laion/r2egym-31600-opt100k__Qwen3-8B
1
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity
Files
41dda9eab08506028b2548f86cb81d78680a9d4f
r2egym-31600-opt100k__Qwen3-8B/trainer_state.json

3674 lines
103 KiB
JSON
Raw Normal View History

初始化项目，由ModelHub XC社区提供模型 Model: laion/r2egym-31600-opt100k__Qwen3-8B Source: Original Platform
2026-04-30 01:24:06 +08:00
{
"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": 7.330185554786115,
"learning_rate": 9.696969696969698e-07,
"loss": 0.384,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1300801783800125,
"step": 5,
"valid_targets_mean": 4714.4,
"valid_targets_min": 2106
},
{
"epoch": 0.030364372469635626,
"grad_norm": 3.3445550275189855,
"learning_rate": 2.181818181818182e-06,
"loss": 0.351,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.09995095431804657,
"step": 10,
"valid_targets_mean": 3845.9,
"valid_targets_min": 1829
},
{
"epoch": 0.04554655870445344,
"grad_norm": 1.8665451513088296,
"learning_rate": 3.3939393939393946e-06,
"loss": 0.3127,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.10031776130199432,
"step": 15,
"valid_targets_mean": 3509.7,
"valid_targets_min": 1820
},
{
"epoch": 0.06072874493927125,
"grad_norm": 0.6936395271280061,
"learning_rate": 4.606060606060606e-06,
"loss": 0.2616,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0811019167304039,
"step": 20,
"valid_targets_mean": 4497.1,
"valid_targets_min": 1716
},
{
"epoch": 0.07591093117408906,
"grad_norm": 0.6220499032835243,
"learning_rate": 5.8181818181818185e-06,
"loss": 0.2401,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.07583131641149521,
"step": 25,
"valid_targets_mean": 2807.7,
"valid_targets_min": 1601
},
{
"epoch": 0.09109311740890688,
"grad_norm": 0.38940013478957664,
"learning_rate": 7.030303030303031e-06,
"loss": 0.1989,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0651925802230835,
"step": 30,
"valid_targets_mean": 3731.5,
"valid_targets_min": 1799
},
{
"epoch": 0.1062753036437247,
"grad_norm": 0.26302391464694386,
"learning_rate": 8.242424242424243e-06,
"loss": 0.1716,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.04892873018980026,
"step": 35,
"valid_targets_mean": 3727.4,
"valid_targets_min": 1884
},
{
"epoch": 0.1214574898785425,
"grad_norm": 0.21091937732454977,
"learning_rate": 9.454545454545456e-06,
"loss": 0.1592,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.051163993775844574,
"step": 40,
"valid_targets_mean": 3958.4,
"valid_targets_min": 1907
},
{
"epoch": 0.13663967611336034,
"grad_norm": 0.18998319529927965,
"learning_rate": 1.0666666666666667e-05,
"loss": 0.1538,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.05154033377766609,
"step": 45,
"valid_targets_mean": 3577.2,
"valid_targets_min": 1914
},
{
"epoch": 0.15182186234817813,
"grad_norm": 0.16050847827157155,
"learning_rate": 1.187878787878788e-05,
"loss": 0.1416,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.04420763999223709,
"step": 50,
"valid_targets_mean": 3796.2,
"valid_targets_min": 2089
},
{
"epoch": 0.16700404858299595,
"grad_norm": 0.17802763431017923,
"learning_rate": 1.3090909090909092e-05,
"loss": 0.1317,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.04160337150096893,
"step": 55,
"valid_targets_mean": 4283.2,
"valid_targets_min": 1982
},
{
"epoch": 0.18218623481781376,
"grad_norm": 0.15511165082528813,
"learning_rate": 1.4303030303030305e-05,
"loss": 0.1237,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.03849010169506073,
"step": 60,
"valid_targets_mean": 3749.1,
"valid_targets_min": 1799
},
{
"epoch": 0.19736842105263158,
"grad_norm": 0.16383205927103164,
"learning_rate": 1.5515151515151516e-05,
"loss": 0.1189,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.033913105726242065,
"step": 65,
"valid_targets_mean": 3822.7,
"valid_targets_min": 1562
},
{
"epoch": 0.2125506072874494,
"grad_norm": 0.1742430101375928,
"learning_rate": 1.672727272727273e-05,
"loss": 0.1159,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.04306407272815704,
"step": 70,
"valid_targets_mean": 4578.4,
"valid_targets_min": 2087
},
{
"epoch": 0.22773279352226722,
"grad_norm": 0.15853754404031653,
"learning_rate": 1.7939393939393942e-05,
"loss": 0.1114,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.03584174066781998,
"step": 75,
"valid_targets_mean": 4356.1,
"valid_targets_min": 1870
},
{
"epoch": 0.242914979757085,
"grad_norm": 0.19401235076777462,
"learning_rate": 1.9151515151515152e-05,
"loss": 0.1089,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.03848009184002876,
"step": 80,
"valid_targets_mean": 3675.0,
"valid_targets_min": 1863
},
{
"epoch": 0.25809716599190285,
"grad_norm": 0.20323002000236698,
"learning_rate": 2.0363636363636365e-05,
"loss": 0.1055,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.03311232104897499,
"step": 85,
"valid_targets_mean": 3828.6,
"valid_targets_min": 1709
},
{
"epoch": 0.2732793522267207,
"grad_norm": 0.1981435700994027,
"learning_rate": 2.1575757575757578e-05,
"loss": 0.1052,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.03980322927236557,
"step": 90,
"valid_targets_mean": 3919.8,
"valid_targets_min": 1982
},
{
"epoch": 0.28846153846153844,
"grad_norm": 0.20391405547861052,
"learning_rate": 2.278787878787879e-05,
"loss": 0.0898,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.028198853135108948,
"step": 95,
"valid_targets_mean": 3938.0,
"valid_targets_min": 1918
},
{
"epoch": 0.30364372469635625,
"grad_norm": 0.2011260592825684,
"learning_rate": 2.4e-05,
"loss": 0.0932,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.031524837017059326,
"step": 100,
"valid_targets_mean": 3767.9,
"valid_targets_min": 1906
},
{
"epoch": 0.3188259109311741,
"grad_norm": 0.20056590507806465,
"learning_rate": 2.5212121212121214e-05,
"loss": 0.0847,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.02495935931801796,
"step": 105,
"valid_targets_mean": 3263.9,
"valid_targets_min": 1662
},
{
"epoch": 0.3340080971659919,
"grad_norm": 0.21140112692239926,
"learning_rate": 2.6424242424242427e-05,
"loss": 0.0797,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.026287691667675972,
"step": 110,
"valid_targets_mean": 3164.8,
"valid_targets_min": 1361
},
{
"epoch": 0.3491902834008097,
"grad_norm": 0.22870318519477917,
"learning_rate": 2.763636363636364e-05,
"loss": 0.079,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.026142392307519913,
"step": 115,
"valid_targets_mean": 3204.0,
"valid_targets_min": 1951
},
{
"epoch": 0.3643724696356275,
"grad_norm": 0.3043038271601291,
"learning_rate": 2.884848484848485e-05,
"loss": 0.0743,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.024183720350265503,
"step": 120,
"valid_targets_mean": 3489.1,
"valid_targets_min": 1662
},
{
"epoch": 0.37955465587044535,
"grad_norm": 0.20759888632242562,
"learning_rate": 3.0060606060606062e-05,
"loss": 0.0674,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.01838626340031624,
"step": 125,
"valid_targets_mean": 2989.6,
"valid_targets_min": 1981
},
{
"epoch": 0.39473684210526316,
"grad_norm": 0.23047055221001278,
"learning_rate": 3.127272727272728e-05,
"loss": 0.0648,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.02019556052982807,
"step": 130,
"valid_targets_mean": 4682.2,
"valid_targets_min": 1652
},
{
"epoch": 0.409919028340081,
"grad_norm": 0.23294010744855131,
"learning_rate": 3.2484848484848485e-05,
"loss": 0.0609,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.016776099801063538,
"step": 135,
"valid_targets_mean": 5853.7,
"valid_targets_min": 1517
},
{
"epoch": 0.4251012145748988,
"grad_norm": 0.2855366270943525,
"learning_rate": 3.36969696969697e-05,
"loss": 0.0617,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.02199370414018631,
"step": 140,
"valid_targets_mean": 3755.6,
"valid_targets_min": 1786
},
{
"epoch": 0.4402834008097166,
"grad_norm": 0.24216134997256672,
"learning_rate": 3.490909090909091e-05,
"loss": 0.0555,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.015389672480523586,
"step": 145,
"valid_targets_mean": 3964.7,
"valid_targets_min": 1919
},
{
"epoch": 0.45546558704453444,
"grad_norm": 0.32949119100037777,
"learning_rate": 3.6121212121212124e-05,
"loss": 0.0528,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.020409779623150826,
"step": 150,
"valid_targets_mean": 2901.4,
"valid_targets_min": 1942
},
{
"epoch": 0.4706477732793522,
"grad_norm": 0.24074948465558085,
"learning_rate": 3.733333333333334e-05,
"loss": 0.0512,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.014248735271394253,
"step": 155,
"valid_targets_mean": 4395.2,
"valid_targets_min": 1858
},
{
"epoch": 0.48582995951417,
"grad_norm": 0.26371267619180006,
"learning_rate": 3.854545454545455e-05,
"loss": 0.0464,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.01621677353978157,
"step": 160,
"valid_targets_mean": 3544.7,
"valid_targets_min": 2025
},
{
"epoch": 0.5010121457489879,
"grad_norm": 0.2303948593638888,
"learning_rate": 3.9757575757575757e-05,
"loss": 0.0421,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.013033179566264153,
"step": 165,
"valid_targets_mean": 3318.6,
"valid_targets_min": 1709
},
{
"epoch": 0.5161943319838057,
"grad_norm": 0.2751485068147365,
"learning_rate": 3.999928391557286e-05,
"loss": 0.0423,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.013866172172129154,
"step": 170,
"valid_targets_mean": 3784.6,
"valid_targets_min": 1877
},
{
"epoch": 0.5313765182186235,
"grad_norm": 0.2162418079826675,
"learning_rate": 3.999637491047052e-05,
"loss": 0.0406,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.010866734199225903,
"step": 175,
"valid_targets_mean": 4622.6,
"valid_targets_min": 1924
},
{
"epoch": 0.5465587044534413,
"grad_norm": 0.19747404288558038,
"learning_rate": 3.999122855464813e-05,
"loss": 0.0363,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.010466480627655983,
"step": 180,
"valid_targets_mean": 4335.8,
"valid_targets_min": 1884
},
{
"epoch": 0.5617408906882592,
"grad_norm": 0.23089319129780744,
"learning_rate": 3.998384542392021e-05,
"loss": 0.036,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.01500026322901249,
"step": 185,
"valid_targets_mean": 3850.8,
"valid_targets_min": 1517
},
{
"epoch": 0.5769230769230769,
"grad_norm": 0.2127962416533006,
"learning_rate": 3.9974226344369124e-05,
"loss": 0.0337,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.008760621771216393,
"step": 190,
"valid_targets_mean": 4441.3,
"valid_targets_min": 1940
},
{
"epoch": 0.5921052631578947,
"grad_norm": 0.2187036886834824,
"learning_rate": 3.996237239225268e-05,
"loss": 0.0308,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.010154707357287407,
"step": 195,
"valid_targets_mean": 3621.6,
"valid_targets_min": 1900
},
{
"epoch": 0.6072874493927125,
"grad_norm": 0.24501759700841205,
"learning_rate": 3.994828489388371e-05,
"loss": 0.0303,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.009506016969680786,
"step": 200,
"valid_targets_mean": 3383.7,
"valid_targets_min": 1601
},
{
"epoch": 0.6224696356275303,
"grad_norm": 0.22959276687097194,
"learning_rate": 3.993196542548162e-05,
"loss": 0.0293,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00975283607840538,
"step": 205,
"valid_targets_mean": 3611.7,
"valid_targets_min": 1763
},
{
"epoch": 0.6376518218623481,
"grad_norm": 0.24080321314548556,
"learning_rate": 3.991341581299609e-05,
"loss": 0.0263,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.011061406694352627,
"step": 210,
"valid_targets_mean": 3533.5,
"valid_targets_min": 1562
},
{
"epoch": 0.652834008097166,
"grad_norm": 0.20805800614271266,
"learning_rate": 3.9892638131902765e-05,
"loss": 0.0246,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.011325137689709663,
"step": 215,
"valid_targets_mean": 3618.1,
"valid_targets_min": 1909
},
{
"epoch": 0.6680161943319838,
"grad_norm": 0.19699851157747658,
"learning_rate": 3.9869634706971e-05,
"loss": 0.0227,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0078015620820224285,
"step": 220,
"valid_targets_mean": 4144.3,
"valid_targets_min": 1705
},
{
"epoch": 0.6831983805668016,
"grad_norm": 0.2185159886324255,
"learning_rate": 3.984440811200379e-05,
"loss": 0.0229,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.008051777258515358,
"step": 225,
"valid_targets_mean": 3167.5,
"valid_targets_min": 1525
},
{
"epoch": 0.6983805668016194,
"grad_norm": 0.22934835342502952,
"learning_rate": 3.981696116954973e-05,
"loss": 0.0204,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0072164274752140045,
"step": 230,
"valid_targets_mean": 4077.3,
"valid_targets_min": 1799
},
{
"epoch": 0.7135627530364372,
"grad_norm": 0.19869122826969568,
"learning_rate": 3.978729695058729e-05,
"loss": 0.0209,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.005740233231335878,
"step": 235,
"valid_targets_mean": 3489.4,
"valid_targets_min": 2089
},
{
"epoch": 0.728744939271255,
"grad_norm": 0.19919625368885566,
"learning_rate": 3.9755418774181146e-05,
"loss": 0.0198,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.007875881157815456,
"step": 240,
"valid_targets_mean": 3947.7,
"valid_targets_min": 1982
},
{
"epoch": 0.7439271255060729,
"grad_norm": 0.20919474516010492,
"learning_rate": 3.9721330207110835e-05,
"loss": 0.0182,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.005854409653693438,
"step": 245,
"valid_targets_mean": 3425.2,
"valid_targets_min": 1763
},
{
"epoch": 0.7591093117408907,
"grad_norm": 0.19447693583220627,
"learning_rate": 3.9685035063471675e-05,
"loss": 0.0181,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.007061396725475788,
"step": 250,
"valid_targets_mean": 4074.4,
"valid_targets_min": 1802
},
{
"epoch": 0.7742914979757085,
"grad_norm": 0.22629827321134982,
"learning_rate": 3.964653740424804e-05,
"loss": 0.0166,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.005824090447276831,
"step": 255,
"valid_targets_mean": 3201.4,
"valid_targets_min": 1771
},
{
"epoch": 0.7894736842105263,
"grad_norm": 0.24577229492094663,
"learning_rate": 3.960584153685895e-05,
"loss": 0.0158,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.005213964730501175,
"step": 260,
"valid_targets_mean": 2662.2,
"valid_targets_min": 1614
},
{
"epoch": 0.8046558704453441,
"grad_norm": 0.21388131919460343,
"learning_rate": 3.9562952014676116e-05,
"loss": 0.0158,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00423480523750186,
"step": 265,
"valid_targets_mean": 3514.4,
"valid_targets_min": 1858
},
{
"epoch": 0.819838056680162,
"grad_norm": 0.178130321151751,
"learning_rate": 3.9517873636514525e-05,
"loss": 0.0136,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.004601405002176762,
"step": 270,
"valid_targets_mean": 4206.3,
"valid_targets_min": 1716
},
{
"epoch": 0.8350202429149798,
"grad_norm": 0.18405218336133228,
"learning_rate": 3.947061144609546e-05,
"loss": 0.0132,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.005071922671049833,
"step": 275,
"valid_targets_mean": 4017.1,
"valid_targets_min": 1839
},
{
"epoch": 0.8502024291497976,
"grad_norm": 0.17657921729737056,
"learning_rate": 3.942117073148221e-05,
"loss": 0.0127,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0038262158632278442,
"step": 280,
"valid_targets_mean": 3661.7,
"valid_targets_min": 1709
},
{
"epoch": 0.8653846153846154,
"grad_norm": 0.1830557185649827,
"learning_rate": 3.9369557024488345e-05,
"loss": 0.0125,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0034842612221837044,
"step": 285,
"valid_targets_mean": 3941.0,
"valid_targets_min": 1906
},
{
"epoch": 0.8805668016194332,
"grad_norm": 0.15062994300292865,
"learning_rate": 3.931577610005883e-05,
"loss": 0.0114,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.004331760574132204,
"step": 290,
"valid_targets_mean": 3859.6,
"valid_targets_min": 2009
},
{
"epoch": 0.895748987854251,
"grad_norm": 0.18322793798575313,
"learning_rate": 3.925983397562385e-05,
"loss": 0.0124,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.004891665652394295,
"step": 295,
"valid_targets_mean": 4143.8,
"valid_targets_min": 1914
},
{
"epoch": 0.9109311740890689,
"grad_norm": 0.1674625166589611,
"learning_rate": 3.920173691042554e-05,
"loss": 0.0115,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0034718886017799377,
"step": 300,
"valid_targets_mean": 3484.5,
"valid_targets_min": 1921
},
{
"epoch": 0.9261133603238867,
"grad_norm": 0.16809713590324474,
"learning_rate": 3.914149140481766e-05,
"loss": 0.0104,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.002860719570890069,
"step": 305,
"valid_targets_mean": 2891.0,
"valid_targets_min": 1662
},
{
"epoch": 0.9412955465587044,
"grad_norm": 0.16774022368034475,
"learning_rate": 3.9079104199538256e-05,
"loss": 0.0111,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.003308693878352642,
"step": 310,
"valid_targets_mean": 3956.6,
"valid_targets_min": 1833
},
{
"epoch": 0.9564777327935222,
"grad_norm": 0.1807349282309273,
"learning_rate": 3.901458227495549e-05,
"loss": 0.0105,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.003315879963338375,
"step": 315,
"valid_targets_mean": 4054.1,
"valid_targets_min": 1914
},
{
"epoch": 0.97165991902834,
"grad_norm": 0.1742099004293395,
"learning_rate": 3.8947932850286585e-05,
"loss": 0.0091,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00291993236169219,
"step": 320,
"valid_targets_mean": 3363.0,
"valid_targets_min": 1952
},
{
"epoch": 0.9868421052631579,
"grad_norm": 0.14500304259148103,
"learning_rate": 3.887916338279014e-05,
"loss": 0.0103,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0026057774666696787,
"step": 325,
"valid_targets_mean": 4190.8,
"valid_targets_min": 2033
},
{
"epoch": 1.0,
"grad_norm": 0.2517151943221558,
"learning_rate": 3.8808281566931675e-05,
"loss": 0.0091,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.008416064083576202,
"step": 330,
"valid_targets_mean": 3570.5,
"valid_targets_min": 1562
},
{
"epoch": 1.0151821862348178,
"grad_norm": 0.164243403283566,
"learning_rate": 3.873529533352277e-05,
"loss": 0.0087,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.003136058570817113,
"step": 335,
"valid_targets_mean": 3419.4,
"valid_targets_min": 1951
},
{
"epoch": 1.0303643724696356,
"grad_norm": 0.14414941894165267,
"learning_rate": 3.8660212848833705e-05,
"loss": 0.0084,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0024711661972105503,
"step": 340,
"valid_targets_mean": 3528.8,
"valid_targets_min": 1868
},
{
"epoch": 1.0455465587044535,
"grad_norm": 0.19586224740728458,
"learning_rate": 3.858304251367972e-05,
"loss": 0.008,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0027735955081880093,
"step": 345,
"valid_targets_mean": 2958.7,
"valid_targets_min": 1361
},
{
"epoch": 1.0607287449392713,
"grad_norm": 0.14897574986322765,
"learning_rate": 3.850379296248107e-05,
"loss": 0.0078,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0028114435262978077,
"step": 350,
"valid_targets_mean": 3791.8,
"valid_targets_min": 1870
},
{
"epoch": 1.075910931174089,
"grad_norm": 0.1558838429784627,
"learning_rate": 3.8422473062297e-05,
"loss": 0.0076,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0024879274424165487,
"step": 355,
"valid_targets_mean": 3638.7,
"valid_targets_min": 2175
},
{
"epoch": 1.091093117408907,
"grad_norm": 0.17308536245807407,
"learning_rate": 3.8339091911833545e-05,
"loss": 0.0076,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0038288887590169907,
"step": 360,
"valid_targets_mean": 3870.2,
"valid_targets_min": 1942
},
{
"epoch": 1.1062753036437247,
"grad_norm": 0.16779799609529814,
"learning_rate": 3.825365884042553e-05,
"loss": 0.0074,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.002971275243908167,
"step": 365,
"valid_targets_mean": 3224.4,
"valid_targets_min": 1680
},
{
"epoch": 1.1214574898785425,
"grad_norm": 0.15172020910248868,
"learning_rate": 3.8166183406992745e-05,
"loss": 0.0065,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0018144859932363033,
"step": 370,
"valid_targets_mean": 2999.0,
"valid_targets_min": 2026
},
{
"epoch": 1.1366396761133604,
"grad_norm": 0.16339142774129306,
"learning_rate": 3.807667539897041e-05,
"loss": 0.0068,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0029783309437334538,
"step": 375,
"valid_targets_mean": 3691.7,
"valid_targets_min": 1984
},
{
"epoch": 1.1518218623481782,
"grad_norm": 0.1292852541633403,
"learning_rate": 3.798514483121408e-05,
"loss": 0.0065,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0017669487278908491,
"step": 380,
"valid_targets_mean": 4103.9,
"valid_targets_min": 1614
},
{
"epoch": 1.167004048582996,
"grad_norm": 0.1371887720281193,
"learning_rate": 3.789160194487908e-05,
"loss": 0.0067,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00210462580434978,
"step": 385,
"valid_targets_mean": 4121.5,
"valid_targets_min": 1972
},
{
"epoch": 1.1821862348178138,
"grad_norm": 0.14485399953221598,
"learning_rate": 3.7796057206274686e-05,
"loss": 0.0063,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.001915339264087379,
"step": 390,
"valid_targets_mean": 4134.5,
"valid_targets_min": 1970
},
{
"epoch": 1.1973684210526316,
"grad_norm": 0.14420148445134826,
"learning_rate": 3.769852130569304e-05,
"loss": 0.0057,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0018860504496842623,
"step": 395,
"valid_targets_mean": 3892.3,
"valid_targets_min": 1909
},
{
"epoch": 1.2125506072874495,
"grad_norm": 0.13878326827529558,
"learning_rate": 3.7599005156213066e-05,
"loss": 0.006,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0019143268000334501,
"step": 400,
"valid_targets_mean": 3761.9,
"valid_targets_min": 1915
},
{
"epoch": 1.2277327935222673,
"grad_norm": 0.11476026427302558,
"learning_rate": 3.74975198924794e-05,
"loss": 0.0052,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0012934381375089288,
"step": 405,
"valid_targets_mean": 5056.4,
"valid_targets_min": 2001
},
{
"epoch": 1.242914979757085,
"grad_norm": 0.1330200894754117,
"learning_rate": 3.739407686945658e-05,
"loss": 0.0054,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0018799215322360396,
"step": 410,
"valid_targets_mean": 3872.5,
"valid_targets_min": 1942
},
{
"epoch": 1.258097165991903,
"grad_norm": 0.13846149867579144,
"learning_rate": 3.728868766115854e-05,
"loss": 0.0059,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0020003884565085173,
"step": 415,
"valid_targets_mean": 3473.5,
"valid_targets_min": 1716
},
{
"epoch": 1.2732793522267207,
"grad_norm": 0.13605245571458235,
"learning_rate": 3.718136405935365e-05,
"loss": 0.0049,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0019030460389330983,
"step": 420,
"valid_targets_mean": 3039.6,
"valid_targets_min": 1900
},
{
"epoch": 1.2884615384615383,
"grad_norm": 0.1275992332798571,
"learning_rate": 3.707211807224534e-05,
"loss": 0.005,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0015746339922770858,
"step": 425,
"valid_targets_mean": 3385.7,
"valid_targets_min": 1847
},
{
"epoch": 1.3036437246963564,
"grad_norm": 0.11948889386818098,
"learning_rate": 3.696096192312852e-05,
"loss": 0.0051,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0015787535812705755,
"step": 430,
"valid_targets_mean": 3911.8,
"valid_targets_min": 1950
},
{
"epoch": 1.318825910931174,
"grad_norm": 0.14109767925238773,
"learning_rate": 3.684790804902199e-05,
"loss": 0.0054,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0016520526260137558,
"step": 435,
"valid_targets_mean": 4149.7,
"valid_targets_min": 1836
},
{
"epoch": 1.334008097165992,
"grad_norm": 0.138897250305289,
"learning_rate": 3.673296909927682e-05,
"loss": 0.0051,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0018126749200746417,
"step": 440,
"valid_targets_mean": 3260.5,
"valid_targets_min": 1839
},
{
"epoch": 1.3491902834008096,
"grad_norm": 0.11611244997870669,
"learning_rate": 3.661615793416109e-05,
"loss": 0.0045,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0014871362363919616,
"step": 445,
"valid_targets_mean": 4510.7,
"valid_targets_min": 1760
},
{
"epoch": 1.3643724696356276,
"grad_norm": 0.120597485658797,
"learning_rate": 3.649748762342098e-05,
"loss": 0.0042,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0010552321327850223,
"step": 450,
"valid_targets_mean": 4055.2,
"valid_targets_min": 2042
},
{
"epoch": 1.3795546558704452,
"grad_norm": 0.11162350623192044,
"learning_rate": 3.637697144481839e-05,
"loss": 0.0048,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0013892888091504574,
"step": 455,
"valid_targets_mean": 4131.2,
"valid_targets_min": 1900
},
{
"epoch": 1.3947368421052633,
"grad_norm": 0.13236130484756595,
"learning_rate": 3.625462288264536e-05,
"loss": 0.0043,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.001717492239549756,
"step": 460,
"valid_targets_mean": 4701.0,
"valid_targets_min": 2136
},
{
"epoch": 1.4099190283400809,
"grad_norm": 0.10634207206921958,
"learning_rate": 3.613045562621533e-05,
"loss": 0.0046,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0014703264459967613,
"step": 465,
"valid_targets_mean": 4877.9,
"valid_targets_min": 1993
},
{
"epoch": 1.425101214574899,
"grad_norm": 0.11833374847627441,
"learning_rate": 3.600448356833146e-05,
"loss": 0.004,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.001124208327382803,
"step": 470,
"valid_targets_mean": 3889.0,
"valid_targets_min": 1933
},
{
"epoch": 1.4402834008097165,
"grad_norm": 0.09905163928566187,
"learning_rate": 3.587672080373219e-05,
"loss": 0.0039,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0014082645066082478,
"step": 475,
"valid_targets_mean": 5741.5,
"valid_targets_min": 1719
},
{
"epoch": 1.4554655870445345,
"grad_norm": 0.10883487374241034,
"learning_rate": 3.574718162751426e-05,
"loss": 0.0039,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0012796204537153244,
"step": 480,
"valid_targets_mean": 3971.8,
"valid_targets_min": 2054
},
{
"epoch": 1.4706477732793521,
"grad_norm": 0.11845317295570815,
"learning_rate": 3.561588053353319e-05,
"loss": 0.0043,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0014125773450359702,
"step": 485,
"valid_targets_mean": 3225.4,
"valid_targets_min": 1763
},
{
"epoch": 1.48582995951417,
"grad_norm": 0.12120113157747212,
"learning_rate": 3.5482832212781655e-05,
"loss": 0.0037,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0011800138745456934,
"step": 490,
"valid_targets_mean": 3134.7,
"valid_targets_min": 1900
},
{
"epoch": 1.5010121457489878,
"grad_norm": 0.11487518503490175,
"learning_rate": 3.53480515517457e-05,
"loss": 0.0043,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.001305198296904564,
"step": 495,
"valid_targets_mean": 3831.7,
"valid_targets_min": 1525
},
{
"epoch": 1.5161943319838058,
"grad_norm": 0.11042030153261334,
"learning_rate": 3.5211553630739166e-05,
"loss": 0.0042,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0014829908031970263,
"step": 500,
"valid_targets_mean": 2815.1,
"valid_targets_min": 1662
},
{
"epoch": 1.5313765182186234,
"grad_norm": 0.09669183557992163,
"learning_rate": 3.5073353722216334e-05,
"loss": 0.0035,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0009500256273895502,
"step": 505,
"valid_targets_mean": 4798.8,
"valid_targets_min": 1891
},
{
"epoch": 1.5465587044534415,
"grad_norm": 0.10258133251776518,
"learning_rate": 3.4933467289063156e-05,
"loss": 0.0039,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.000988983316347003,
"step": 510,
"valid_targets_mean": 3039.0,
"valid_targets_min": 1857
},
{
"epoch": 1.561740890688259,
"grad_norm": 0.10410092818256382,
"learning_rate": 3.4791909982867175e-05,
"loss": 0.0036,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0013444151263684034,
"step": 515,
"valid_targets_mean": 4029.6,
"valid_targets_min": 1361
},
{
"epoch": 1.5769230769230769,
"grad_norm": 0.09960156518662373,
"learning_rate": 3.464869764216622e-05,
"loss": 0.0039,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00153966317884624,
"step": 520,
"valid_targets_mean": 5110.7,
"valid_targets_min": 1988
},
{
"epoch": 1.5921052631578947,
"grad_norm": 0.09983557988159186,
"learning_rate": 3.450384629067635e-05,
"loss": 0.0034,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.001081749564036727,
"step": 525,
"valid_targets_mean": 3048.0,
"valid_targets_min": 1662
},
{
"epoch": 1.6072874493927125,
"grad_norm": 0.10120168106576323,
"learning_rate": 3.435737213549896e-05,
"loss": 0.0033,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0008452935726381838,
"step": 530,
"valid_targets_mean": 3952.0,
"valid_targets_min": 1562
},
{
"epoch": 1.6224696356275303,
"grad_norm": 0.08493587072597682,
"learning_rate": 3.420929156530738e-05,
"loss": 0.0036,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0011547571048140526,
"step": 535,
"valid_targets_mean": 4687.3,
"valid_targets_min": 1820
},
{
"epoch": 1.6376518218623481,
"grad_norm": 0.10523126139385196,
"learning_rate": 3.405962114851324e-05,
"loss": 0.0034,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0011256784200668335,
"step": 540,
"valid_targets_mean": 3631.1,
"valid_targets_min": 2011
},
{
"epoch": 1.652834008097166,
"grad_norm": 0.11195712798002991,
"learning_rate": 3.390837763141261e-05,
"loss": 0.0035,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0013444002252072096,
"step": 545,
"valid_targets_mean": 2856.7,
"valid_targets_min": 1966
},
{
"epoch": 1.6680161943319838,
"grad_norm": 0.08610234523684804,
"learning_rate": 3.3755577936312344e-05,
"loss": 0.0032,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0011940321419388056,
"step": 550,
"valid_targets_mean": 4242.0,
"valid_targets_min": 1719
},
{
"epoch": 1.6831983805668016,
"grad_norm": 0.1051662009692625,
"learning_rate": 3.360123915963662e-05,
"loss": 0.003,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0009770877659320831,
"step": 555,
"valid_targets_mean": 4124.0,
"valid_targets_min": 2062
},
{
"epoch": 1.6983805668016194,
"grad_norm": 0.11776081274186186,
"learning_rate": 3.3445378570014125e-05,
"loss": 0.0031,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0013557838974520564,
"step": 560,
"valid_targets_mean": 3496.0,
"valid_targets_min": 2161
},
{
"epoch": 1.7135627530364372,
"grad_norm": 0.10165976143565304,
"learning_rate": 3.328801360634585e-05,
"loss": 0.003,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0008353482699021697,
"step": 565,
"valid_targets_mean": 3297.1,
"valid_targets_min": 1919
},
{
"epoch": 1.728744939271255,
"grad_norm": 0.10319162200707634,
"learning_rate": 3.312916187585392e-05,
"loss": 0.0032,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.001077563501894474,
"step": 570,
"valid_targets_mean": 3406.3,
"valid_targets_min": 1562
},
{
"epoch": 1.7439271255060729,
"grad_norm": 0.08743872003639647,
"learning_rate": 3.296884115211157e-05,
"loss": 0.003,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0007602584082633257,
"step": 575,
"valid_targets_mean": 3701.0,
"valid_targets_min": 1751
},
{
"epoch": 1.7591093117408907,
"grad_norm": 0.10299599098322705,
"learning_rate": 3.280706937305445e-05,
"loss": 0.0031,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00123096804600209,
"step": 580,
"valid_targets_mean": 3627.1,
"valid_targets_min": 1601
},
{
"epoch": 1.7742914979757085,
"grad_norm": 0.09201994600780483,
"learning_rate": 3.2643864638973645e-05,
"loss": 0.003,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0009430198697373271,
"step": 585,
"valid_targets_mean": 3924.4,
"valid_targets_min": 1928
},
{
"epoch": 1.7894736842105263,
"grad_norm": 0.09185507240932078,
"learning_rate": 3.2479245210490434e-05,
"loss": 0.0029,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0010372858960181475,
"step": 590,
"valid_targets_mean": 5039.9,
"valid_targets_min": 1716
},
{
"epoch": 1.8046558704453441,
"grad_norm": 0.0849595096314164,
"learning_rate": 3.2313229506513167e-05,
"loss": 0.0027,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0008949165930971503,
"step": 595,
"valid_targets_mean": 4666.0,
"valid_targets_min": 1614
},
{
"epoch": 1.819838056680162,
"grad_norm": 0.08745985433128926,
"learning_rate": 3.2145836102176424e-05,
"loss": 0.0025,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0006510005332529545,
"step": 600,
"valid_targets_mean": 3548.5,
"valid_targets_min": 1705
},
{
"epoch": 1.8350202429149798,
"grad_norm": 0.08825374567222655,
"learning_rate": 3.197708372676265e-05,
"loss": 0.0028,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0010350134689360857,
"step": 605,
"valid_targets_mean": 4250.8,
"valid_targets_min": 1987
},
{
"epoch": 1.8502024291497976,
"grad_norm": 0.09666617435790727,
"learning_rate": 3.1806991261606604e-05,
"loss": 0.0024,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0008731918642297387,
"step": 610,
"valid_targets_mean": 4138.9,
"valid_targets_min": 1786
},
{
"epoch": 1.8653846153846154,
"grad_norm": 0.08891992368198291,
"learning_rate": 3.163557773798276e-05,
"loss": 0.0025,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0007548497524112463,
"step": 615,
"valid_targets_mean": 3596.1,
"valid_targets_min": 1915
},
{
"epoch": 1.8805668016194332,
"grad_norm": 0.10298148057397864,
"learning_rate": 3.146286233497593e-05,
"loss": 0.0027,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0012539434246718884,
"step": 620,
"valid_targets_mean": 3995.3,
"valid_targets_min": 1771
},
{
"epoch": 1.895748987854251,
"grad_norm": 0.09072979307646942,
"learning_rate": 3.128886437733539e-05,
"loss": 0.0026,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0008174719987437129,
"step": 625,
"valid_targets_mean": 4311.0,
"valid_targets_min": 1952
},
{
"epoch": 1.9109311740890689,
"grad_norm": 0.10926696533151599,
"learning_rate": 3.111360333331263e-05,
"loss": 0.0026,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0006682357052341104,
"step": 630,
"valid_targets_mean": 3494.0,
"valid_targets_min": 1933
},
{
"epoch": 1.9261133603238867,
"grad_norm": 0.07708358428359662,
"learning_rate": 3.093709881248312e-05,
"loss": 0.0025,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0006784072029404342,
"step": 635,
"valid_targets_mean": 4398.0,
"valid_targets_min": 2108
},
{
"epoch": 1.9412955465587043,
"grad_norm": 0.09782921316490462,
"learning_rate": 3.075937056355225e-05,
"loss": 0.0027,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0006533457199111581,
"step": 640,
"valid_targets_mean": 3165.5,
"valid_targets_min": 1562
},
{
"epoch": 1.9564777327935223,
"grad_norm": 0.07874241058965538,
"learning_rate": 3.0580438472145665e-05,
"loss": 0.0024,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0005525833694264293,
"step": 645,
"valid_targets_mean": 4149.3,
"valid_targets_min": 1870
},
{
"epoch": 1.97165991902834,
"grad_norm": 0.07242269740889695,
"learning_rate": 3.0400322558584308e-05,
"loss": 0.0022,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00043955055298283696,
"step": 650,
"valid_targets_mean": 3438.7,
"valid_targets_min": 1799
},
{
"epoch": 1.986842105263158,
"grad_norm": 0.08322163172521595,
"learning_rate": 3.0219042975644415e-05,
"loss": 0.0022,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0008516611997038126,
"step": 655,
"valid_targets_mean": 4353.1,
"valid_targets_min": 2250
},
{
"epoch": 2.0,
"grad_norm": 0.12405167808565279,
"learning_rate": 3.0036620006302624e-05,
"loss": 0.002,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.001781097729690373,
"step": 660,
"valid_targets_mean": 5059.7,
"valid_targets_min": 2012
},
{
"epoch": 2.0151821862348176,
"grad_norm": 0.07105519248520684,
"learning_rate": 2.9853074061466602e-05,
"loss": 0.002,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0005106439930386841,
"step": 665,
"valid_targets_mean": 4679.2,
"valid_targets_min": 1914
},
{
"epoch": 2.0303643724696356,
"grad_norm": 0.07612946301913277,
"learning_rate": 2.9668425677691278e-05,
"loss": 0.002,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0007456626044586301,
"step": 670,
"valid_targets_mean": 4227.5,
"valid_targets_min": 1839
},
{
"epoch": 2.0455465587044532,
"grad_norm": 0.09095623858759358,
"learning_rate": 2.948269551488108e-05,
"loss": 0.0021,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0008456467185169458,
"step": 675,
"valid_targets_mean": 2652.4,
"valid_targets_min": 1908
},
{
"epoch": 2.0607287449392713,
"grad_norm": 0.078083688802221,
"learning_rate": 2.929590435397832e-05,
"loss": 0.0019,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0005976787069812417,
"step": 680,
"valid_targets_mean": 3305.4,
"valid_targets_min": 1838
},
{
"epoch": 2.075910931174089,
"grad_norm": 0.08699014214754179,
"learning_rate": 2.9108073094638066e-05,
"loss": 0.0021,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.000658693490549922,
"step": 685,
"valid_targets_mean": 4065.9,
"valid_targets_min": 1361
},
{
"epoch": 2.091093117408907,
"grad_norm": 0.06937637279552442,
"learning_rate": 2.8919222752889727e-05,
"loss": 0.002,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0004983286489732563,
"step": 690,
"valid_targets_mean": 3370.7,
"valid_targets_min": 1919
},
{
"epoch": 2.1062753036437245,
"grad_norm": 0.07691820761499013,
"learning_rate": 2.8729374458785647e-05,
"loss": 0.0019,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0006680118967778981,
"step": 695,
"valid_targets_mean": 3699.5,
"valid_targets_min": 1889
},
{
"epoch": 2.1214574898785425,
"grad_norm": 0.0904290166328251,
"learning_rate": 2.8538549454036838e-05,
"loss": 0.0021,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0008592807571403682,
"step": 700,
"valid_targets_mean": 3886.0,
"valid_targets_min": 1709
},
{
"epoch": 2.13663967611336,
"grad_norm": 0.07641089925208772,
"learning_rate": 2.834676908963636e-05,
"loss": 0.002,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0005071223713457584,
"step": 705,
"valid_targets_mean": 3066.1,
"valid_targets_min": 1914
},
{
"epoch": 2.151821862348178,
"grad_norm": 0.08676223294976994,
"learning_rate": 2.815405482347037e-05,
"loss": 0.0018,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0006401725113391876,
"step": 710,
"valid_targets_mean": 3339.6,
"valid_targets_min": 1562
},
{
"epoch": 2.167004048582996,
"grad_norm": 0.08150309039261369,
"learning_rate": 2.796042821791725e-05,
"loss": 0.002,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0007411183905787766,
"step": 715,
"valid_targets_mean": 4100.4,
"valid_targets_min": 1917
},
{
"epoch": 2.182186234817814,
"grad_norm": 0.06525762409218074,
"learning_rate": 2.776591093743505e-05,
"loss": 0.0017,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0004551703459583223,
"step": 720,
"valid_targets_mean": 3649.8,
"valid_targets_min": 1833
},
{
"epoch": 2.1973684210526314,
"grad_norm": 0.08423433753590777,
"learning_rate": 2.7570524746137485e-05,
"loss": 0.0016,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0005831866292282939,
"step": 725,
"valid_targets_mean": 3069.6,
"valid_targets_min": 1562
},
{
"epoch": 2.2125506072874495,
"grad_norm": 0.08427119680751806,
"learning_rate": 2.7374291505358818e-05,
"loss": 0.0018,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0006087704095989466,
"step": 730,
"valid_targets_mean": 3375.6,
"valid_targets_min": 1662
},
{
"epoch": 2.227732793522267,
"grad_norm": 0.07080323308041764,
"learning_rate": 2.7177233171207817e-05,
"loss": 0.0021,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0004960486548952758,
"step": 735,
"valid_targets_mean": 3955.1,
"valid_targets_min": 1952
},
{
"epoch": 2.242914979757085,
"grad_norm": 0.06754870894326025,
"learning_rate": 2.6979371792111147e-05,
"loss": 0.0017,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0006193451699800789,
"step": 740,
"valid_targets_mean": 4175.6,
"valid_targets_min": 1799
},
{
"epoch": 2.2580971659919027,
"grad_norm": 0.08387447697414334,
"learning_rate": 2.678072950634641e-05,
"loss": 0.0018,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.000532821228262037,
"step": 745,
"valid_targets_mean": 3796.0,
"valid_targets_min": 1972
},
{
"epoch": 2.2732793522267207,
"grad_norm": 0.07492202664392031,
"learning_rate": 2.6581328539565184e-05,
"loss": 0.0017,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0007153525366447866,
"step": 750,
"valid_targets_mean": 3809.5,
"valid_targets_min": 2012
},
{
"epoch": 2.2884615384615383,
"grad_norm": 0.07901447382038594,
"learning_rate": 2.638119120230616e-05,
"loss": 0.0017,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00038476218469440937,
"step": 755,
"valid_targets_mean": 3084.6,
"valid_targets_min": 1942
},
{
"epoch": 2.3036437246963564,
"grad_norm": 0.07009381327273641,
"learning_rate": 2.618033988749895e-05,
"loss": 0.0018,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0004404044011607766,
"step": 760,
"valid_targets_mean": 3970.5,
"valid_targets_min": 2022
},
{
"epoch": 2.318825910931174,
"grad_norm": 0.06088519307445401,
"learning_rate": 2.5978797067958542e-05,
"loss": 0.0015,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0003948418889194727,
"step": 765,
"valid_targets_mean": 3346.2,
"valid_targets_min": 1680
},
{
"epoch": 2.334008097165992,
"grad_norm": 0.08152602589343329,
"learning_rate": 2.5776585293870877e-05,
"loss": 0.0016,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0007163034752011299,
"step": 770,
"valid_targets_mean": 3709.0,
"valid_targets_min": 1868
},
{
"epoch": 2.3491902834008096,
"grad_norm": 0.07710624633342333,
"learning_rate": 2.557372719026976e-05,
"loss": 0.0017,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0005486889276653528,
"step": 775,
"valid_targets_mean": 3343.7,
"valid_targets_min": 1716
},
{
"epoch": 2.3643724696356276,
"grad_norm": 0.06266729922368013,
"learning_rate": 2.537024545450539e-05,
"loss": 0.0015,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0006524648051708937,
"step": 780,
"valid_targets_mean": 4165.9,
"valid_targets_min": 1891
},
{
"epoch": 2.3795546558704452,
"grad_norm": 0.06468733900085777,
"learning_rate": 2.5166162853704825e-05,
"loss": 0.0017,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0004791424435097724,
"step": 785,
"valid_targets_mean": 3558.7,
"valid_targets_min": 1652
},
{
"epoch": 2.3947368421052633,
"grad_norm": 0.06898761464531536,
"learning_rate": 2.496150222222458e-05,
"loss": 0.0017,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00033683591755107045,
"step": 790,
"valid_targets_mean": 3842.0,
"valid_targets_min": 1839
},
{
"epoch": 2.409919028340081,
"grad_norm": 0.056657057717980436,
"learning_rate": 2.475628645909576e-05,
"loss": 0.0015,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0004480042844079435,
"step": 795,
"valid_targets_mean": 4623.7,
"valid_targets_min": 1709
},
{
"epoch": 2.425101214574899,
"grad_norm": 0.07727531147134924,
"learning_rate": 2.4550538525461963e-05,
"loss": 0.0014,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0006690880982205272,
"step": 800,
"valid_targets_mean": 4689.0,
"valid_targets_min": 2002
},
{
"epoch": 2.4402834008097165,
"grad_norm": 0.0571678762573575,
"learning_rate": 2.434428144201016e-05,
"loss": 0.0013,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0005046890582889318,
"step": 805,
"valid_targets_mean": 3675.2,
"valid_targets_min": 1763
},
{
"epoch": 2.4554655870445345,
"grad_norm": 0.07715049920265724,
"learning_rate": 2.4137538286394976e-05,
"loss": 0.0014,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0004507353878580034,
"step": 810,
"valid_targets_mean": 3737.3,
"valid_targets_min": 2033
},
{
"epoch": 2.470647773279352,
"grad_norm": 0.05356937759055232,
"learning_rate": 2.3930332190656604e-05,
"loss": 0.0013,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00032879202626645565,
"step": 815,
"valid_targets_mean": 3223.5,
"valid_targets_min": 1884
},
{
"epoch": 2.48582995951417,
"grad_norm": 0.07714930437893976,
"learning_rate": 2.3722686338632602e-05,
"loss": 0.0014,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0005657218862324953,
"step": 820,
"valid_targets_mean": 3412.6,
"valid_targets_min": 1760
},
{
"epoch": 2.501012145748988,
"grad_norm": 0.06024036597283588,
"learning_rate": 2.3514623963363886e-05,
"loss": 0.0012,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00045516961836256087,
"step": 825,
"valid_targets_mean": 3558.2,
"valid_targets_min": 1940
},
{
"epoch": 2.516194331983806,
"grad_norm": 0.06292856358359337,
"learning_rate": 2.330616834449525e-05,
"loss": 0.0012,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0004074290336575359,
"step": 830,
"valid_targets_mean": 3850.9,
"valid_targets_min": 1919
},
{
"epoch": 2.5313765182186234,
"grad_norm": 0.07449928244101439,
"learning_rate": 2.309734280567065e-05,
"loss": 0.0013,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0006758588715456426,
"step": 835,
"valid_targets_mean": 3555.7,
"valid_targets_min": 1847
},
{
"epoch": 2.5465587044534415,
"grad_norm": 0.048901154793971606,
"learning_rate": 2.28881707119236e-05,
"loss": 0.0013,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00026356778107583523,
"step": 840,
"valid_targets_mean": 3955.5,
"valid_targets_min": 1999
},
{
"epoch": 2.561740890688259,
"grad_norm": 0.05986272043977923,
"learning_rate": 2.267867546706287e-05,
"loss": 0.0012,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0004422905622050166,
"step": 845,
"valid_targets_mean": 3143.4,
"valid_targets_min": 1839
},
{
"epoch": 2.5769230769230766,
"grad_norm": 0.06674161610232071,
"learning_rate": 2.2468880511053896e-05,
"loss": 0.0013,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00025287189055234194,
"step": 850,
"valid_targets_mean": 4546.0,
"valid_targets_min": 1913
},
{
"epoch": 2.5921052631578947,
"grad_norm": 0.05409182676999133,
"learning_rate": 2.2258809317396163e-05,
"loss": 0.0013,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00036123377503827214,
"step": 855,
"valid_targets_mean": 3890.5,
"valid_targets_min": 2091
},
{
"epoch": 2.6072874493927127,
"grad_norm": 0.057302284047087634,
"learning_rate": 2.2048485390496757e-05,
"loss": 0.0012,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00038291895180009305,
"step": 860,
"valid_targets_mean": 3662.3,
"valid_targets_min": 1601
},
{
"epoch": 2.6224696356275303,
"grad_norm": 0.05224407807013093,
"learning_rate": 2.1837932263040553e-05,
"loss": 0.001,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0002840733213815838,
"step": 865,
"valid_targets_mean": 3687.8,
"valid_targets_min": 1680
},
{
"epoch": 2.637651821862348,
"grad_norm": 0.05179734929492071,
"learning_rate": 2.1627173493357167e-05,
"loss": 0.0011,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.000430211890488863,
"step": 870,
"valid_targets_mean": 3984.7,
"valid_targets_min": 1525
},
{
"epoch": 2.652834008097166,
"grad_norm": 0.05164540146149995,
"learning_rate": 2.1416232662785084e-05,
"loss": 0.0011,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00040516024455428123,
"step": 875,
"valid_targets_mean": 3974.6,
"valid_targets_min": 1959
},
{
"epoch": 2.668016194331984,
"grad_norm": 0.050801654869212585,
"learning_rate": 2.1205133373033173e-05,
"loss": 0.0011,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00018009051564149559,
"step": 880,
"valid_targets_mean": 3982.6,
"valid_targets_min": 1719
},
{
"epoch": 2.6831983805668016,
"grad_norm": 0.054723217913153176,
"learning_rate": 2.0993899243539953e-05,
"loss": 0.0011,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00042076894897036254,
"step": 885,
"valid_targets_mean": 2904.8,
"valid_targets_min": 1829
},
{
"epoch": 2.698380566801619,
"grad_norm": 0.042261330830379425,
"learning_rate": 2.0782553908830887e-05,
"loss": 0.001,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0002117999829351902,
"step": 890,
"valid_targets_mean": 3374.6,
"valid_targets_min": 1562
},
{
"epoch": 2.7135627530364372,
"grad_norm": 0.06376816464192013,
"learning_rate": 2.0571121015873924e-05,
"loss": 0.001,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00032623833976686,
"step": 895,
"valid_targets_mean": 3469.3,
"valid_targets_min": 1716
},
{
"epoch": 2.7287449392712553,
"grad_norm": 0.0475123913429236,
"learning_rate": 2.0359624221433728e-05,
"loss": 0.001,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00030931096989661455,
"step": 900,
"valid_targets_mean": 3838.9,
"valid_targets_min": 1928
},
{
"epoch": 2.743927125506073,
"grad_norm": 0.045018407359695714,
"learning_rate": 2.014808718942476e-05,
"loss": 0.0009,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0002605599584057927,
"step": 905,
"valid_targets_mean": 3641.1,
"valid_targets_min": 1601
},
{
"epoch": 2.7591093117408905,
"grad_norm": 0.05851114372504928,
"learning_rate": 1.9936533588263557e-05,
"loss": 0.001,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0003606117097660899,
"step": 910,
"valid_targets_mean": 3916.1,
"valid_targets_min": 2119
},
{
"epoch": 2.7742914979757085,
"grad_norm": 0.051074888944313154,
"learning_rate": 1.9724987088220565e-05,
"loss": 0.0009,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.000338810175890103,
"step": 915,
"valid_targets_mean": 3880.7,
"valid_targets_min": 1909
},
{
"epoch": 2.7894736842105265,
"grad_norm": 0.04406347960449992,
"learning_rate": 1.951347135877169e-05,
"loss": 0.001,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00024635277804918587,
"step": 920,
"valid_targets_mean": 3948.9,
"valid_targets_min": 1764
},
{
"epoch": 2.804655870445344,
"grad_norm": 0.05170895282858387,
"learning_rate": 1.930201006594999e-05,
"loss": 0.0008,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0003006479237228632,
"step": 925,
"valid_targets_mean": 4018.8,
"valid_targets_min": 1939
},
{
"epoch": 2.8198380566801617,
"grad_norm": 0.05037854072436599,
"learning_rate": 1.9090626869697714e-05,
"loss": 0.0009,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00037075512227602303,
"step": 930,
"valid_targets_mean": 3843.9,
"valid_targets_min": 1970
},
{
"epoch": 2.83502024291498,
"grad_norm": 0.046536741186148624,
"learning_rate": 1.8879345421219063e-05,
"loss": 0.0008,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00021453885710798204,
"step": 935,
"valid_targets_mean": 3863.4,
"valid_targets_min": 1802
},
{
"epoch": 2.850202429149798,
"grad_norm": 0.04254469723076537,
"learning_rate": 1.8668189360333923e-05,
"loss": 0.0008,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00018978436128236353,
"step": 940,
"valid_targets_mean": 3717.2,
"valid_targets_min": 1820
},
{
"epoch": 2.8653846153846154,
"grad_norm": 0.06097694340781264,
"learning_rate": 1.845718231283281e-05,
"loss": 0.0008,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0002881702675949782,
"step": 945,
"valid_targets_mean": 3184.4,
"valid_targets_min": 1933
},
{
"epoch": 2.880566801619433,
"grad_norm": 0.051752449780132236,
"learning_rate": 1.8246347887833457e-05,
"loss": 0.0008,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00018924723553936929,
"step": 950,
"valid_targets_mean": 3851.3,
"valid_targets_min": 1857
},
{
"epoch": 2.895748987854251,
"grad_norm": 0.05003241285691749,
"learning_rate": 1.8035709675139258e-05,
"loss": 0.0008,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0002978502307087183,
"step": 955,
"valid_targets_mean": 2977.5,
"valid_targets_min": 1930
},
{
"epoch": 2.910931174089069,
"grad_norm": 0.06589700174277081,
"learning_rate": 1.7825291242599837e-05,
"loss": 0.0009,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00048012990737333894,
"step": 960,
"valid_targets_mean": 3793.8,
"valid_targets_min": 2108
},
{
"epoch": 2.9261133603238867,
"grad_norm": 0.03912630718400104,
"learning_rate": 1.7615116133474084e-05,
"loss": 0.0008,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00016061097267083824,
"step": 965,
"valid_targets_mean": 3659.3,
"valid_targets_min": 2040
},
{
"epoch": 2.9412955465587043,
"grad_norm": 0.04197034155046722,
"learning_rate": 1.7405207863795966e-05,
"loss": 0.0007,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00021663053485099226,
"step": 970,
"valid_targets_mean": 4265.0,
"valid_targets_min": 2112
},
{
"epoch": 2.9564777327935223,
"grad_norm": 0.045961798865801484,
"learning_rate": 1.719558991974339e-05,
"loss": 0.0007,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00019311830692458898,
"step": 975,
"valid_targets_mean": 3668.4,
"valid_targets_min": 1908
},
{
"epoch": 2.97165991902834,
"grad_norm": 0.04185916928496241,
"learning_rate": 1.698628575501034e-05,
"loss": 0.0007,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0001580404059495777,
"step": 980,
"valid_targets_mean": 3805.8,
"valid_targets_min": 1982
},
{
"epoch": 2.986842105263158,
"grad_norm": 0.058233491725864076,
"learning_rate": 1.6777318788182723e-05,
"loss": 0.0008,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0002382812963332981,
"step": 985,
"valid_targets_mean": 3765.7,
"valid_targets_min": 1993
},
{
"epoch": 3.0,
"grad_norm": 0.1297404243611315,
"learning_rate": 1.6568712400118102e-05,
"loss": 0.0007,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0008972192881628871,
"step": 990,
"valid_targets_mean": 3093.5,
"valid_targets_min": 1601
},
{
"epoch": 3.0151821862348176,
"grad_norm": 0.04791864517076802,
"learning_rate": 1.636048993132969e-05,
"loss": 0.0006,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0001960406661964953,
"step": 995,
"valid_targets_mean": 3804.4,
"valid_targets_min": 1995
},
{
"epoch": 3.0303643724696356,
"grad_norm": 0.040441060760252036,
"learning_rate": 1.615267467937479e-05,
"loss": 0.0006,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0002337736077606678,
"step": 1000,
"valid_targets_mean": 3500.3,
"valid_targets_min": 1993
},
{
"epoch": 3.0455465587044532,
"grad_norm": 0.03700089711220268,
"learning_rate": 1.59452898962481e-05,
"loss": 0.0007,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00014921928232070059,
"step": 1005,
"valid_targets_mean": 4355.8,
"valid_targets_min": 1719
},
{
"epoch": 3.0607287449392713,
"grad_norm": 0.03728508054263674,
"learning_rate": 1.573835878578013e-05,
"loss": 0.0006,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0002580939035397023,
"step": 1010,
"valid_targets_mean": 3384.2,
"valid_targets_min": 1836
},
{
"epoch": 3.075910931174089,
"grad_norm": 0.033391122228532424,
"learning_rate": 1.5531904501040917e-05,
"loss": 0.0007,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00012008862540824339,
"step": 1015,
"valid_targets_mean": 3431.9,
"valid_targets_min": 2012
},
{
"epoch": 3.091093117408907,
"grad_norm": 0.04681189312643591,
"learning_rate": 1.5325950141749522e-05,
"loss": 0.0006,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0001886932586785406,
"step": 1020,
"valid_targets_mean": 2928.0,
"valid_targets_min": 1775
},
{
"epoch": 3.1062753036437245,
"grad_norm": 0.044443816642985745,
"learning_rate": 1.5120518751689438e-05,
"loss": 0.0007,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00016409529780503362,
"step": 1025,
"valid_targets_mean": 3362.0,
"valid_targets_min": 1925
},
{
"epoch": 3.1214574898785425,
"grad_norm": 0.043336428093634194,
"learning_rate": 1.4915633316130267e-05,
"loss": 0.0005,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00020097648666705936,
"step": 1030,
"valid_targets_mean": 4062.7,
"valid_targets_min": 1995
},
{
"epoch": 3.13663967611336,
"grad_norm": 0.0459329657416429,
"learning_rate": 1.4711316759255963e-05,
"loss": 0.0006,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0001739387516863644,
"step": 1035,
"valid_targets_mean": 3088.9,
"valid_targets_min": 1525
},
{
"epoch": 3.151821862348178,
"grad_norm": 0.03722667539494412,
"learning_rate": 1.450759194159987e-05,
"loss": 0.0006,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00026353122666478157,
"step": 1040,
"valid_targets_mean": 3796.6,
"valid_targets_min": 1562
},
{
"epoch": 3.167004048582996,
"grad_norm": 0.04015879003618351,
"learning_rate": 1.4304481657486955e-05,
"loss": 0.0005,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0002373484312556684,
"step": 1045,
"valid_targets_mean": 3303.8,
"valid_targets_min": 1913
},
{
"epoch": 3.182186234817814,
"grad_norm": 0.038621834438852996,
"learning_rate": 1.4102008632483344e-05,
"loss": 0.0005,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00011154404637636617,
"step": 1050,
"valid_targets_mean": 4531.4,
"valid_targets_min": 2143
},
{
"epoch": 3.1973684210526314,
"grad_norm": 0.035586402549734227,
"learning_rate": 1.3900195520853628e-05,
"loss": 0.0006,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00015819785767234862,
"step": 1055,
"valid_targets_mean": 3401.1,
"valid_targets_min": 1907
},
{
"epoch": 3.2125506072874495,
"grad_norm": 0.037916654314973396,
"learning_rate": 1.3699064903026149e-05,
"loss": 0.0005,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00013206344738136977,
"step": 1060,
"valid_targets_mean": 4700.9,
"valid_targets_min": 1884
},
{
"epoch": 3.227732793522267,
"grad_norm": 0.04122170067545708,
"learning_rate": 1.34986392830665e-05,
"loss": 0.0006,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0002938436227850616,
"step": 1065,
"valid_targets_mean": 3508.2,
"valid_targets_min": 1709
},
{
"epoch": 3.242914979757085,
"grad_norm": 0.04402273062092337,
"learning_rate": 1.3298941086159598e-05,
"loss": 0.0005,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00011752369755413383,
"step": 1070,
"valid_targets_mean": 2727.9,
"valid_targets_min": 1833
},
{
"epoch": 3.2580971659919027,
"grad_norm": 0.028137948471880926,
"learning_rate": 1.3099992656100592e-05,
"loss": 0.0004,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00014230958186089993,
"step": 1075,
"valid_targets_mean": 3507.3,
"valid_targets_min": 1891
},
{
"epoch": 3.2732793522267207,
"grad_norm": 0.03254307517941282,
"learning_rate": 1.2901816252794848e-05,
"loss": 0.0006,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00016944081289693713,
"step": 1080,
"valid_targets_mean": 4335.8,
"valid_targets_min": 1908
},
{
"epoch": 3.2884615384615383,
"grad_norm": 0.02929668975309448,
"learning_rate": 1.2704434049767356e-05,
"loss": 0.0004,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00026809610426425934,
"step": 1085,
"valid_targets_mean": 4831.5,
"valid_targets_min": 1863
},
{
"epoch": 3.3036437246963564,
"grad_norm": 0.03393895091615616,
"learning_rate": 1.250786813168176e-05,
"loss": 0.0005,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00018053094390779734,
"step": 1090,
"valid_targets_mean": 3571.0,
"valid_targets_min": 1525
},
{
"epoch": 3.318825910931174,
"grad_norm": 0.028197205291907283,
"learning_rate": 1.2312140491869369e-05,
"loss": 0.0004,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0001489449932705611,
"step": 1095,
"valid_targets_mean": 4507.4,
"valid_targets_min": 1942
},
{
"epoch": 3.334008097165992,
"grad_norm": 0.02836106119741283,
"learning_rate": 1.2117273029868362e-05,
"loss": 0.0005,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00014065299183130264,
"step": 1100,
"valid_targets_mean": 3763.5,
"valid_targets_min": 1943
},
{
"epoch": 3.3491902834008096,
"grad_norm": 0.02528925372442717,
"learning_rate": 1.1923287548973508e-05,
"loss": 0.0005,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0001509625290054828,
"step": 1105,
"valid_targets_mean": 4109.8,
"valid_targets_min": 1972
},
{
"epoch": 3.3643724696356276,
"grad_norm": 0.03166602555558615,
"learning_rate": 1.1730205753796631e-05,
"loss": 0.0004,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00012184908700874075,
"step": 1110,
"valid_targets_mean": 3165.5,
"valid_targets_min": 1908
},
{
"epoch": 3.3795546558704452,
"grad_norm": 0.028549342217324303,
"learning_rate": 1.1538049247838128e-05,
"loss": 0.0004,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0001959708024514839,
"step": 1115,
"valid_targets_mean": 4969.4,
"valid_targets_min": 1914
},
{
"epoch": 3.3947368421052633,
"grad_norm": 0.030028711778133543,
"learning_rate": 1.134683953106983e-05,
"loss": 0.0003,
"loss_nan_ranks": 0,
"loss_rank_avg": 9.96004237094894e-05,
"step": 1120,
"valid_targets_mean": 4512.2,
"valid_targets_min": 1863
},
{
"epoch": 3.409919028340081,
"grad_norm": 0.04370869946081909,
"learning_rate": 1.115659799752938e-05,
"loss": 0.0005,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.000138857401907444,
"step": 1125,
"valid_targets_mean": 3083.2,
"valid_targets_min": 1960
},
{
"epoch": 3.425101214574899,
"grad_norm": 0.030000789299433693,
"learning_rate": 1.096734593292649e-05,
"loss": 0.0004,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0001483315136283636,
"step": 1130,
"valid_targets_mean": 3425.8,
"valid_targets_min": 1982
},
{
"epoch": 3.4402834008097165,
"grad_norm": 0.028224513209535406,
"learning_rate": 1.077910451226138e-05,
"loss": 0.0004,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00014747094246558845,
"step": 1135,
"valid_targets_mean": 3538.1,
"valid_targets_min": 1877
},
{
"epoch": 3.4554655870445345,
"grad_norm": 0.022426690960121926,
"learning_rate": 1.0591894797455526e-05,
"loss": 0.0004,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00011917293886654079,
"step": 1140,
"valid_targets_mean": 4210.4,
"valid_targets_min": 2102
},
{
"epoch": 3.470647773279352,
"grad_norm": 0.06303109109212526,
"learning_rate": 1.0405737734995083e-05,
"loss": 0.0005,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00022693569189868867,
"step": 1145,
"valid_targets_mean": 2992.8,
"valid_targets_min": 1921
},
{
"epoch": 3.48582995951417,
"grad_norm": 0.028066996750550254,
"learning_rate": 1.0220654153587225e-05,
"loss": 0.0003,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00015031444490887225,
"step": 1150,
"valid_targets_mean": 5191.8,
"valid_targets_min": 1929
},
{
"epoch": 3.501012145748988,
"grad_norm": 0.024160611003187605,
"learning_rate": 1.00366647618297e-05,
"loss": 0.0004,
"loss_nan_ranks": 0,
"loss_rank_avg": 5.5384705774486065e-05,
"step": 1155,
"valid_targets_mean": 5001.4,
"valid_targets_min": 2055
},
{
"epoch": 3.516194331983806,
"grad_norm": 0.03364398807924398,
"learning_rate": 9.853790145893742e-06,
"loss": 0.0004,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00012208351108711213,
"step": 1160,
"valid_targets_mean": 3204.1,
"valid_targets_min": 1966
},
{
"epoch": 3.5313765182186234,
"grad_norm": 0.03295170759689548,
"learning_rate": 9.672050767220765e-06,
"loss": 0.0003,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0001380308240186423,
"step": 1165,
"valid_targets_mean": 2981.7,
"valid_targets_min": 1900
},
{
"epoch": 3.5465587044534415,
"grad_norm": 0.030268626306576136,
"learning_rate": 9.491466960232955e-06,
"loss": 0.0004,
"loss_nan_ranks": 0,
"loss_rank_avg": 9.681130177341402e-05,
"step": 1170,
"valid_targets_mean": 4048.1,
"valid_targets_min": 1972
},
{
"epoch": 3.561740890688259,
"grad_norm": 0.03608657038512924,
"learning_rate": 9.312058930058114e-06,
"loss": 0.0003,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00018855561211239547,
"step": 1175,
"valid_targets_mean": 3406.9,
"valid_targets_min": 1361
},
{
"epoch": 3.5769230769230766,
"grad_norm": 0.022869933499966452,
"learning_rate": 9.133846750268945e-06,
"loss": 0.0003,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00011784023081418127,
"step": 1180,
"valid_targets_mean": 3454.8,
"valid_targets_min": 1863
},
{
"epoch": 3.5921052631578947,
"grad_norm": 0.028923130050734012,
"learning_rate": 8.956850360637046e-06,
"loss": 0.0003,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00016331742517650127,
"step": 1185,
"valid_targets_mean": 3790.9,
"valid_targets_min": 1889
},
{
"epoch": 3.6072874493927127,
"grad_norm": 0.02605894010229494,
"learning_rate": 8.78108956490194e-06,
"loss": 0.0003,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00012114306446164846,
"step": 1190,
"valid_targets_mean": 3362.1,
"valid_targets_min": 1900
},
{
"epoch": 3.6224696356275303,
"grad_norm": 0.023018268228659888,
"learning_rate": 8.606584028555225e-06,
"loss": 0.0003,
"loss_nan_ranks": 0,
"loss_rank_avg": 7.702252332819626e-05,
"step": 1195,
"valid_targets_mean": 3200.9,
"valid_targets_min": 1933
},
{
"epoch": 3.637651821862348,
"grad_norm": 0.020059240240159412,
"learning_rate": 8.43335327664027e-06,
"loss": 0.0003,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00010392841068096459,
"step": 1200,
"valid_targets_mean": 3440.0,
"valid_targets_min": 2128
},
{
"epoch": 3.652834008097166,
"grad_norm": 0.02131452772328446,
"learning_rate": 8.261416691567601e-06,
"loss": 0.0003,
"loss_nan_ranks": 0,
"loss_rank_avg": 8.232158870669082e-05,
"step": 1205,
"valid_targets_mean": 3654.4,
"valid_targets_min": 1943
},
{
"epoch": 3.668016194331984,
"grad_norm": 0.019373195524993995,
"learning_rate": 8.090793510946242e-06,
"loss": 0.0003,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00010396288416814059,
"step": 1210,
"valid_targets_mean": 3880.0,
"valid_targets_min": 1771
},
{
"epoch": 3.6831983805668016,
"grad_norm": 0.02917989499051033,
"learning_rate": 7.921502825431258e-06,
"loss": 0.0003,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0001521241501905024,
"step": 1215,
"valid_targets_mean": 4569.3,
"valid_targets_min": 1680
},
{
"epoch": 3.698380566801619,
"grad_norm": 0.024589760224471912,
"learning_rate": 7.753563576587753e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 8.788294508121908e-05,
"step": 1220,
"valid_targets_mean": 4094.7,
"valid_targets_min": 1987
},
{
"epoch": 3.7135627530364372,
"grad_norm": 0.017200210728940582,
"learning_rate": 7.5869945547715275e-06,
"loss": 0.0003,
"loss_nan_ranks": 0,
"loss_rank_avg": 7.513599848607555e-05,
"step": 1225,
"valid_targets_mean": 4293.6,
"valid_targets_min": 1820
},
{
"epoch": 3.7287449392712553,
"grad_norm": 0.017930698287844393,
"learning_rate": 7.421814397026674e-06,
"loss": 0.0003,
"loss_nan_ranks": 0,
"loss_rank_avg": 8.184764737961814e-05,
"step": 1230,
"valid_targets_mean": 3846.2,
"valid_targets_min": 1951
},
{
"epoch": 3.743927125506073,
"grad_norm": 0.02890630560957326,
"learning_rate": 7.258041585000317e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 9.473184763919562e-05,
"step": 1235,
"valid_targets_mean": 3583.5,
"valid_targets_min": 1847
},
{
"epoch": 3.7591093117408905,
"grad_norm": 0.01740832457431731,
"learning_rate": 7.095694442874743e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 8.939133840613067e-05,
"step": 1240,
"valid_targets_mean": 3513.1,
"valid_targets_min": 1740
},
{
"epoch": 3.7742914979757085,
"grad_norm": 0.02101575750087779,
"learning_rate": 6.934791135317147e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00012081398745067418,
"step": 1245,
"valid_targets_mean": 4042.1,
"valid_targets_min": 1361
},
{
"epoch": 3.7894736842105265,
"grad_norm": 0.0328870324904059,
"learning_rate": 6.775349665447222e-06,
"loss": 0.0003,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00016577565111219883,
"step": 1250,
"valid_targets_mean": 3242.9,
"valid_targets_min": 1562
},
{
"epoch": 3.804655870445344,
"grad_norm": 0.017947798543351427,
"learning_rate": 6.617387872822842e-06,
"loss": 0.0003,
"loss_nan_ranks": 0,
"loss_rank_avg": 4.809114761883393e-05,
"step": 1255,
"valid_targets_mean": 4447.4,
"valid_targets_min": 1952
},
{
"epoch": 3.8198380566801617,
"grad_norm": 0.023614675723710878,
"learning_rate": 6.460923431444015e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 5.9992085880367085e-05,
"step": 1260,
"valid_targets_mean": 4193.2,
"valid_targets_min": 1974
},
{
"epoch": 3.83502024291498,
"grad_norm": 0.013798886144374323,
"learning_rate": 6.305973847775406e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 7.058361370582134e-05,
"step": 1265,
"valid_targets_mean": 4812.4,
"valid_targets_min": 1601
},
{
"epoch": 3.850202429149798,
"grad_norm": 0.019822363797309384,
"learning_rate": 6.152556458787546e-06,
"loss": 0.0003,
"loss_nan_ranks": 0,
"loss_rank_avg": 9.843902080319822e-05,
"step": 1270,
"valid_targets_mean": 3342.8,
"valid_targets_min": 1802
},
{
"epoch": 3.8653846153846154,
"grad_norm": 0.008453341745438561,
"learning_rate": 6.000688430017048e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 6.605287489946932e-05,
"step": 1275,
"valid_targets_mean": 3518.0,
"valid_targets_min": 1906
},
{
"epoch": 3.880566801619433,
"grad_norm": 0.013488398196674056,
"learning_rate": 5.850386753645998e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 7.022082718322054e-05,
"step": 1280,
"valid_targets_mean": 3390.8,
"valid_targets_min": 1919
},
{
"epoch": 3.895748987854251,
"grad_norm": 0.013266846971383653,
"learning_rate": 5.701668246600731e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 5.2952855185139924e-05,
"step": 1285,
"valid_targets_mean": 3293.5,
"valid_targets_min": 1601
},
{
"epoch": 3.910931174089069,
"grad_norm": 0.016227940337685064,
"learning_rate": 5.554549548670227e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 5.36451734660659e-05,
"step": 1290,
"valid_targets_mean": 4061.6,
"valid_targets_min": 1930
},
{
"epoch": 3.9261133603238867,
"grad_norm": 0.014053296860649007,
"learning_rate": 5.409047120644307e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00010678240505512804,
"step": 1295,
"valid_targets_mean": 2907.3,
"valid_targets_min": 2055
},
{
"epoch": 3.9412955465587043,
"grad_norm": 0.007549038743560788,
"learning_rate": 5.265177242471899e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 7.16036738594994e-05,
"step": 1300,
"valid_targets_mean": 3240.8,
"valid_targets_min": 2010
},
{
"epoch": 3.9564777327935223,
"grad_norm": 0.013485563062924482,
"learning_rate": 5.122956011439486e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 6.303354166448116e-05,
"step": 1305,
"valid_targets_mean": 3311.7,
"valid_targets_min": 1993
},
{
"epoch": 3.97165991902834,
"grad_norm": 0.010379116169811794,
"learning_rate": 4.982399340370017e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 4.860375338466838e-05,
"step": 1310,
"valid_targets_mean": 3747.0,
"valid_targets_min": 2000
},
{
"epoch": 3.986842105263158,
"grad_norm": 0.021410637385460316,
"learning_rate": 4.843522955842464e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 7.024600927252322e-05,
"step": 1315,
"valid_targets_mean": 3631.0,
"valid_targets_min": 1884
},
{
"epoch": 4.0,
"grad_norm": 0.01363656750259932,
"learning_rate": 4.706342396432213e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00019095477182418108,
"step": 1320,
"valid_targets_mean": 3038.1,
"valid_targets_min": 1836
},
{
"epoch": 4.015182186234818,
"grad_norm": 0.022982623600982004,
"learning_rate": 4.570873010972477e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 6.46241387585178e-05,
"step": 1325,
"valid_targets_mean": 3446.2,
"valid_targets_min": 1970
},
{
"epoch": 4.030364372469635,
"grad_norm": 0.011251477178822506,
"learning_rate": 4.43712995683695e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 5.4794454626971856e-05,
"step": 1330,
"valid_targets_mean": 6213.4,
"valid_targets_min": 2161
},
{
"epoch": 4.045546558704453,
"grad_norm": 0.007165946969264405,
"learning_rate": 4.305128198243888e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 7.171961624408141e-05,
"step": 1335,
"valid_targets_mean": 3990.1,
"valid_targets_min": 1917
},
{
"epoch": 4.060728744939271,
"grad_norm": 0.007957987522309468,
"learning_rate": 4.174882504581794e-06,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.8172773201949894e-05,
"step": 1340,
"valid_targets_mean": 3745.6,
"valid_targets_min": 1942
},
{
"epoch": 4.075910931174089,
"grad_norm": 0.009078057067041805,
"learning_rate": 4.046407448756895e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 6.138246681075543e-05,
"step": 1345,
"valid_targets_mean": 4506.6,
"valid_targets_min": 1998
},
{
"epoch": 4.0910931174089065,
"grad_norm": 0.010257378819494439,
"learning_rate": 3.91971740556262e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 5.178306673769839e-05,
"step": 1350,
"valid_targets_mean": 4066.5,
"valid_targets_min": 1977
},
{
"epoch": 4.1062753036437245,
"grad_norm": 0.011789041059906483,
"learning_rate": 3.7948265500712313e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.0001629288017284125,
"step": 1355,
"valid_targets_mean": 4395.5,
"valid_targets_min": 1763
},
{
"epoch": 4.1214574898785425,
"grad_norm": 0.004437573848554222,
"learning_rate": 3.6717488560478096e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 6.353334174491465e-05,
"step": 1360,
"valid_targets_mean": 3288.6,
"valid_targets_min": 1709
},
{
"epoch": 4.136639676113361,
"grad_norm": 0.008315509009432932,
"learning_rate": 3.5504980943867538e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 4.049144627060741e-05,
"step": 1365,
"valid_targets_mean": 4831.8,
"valid_targets_min": 1921
},
{
"epoch": 4.151821862348178,
"grad_norm": 0.015457446077583531,
"learning_rate": 3.4310878315710074e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 6.23580053797923e-05,
"step": 1370,
"valid_targets_mean": 2824.3,
"valid_targets_min": 1760
},
{
"epoch": 4.167004048582996,
"grad_norm": 0.00807850067261138,
"learning_rate": 3.3135314281540954e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 5.613297616946511e-05,
"step": 1375,
"valid_targets_mean": 4037.2,
"valid_targets_min": 1909
},
{
"epoch": 4.182186234817814,
"grad_norm": 0.017099145172780647,
"learning_rate": 3.1978420372652776e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 4.21667646151036e-05,
"step": 1380,
"valid_targets_mean": 3238.4,
"valid_targets_min": 2063
},
{
"epoch": 4.197368421052632,
"grad_norm": 0.010828215109866889,
"learning_rate": 3.084032603137852e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 7.516544428654015e-05,
"step": 1385,
"valid_targets_mean": 3698.3,
"valid_targets_min": 2143
},
{
"epoch": 4.212550607287449,
"grad_norm": 0.004016329731593419,
"learning_rate": 2.9721158596608622e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 4.366480061435141e-05,
"step": 1390,
"valid_targets_mean": 3935.9,
"valid_targets_min": 1863
},
{
"epoch": 4.227732793522267,
"grad_norm": 0.007907605733652394,
"learning_rate": 2.8621043289543314e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 6.708363071084023e-05,
"step": 1395,
"valid_targets_mean": 4472.4,
"valid_targets_min": 1517
},
{
"epoch": 4.242914979757085,
"grad_norm": 0.007905154034223622,
"learning_rate": 2.754010319968181e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 5.4375639592763036e-05,
"step": 1400,
"valid_targets_mean": 3555.1,
"valid_targets_min": 1970
},
{
"epoch": 4.258097165991903,
"grad_norm": 0.0035784372258035524,
"learning_rate": 2.647845927105015e-06,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.880778967868537e-05,
"step": 1405,
"valid_targets_mean": 3965.3,
"valid_targets_min": 1799
},
{
"epoch": 4.27327935222672,
"grad_norm": 0.0032621289678014137,
"learning_rate": 2.543623028866915e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 4.033786535728723e-05,
"step": 1410,
"valid_targets_mean": 4653.7,
"valid_targets_min": 1847
},
{
"epoch": 4.288461538461538,
"grad_norm": 0.008056661917228328,
"learning_rate": 2.4413532865263533e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 8.257682202383876e-05,
"step": 1415,
"valid_targets_mean": 4110.0,
"valid_targets_min": 2072
},
{
"epoch": 4.303643724696356,
"grad_norm": 0.007271066373413746,
"learning_rate": 2.3410481428214602e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 6.296165520325303e-05,
"step": 1420,
"valid_targets_mean": 2792.3,
"valid_targets_min": 1933
},
{
"epoch": 4.318825910931174,
"grad_norm": 0.006110931303422554,
"learning_rate": 2.242718820675718e-06,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 4.118061769986525e-05,
"step": 1425,
"valid_targets_mean": 3969.2,
"valid_targets_min": 1919
},
{
"epoch": 4.334008097165992,
"grad_norm": 0.003681031278942579,
"learning_rate": 2.1463763219422495e-06,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 5.706408410333097e-05,
"step": 1430,
"valid_targets_mean": 3094.1,
"valid_targets_min": 1921
},
{
"epoch": 4.34919028340081,
"grad_norm": 0.0030366716957783745,
"learning_rate": 2.0520314261728357e-06,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.24454449582845e-05,
"step": 1435,
"valid_targets_mean": 4022.7,
"valid_targets_min": 1949
},
{
"epoch": 4.364372469635628,
"grad_norm": 0.010468443300618788,
"learning_rate": 1.9596946894118306e-06,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.8533806218765676e-05,
"step": 1440,
"valid_targets_mean": 3913.8,
"valid_targets_min": 1948
},
{
"epoch": 4.379554655870446,
"grad_norm": 0.003850036234542222,
"learning_rate": 1.8693764430150696e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.789378752117045e-05,
"step": 1445,
"valid_targets_mean": 5511.4,
"valid_targets_min": 2124
},
{
"epoch": 4.394736842105263,
"grad_norm": 0.004034126824294064,
"learning_rate": 1.7810867924938978e-06,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 6.700120866298676e-05,
"step": 1450,
"valid_targets_mean": 3396.1,
"valid_targets_min": 1981
},
{
"epoch": 4.409919028340081,
"grad_norm": 0.003021561239286034,
"learning_rate": 1.6948356163845048e-06,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 5.223594780545682e-05,
"step": 1455,
"valid_targets_mean": 4130.1,
"valid_targets_min": 1614
},
{
"epoch": 4.425101214574899,
"grad_norm": 0.003789678583252369,
"learning_rate": 1.610632565142627e-06,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 4.0959115722216666e-05,
"step": 1460,
"valid_targets_mean": 3068.0,
"valid_targets_min": 1709
},
{
"epoch": 4.440283400809717,
"grad_norm": 0.0028920259481009136,
"learning_rate": 1.5284870600637813e-06,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.99993805331178e-05,
"step": 1465,
"valid_targets_mean": 4319.3,
"valid_targets_min": 1680
},
{
"epoch": 4.455465587044534,
"grad_norm": 0.0028155440983919466,
"learning_rate": 1.4484082922291376e-06,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.746398942894302e-05,
"step": 1470,
"valid_targets_mean": 3233.0,
"valid_targets_min": 1983
},
{
"epoch": 4.470647773279352,
"grad_norm": 0.002966778230136015,
"learning_rate": 1.3704052214771513e-06,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.146575909340754e-05,
"step": 1475,
"valid_targets_mean": 3540.2,
"valid_targets_min": 2012
},
{
"epoch": 4.48582995951417,
"grad_norm": 0.0029634453995010597,
"learning_rate": 1.2944865754010682e-06,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 5.132453588885255e-05,
"step": 1480,
"valid_targets_mean": 3974.7,
"valid_targets_min": 1943
},
{
"epoch": 4.501012145748988,
"grad_norm": 0.0026200053720884455,
"learning_rate": 1.2206608483724013e-06,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.260865923948586e-05,
"step": 1485,
"valid_targets_mean": 4719.4,
"valid_targets_min": 1517
},
{
"epoch": 4.516194331983805,
"grad_norm": 0.0033240782127571516,
"learning_rate": 1.1489363005905241e-06,
"loss": 0.0002,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.350360202603042e-05,
"step": 1490,
"valid_targets_mean": 4807.5,
"valid_targets_min": 2007
},
{
"epoch": 4.531376518218623,
"grad_norm": 0.0027890978050445336,
"learning_rate": 1.0793209571584562e-06,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 4.99808847962413e-05,
"step": 1495,
"valid_targets_mean": 2937.9,
"valid_targets_min": 1998
},
{
"epoch": 4.5465587044534415,
"grad_norm": 0.003752056507144487,
"learning_rate": 1.0118226071849424e-06,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 4.8986348701873794e-05,
"step": 1500,
"valid_targets_mean": 4077.5,
"valid_targets_min": 1820
},
{
"epoch": 4.5617408906882595,
"grad_norm": 0.0026515347084138004,
"learning_rate": 9.464488029129581e-07,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.3156135032186285e-05,
"step": 1505,
"valid_targets_mean": 3627.2,
"valid_targets_min": 1802
},
{
"epoch": 4.576923076923077,
"grad_norm": 0.002416928744029068,
"learning_rate": 8.832068588746945e-07,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.0431685445364565e-05,
"step": 1510,
"valid_targets_mean": 3156.1,
"valid_targets_min": 1974
},
{
"epoch": 4.592105263157895,
"grad_norm": 0.007540725085597577,
"learning_rate": 8.221038510731704e-07,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.2028783607529476e-05,
"step": 1515,
"valid_targets_mean": 3551.4,
"valid_targets_min": 1662
},
{
"epoch": 4.607287449392713,
"grad_norm": 0.0021090127606232228,
"learning_rate": 7.631466161904821e-07,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 2.609085277072154e-05,
"step": 1520,
"valid_targets_mean": 4936.9,
"valid_targets_min": 2098
},
{
"epoch": 4.62246963562753,
"grad_norm": 0.003584893184950617,
"learning_rate": 7.063417508228876e-07,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 4.792536492459476e-05,
"step": 1525,
"valid_targets_mean": 3021.8,
"valid_targets_min": 1838
},
{
"epoch": 4.637651821862348,
"grad_norm": 0.002614821973292812,
"learning_rate": 6.516956107427241e-07,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 4.284080205252394e-05,
"step": 1530,
"valid_targets_mean": 4926.7,
"valid_targets_min": 1562
},
{
"epoch": 4.652834008097166,
"grad_norm": 0.003217559206306655,
"learning_rate": 5.992143101872638e-07,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 4.4474909373093396e-05,
"step": 1535,
"valid_targets_mean": 4893.9,
"valid_targets_min": 1940
},
{
"epoch": 4.668016194331984,
"grad_norm": 0.003040470808898774,
"learning_rate": 5.489037211746184e-07,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 5.6487737310817465e-05,
"step": 1540,
"valid_targets_mean": 3321.6,
"valid_targets_min": 2062
},
{
"epoch": 4.683198380566802,
"grad_norm": 0.0031116812446011096,
"learning_rate": 5.007694728467228e-07,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 5.5639928177697584e-05,
"step": 1545,
"valid_targets_mean": 4212.1,
"valid_targets_min": 2004
},
{
"epoch": 4.698380566801619,
"grad_norm": 0.003738905720370584,
"learning_rate": 4.548169508395028e-07,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 4.869704571319744e-05,
"step": 1550,
"valid_targets_mean": 3816.8,
"valid_targets_min": 1763
},
{
"epoch": 4.713562753036437,
"grad_norm": 0.0030333477332079033,
"learning_rate": 4.1105129668029595e-07,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 4.843453280045651e-05,
"step": 1555,
"valid_targets_mean": 3726.7,
"valid_targets_min": 1930
},
{
"epoch": 4.728744939271255,
"grad_norm": 0.002526690207219582,
"learning_rate": 3.6947740721257066e-07,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.6313707823865116e-05,
"step": 1560,
"valid_targets_mean": 3447.2,
"valid_targets_min": 1928
},
{
"epoch": 4.743927125506072,
"grad_norm": 0.00289699125410512,
"learning_rate": 3.3009993404802486e-07,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 4.3805579480249435e-05,
"step": 1565,
"valid_targets_mean": 3905.7,
"valid_targets_min": 1833
},
{
"epoch": 4.7591093117408905,
"grad_norm": 0.0023487543714410335,
"learning_rate": 2.929232830461404e-07,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 2.8654414563789032e-05,
"step": 1570,
"valid_targets_mean": 3416.5,
"valid_targets_min": 1775
},
{
"epoch": 4.7742914979757085,
"grad_norm": 0.0021418350241265596,
"learning_rate": 2.579516138212101e-07,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.0178223823895678e-05,
"step": 1575,
"valid_targets_mean": 3723.2,
"valid_targets_min": 1870
},
{
"epoch": 4.7894736842105265,
"grad_norm": 0.00260761157427604,
"learning_rate": 2.2518883927692857e-07,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 4.43345561507158e-05,
"step": 1580,
"valid_targets_mean": 3169.7,
"valid_targets_min": 1980
},
{
"epoch": 4.804655870445345,
"grad_norm": 0.0024886501727788973,
"learning_rate": 1.9463862516859277e-07,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.9659506001044065e-05,
"step": 1585,
"valid_targets_mean": 3693.8,
"valid_targets_min": 1601
},
{
"epoch": 4.819838056680162,
"grad_norm": 0.002691952961837786,
"learning_rate": 1.6630438969294615e-07,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 4.064635868417099e-05,
"step": 1590,
"valid_targets_mean": 4679.5,
"valid_targets_min": 1983
},
{
"epoch": 4.83502024291498,
"grad_norm": 0.00375571542227193,
"learning_rate": 1.4018930310571553e-07,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 5.9733189118560404e-05,
"step": 1595,
"valid_targets_mean": 2894.2,
"valid_targets_min": 1950
},
{
"epoch": 4.850202429149798,
"grad_norm": 0.002831403381709464,
"learning_rate": 1.1629628736690824e-07,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.2585434382781386e-05,
"step": 1600,
"valid_targets_mean": 3363.4,
"valid_targets_min": 1933
},
{
"epoch": 4.865384615384615,
"grad_norm": 0.002612992177486836,
"learning_rate": 9.46280158138757e-08,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 4.9785296141635627e-05,
"step": 1605,
"valid_targets_mean": 3934.4,
"valid_targets_min": 1908
},
{
"epoch": 4.880566801619433,
"grad_norm": 0.0024813973327434376,
"learning_rate": 7.518691286220625e-08,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.368567558936775e-05,
"step": 1610,
"valid_targets_mean": 3558.2,
"valid_targets_min": 1913
},
{
"epoch": 4.895748987854251,
"grad_norm": 0.0033059333107141988,
"learning_rate": 5.797515373445084e-08,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 5.2559720643330365e-05,
"step": 1615,
"valid_targets_mean": 2949.1,
"valid_targets_min": 1820
},
{
"epoch": 4.910931174089069,
"grad_norm": 0.00284366915030718,
"learning_rate": 4.299466421675113e-08,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.843402737402357e-05,
"step": 1620,
"valid_targets_mean": 4162.7,
"valid_targets_min": 1900
},
{
"epoch": 4.926113360323887,
"grad_norm": 0.0021305526376056393,
"learning_rate": 3.0247120443362976e-08,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.438566636759788e-05,
"step": 1625,
"valid_targets_mean": 4390.2,
"valid_targets_min": 1870
},
{
"epoch": 4.941295546558704,
"grad_norm": 0.0022833703559342793,
"learning_rate": 1.973394870912193e-08,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 2.7414684154791757e-05,
"step": 1630,
"valid_targets_mean": 4535.5,
"valid_targets_min": 1950
},
{
"epoch": 4.956477732793522,
"grad_norm": 0.002431360255696445,
"learning_rate": 1.145632530985541e-08,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.88288899557665e-05,
"step": 1635,
"valid_targets_mean": 4676.2,
"valid_targets_min": 2013
},
{
"epoch": 4.97165991902834,
"grad_norm": 0.0031995859805627838,
"learning_rate": 5.415176410765721e-09,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.984987051808275e-05,
"step": 1640,
"valid_targets_mean": 4557.1,
"valid_targets_min": 1680
},
{
"epoch": 4.9868421052631575,
"grad_norm": 0.002354505847718645,
"learning_rate": 1.611177942812958e-09,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 2.9670020012417808e-05,
"step": 1645,
"valid_targets_mean": 4601.4,
"valid_targets_min": 1940
},
{
"epoch": 5.0,
"grad_norm": 0.004185601226855871,
"learning_rate": 4.475552707772224e-11,
"loss": 0.0001,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.00010255213419441134,
"step": 1650,
"valid_targets_mean": 4467.4,
"valid_targets_min": 1993
},
{
"epoch": 5.0,
"step": 1650,
"total_flos": 2.127078594457895e+18,
"train_loss": 0.0,
"train_runtime": 0.8508,
"train_samples_per_second": 185705.011,
"train_steps_per_second": 1939.324
}
],
"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": 2.127078594457895e+18,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}
Reference in New Issue Copy Permalink