": 9762.8,
"valid_targets_min": 245
},
{
"epoch": 3.318825910931174,
"grad_norm": 0.20045272092400518,
"learning_rate": 1.2312140491869369e-05,
"loss": 0.3785,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11439470946788788,
"step": 1095,
"valid_targets_mean": 9937.9,
"valid_targets_min": 2642
},
{
"epoch": 3.334008097165992,
"grad_norm": 0.21007599323553236,
"learning_rate": 1.2117273029868362e-05,
"loss": 0.3802,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.10896288603544235,
"step": 1100,
"valid_targets_mean": 9575.8,
"valid_targets_min": 2482
},
{
"epoch": 3.3491902834008096,
"grad_norm": 0.19927323498968952,
"learning_rate": 1.1923287548973508e-05,
"loss": 0.3818,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12662863731384277,
"step": 1105,
"valid_targets_mean": 11785.4,
"valid_targets_min": 1696
},
{
"epoch": 3.3643724696356276,
"grad_norm": 0.19519616335018541,
"learning_rate": 1.1730205753796631e-05,
"loss": 0.3759,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12134730815887451,
"step": 1110,
"valid_targets_mean": 10729.1,
"valid_targets_min": 3001
},
{
"epoch": 3.3795546558704452,
"grad_norm": 0.21028694968176903,
"learning_rate": 1.1538049247838128e-05,
"loss": 0.3871,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13904663920402527,
"step": 1115,
"valid_targets_mean": 10362.8,
"valid_targets_min": 1519
},
{
"epoch": 3.3947368421052633,
"grad_norm": 0.19912225165901493,
"learning_rate": 1.134683953106983e-05,
"loss": 0.3802,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12339392304420471,
"step": 1120,
"valid_targets_mean": 9278.3,
"valid_targets_min": 2476
},
{
"epoch": 3.409919028340081,
"grad_norm": 0.19327825454771747,
"learning_rate": 1.115659799752938e-05,
"loss": 0.3733,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1173870712518692,
"step": 1125,
"valid_targets_mean": 10774.1,
"valid_targets_min": 413
},
{
"epoch": 3.425101214574899,
"grad_norm": 0.1844225077898535,
"learning_rate": 1.096734593292649e-05,
"loss": 0.3823,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1067117229104042,
"step": 1130,
"valid_targets_mean": 9642.9,
"valid_targets_min": 2019
},
{
"epoch": 3.4402834008097165,
"grad_norm": 0.2066798200776645,
"learning_rate": 1.077910451226138e-05,
"loss": 0.3789,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11906486749649048,
"step": 1135,
"valid_targets_mean": 10392.2,
"valid_targets_min": 2625
},
{
"epoch": 3.4554655870445345,
"grad_norm": 0.20827790359398754,
"learning_rate": 1.0591894797455526e-05,
"loss": 0.3822,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11905896663665771,
"step": 1140,
"valid_targets_mean": 10585.9,
"valid_targets_min": 1563
},
{
"epoch": 3.470647773279352,
"grad_norm": 0.20902598407121353,
"learning_rate": 1.0405737734995083e-05,
"loss": 0.3762,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13168013095855713,
"step": 1145,
"valid_targets_mean": 11115.3,
"valid_targets_min": 2500
},
{
"epoch": 3.48582995951417,
"grad_norm": 0.17996554685296223,
"learning_rate": 1.0220654153587225e-05,
"loss": 0.3864,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11831013113260269,
"step": 1150,
"valid_targets_mean": 9677.0,
"valid_targets_min": 1782
},
{
"epoch": 3.501012145748988,
"grad_norm": 0.18994778826070668,
"learning_rate": 1.00366647618297e-05,
"loss": 0.3749,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.10179728269577026,
"step": 1155,
"valid_targets_mean": 8558.6,
"valid_targets_min": 1210
},
{
"epoch": 3.516194331983806,
"grad_norm": 0.19546690886244628,
"learning_rate": 9.853790145893742e-06,
"loss": 0.3789,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14318439364433289,
"step": 1160,
"valid_targets_mean": 11218.9,
"valid_targets_min": 3131
},
{
"epoch": 3.5313765182186234,
"grad_norm": 0.18549801901119017,
"learning_rate": 9.672050767220765e-06,
"loss": 0.3783,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12476805597543716,
"step": 1165,
"valid_targets_mean": 10252.2,
"valid_targets_min": 2786
},
{
"epoch": 3.5465587044534415,
"grad_norm": 0.18732516786072945,
"learning_rate": 9.491466960232955e-06,
"loss": 0.3723,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11273340880870819,
"step": 1170,
"valid_targets_mean": 8616.5,
"valid_targets_min": 1453
},
{
"epoch": 3.561740890688259,
"grad_norm": 0.20149643791941338,
"learning_rate": 9.312058930058114e-06,
"loss": 0.3785,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11582259833812714,
"step": 1175,
"valid_targets_mean": 9118.4,
"valid_targets_min": 787
},
{
"epoch": 3.5769230769230766,
"grad_norm": 0.21085059562942834,
"learning_rate": 9.133846750268945e-06,
"loss": 0.3784,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13565123081207275,
"step": 1180,
"valid_targets_mean": 10716.3,
"valid_targets_min": 1933
},
{
"epoch": 3.5921052631578947,
"grad_norm": 0.19440857465551512,
"learning_rate": 8.956850360637046e-06,
"loss": 0.3806,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11549557745456696,
"step": 1185,
"valid_targets_mean": 9663.7,
"valid_targets_min": 2336
},
{
"epoch": 3.6072874493927127,
"grad_norm": 0.18580297908502816,
"learning_rate": 8.78108956490194e-06,
"loss": 0.3822,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1174996942281723,
"step": 1190,
"valid_targets_mean": 9356.2,
"valid_targets_min": 1694
},
{
"epoch": 3.6224696356275303,
"grad_norm": 0.19481766458314584,
"learning_rate": 8.606584028555225e-06,
"loss": 0.3769,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13105148077011108,
"step": 1195,
"valid_targets_mean": 10477.2,
"valid_targets_min": 2939
},
{
"epoch": 3.637651821862348,
"grad_norm": 0.18881391921278226,
"learning_rate": 8.43335327664027e-06,
"loss": 0.3765,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.15695153176784515,
"step": 1200,
"valid_targets_mean": 13012.8,
"valid_targets_min": 417
},
{
"epoch": 3.652834008097166,
"grad_norm": 0.1994823937746155,
"learning_rate": 8.261416691567601e-06,
"loss": 0.3691,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12895889580249786,
"step": 1205,
"valid_targets_mean": 11328.2,
"valid_targets_min": 1957
},
{
"epoch": 3.668016194331984,
"grad_norm": 0.19695915654379229,
"learning_rate": 8.090793510946242e-06,
"loss": 0.3647,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.10856752097606659,
"step": 1210,
"valid_targets_mean": 9687.9,
"valid_targets_min": 3468
},
{
"epoch": 3.6831983805668016,
"grad_norm": 0.20405732090532516,
"learning_rate": 7.921502825431258e-06,
"loss": 0.3583,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11551082879304886,
"step": 1215,
"valid_targets_mean": 9392.2,
"valid_targets_min": 2945
},
{
"epoch": 3.698380566801619,
"grad_norm": 0.19864256408199005,
"learning_rate": 7.753563576587753e-06,
"loss": 0.3747,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12581707537174225,
"step": 1220,
"valid_targets_mean": 11870.0,
"valid_targets_min": 3248
},
{
"epoch": 3.7135627530364372,
"grad_norm": 0.2153470946121616,
"learning_rate": 7.5869945547715275e-06,
"loss": 0.3732,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1357821673154831,
"step": 1225,
"valid_targets_mean": 10057.5,
"valid_targets_min": 755
},
{
"epoch": 3.7287449392712553,
"grad_norm": 0.19143348370922542,
"learning_rate": 7.421814397026674e-06,
"loss": 0.3683,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12181690335273743,
"step": 1230,
"valid_targets_mean": 10055.2,
"valid_targets_min": 1140
},
{
"epoch": 3.743927125506073,
"grad_norm": 0.18437116959705785,
"learning_rate": 7.258041585000317e-06,
"loss": 0.3608,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.10211285203695297,
"step": 1235,
"valid_targets_mean": 8725.9,
"valid_targets_min": 1518
},
{
"epoch": 3.7591093117408905,
"grad_norm": 0.18527155499014733,
"learning_rate": 7.095694442874743e-06,
"loss": 0.3689,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13864530622959137,
"step": 1240,
"valid_targets_mean": 11871.8,
"valid_targets_min": 1440
},
{
"epoch": 3.7742914979757085,
"grad_norm": 0.20256166841034448,
"learning_rate": 6.934791135317147e-06,
"loss": 0.3684,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12536349892616272,
"step": 1245,
"valid_targets_mean": 9864.8,
"valid_targets_min": 1364
},
{
"epoch": 3.7894736842105265,
"grad_norm": 0.1885365721274064,
"learning_rate": 6.775349665447222e-06,
"loss": 0.3612,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.10805296897888184,
"step": 1250,
"valid_targets_mean": 10663.3,
"valid_targets_min": 1778
},
{
"epoch": 3.804655870445344,
"grad_norm": 0.19117386431021458,
"learning_rate": 6.617387872822842e-06,
"loss": 0.3707,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.10408435761928558,
"step": 1255,
"valid_targets_mean": 10560.4,
"valid_targets_min": 3376
},
{
"epoch": 3.8198380566801617,
"grad_norm": 0.1887911686679957,
"learning_rate": 6.460923431444015e-06,
"loss": 0.3634,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11224344372749329,
"step": 1260,
"valid_targets_mean": 10806.7,
"valid_targets_min": 2834
},
{
"epoch": 3.83502024291498,
"grad_norm": 0.2511629707679848,
"learning_rate": 6.305973847775406e-06,
"loss": 0.3611,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11038510501384735,
"step": 1265,
"valid_targets_mean": 9412.5,
"valid_targets_min": 1636
},
{
"epoch": 3.850202429149798,
"grad_norm": 0.19524260594282525,
"learning_rate": 6.152556458787546e-06,
"loss": 0.3691,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13783195614814758,
"step": 1270,
"valid_targets_mean": 11960.5,
"valid_targets_min": 3625
},
{
"epoch": 3.8653846153846154,
"grad_norm": 0.20830638946837632,
"learning_rate": 6.000688430017048e-06,
"loss": 0.3656,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1059221401810646,
"step": 1275,
"valid_targets_mean": 8446.1,
"valid_targets_min": 2648
},
{
"epoch": 3.880566801619433,
"grad_norm": 0.18358339557954717,
"learning_rate": 5.850386753645998e-06,
"loss": 0.367,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12444412708282471,
"step": 1280,
"valid_targets_mean": 10544.0,
"valid_targets_min": 2552
},
{
"epoch": 3.895748987854251,
"grad_norm": 0.1913134429816862,
"learning_rate": 5.701668246600731e-06,
"loss": 0.365,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11741238832473755,
"step": 1285,
"valid_targets_mean": 9534.8,
"valid_targets_min": 322
},
{
"epoch": 3.910931174089069,
"grad_norm": 0.19737214526506686,
"learning_rate": 5.554549548670227e-06,
"loss": 0.3644,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12340462952852249,
"step": 1290,
"valid_targets_mean": 9826.8,
"valid_targets_min": 2599
},
{
"epoch": 3.9261133603238867,
"grad_norm": 0.19322680874339637,
"learning_rate": 5.409047120644307e-06,
"loss": 0.3686,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13688233494758606,
"step": 1295,
"valid_targets_mean": 11182.3,
"valid_targets_min": 1557
},
{
"epoch": 3.9412955465587043,
"grad_norm": 0.19837836767468806,
"learning_rate": 5.265177242471899e-06,
"loss": 0.3722,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12162747979164124,
"step": 1300,
"valid_targets_mean": 9966.4,
"valid_targets_min": 2376
},
{
"epoch": 3.9564777327935223,
"grad_norm": 0.1980287432099399,
"learning_rate": 5.122956011439486e-06,
"loss": 0.3574,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12239546328783035,
"step": 1305,
"valid_targets_mean": 10853.2,
"valid_targets_min": 4115
},
{
"epoch": 3.97165991902834,
"grad_norm": 0.7212715678474377,
"learning_rate": 4.982399340370017e-06,
"loss": 0.3676,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1188947781920433,
"step": 1310,
"valid_targets_mean": 10789.0,
"valid_targets_min": 1678
},
{
"epoch": 3.986842105263158,
"grad_norm": 0.1958031682699217,
"learning_rate": 4.843522955842464e-06,
"loss": 0.3719,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14273865520954132,
"step": 1315,
"valid_targets_mean": 10936.7,
"valid_targets_min": 1649
},
{
"epoch": 4.0,
"grad_norm": 0.31774916038719164,
"learning_rate": 4.706342396432213e-06,
"loss": 0.3496,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.29835373163223267,
"step": 1320,
"valid_targets_mean": 10492.2,
"valid_targets_min": 2847
},
{
"epoch": 4.015182186234818,
"grad_norm": 0.1843924113718929,
"learning_rate": 4.570873010972477e-06,
"loss": 0.3652,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.10512104630470276,
"step": 1325,
"valid_targets_mean": 10884.8,
"valid_targets_min": 2671
},
{
"epoch": 4.030364372469635,
"grad_norm": 0.18018854644817908,
"learning_rate": 4.43712995683695e-06,
"loss": 0.3681,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.09973619133234024,
"step": 1330,
"valid_targets_mean": 8272.1,
"valid_targets_min": 2067
},
{
"epoch": 4.045546558704453,
"grad_norm": 0.17571816112831184,
"learning_rate": 4.305128198243888e-06,
"loss": 0.3689,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12737268209457397,
"step": 1335,
"valid_targets_mean": 11352.1,
"valid_targets_min": 1795
},
{
"epoch": 4.060728744939271,
"grad_norm": 0.18330154031717863,
"learning_rate": 4.174882504581794e-06,
"loss": 0.3625,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.10680342465639114,
"step": 1340,
"valid_targets_mean": 8873.7,
"valid_targets_min": 2326
},
{
"epoch": 4.075910931174089,
"grad_norm": 0.18429043131384004,
"learning_rate": 4.046407448756895e-06,
"loss": 0.3685,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11359091103076935,
"step": 1345,
"valid_targets_mean": 9797.7,
"valid_targets_min": 1810
},
{
"epoch": 4.0910931174089065,
"grad_norm": 0.18897272974624404,
"learning_rate": 3.91971740556262e-06,
"loss": 0.3668,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.10602220892906189,
"step": 1350,
"valid_targets_mean": 9621.8,
"valid_targets_min": 2154
},
{
"epoch": 4.1062753036437245,
"grad_norm": 0.1928917574346044,
"learning_rate": 3.7948265500712313e-06,
"loss": 0.3624,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12940697371959686,
"step": 1355,
"valid_targets_mean": 10877.2,
"valid_targets_min": 2615
},
{
"epoch": 4.1214574898785425,
"grad_norm": 0.18005268626554144,
"learning_rate": 3.6717488560478096e-06,
"loss": 0.3664,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12552385032176971,
"step": 1360,
"valid_targets_mean": 10077.4,
"valid_targets_min": 2879
},
{
"epoch": 4.136639676113361,
"grad_norm": 0.1788632112469236,
"learning_rate": 3.5504980943867538e-06,
"loss": 0.3746,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.09735994040966034,
"step": 1365,
"valid_targets_mean": 9807.6,
"valid_targets_min": 684
},
{
"epoch": 4.151821862348178,
"grad_norm": 0.21241167381424772,
"learning_rate": 3.4310878315710074e-06,
"loss": 0.3689,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11332625150680542,
"step": 1370,
"valid_targets_mean": 9672.8,
"valid_targets_min": 2404
},
{
"epoch": 4.167004048582996,
"grad_norm": 0.19299307837990345,
"learning_rate": 3.3135314281540954e-06,
"loss": 0.3724,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13851384818553925,
"step": 1375,
"valid_targets_mean": 11021.4,
"valid_targets_min": 2349
},
{
"epoch": 4.182186234817814,
"grad_norm": 0.17435774249387354,
"learning_rate": 3.1978420372652776e-06,
"loss": 0.3653,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1209459900856018,
"step": 1380,
"valid_targets_mean": 11227.0,
"valid_targets_min": 1843
},
{
"epoch": 4.197368421052632,
"grad_norm": 0.17793401384716034,
"learning_rate": 3.084032603137852e-06,
"loss": 0.3728,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12819677591323853,
"step": 1385,
"valid_targets_mean": 10938.0,
"valid_targets_min": 2115
},
{
"epoch": 4.212550607287449,
"grad_norm": 0.1847368183158193,
"learning_rate": 2.9721158596608622e-06,
"loss": 0.3681,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12626275420188904,
"step": 1390,
"valid_targets_mean": 10809.4,
"valid_targets_min": 1640
},
{
"epoch": 4.227732793522267,
"grad_norm": 0.18692877804023675,
"learning_rate": 2.8621043289543314e-06,
"loss": 0.374,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12628719210624695,
"step": 1395,
"valid_targets_mean": 11386.6,
"valid_targets_min": 818
},
{
"epoch": 4.242914979757085,
"grad_norm": 0.18025836952356916,
"learning_rate": 2.754010319968181e-06,
"loss": 0.3722,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12013980001211166,
"step": 1400,
"valid_targets_mean": 10630.8,
"valid_targets_min": 2219
},
{
"epoch": 4.258097165991903,
"grad_norm": 0.1877232221667285,
"learning_rate": 2.647845927105015e-06,
"loss": 0.3717,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1244920939207077,
"step": 1405,
"valid_targets_mean": 9607.3,
"valid_targets_min": 1754
},
{
"epoch": 4.27327935222672,
"grad_norm": 0.18249632818477718,
"learning_rate": 2.543623028866915e-06,
"loss": 0.361,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13994036614894867,
"step": 1410,
"valid_targets_mean": 11417.1,
"valid_targets_min": 2408
},
{
"epoch": 4.288461538461538,
"grad_norm": 0.18120025432339984,
"learning_rate": 2.4413532865263533e-06,
"loss": 0.3751,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11695703864097595,
"step": 1415,
"valid_targets_mean": 9500.3,
"valid_targets_min": 1626
},
{
"epoch": 4.303643724696356,
"grad_norm": 0.18038827988155628,
"learning_rate": 2.3410481428214602e-06,
"loss": 0.3653,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1068282350897789,
"step": 1420,
"valid_targets_mean": 9340.8,
"valid_targets_min": 1542
},
{
"epoch": 4.318825910931174,
"grad_norm": 0.19901656665636938,
"learning_rate": 2.242718820675718e-06,
"loss": 0.3696,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11531911790370941,
"step": 1425,
"valid_targets_mean": 10091.9,
"valid_targets_min": 1991
},
{
"epoch": 4.334008097165992,
"grad_norm": 0.17656725900990108,
"learning_rate": 2.1463763219422495e-06,
"loss": 0.3654,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.10796954482793808,
"step": 1430,
"valid_targets_mean": 9848.0,
"valid_targets_min": 2599
},
{
"epoch": 4.34919028340081,
"grad_norm": 0.17833357555078458,
"learning_rate": 2.0520314261728357e-06,
"loss": 0.3643,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.10144203156232834,
"step": 1435,
"valid_targets_mean": 8776.6,
"valid_targets_min": 2066
},
{
"epoch": 4.364372469635628,
"grad_norm": 0.18607221555908898,
"learning_rate": 1.9596946894118306e-06,
"loss": 0.3585,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11383245885372162,
"step": 1440,
"valid_targets_mean": 9263.6,
"valid_targets_min": 1440
},
{
"epoch": 4.379554655870446,
"grad_norm": 0.18577021193639703,
"learning_rate": 1.8693764430150696e-06,
"loss": 0.3697,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13473168015480042,
"step": 1445,
"valid_targets_mean": 11097.1,
"valid_targets_min": 2380
},
{
"epoch": 4.394736842105263,
"grad_norm": 0.1709129995577021,
"learning_rate": 1.7810867924938978e-06,
"loss": 0.3661,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12228719145059586,
"step": 1450,
"valid_targets_mean": 9294.8,
"valid_targets_min": 477
},
{
"epoch": 4.409919028340081,
"grad_norm": 0.1810787994190016,
"learning_rate": 1.6948356163845048e-06,
"loss": 0.3677,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14793120324611664,
"step": 1455,
"valid_targets_mean": 11832.1,
"valid_targets_min": 2338
},
{
"epoch": 4.425101214574899,
"grad_norm": 0.18892851687010215,
"learning_rate": 1.610632565142627e-06,
"loss": 0.3691,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12090542912483215,
"step": 1460,
"valid_targets_mean": 10259.0,
"valid_targets_min": 3516
},
{
"epoch": 4.440283400809717,
"grad_norm": 0.17894691933426599,
"learning_rate": 1.5284870600637813e-06,
"loss": 0.3648,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11216691136360168,
"step": 1465,
"valid_targets_mean": 9150.0,
"valid_targets_min": 2502
},
{
"epoch": 4.455465587044534,
"grad_norm": 0.18757418087454025,
"learning_rate": 1.4484082922291376e-06,
"loss": 0.3624,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1409149169921875,
"step": 1470,
"valid_targets_mean": 12013.4,
"valid_targets_min": 1381
},
{
"epoch": 4.470647773279352,
"grad_norm": 0.1954402311039142,
"learning_rate": 1.3704052214771513e-06,
"loss": 0.3682,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12278928607702255,
"step": 1475,
"valid_targets_mean": 9280.6,
"valid_targets_min": 2312
},
{
"epoch": 4.48582995951417,
"grad_norm": 0.1727940865183207,
"learning_rate": 1.2944865754010682e-06,
"loss": 0.3662,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14360244572162628,
"step": 1480,
"valid_targets_mean": 12639.5,
"valid_targets_min": 1846
},
{
"epoch": 4.501012145748988,
"grad_norm": 0.19104104864167198,
"learning_rate": 1.2206608483724013e-06,
"loss": 0.3665,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11449052393436432,
"step": 1485,
"valid_targets_mean": 9093.4,
"valid_targets_min": 1659
},
{
"epoch": 4.516194331983805,
"grad_norm": 0.1867664116957937,
"learning_rate": 1.1489363005905241e-06,
"loss": 0.3598,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12641845643520355,
"step": 1490,
"valid_targets_mean": 9890.8,
"valid_targets_min": 2715
},
{
"epoch": 4.531376518218623,
"grad_norm": 0.18476830111294615,
"learning_rate": 1.0793209571584562e-06,
"loss": 0.3649,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12984566390514374,
"step": 1495,
"valid_targets_mean": 11103.8,
"valid_targets_min": 1663
},
{
"epoch": 4.5465587044534415,
"grad_norm": 0.18066626607720007,
"learning_rate": 1.0118226071849424e-06,
"loss": 0.3616,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.10867150127887726,
"step": 1500,
"valid_targets_mean": 9276.9,
"valid_targets_min": 1694
},
{
"epoch": 4.5617408906882595,
"grad_norm": 0.16173971571301107,
"learning_rate": 9.464488029129581e-07,
"loss": 0.3729,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.10973206907510757,
"step": 1505,
"valid_targets_mean": 10314.8,
"valid_targets_min": 2799
},
{
"epoch": 4.576923076923077,
"grad_norm": 0.17897671753750052,
"learning_rate": 8.832068588746945e-07,
"loss": 0.3605,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11189191043376923,
"step": 1510,
"valid_targets_mean": 10190.8,
"valid_targets_min": 1649
},
{
"epoch": 4.592105263157895,
"grad_norm": 0.18012360756114543,
"learning_rate": 8.221038510731704e-07,
"loss": 0.3714,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.10856941342353821,
"step": 1515,
"valid_targets_mean": 8771.6,
"valid_targets_min": 1901
},
{
"epoch": 4.607287449392713,
"grad_norm": 0.1547261000548381,
"learning_rate": 7.631466161904821e-07,
"loss": 0.3645,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12571071088314056,
"step": 1520,
"valid_targets_mean": 12577.9,
"valid_targets_min": 1948
},
{
"epoch": 4.62246963562753,
"grad_norm": 0.17262392891010045,
"learning_rate": 7.063417508228876e-07,
"loss": 0.3654,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11583705246448517,
"step": 1525,
"valid_targets_mean": 9228.4,
"valid_targets_min": 1600
},
{
"epoch": 4.637651821862348,
"grad_norm": 0.17161338010550106,
"learning_rate": 6.516956107427241e-07,
"loss": 0.3691,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12043744325637817,
"step": 1530,
"valid_targets_mean": 10210.4,
"valid_targets_min": 1719
},
{
"epoch": 4.652834008097166,
"grad_norm": 0.18234822744394527,
"learning_rate": 5.992143101872638e-07,
"loss": 0.3648,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.09593389928340912,
"step": 1535,
"valid_targets_mean": 7271.0,
"valid_targets_min": 1735
},
{
"epoch": 4.668016194331984,
"grad_norm": 0.2162606197737493,
"learning_rate": 5.489037211746184e-07,
"loss": 0.3678,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1457173228263855,
"step": 1540,
"valid_targets_mean": 11809.0,
"valid_targets_min": 4250
},
{
"epoch": 4.683198380566802,
"grad_norm": 0.16995590048199882,
"learning_rate": 5.007694728467228e-07,
"loss": 0.3676,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1253059357404709,
"step": 1545,
"valid_targets_mean": 10858.9,
"valid_targets_min": 1230
},
{
"epoch": 4.698380566801619,
"grad_norm": 0.17734059670619579,
"learning_rate": 4.548169508395028e-07,
"loss": 0.3639,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12860271334648132,
"step": 1550,
"valid_targets_mean": 11737.7,
"valid_targets_min": 3485
},
{
"epoch": 4.713562753036437,
"grad_norm": 0.18050233010338437,
"learning_rate": 4.1105129668029595e-07,
"loss": 0.3703,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1319669783115387,
"step": 1555,
"valid_targets_mean": 10501.0,
"valid_targets_min": 1372
},
{
"epoch": 4.728744939271255,
"grad_norm": 0.18215765373981796,
"learning_rate": 3.6947740721257066e-07,
"loss": 0.3644,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12586212158203125,
"step": 1560,
"valid_targets_mean": 11023.1,
"valid_targets_min": 2636
},
{
"epoch": 4.743927125506072,
"grad_norm": 0.18748325536702345,
"learning_rate": 3.3009993404802486e-07,
"loss": 0.3654,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.10447224229574203,
"step": 1565,
"valid_targets_mean": 9109.5,
"valid_targets_min": 1941
},
{
"epoch": 4.7591093117408905,
"grad_norm": 0.18050363980638168,
"learning_rate": 2.929232830461404e-07,
"loss": 0.3681,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14107687771320343,
"step": 1570,
"valid_targets_mean": 11744.7,
"valid_targets_min": 2422
},
{
"epoch": 4.7742914979757085,
"grad_norm": 0.2023921728351548,
"learning_rate": 2.579516138212101e-07,
"loss": 0.3736,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12674914300441742,
"step": 1575,
"valid_targets_mean": 11620.3,
"valid_targets_min": 1761
},
{
"epoch": 4.7894736842105265,
"grad_norm": 0.1622695741435569,
"learning_rate": 2.2518883927692857e-07,
"loss": 0.3763,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12355431914329529,
"step": 1580,
"valid_targets_mean": 11819.7,
"valid_targets_min": 2489
},
{
"epoch": 4.804655870445345,
"grad_norm": 0.20101474821198356,
"learning_rate": 1.9463862516859277e-07,
"loss": 0.3674,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1243324801325798,
"step": 1585,
"valid_targets_mean": 10348.0,
"valid_targets_min": 1587
},
{
"epoch": 4.819838056680162,
"grad_norm": 0.1675194942295015,
"learning_rate": 1.6630438969294615e-07,
"loss": 0.3729,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13866698741912842,
"step": 1590,
"valid_targets_mean": 12737.4,
"valid_targets_min": 3208
},
{
"epoch": 4.83502024291498,
"grad_norm": 0.17086348937435433,
"learning_rate": 1.4018930310571553e-07,
"loss": 0.3691,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1291865110397339,
"step": 1595,
"valid_targets_mean": 11653.0,
"valid_targets_min": 2275
},
{
"epoch": 4.850202429149798,
"grad_norm": 0.17068101788111856,
"learning_rate": 1.1629628736690824e-07,
"loss": 0.3576,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1036512702703476,
"step": 1600,
"valid_targets_mean": 9271.8,
"valid_targets_min": 1083
},
{
"epoch": 4.865384615384615,
"grad_norm": 0.17230658064627632,
"learning_rate": 9.46280158138757e-08,
"loss": 0.3727,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11794915050268173,
"step": 1605,
"valid_targets_mean": 10087.9,
"valid_targets_min": 1841
},
{
"epoch": 4.880566801619433,
"grad_norm": 0.17026402977551577,
"learning_rate": 7.518691286220625e-08,
"loss": 0.3684,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12925563752651215,
"step": 1610,
"valid_targets_mean": 11449.4,
"valid_targets_min": 2402
},
{
"epoch": 4.895748987854251,
"grad_norm": 0.16996512113902293,
"learning_rate": 5.797515373445084e-08,
"loss": 0.3747,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12282838672399521,
"step": 1615,
"valid_targets_mean": 11293.9,
"valid_targets_min": 2486
},
{
"epoch": 4.910931174089069,
"grad_norm": 0.1827513479377631,
"learning_rate": 4.299466421675113e-08,
"loss": 0.3668,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12446284294128418,
"step": 1620,
"valid_targets_mean": 10751.7,
"valid_targets_min": 2692
},
{
"epoch": 4.926113360323887,
"grad_norm": 0.18688189150800386,
"learning_rate": 3.0247120443362976e-08,
"loss": 0.3719,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.10103616118431091,
"step": 1625,
"valid_targets_mean": 9804.4,
"valid_targets_min": 1363
},
{
"epoch": 4.941295546558704,
"grad_norm": 0.17387094857705057,
"learning_rate": 1.973394870912193e-08,
"loss": 0.3708,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12511810660362244,
"step": 1630,
"valid_targets_mean": 10579.7,
"valid_targets_min": 1087
},
{
"epoch": 4.956477732793522,
"grad_norm": 0.18424632034257762,
"learning_rate": 1.145632530985541e-08,
"loss": 0.3659,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.13365338742733002,
"step": 1635,
"valid_targets_mean": 10992.5,
"valid_targets_min": 1896
},
{
"epoch": 4.97165991902834,
"grad_norm": 0.16930897925302424,
"learning_rate": 5.415176410765721e-09,
"loss": 0.3652,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14591088891029358,
"step": 1640,
"valid_targets_mean": 11615.2,
"valid_targets_min": 1369
},
{
"epoch": 4.9868421052631575,
"grad_norm": 0.1867459162363072,
"learning_rate": 1.611177942812958e-09,
"loss": 0.365,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.148482084274292,
"step": 1645,
"valid_targets_mean": 11537.9,
"valid_targets_min": 850
},
{
"epoch": 5.0,
"grad_norm": 0.2764448693570558,
"learning_rate": 4.475552707772224e-11,
"loss": 0.3697,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.38277196884155273,
"step": 1650,
"valid_targets_mean": 11949.7,
"valid_targets_min": 2234
},
{
"epoch": 5.0,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.38277196884155273,
"step": 1650,
"total_flos": 6.453442034819662e+18,
"train_loss": 0.1001221828027205,
"train_runtime": 13187.2715,
"train_samples_per_second": 11.981,
"train_steps_per_second": 0.125,
"valid_targets_mean": 11949.7,
"valid_targets_min": 2234
}
],
"logging_steps": 5,
"max_steps": 1650,
"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": 6.453442034819662e+18,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}