2125 lines
59 KiB
JSON
2125 lines
59 KiB
JSON
|
{
|
||
|
|
"best_global_step": null,
|
||
|
|
"best_metric": null,
|
||
|
|
"best_model_checkpoint": null,
|
||
|
|
"epoch": 0.9989484752891693,
|
||
|
|
"eval_steps": 500,
|
||
|
|
"global_step": 950,
|
||
|
|
"is_hyper_param_search": false,
|
||
|
|
"is_local_process_zero": true,
|
||
|
|
"is_world_process_zero": true,
|
||
|
|
"log_history": [
|
||
|
|
{
|
||
|
|
"epoch": 0.005257623554153523,
|
||
|
|
"grad_norm": 10.237513252992448,
|
||
|
|
"learning_rate": 2.4024024024024026e-07,
|
||
|
|
"loss": 0.599,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5694050788879395,
|
||
|
|
"step": 5,
|
||
|
|
"valid_targets_mean": 2187.3,
|
||
|
|
"valid_targets_min": 652
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.010515247108307046,
|
||
|
|
"grad_norm": 10.32396356889483,
|
||
|
|
"learning_rate": 5.405405405405406e-07,
|
||
|
|
"loss": 0.586,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.628214418888092,
|
||
|
|
"step": 10,
|
||
|
|
"valid_targets_mean": 2250.8,
|
||
|
|
"valid_targets_min": 825
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.015772870662460567,
|
||
|
|
"grad_norm": 8.774135740887447,
|
||
|
|
"learning_rate": 8.40840840840841e-07,
|
||
|
|
"loss": 0.6115,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5322482585906982,
|
||
|
|
"step": 15,
|
||
|
|
"valid_targets_mean": 2316.7,
|
||
|
|
"valid_targets_min": 518
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.02103049421661409,
|
||
|
|
"grad_norm": 8.845664788123988,
|
||
|
|
"learning_rate": 1.1411411411411411e-06,
|
||
|
|
"loss": 0.5806,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5724717378616333,
|
||
|
|
"step": 20,
|
||
|
|
"valid_targets_mean": 2757.2,
|
||
|
|
"valid_targets_min": 826
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.026288117770767613,
|
||
|
|
"grad_norm": 6.9709639457139,
|
||
|
|
"learning_rate": 1.4414414414414416e-06,
|
||
|
|
"loss": 0.5691,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5152867436408997,
|
||
|
|
"step": 25,
|
||
|
|
"valid_targets_mean": 2158.8,
|
||
|
|
"valid_targets_min": 655
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.031545741324921134,
|
||
|
|
"grad_norm": 5.485739926893386,
|
||
|
|
"learning_rate": 1.7417417417417418e-06,
|
||
|
|
"loss": 0.5258,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4879477620124817,
|
||
|
|
"step": 30,
|
||
|
|
"valid_targets_mean": 2350.6,
|
||
|
|
"valid_targets_min": 771
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03680336487907466,
|
||
|
|
"grad_norm": 4.926799072164588,
|
||
|
|
"learning_rate": 2.0420420420420424e-06,
|
||
|
|
"loss": 0.5395,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5488409996032715,
|
||
|
|
"step": 35,
|
||
|
|
"valid_targets_mean": 1980.2,
|
||
|
|
"valid_targets_min": 663
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.04206098843322818,
|
||
|
|
"grad_norm": 3.8065581813352596,
|
||
|
|
"learning_rate": 2.3423423423423424e-06,
|
||
|
|
"loss": 0.4583,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.485211580991745,
|
||
|
|
"step": 40,
|
||
|
|
"valid_targets_mean": 2417.9,
|
||
|
|
"valid_targets_min": 750
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.0473186119873817,
|
||
|
|
"grad_norm": 3.0947838801973044,
|
||
|
|
"learning_rate": 2.642642642642643e-06,
|
||
|
|
"loss": 0.4327,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.41910433769226074,
|
||
|
|
"step": 45,
|
||
|
|
"valid_targets_mean": 2325.9,
|
||
|
|
"valid_targets_min": 739
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.052576235541535225,
|
||
|
|
"grad_norm": 2.998980321024414,
|
||
|
|
"learning_rate": 2.942942942942943e-06,
|
||
|
|
"loss": 0.4287,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4194478988647461,
|
||
|
|
"step": 50,
|
||
|
|
"valid_targets_mean": 2370.1,
|
||
|
|
"valid_targets_min": 683
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.05783385909568875,
|
||
|
|
"grad_norm": 1.5704868282174143,
|
||
|
|
"learning_rate": 3.2432432432432437e-06,
|
||
|
|
"loss": 0.375,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3745213747024536,
|
||
|
|
"step": 55,
|
||
|
|
"valid_targets_mean": 2348.0,
|
||
|
|
"valid_targets_min": 789
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06309148264984227,
|
||
|
|
"grad_norm": 1.2290388147514733,
|
||
|
|
"learning_rate": 3.5435435435435437e-06,
|
||
|
|
"loss": 0.3981,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.35347527265548706,
|
||
|
|
"step": 60,
|
||
|
|
"valid_targets_mean": 2349.1,
|
||
|
|
"valid_targets_min": 612
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.0683491062039958,
|
||
|
|
"grad_norm": 0.955138369025997,
|
||
|
|
"learning_rate": 3.843843843843844e-06,
|
||
|
|
"loss": 0.3185,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2992398738861084,
|
||
|
|
"step": 65,
|
||
|
|
"valid_targets_mean": 2483.2,
|
||
|
|
"valid_targets_min": 717
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07360672975814932,
|
||
|
|
"grad_norm": 0.9495894921526505,
|
||
|
|
"learning_rate": 4.1441441441441446e-06,
|
||
|
|
"loss": 0.3338,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28637954592704773,
|
||
|
|
"step": 70,
|
||
|
|
"valid_targets_mean": 2394.4,
|
||
|
|
"valid_targets_min": 652
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07886435331230283,
|
||
|
|
"grad_norm": 0.9301530231463971,
|
||
|
|
"learning_rate": 4.444444444444444e-06,
|
||
|
|
"loss": 0.3322,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.31543201208114624,
|
||
|
|
"step": 75,
|
||
|
|
"valid_targets_mean": 2335.7,
|
||
|
|
"valid_targets_min": 699
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08412197686645637,
|
||
|
|
"grad_norm": 0.8233168540866798,
|
||
|
|
"learning_rate": 4.7447447447447454e-06,
|
||
|
|
"loss": 0.2851,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28866514563560486,
|
||
|
|
"step": 80,
|
||
|
|
"valid_targets_mean": 2259.5,
|
||
|
|
"valid_targets_min": 711
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08937960042060988,
|
||
|
|
"grad_norm": 0.8435413907307819,
|
||
|
|
"learning_rate": 5.045045045045045e-06,
|
||
|
|
"loss": 0.3145,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.32431674003601074,
|
||
|
|
"step": 85,
|
||
|
|
"valid_targets_mean": 2552.5,
|
||
|
|
"valid_targets_min": 559
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.0946372239747634,
|
||
|
|
"grad_norm": 0.8815469704608869,
|
||
|
|
"learning_rate": 5.345345345345346e-06,
|
||
|
|
"loss": 0.2827,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.321563720703125,
|
||
|
|
"step": 90,
|
||
|
|
"valid_targets_mean": 2037.8,
|
||
|
|
"valid_targets_min": 618
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09989484752891693,
|
||
|
|
"grad_norm": 0.7392909357933146,
|
||
|
|
"learning_rate": 5.645645645645647e-06,
|
||
|
|
"loss": 0.2849,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.29773595929145813,
|
||
|
|
"step": 95,
|
||
|
|
"valid_targets_mean": 2283.9,
|
||
|
|
"valid_targets_min": 488
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.10515247108307045,
|
||
|
|
"grad_norm": 0.6695825996100343,
|
||
|
|
"learning_rate": 5.945945945945947e-06,
|
||
|
|
"loss": 0.2892,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25371721386909485,
|
||
|
|
"step": 100,
|
||
|
|
"valid_targets_mean": 2559.4,
|
||
|
|
"valid_targets_min": 788
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11041009463722397,
|
||
|
|
"grad_norm": 0.6633556316344058,
|
||
|
|
"learning_rate": 6.246246246246247e-06,
|
||
|
|
"loss": 0.2689,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22198472917079926,
|
||
|
|
"step": 105,
|
||
|
|
"valid_targets_mean": 2259.7,
|
||
|
|
"valid_targets_min": 615
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1156677181913775,
|
||
|
|
"grad_norm": 0.6784867744444589,
|
||
|
|
"learning_rate": 6.546546546546547e-06,
|
||
|
|
"loss": 0.2648,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2438749521970749,
|
||
|
|
"step": 110,
|
||
|
|
"valid_targets_mean": 2373.6,
|
||
|
|
"valid_targets_min": 646
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12092534174553102,
|
||
|
|
"grad_norm": 0.7632921499769664,
|
||
|
|
"learning_rate": 6.846846846846848e-06,
|
||
|
|
"loss": 0.2692,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26377084851264954,
|
||
|
|
"step": 115,
|
||
|
|
"valid_targets_mean": 1854.9,
|
||
|
|
"valid_targets_min": 697
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12618296529968454,
|
||
|
|
"grad_norm": 0.8842980073128638,
|
||
|
|
"learning_rate": 7.147147147147148e-06,
|
||
|
|
"loss": 0.2571,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25346481800079346,
|
||
|
|
"step": 120,
|
||
|
|
"valid_targets_mean": 1748.6,
|
||
|
|
"valid_targets_min": 677
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13144058885383805,
|
||
|
|
"grad_norm": 0.6407129393126573,
|
||
|
|
"learning_rate": 7.447447447447448e-06,
|
||
|
|
"loss": 0.257,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21873721480369568,
|
||
|
|
"step": 125,
|
||
|
|
"valid_targets_mean": 2439.9,
|
||
|
|
"valid_targets_min": 830
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1366982124079916,
|
||
|
|
"grad_norm": 0.7274148050845026,
|
||
|
|
"learning_rate": 7.747747747747749e-06,
|
||
|
|
"loss": 0.2475,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24801094830036163,
|
||
|
|
"step": 130,
|
||
|
|
"valid_targets_mean": 2037.8,
|
||
|
|
"valid_targets_min": 749
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.14195583596214512,
|
||
|
|
"grad_norm": 0.7795039618222616,
|
||
|
|
"learning_rate": 8.048048048048048e-06,
|
||
|
|
"loss": 0.2489,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2689794898033142,
|
||
|
|
"step": 135,
|
||
|
|
"valid_targets_mean": 2109.6,
|
||
|
|
"valid_targets_min": 522
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.14721345951629863,
|
||
|
|
"grad_norm": 0.8549176367033016,
|
||
|
|
"learning_rate": 8.348348348348348e-06,
|
||
|
|
"loss": 0.2358,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23591774702072144,
|
||
|
|
"step": 140,
|
||
|
|
"valid_targets_mean": 2099.1,
|
||
|
|
"valid_targets_min": 741
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15247108307045215,
|
||
|
|
"grad_norm": 0.7848864380128985,
|
||
|
|
"learning_rate": 8.64864864864865e-06,
|
||
|
|
"loss": 0.2455,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2527945637702942,
|
||
|
|
"step": 145,
|
||
|
|
"valid_targets_mean": 2100.0,
|
||
|
|
"valid_targets_min": 881
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15772870662460567,
|
||
|
|
"grad_norm": 0.7935620289825588,
|
||
|
|
"learning_rate": 8.94894894894895e-06,
|
||
|
|
"loss": 0.2453,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2635729908943176,
|
||
|
|
"step": 150,
|
||
|
|
"valid_targets_mean": 1760.9,
|
||
|
|
"valid_targets_min": 658
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16298633017875921,
|
||
|
|
"grad_norm": 0.7180384253902464,
|
||
|
|
"learning_rate": 9.24924924924925e-06,
|
||
|
|
"loss": 0.2449,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2314262092113495,
|
||
|
|
"step": 155,
|
||
|
|
"valid_targets_mean": 2128.6,
|
||
|
|
"valid_targets_min": 589
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16824395373291273,
|
||
|
|
"grad_norm": 0.6179704185250057,
|
||
|
|
"learning_rate": 9.54954954954955e-06,
|
||
|
|
"loss": 0.2217,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2035289704799652,
|
||
|
|
"step": 160,
|
||
|
|
"valid_targets_mean": 2399.4,
|
||
|
|
"valid_targets_min": 458
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17350157728706625,
|
||
|
|
"grad_norm": 0.6996448845379029,
|
||
|
|
"learning_rate": 9.849849849849851e-06,
|
||
|
|
"loss": 0.2478,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26244014501571655,
|
||
|
|
"step": 165,
|
||
|
|
"valid_targets_mean": 2430.4,
|
||
|
|
"valid_targets_min": 772
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17875920084121977,
|
||
|
|
"grad_norm": 0.6325925415766741,
|
||
|
|
"learning_rate": 1.015015015015015e-05,
|
||
|
|
"loss": 0.2135,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1965426355600357,
|
||
|
|
"step": 170,
|
||
|
|
"valid_targets_mean": 2616.8,
|
||
|
|
"valid_targets_min": 762
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18401682439537329,
|
||
|
|
"grad_norm": 0.7251871265641249,
|
||
|
|
"learning_rate": 1.0450450450450452e-05,
|
||
|
|
"loss": 0.2378,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22762617468833923,
|
||
|
|
"step": 175,
|
||
|
|
"valid_targets_mean": 2247.9,
|
||
|
|
"valid_targets_min": 577
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1892744479495268,
|
||
|
|
"grad_norm": 0.7642298318352941,
|
||
|
|
"learning_rate": 1.0750750750750751e-05,
|
||
|
|
"loss": 0.2378,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23049914836883545,
|
||
|
|
"step": 180,
|
||
|
|
"valid_targets_mean": 1950.9,
|
||
|
|
"valid_targets_min": 759
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19453207150368035,
|
||
|
|
"grad_norm": 0.6712837538625741,
|
||
|
|
"learning_rate": 1.1051051051051051e-05,
|
||
|
|
"loss": 0.2155,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19793814420700073,
|
||
|
|
"step": 185,
|
||
|
|
"valid_targets_mean": 2341.1,
|
||
|
|
"valid_targets_min": 802
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19978969505783387,
|
||
|
|
"grad_norm": 0.761223347104956,
|
||
|
|
"learning_rate": 1.1351351351351352e-05,
|
||
|
|
"loss": 0.2257,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23351767659187317,
|
||
|
|
"step": 190,
|
||
|
|
"valid_targets_mean": 1832.2,
|
||
|
|
"valid_targets_min": 516
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.20504731861198738,
|
||
|
|
"grad_norm": 0.6313012086267654,
|
||
|
|
"learning_rate": 1.1651651651651652e-05,
|
||
|
|
"loss": 0.2319,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18675370514392853,
|
||
|
|
"step": 195,
|
||
|
|
"valid_targets_mean": 2573.8,
|
||
|
|
"valid_targets_min": 883
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2103049421661409,
|
||
|
|
"grad_norm": 0.6670267290569365,
|
||
|
|
"learning_rate": 1.1951951951951951e-05,
|
||
|
|
"loss": 0.2016,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22376933693885803,
|
||
|
|
"step": 200,
|
||
|
|
"valid_targets_mean": 2668.7,
|
||
|
|
"valid_targets_min": 804
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.21556256572029442,
|
||
|
|
"grad_norm": 0.6149515856013064,
|
||
|
|
"learning_rate": 1.2252252252252253e-05,
|
||
|
|
"loss": 0.1978,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18526947498321533,
|
||
|
|
"step": 205,
|
||
|
|
"valid_targets_mean": 2980.8,
|
||
|
|
"valid_targets_min": 919
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22082018927444794,
|
||
|
|
"grad_norm": 0.6454445504261381,
|
||
|
|
"learning_rate": 1.2552552552552552e-05,
|
||
|
|
"loss": 0.2081,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19164514541625977,
|
||
|
|
"step": 210,
|
||
|
|
"valid_targets_mean": 2569.8,
|
||
|
|
"valid_targets_min": 898
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22607781282860148,
|
||
|
|
"grad_norm": 0.6824518543715602,
|
||
|
|
"learning_rate": 1.2852852852852854e-05,
|
||
|
|
"loss": 0.211,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23652751743793488,
|
||
|
|
"step": 215,
|
||
|
|
"valid_targets_mean": 2380.8,
|
||
|
|
"valid_targets_min": 673
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.231335436382755,
|
||
|
|
"grad_norm": 0.7156518074476397,
|
||
|
|
"learning_rate": 1.3153153153153155e-05,
|
||
|
|
"loss": 0.2176,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21512779593467712,
|
||
|
|
"step": 220,
|
||
|
|
"valid_targets_mean": 2281.6,
|
||
|
|
"valid_targets_min": 582
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23659305993690852,
|
||
|
|
"grad_norm": 0.5582883876210647,
|
||
|
|
"learning_rate": 1.3453453453453456e-05,
|
||
|
|
"loss": 0.2134,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18910522758960724,
|
||
|
|
"step": 225,
|
||
|
|
"valid_targets_mean": 2919.2,
|
||
|
|
"valid_targets_min": 740
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24185068349106204,
|
||
|
|
"grad_norm": 0.7212159877317216,
|
||
|
|
"learning_rate": 1.3753753753753756e-05,
|
||
|
|
"loss": 0.2278,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24723276495933533,
|
||
|
|
"step": 230,
|
||
|
|
"valid_targets_mean": 2381.1,
|
||
|
|
"valid_targets_min": 734
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24710830704521555,
|
||
|
|
"grad_norm": 0.5793329358961439,
|
||
|
|
"learning_rate": 1.4054054054054055e-05,
|
||
|
|
"loss": 0.2115,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18832668662071228,
|
||
|
|
"step": 235,
|
||
|
|
"valid_targets_mean": 3058.8,
|
||
|
|
"valid_targets_min": 855
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.25236593059936907,
|
||
|
|
"grad_norm": 0.7963974619566586,
|
||
|
|
"learning_rate": 1.4354354354354357e-05,
|
||
|
|
"loss": 0.2185,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2544074058532715,
|
||
|
|
"step": 240,
|
||
|
|
"valid_targets_mean": 1978.3,
|
||
|
|
"valid_targets_min": 782
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2576235541535226,
|
||
|
|
"grad_norm": 0.57197082895393,
|
||
|
|
"learning_rate": 1.4654654654654656e-05,
|
||
|
|
"loss": 0.1943,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18041585385799408,
|
||
|
|
"step": 245,
|
||
|
|
"valid_targets_mean": 2629.1,
|
||
|
|
"valid_targets_min": 529
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2628811777076761,
|
||
|
|
"grad_norm": 0.6742356966988919,
|
||
|
|
"learning_rate": 1.4954954954954957e-05,
|
||
|
|
"loss": 0.213,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21414026618003845,
|
||
|
|
"step": 250,
|
||
|
|
"valid_targets_mean": 2684.2,
|
||
|
|
"valid_targets_min": 596
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.26813880126182965,
|
||
|
|
"grad_norm": 0.6640561192745676,
|
||
|
|
"learning_rate": 1.5255255255255257e-05,
|
||
|
|
"loss": 0.2041,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19483071565628052,
|
||
|
|
"step": 255,
|
||
|
|
"valid_targets_mean": 2610.5,
|
||
|
|
"valid_targets_min": 699
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2733964248159832,
|
||
|
|
"grad_norm": 0.6069260836811652,
|
||
|
|
"learning_rate": 1.555555555555556e-05,
|
||
|
|
"loss": 0.1899,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1980745494365692,
|
||
|
|
"step": 260,
|
||
|
|
"valid_targets_mean": 2579.2,
|
||
|
|
"valid_targets_min": 765
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2786540483701367,
|
||
|
|
"grad_norm": 0.5700078730790441,
|
||
|
|
"learning_rate": 1.5855855855855858e-05,
|
||
|
|
"loss": 0.2159,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19042900204658508,
|
||
|
|
"step": 265,
|
||
|
|
"valid_targets_mean": 2987.2,
|
||
|
|
"valid_targets_min": 1078
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.28391167192429023,
|
||
|
|
"grad_norm": 0.6960137417429887,
|
||
|
|
"learning_rate": 1.6156156156156157e-05,
|
||
|
|
"loss": 0.2031,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1713910698890686,
|
||
|
|
"step": 270,
|
||
|
|
"valid_targets_mean": 2192.1,
|
||
|
|
"valid_targets_min": 871
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2891692954784437,
|
||
|
|
"grad_norm": 0.622428122153235,
|
||
|
|
"learning_rate": 1.6456456456456457e-05,
|
||
|
|
"loss": 0.2005,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20279918611049652,
|
||
|
|
"step": 275,
|
||
|
|
"valid_targets_mean": 2866.9,
|
||
|
|
"valid_targets_min": 538
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.29442691903259727,
|
||
|
|
"grad_norm": 0.7354920044007712,
|
||
|
|
"learning_rate": 1.6756756756756757e-05,
|
||
|
|
"loss": 0.2102,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20906475186347961,
|
||
|
|
"step": 280,
|
||
|
|
"valid_targets_mean": 2221.4,
|
||
|
|
"valid_targets_min": 727
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2996845425867508,
|
||
|
|
"grad_norm": 0.8923865333855707,
|
||
|
|
"learning_rate": 1.705705705705706e-05,
|
||
|
|
"loss": 0.2155,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2942986488342285,
|
||
|
|
"step": 285,
|
||
|
|
"valid_targets_mean": 1926.3,
|
||
|
|
"valid_targets_min": 734
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3049421661409043,
|
||
|
|
"grad_norm": 1.1412843155278614,
|
||
|
|
"learning_rate": 1.735735735735736e-05,
|
||
|
|
"loss": 0.2089,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23976126313209534,
|
||
|
|
"step": 290,
|
||
|
|
"valid_targets_mean": 2329.9,
|
||
|
|
"valid_targets_min": 716
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.31019978969505785,
|
||
|
|
"grad_norm": 0.7251730869917394,
|
||
|
|
"learning_rate": 1.765765765765766e-05,
|
||
|
|
"loss": 0.2142,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21211419999599457,
|
||
|
|
"step": 295,
|
||
|
|
"valid_targets_mean": 2236.8,
|
||
|
|
"valid_targets_min": 333
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.31545741324921134,
|
||
|
|
"grad_norm": 0.6884671855375899,
|
||
|
|
"learning_rate": 1.795795795795796e-05,
|
||
|
|
"loss": 0.1932,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18856525421142578,
|
||
|
|
"step": 300,
|
||
|
|
"valid_targets_mean": 2204.8,
|
||
|
|
"valid_targets_min": 543
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3207150368033649,
|
||
|
|
"grad_norm": 0.6726161803752185,
|
||
|
|
"learning_rate": 1.8258258258258258e-05,
|
||
|
|
"loss": 0.182,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17828920483589172,
|
||
|
|
"step": 305,
|
||
|
|
"valid_targets_mean": 2458.4,
|
||
|
|
"valid_targets_min": 816
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.32597266035751843,
|
||
|
|
"grad_norm": 0.8043219816821761,
|
||
|
|
"learning_rate": 1.855855855855856e-05,
|
||
|
|
"loss": 0.2253,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21263109147548676,
|
||
|
|
"step": 310,
|
||
|
|
"valid_targets_mean": 2375.6,
|
||
|
|
"valid_targets_min": 740
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3312302839116719,
|
||
|
|
"grad_norm": 0.5904188800340215,
|
||
|
|
"learning_rate": 1.885885885885886e-05,
|
||
|
|
"loss": 0.2043,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18243391811847687,
|
||
|
|
"step": 315,
|
||
|
|
"valid_targets_mean": 2669.8,
|
||
|
|
"valid_targets_min": 961
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.33648790746582546,
|
||
|
|
"grad_norm": 0.7354619498470438,
|
||
|
|
"learning_rate": 1.915915915915916e-05,
|
||
|
|
"loss": 0.1983,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20419487357139587,
|
||
|
|
"step": 320,
|
||
|
|
"valid_targets_mean": 2061.4,
|
||
|
|
"valid_targets_min": 633
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.34174553101997895,
|
||
|
|
"grad_norm": 0.811522005805362,
|
||
|
|
"learning_rate": 1.9459459459459463e-05,
|
||
|
|
"loss": 0.1747,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18358461558818817,
|
||
|
|
"step": 325,
|
||
|
|
"valid_targets_mean": 2918.9,
|
||
|
|
"valid_targets_min": 727
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3470031545741325,
|
||
|
|
"grad_norm": 0.749914914215732,
|
||
|
|
"learning_rate": 1.9759759759759763e-05,
|
||
|
|
"loss": 0.2144,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2509661018848419,
|
||
|
|
"step": 330,
|
||
|
|
"valid_targets_mean": 2102.5,
|
||
|
|
"valid_targets_min": 570
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.352260778128286,
|
||
|
|
"grad_norm": 0.6839519896818547,
|
||
|
|
"learning_rate": 2.0060060060060062e-05,
|
||
|
|
"loss": 0.1808,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19201715290546417,
|
||
|
|
"step": 335,
|
||
|
|
"valid_targets_mean": 2207.9,
|
||
|
|
"valid_targets_min": 712
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.35751840168243953,
|
||
|
|
"grad_norm": 0.7651917500393488,
|
||
|
|
"learning_rate": 2.0360360360360362e-05,
|
||
|
|
"loss": 0.1843,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20671632885932922,
|
||
|
|
"step": 340,
|
||
|
|
"valid_targets_mean": 2240.0,
|
||
|
|
"valid_targets_min": 816
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3627760252365931,
|
||
|
|
"grad_norm": 0.6557513064663701,
|
||
|
|
"learning_rate": 2.066066066066066e-05,
|
||
|
|
"loss": 0.1875,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.177345409989357,
|
||
|
|
"step": 345,
|
||
|
|
"valid_targets_mean": 2483.1,
|
||
|
|
"valid_targets_min": 530
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.36803364879074657,
|
||
|
|
"grad_norm": 0.6001314272931987,
|
||
|
|
"learning_rate": 2.0960960960960964e-05,
|
||
|
|
"loss": 0.2087,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17004337906837463,
|
||
|
|
"step": 350,
|
||
|
|
"valid_targets_mean": 2461.9,
|
||
|
|
"valid_targets_min": 704
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3732912723449001,
|
||
|
|
"grad_norm": 0.867896435065951,
|
||
|
|
"learning_rate": 2.1261261261261264e-05,
|
||
|
|
"loss": 0.2088,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17218470573425293,
|
||
|
|
"step": 355,
|
||
|
|
"valid_targets_mean": 1978.3,
|
||
|
|
"valid_targets_min": 662
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3785488958990536,
|
||
|
|
"grad_norm": 0.7925560317512116,
|
||
|
|
"learning_rate": 2.1561561561561564e-05,
|
||
|
|
"loss": 0.2279,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25388026237487793,
|
||
|
|
"step": 360,
|
||
|
|
"valid_targets_mean": 2015.7,
|
||
|
|
"valid_targets_min": 720
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.38380651945320715,
|
||
|
|
"grad_norm": 0.643398855161114,
|
||
|
|
"learning_rate": 2.1861861861861863e-05,
|
||
|
|
"loss": 0.204,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2091820240020752,
|
||
|
|
"step": 365,
|
||
|
|
"valid_targets_mean": 2631.3,
|
||
|
|
"valid_targets_min": 818
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3890641430073607,
|
||
|
|
"grad_norm": 0.7607898637671435,
|
||
|
|
"learning_rate": 2.2162162162162163e-05,
|
||
|
|
"loss": 0.2108,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2094314992427826,
|
||
|
|
"step": 370,
|
||
|
|
"valid_targets_mean": 2089.7,
|
||
|
|
"valid_targets_min": 741
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3943217665615142,
|
||
|
|
"grad_norm": 0.5408012409103932,
|
||
|
|
"learning_rate": 2.2462462462462466e-05,
|
||
|
|
"loss": 0.1805,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13508717715740204,
|
||
|
|
"step": 375,
|
||
|
|
"valid_targets_mean": 2720.9,
|
||
|
|
"valid_targets_min": 485
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.39957939011566773,
|
||
|
|
"grad_norm": 0.6462442411946696,
|
||
|
|
"learning_rate": 2.2762762762762765e-05,
|
||
|
|
"loss": 0.1767,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18138816952705383,
|
||
|
|
"step": 380,
|
||
|
|
"valid_targets_mean": 2664.4,
|
||
|
|
"valid_targets_min": 927
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4048370136698212,
|
||
|
|
"grad_norm": 0.73375016894413,
|
||
|
|
"learning_rate": 2.3063063063063065e-05,
|
||
|
|
"loss": 0.2003,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19467218220233917,
|
||
|
|
"step": 385,
|
||
|
|
"valid_targets_mean": 2010.2,
|
||
|
|
"valid_targets_min": 534
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.41009463722397477,
|
||
|
|
"grad_norm": 0.627096732110938,
|
||
|
|
"learning_rate": 2.3363363363363364e-05,
|
||
|
|
"loss": 0.1945,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.192593514919281,
|
||
|
|
"step": 390,
|
||
|
|
"valid_targets_mean": 2467.2,
|
||
|
|
"valid_targets_min": 847
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4153522607781283,
|
||
|
|
"grad_norm": 0.6374795850905662,
|
||
|
|
"learning_rate": 2.3663663663663664e-05,
|
||
|
|
"loss": 0.1833,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1870959848165512,
|
||
|
|
"step": 395,
|
||
|
|
"valid_targets_mean": 2629.9,
|
||
|
|
"valid_targets_min": 940
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4206098843322818,
|
||
|
|
"grad_norm": 0.592275147113119,
|
||
|
|
"learning_rate": 2.3963963963963967e-05,
|
||
|
|
"loss": 0.1777,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17953616380691528,
|
||
|
|
"step": 400,
|
||
|
|
"valid_targets_mean": 2703.4,
|
||
|
|
"valid_targets_min": 689
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.42586750788643535,
|
||
|
|
"grad_norm": 0.7884806996166527,
|
||
|
|
"learning_rate": 2.4264264264264267e-05,
|
||
|
|
"loss": 0.1994,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2150239646434784,
|
||
|
|
"step": 405,
|
||
|
|
"valid_targets_mean": 2128.2,
|
||
|
|
"valid_targets_min": 712
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.43112513144058884,
|
||
|
|
"grad_norm": 0.7625877916283756,
|
||
|
|
"learning_rate": 2.4564564564564566e-05,
|
||
|
|
"loss": 0.175,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1964799463748932,
|
||
|
|
"step": 410,
|
||
|
|
"valid_targets_mean": 1751.6,
|
||
|
|
"valid_targets_min": 638
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4363827549947424,
|
||
|
|
"grad_norm": 0.7190501254301566,
|
||
|
|
"learning_rate": 2.4864864864864866e-05,
|
||
|
|
"loss": 0.2051,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20976972579956055,
|
||
|
|
"step": 415,
|
||
|
|
"valid_targets_mean": 2239.5,
|
||
|
|
"valid_targets_min": 478
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4416403785488959,
|
||
|
|
"grad_norm": 0.5887915880067021,
|
||
|
|
"learning_rate": 2.5165165165165165e-05,
|
||
|
|
"loss": 0.2154,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15709666907787323,
|
||
|
|
"step": 420,
|
||
|
|
"valid_targets_mean": 2530.4,
|
||
|
|
"valid_targets_min": 816
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4468980021030494,
|
||
|
|
"grad_norm": 0.6188818015238577,
|
||
|
|
"learning_rate": 2.5465465465465465e-05,
|
||
|
|
"loss": 0.1803,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17590877413749695,
|
||
|
|
"step": 425,
|
||
|
|
"valid_targets_mean": 2636.4,
|
||
|
|
"valid_targets_min": 744
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.45215562565720296,
|
||
|
|
"grad_norm": 0.7206788274447538,
|
||
|
|
"learning_rate": 2.5765765765765768e-05,
|
||
|
|
"loss": 0.182,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18649278581142426,
|
||
|
|
"step": 430,
|
||
|
|
"valid_targets_mean": 2198.9,
|
||
|
|
"valid_targets_min": 649
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.45741324921135645,
|
||
|
|
"grad_norm": 0.6865336128025249,
|
||
|
|
"learning_rate": 2.6066066066066067e-05,
|
||
|
|
"loss": 0.196,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20152923464775085,
|
||
|
|
"step": 435,
|
||
|
|
"valid_targets_mean": 2388.2,
|
||
|
|
"valid_targets_min": 718
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.46267087276551,
|
||
|
|
"grad_norm": 0.7574500633972526,
|
||
|
|
"learning_rate": 2.6366366366366367e-05,
|
||
|
|
"loss": 0.2031,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19476348161697388,
|
||
|
|
"step": 440,
|
||
|
|
"valid_targets_mean": 2251.8,
|
||
|
|
"valid_targets_min": 682
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4679284963196635,
|
||
|
|
"grad_norm": 0.7868468166673843,
|
||
|
|
"learning_rate": 2.6666666666666667e-05,
|
||
|
|
"loss": 0.1702,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20086446404457092,
|
||
|
|
"step": 445,
|
||
|
|
"valid_targets_mean": 2351.4,
|
||
|
|
"valid_targets_min": 797
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.47318611987381703,
|
||
|
|
"grad_norm": 0.8060949599740148,
|
||
|
|
"learning_rate": 2.6966966966966966e-05,
|
||
|
|
"loss": 0.182,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18876756727695465,
|
||
|
|
"step": 450,
|
||
|
|
"valid_targets_mean": 1894.8,
|
||
|
|
"valid_targets_min": 909
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4784437434279706,
|
||
|
|
"grad_norm": 0.6734677372719962,
|
||
|
|
"learning_rate": 2.726726726726727e-05,
|
||
|
|
"loss": 0.193,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18098291754722595,
|
||
|
|
"step": 455,
|
||
|
|
"valid_targets_mean": 2314.8,
|
||
|
|
"valid_targets_min": 645
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.48370136698212407,
|
||
|
|
"grad_norm": 0.5748017571039128,
|
||
|
|
"learning_rate": 2.756756756756757e-05,
|
||
|
|
"loss": 0.1775,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15493184328079224,
|
||
|
|
"step": 460,
|
||
|
|
"valid_targets_mean": 2608.8,
|
||
|
|
"valid_targets_min": 526
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4889589905362776,
|
||
|
|
"grad_norm": 0.8504204969787917,
|
||
|
|
"learning_rate": 2.786786786786787e-05,
|
||
|
|
"loss": 0.1913,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17767800390720367,
|
||
|
|
"step": 465,
|
||
|
|
"valid_targets_mean": 2493.2,
|
||
|
|
"valid_targets_min": 734
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4942166140904311,
|
||
|
|
"grad_norm": 0.7408464845892841,
|
||
|
|
"learning_rate": 2.8168168168168168e-05,
|
||
|
|
"loss": 0.185,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16371986269950867,
|
||
|
|
"step": 470,
|
||
|
|
"valid_targets_mean": 1887.7,
|
||
|
|
"valid_targets_min": 722
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.49947423764458465,
|
||
|
|
"grad_norm": 0.7405124432700707,
|
||
|
|
"learning_rate": 2.8468468468468467e-05,
|
||
|
|
"loss": 0.1704,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1891363561153412,
|
||
|
|
"step": 475,
|
||
|
|
"valid_targets_mean": 1849.9,
|
||
|
|
"valid_targets_min": 711
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5047318611987381,
|
||
|
|
"grad_norm": 0.6085263821300575,
|
||
|
|
"learning_rate": 2.8768768768768774e-05,
|
||
|
|
"loss": 0.1779,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17886114120483398,
|
||
|
|
"step": 480,
|
||
|
|
"valid_targets_mean": 2606.5,
|
||
|
|
"valid_targets_min": 757
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5099894847528917,
|
||
|
|
"grad_norm": 0.7263272310036324,
|
||
|
|
"learning_rate": 2.9069069069069073e-05,
|
||
|
|
"loss": 0.184,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21606990694999695,
|
||
|
|
"step": 485,
|
||
|
|
"valid_targets_mean": 2144.4,
|
||
|
|
"valid_targets_min": 591
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5152471083070452,
|
||
|
|
"grad_norm": 0.8043786243044231,
|
||
|
|
"learning_rate": 2.9369369369369373e-05,
|
||
|
|
"loss": 0.1914,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20415738224983215,
|
||
|
|
"step": 490,
|
||
|
|
"valid_targets_mean": 1992.6,
|
||
|
|
"valid_targets_min": 851
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5205047318611987,
|
||
|
|
"grad_norm": 0.6413800286912905,
|
||
|
|
"learning_rate": 2.9669669669669673e-05,
|
||
|
|
"loss": 0.1872,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24239026010036469,
|
||
|
|
"step": 495,
|
||
|
|
"valid_targets_mean": 2486.3,
|
||
|
|
"valid_targets_min": 770
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5257623554153522,
|
||
|
|
"grad_norm": 0.5718136458682129,
|
||
|
|
"learning_rate": 2.9969969969969976e-05,
|
||
|
|
"loss": 0.17,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13713397085666656,
|
||
|
|
"step": 500,
|
||
|
|
"valid_targets_mean": 2497.6,
|
||
|
|
"valid_targets_min": 862
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5310199789695058,
|
||
|
|
"grad_norm": 0.65563223773779,
|
||
|
|
"learning_rate": 3.0270270270270275e-05,
|
||
|
|
"loss": 0.1731,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16426986455917358,
|
||
|
|
"step": 505,
|
||
|
|
"valid_targets_mean": 1927.3,
|
||
|
|
"valid_targets_min": 614
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5362776025236593,
|
||
|
|
"grad_norm": 0.615993728997519,
|
||
|
|
"learning_rate": 3.0570570570570575e-05,
|
||
|
|
"loss": 0.1837,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15791280567646027,
|
||
|
|
"step": 510,
|
||
|
|
"valid_targets_mean": 2199.2,
|
||
|
|
"valid_targets_min": 550
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5415352260778128,
|
||
|
|
"grad_norm": 0.6666125570411275,
|
||
|
|
"learning_rate": 3.0870870870870874e-05,
|
||
|
|
"loss": 0.2183,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18872445821762085,
|
||
|
|
"step": 515,
|
||
|
|
"valid_targets_mean": 2238.9,
|
||
|
|
"valid_targets_min": 736
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5467928496319664,
|
||
|
|
"grad_norm": 0.6432657218526656,
|
||
|
|
"learning_rate": 3.1171171171171174e-05,
|
||
|
|
"loss": 0.1582,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16147862374782562,
|
||
|
|
"step": 520,
|
||
|
|
"valid_targets_mean": 2443.1,
|
||
|
|
"valid_targets_min": 909
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5520504731861199,
|
||
|
|
"grad_norm": 0.5957553693364401,
|
||
|
|
"learning_rate": 3.1471471471471473e-05,
|
||
|
|
"loss": 0.1728,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16299906373023987,
|
||
|
|
"step": 525,
|
||
|
|
"valid_targets_mean": 2472.2,
|
||
|
|
"valid_targets_min": 424
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5573080967402734,
|
||
|
|
"grad_norm": 0.6117914374407668,
|
||
|
|
"learning_rate": 3.177177177177177e-05,
|
||
|
|
"loss": 0.1777,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16739054024219513,
|
||
|
|
"step": 530,
|
||
|
|
"valid_targets_mean": 2387.2,
|
||
|
|
"valid_targets_min": 676
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.562565720294427,
|
||
|
|
"grad_norm": 0.6342208187674023,
|
||
|
|
"learning_rate": 3.207207207207207e-05,
|
||
|
|
"loss": 0.1657,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17585764825344086,
|
||
|
|
"step": 535,
|
||
|
|
"valid_targets_mean": 2111.0,
|
||
|
|
"valid_targets_min": 692
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5678233438485805,
|
||
|
|
"grad_norm": 0.6757674724831666,
|
||
|
|
"learning_rate": 3.237237237237238e-05,
|
||
|
|
"loss": 0.1779,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1941223293542862,
|
||
|
|
"step": 540,
|
||
|
|
"valid_targets_mean": 2259.1,
|
||
|
|
"valid_targets_min": 767
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.573080967402734,
|
||
|
|
"grad_norm": 0.7126958880160293,
|
||
|
|
"learning_rate": 3.267267267267268e-05,
|
||
|
|
"loss": 0.1992,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19136236608028412,
|
||
|
|
"step": 545,
|
||
|
|
"valid_targets_mean": 2100.1,
|
||
|
|
"valid_targets_min": 593
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5783385909568874,
|
||
|
|
"grad_norm": 0.7656416888764622,
|
||
|
|
"learning_rate": 3.297297297297298e-05,
|
||
|
|
"loss": 0.1806,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18279878795146942,
|
||
|
|
"step": 550,
|
||
|
|
"valid_targets_mean": 1668.4,
|
||
|
|
"valid_targets_min": 673
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.583596214511041,
|
||
|
|
"grad_norm": 0.5300670190789428,
|
||
|
|
"learning_rate": 3.327327327327328e-05,
|
||
|
|
"loss": 0.1747,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15574492514133453,
|
||
|
|
"step": 555,
|
||
|
|
"valid_targets_mean": 2771.8,
|
||
|
|
"valid_targets_min": 710
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5888538380651945,
|
||
|
|
"grad_norm": 0.6881065397072038,
|
||
|
|
"learning_rate": 3.357357357357358e-05,
|
||
|
|
"loss": 0.1729,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18636192381381989,
|
||
|
|
"step": 560,
|
||
|
|
"valid_targets_mean": 2491.6,
|
||
|
|
"valid_targets_min": 658
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.594111461619348,
|
||
|
|
"grad_norm": 0.5728887028567489,
|
||
|
|
"learning_rate": 3.387387387387388e-05,
|
||
|
|
"loss": 0.1871,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17813405394554138,
|
||
|
|
"step": 565,
|
||
|
|
"valid_targets_mean": 2598.2,
|
||
|
|
"valid_targets_min": 1126
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5993690851735016,
|
||
|
|
"grad_norm": 0.6026615039606725,
|
||
|
|
"learning_rate": 3.4174174174174176e-05,
|
||
|
|
"loss": 0.1735,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1661795675754547,
|
||
|
|
"step": 570,
|
||
|
|
"valid_targets_mean": 2114.4,
|
||
|
|
"valid_targets_min": 632
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6046267087276551,
|
||
|
|
"grad_norm": 0.7651311697389674,
|
||
|
|
"learning_rate": 3.4474474474474476e-05,
|
||
|
|
"loss": 0.1933,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16414538025856018,
|
||
|
|
"step": 575,
|
||
|
|
"valid_targets_mean": 2169.0,
|
||
|
|
"valid_targets_min": 851
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6098843322818086,
|
||
|
|
"grad_norm": 0.9063191051487169,
|
||
|
|
"learning_rate": 3.4774774774774776e-05,
|
||
|
|
"loss": 0.1717,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21756865084171295,
|
||
|
|
"step": 580,
|
||
|
|
"valid_targets_mean": 1645.6,
|
||
|
|
"valid_targets_min": 781
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6151419558359621,
|
||
|
|
"grad_norm": 0.6398259399691795,
|
||
|
|
"learning_rate": 3.5075075075075075e-05,
|
||
|
|
"loss": 0.1842,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2054351270198822,
|
||
|
|
"step": 585,
|
||
|
|
"valid_targets_mean": 2567.9,
|
||
|
|
"valid_targets_min": 523
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6203995793901157,
|
||
|
|
"grad_norm": 0.5111970959159889,
|
||
|
|
"learning_rate": 3.5375375375375375e-05,
|
||
|
|
"loss": 0.1645,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13696236908435822,
|
||
|
|
"step": 590,
|
||
|
|
"valid_targets_mean": 2637.8,
|
||
|
|
"valid_targets_min": 484
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6256572029442692,
|
||
|
|
"grad_norm": 0.6883255442754,
|
||
|
|
"learning_rate": 3.567567567567568e-05,
|
||
|
|
"loss": 0.198,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19190296530723572,
|
||
|
|
"step": 595,
|
||
|
|
"valid_targets_mean": 2228.6,
|
||
|
|
"valid_targets_min": 786
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6309148264984227,
|
||
|
|
"grad_norm": 0.5554373149854719,
|
||
|
|
"learning_rate": 3.597597597597598e-05,
|
||
|
|
"loss": 0.1947,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18401506543159485,
|
||
|
|
"step": 600,
|
||
|
|
"valid_targets_mean": 2788.6,
|
||
|
|
"valid_targets_min": 1249
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6361724500525763,
|
||
|
|
"grad_norm": 0.8451481806212515,
|
||
|
|
"learning_rate": 3.627627627627628e-05,
|
||
|
|
"loss": 0.1838,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2086559534072876,
|
||
|
|
"step": 605,
|
||
|
|
"valid_targets_mean": 1758.6,
|
||
|
|
"valid_targets_min": 559
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6414300736067298,
|
||
|
|
"grad_norm": 0.7053365690537642,
|
||
|
|
"learning_rate": 3.657657657657658e-05,
|
||
|
|
"loss": 0.1831,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2021399885416031,
|
||
|
|
"step": 610,
|
||
|
|
"valid_targets_mean": 1882.8,
|
||
|
|
"valid_targets_min": 788
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6466876971608833,
|
||
|
|
"grad_norm": 0.6868031046340202,
|
||
|
|
"learning_rate": 3.687687687687688e-05,
|
||
|
|
"loss": 0.1938,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19633038341999054,
|
||
|
|
"step": 615,
|
||
|
|
"valid_targets_mean": 2132.7,
|
||
|
|
"valid_targets_min": 724
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6519453207150369,
|
||
|
|
"grad_norm": 0.4697823431901432,
|
||
|
|
"learning_rate": 3.717717717717718e-05,
|
||
|
|
"loss": 0.1658,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13606944680213928,
|
||
|
|
"step": 620,
|
||
|
|
"valid_targets_mean": 3075.2,
|
||
|
|
"valid_targets_min": 788
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6572029442691903,
|
||
|
|
"grad_norm": 0.7369216315165371,
|
||
|
|
"learning_rate": 3.747747747747748e-05,
|
||
|
|
"loss": 0.1636,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18576949834823608,
|
||
|
|
"step": 625,
|
||
|
|
"valid_targets_mean": 2183.8,
|
||
|
|
"valid_targets_min": 700
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6624605678233438,
|
||
|
|
"grad_norm": 0.5975385901111251,
|
||
|
|
"learning_rate": 3.777777777777778e-05,
|
||
|
|
"loss": 0.1832,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17303280532360077,
|
||
|
|
"step": 630,
|
||
|
|
"valid_targets_mean": 2362.7,
|
||
|
|
"valid_targets_min": 804
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6677181913774973,
|
||
|
|
"grad_norm": 0.6369907123556227,
|
||
|
|
"learning_rate": 3.807807807807808e-05,
|
||
|
|
"loss": 0.1711,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1670917272567749,
|
||
|
|
"step": 635,
|
||
|
|
"valid_targets_mean": 1980.8,
|
||
|
|
"valid_targets_min": 592
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6729758149316509,
|
||
|
|
"grad_norm": 0.6085431308713706,
|
||
|
|
"learning_rate": 3.837837837837838e-05,
|
||
|
|
"loss": 0.159,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18237867951393127,
|
||
|
|
"step": 640,
|
||
|
|
"valid_targets_mean": 2670.2,
|
||
|
|
"valid_targets_min": 750
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6782334384858044,
|
||
|
|
"grad_norm": 0.5994039827911364,
|
||
|
|
"learning_rate": 3.8678678678678684e-05,
|
||
|
|
"loss": 0.174,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18501517176628113,
|
||
|
|
"step": 645,
|
||
|
|
"valid_targets_mean": 2278.9,
|
||
|
|
"valid_targets_min": 787
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6834910620399579,
|
||
|
|
"grad_norm": 0.5795035925542981,
|
||
|
|
"learning_rate": 3.897897897897898e-05,
|
||
|
|
"loss": 0.1644,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14112544059753418,
|
||
|
|
"step": 650,
|
||
|
|
"valid_targets_mean": 2230.8,
|
||
|
|
"valid_targets_min": 722
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6887486855941115,
|
||
|
|
"grad_norm": 0.6460639655071514,
|
||
|
|
"learning_rate": 3.927927927927928e-05,
|
||
|
|
"loss": 0.1794,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1670469045639038,
|
||
|
|
"step": 655,
|
||
|
|
"valid_targets_mean": 2240.4,
|
||
|
|
"valid_targets_min": 773
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.694006309148265,
|
||
|
|
"grad_norm": 0.5402195702845815,
|
||
|
|
"learning_rate": 3.957957957957958e-05,
|
||
|
|
"loss": 0.1925,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1540539562702179,
|
||
|
|
"step": 660,
|
||
|
|
"valid_targets_mean": 2391.4,
|
||
|
|
"valid_targets_min": 785
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6992639327024185,
|
||
|
|
"grad_norm": 0.5729195812303945,
|
||
|
|
"learning_rate": 3.987987987987988e-05,
|
||
|
|
"loss": 0.1792,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17357392609119415,
|
||
|
|
"step": 665,
|
||
|
|
"valid_targets_mean": 2328.1,
|
||
|
|
"valid_targets_min": 841
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.704521556256572,
|
||
|
|
"grad_norm": 0.6527281888817938,
|
||
|
|
"learning_rate": 3.9999975251805184e-05,
|
||
|
|
"loss": 0.1712,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16926416754722595,
|
||
|
|
"step": 670,
|
||
|
|
"valid_targets_mean": 2042.2,
|
||
|
|
"valid_targets_min": 777
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7097791798107256,
|
||
|
|
"grad_norm": 0.545867818845897,
|
||
|
|
"learning_rate": 3.9999824013058675e-05,
|
||
|
|
"loss": 0.1981,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1680353730916977,
|
||
|
|
"step": 675,
|
||
|
|
"valid_targets_mean": 2518.4,
|
||
|
|
"valid_targets_min": 760
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7150368033648791,
|
||
|
|
"grad_norm": 0.6251591067490714,
|
||
|
|
"learning_rate": 3.99995352856012e-05,
|
||
|
|
"loss": 0.1664,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20630434155464172,
|
||
|
|
"step": 680,
|
||
|
|
"valid_targets_mean": 2519.9,
|
||
|
|
"valid_targets_min": 985
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7202944269190326,
|
||
|
|
"grad_norm": 0.5110368474492926,
|
||
|
|
"learning_rate": 3.999910907141761e-05,
|
||
|
|
"loss": 0.1793,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16521261632442474,
|
||
|
|
"step": 685,
|
||
|
|
"valid_targets_mean": 2938.4,
|
||
|
|
"valid_targets_min": 921
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7255520504731862,
|
||
|
|
"grad_norm": 0.6222219346656122,
|
||
|
|
"learning_rate": 3.9998545373437924e-05,
|
||
|
|
"loss": 0.1797,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19978000223636627,
|
||
|
|
"step": 690,
|
||
|
|
"valid_targets_mean": 2286.5,
|
||
|
|
"valid_targets_min": 623
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7308096740273397,
|
||
|
|
"grad_norm": 0.6142153586570709,
|
||
|
|
"learning_rate": 3.999784419553728e-05,
|
||
|
|
"loss": 0.1675,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15856142342090607,
|
||
|
|
"step": 695,
|
||
|
|
"valid_targets_mean": 1922.4,
|
||
|
|
"valid_targets_min": 563
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7360672975814931,
|
||
|
|
"grad_norm": 0.44756085044101696,
|
||
|
|
"learning_rate": 3.9997005542535916e-05,
|
||
|
|
"loss": 0.156,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13361108303070068,
|
||
|
|
"step": 700,
|
||
|
|
"valid_targets_mean": 3132.7,
|
||
|
|
"valid_targets_min": 777
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7413249211356467,
|
||
|
|
"grad_norm": 0.4561829356226844,
|
||
|
|
"learning_rate": 3.9996029420199154e-05,
|
||
|
|
"loss": 0.1734,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1427106261253357,
|
||
|
|
"step": 705,
|
||
|
|
"valid_targets_mean": 3116.9,
|
||
|
|
"valid_targets_min": 768
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7465825446898002,
|
||
|
|
"grad_norm": 0.5804976303987642,
|
||
|
|
"learning_rate": 3.9994915835237336e-05,
|
||
|
|
"loss": 0.1759,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.165114164352417,
|
||
|
|
"step": 710,
|
||
|
|
"valid_targets_mean": 1991.9,
|
||
|
|
"valid_targets_min": 622
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7518401682439537,
|
||
|
|
"grad_norm": 0.6027950423183488,
|
||
|
|
"learning_rate": 3.999366479530581e-05,
|
||
|
|
"loss": 0.1785,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21031297743320465,
|
||
|
|
"step": 715,
|
||
|
|
"valid_targets_mean": 2443.0,
|
||
|
|
"valid_targets_min": 750
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7570977917981072,
|
||
|
|
"grad_norm": 0.5464640777168759,
|
||
|
|
"learning_rate": 3.999227630900483e-05,
|
||
|
|
"loss": 0.177,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14772871136665344,
|
||
|
|
"step": 720,
|
||
|
|
"valid_targets_mean": 2187.8,
|
||
|
|
"valid_targets_min": 645
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7623554153522608,
|
||
|
|
"grad_norm": 0.556873706552765,
|
||
|
|
"learning_rate": 3.9990750385879554e-05,
|
||
|
|
"loss": 0.1759,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18065491318702698,
|
||
|
|
"step": 725,
|
||
|
|
"valid_targets_mean": 2686.9,
|
||
|
|
"valid_targets_min": 759
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7676130389064143,
|
||
|
|
"grad_norm": 0.7821326005174354,
|
||
|
|
"learning_rate": 3.998908703641993e-05,
|
||
|
|
"loss": 0.1828,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1998850405216217,
|
||
|
|
"step": 730,
|
||
|
|
"valid_targets_mean": 1949.1,
|
||
|
|
"valid_targets_min": 544
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7728706624605678,
|
||
|
|
"grad_norm": 0.5533353988561285,
|
||
|
|
"learning_rate": 3.9987286272060644e-05,
|
||
|
|
"loss": 0.1664,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14575821161270142,
|
||
|
|
"step": 735,
|
||
|
|
"valid_targets_mean": 2426.7,
|
||
|
|
"valid_targets_min": 794
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7781282860147214,
|
||
|
|
"grad_norm": 0.5462930761701978,
|
||
|
|
"learning_rate": 3.998534810518104e-05,
|
||
|
|
"loss": 0.1505,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14197488129138947,
|
||
|
|
"step": 740,
|
||
|
|
"valid_targets_mean": 2386.8,
|
||
|
|
"valid_targets_min": 1117
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7833859095688749,
|
||
|
|
"grad_norm": 0.7381314277295772,
|
||
|
|
"learning_rate": 3.998327254910504e-05,
|
||
|
|
"loss": 0.1747,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21469098329544067,
|
||
|
|
"step": 745,
|
||
|
|
"valid_targets_mean": 1959.9,
|
||
|
|
"valid_targets_min": 1044
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7886435331230284,
|
||
|
|
"grad_norm": 0.5397978816820115,
|
||
|
|
"learning_rate": 3.998105961810105e-05,
|
||
|
|
"loss": 0.1611,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1522783488035202,
|
||
|
|
"step": 750,
|
||
|
|
"valid_targets_mean": 2712.9,
|
||
|
|
"valid_targets_min": 691
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7939011566771819,
|
||
|
|
"grad_norm": 0.4895922781852716,
|
||
|
|
"learning_rate": 3.997870932738187e-05,
|
||
|
|
"loss": 0.1877,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16068905591964722,
|
||
|
|
"step": 755,
|
||
|
|
"valid_targets_mean": 2521.6,
|
||
|
|
"valid_targets_min": 731
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7991587802313355,
|
||
|
|
"grad_norm": 0.623876405157066,
|
||
|
|
"learning_rate": 3.997622169310454e-05,
|
||
|
|
"loss": 0.1808,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18478304147720337,
|
||
|
|
"step": 760,
|
||
|
|
"valid_targets_mean": 2134.2,
|
||
|
|
"valid_targets_min": 655
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.804416403785489,
|
||
|
|
"grad_norm": 0.6148092077121673,
|
||
|
|
"learning_rate": 3.9973596732370296e-05,
|
||
|
|
"loss": 0.1798,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17871780693531036,
|
||
|
|
"step": 765,
|
||
|
|
"valid_targets_mean": 1782.9,
|
||
|
|
"valid_targets_min": 490
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8096740273396424,
|
||
|
|
"grad_norm": 0.7248805267657951,
|
||
|
|
"learning_rate": 3.997083446322443e-05,
|
||
|
|
"loss": 0.1969,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18811409175395966,
|
||
|
|
"step": 770,
|
||
|
|
"valid_targets_mean": 1747.4,
|
||
|
|
"valid_targets_min": 697
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.814931650893796,
|
||
|
|
"grad_norm": 0.7541380235031304,
|
||
|
|
"learning_rate": 3.9967934904656145e-05,
|
||
|
|
"loss": 0.1707,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19300444424152374,
|
||
|
|
"step": 775,
|
||
|
|
"valid_targets_mean": 1872.0,
|
||
|
|
"valid_targets_min": 754
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8201892744479495,
|
||
|
|
"grad_norm": 0.6377312536651119,
|
||
|
|
"learning_rate": 3.9964898076598445e-05,
|
||
|
|
"loss": 0.1676,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1708446890115738,
|
||
|
|
"step": 780,
|
||
|
|
"valid_targets_mean": 1849.5,
|
||
|
|
"valid_targets_min": 717
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.825446898002103,
|
||
|
|
"grad_norm": 0.5723814500895619,
|
||
|
|
"learning_rate": 3.996172399992799e-05,
|
||
|
|
"loss": 0.1782,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18040908873081207,
|
||
|
|
"step": 785,
|
||
|
|
"valid_targets_mean": 2206.8,
|
||
|
|
"valid_targets_min": 721
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8307045215562566,
|
||
|
|
"grad_norm": 0.535003518787969,
|
||
|
|
"learning_rate": 3.995841269646496e-05,
|
||
|
|
"loss": 0.2067,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1446647047996521,
|
||
|
|
"step": 790,
|
||
|
|
"valid_targets_mean": 2437.4,
|
||
|
|
"valid_targets_min": 711
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8359621451104101,
|
||
|
|
"grad_norm": 0.5672473401592892,
|
||
|
|
"learning_rate": 3.995496418897291e-05,
|
||
|
|
"loss": 0.1864,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1599861979484558,
|
||
|
|
"step": 795,
|
||
|
|
"valid_targets_mean": 2093.8,
|
||
|
|
"valid_targets_min": 573
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8412197686645636,
|
||
|
|
"grad_norm": 0.4912116094638417,
|
||
|
|
"learning_rate": 3.995137850115856e-05,
|
||
|
|
"loss": 0.16,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15672048926353455,
|
||
|
|
"step": 800,
|
||
|
|
"valid_targets_mean": 2729.2,
|
||
|
|
"valid_targets_min": 1026
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8464773922187171,
|
||
|
|
"grad_norm": 0.4664566983101438,
|
||
|
|
"learning_rate": 3.994765565767174e-05,
|
||
|
|
"loss": 0.1567,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15762017667293549,
|
||
|
|
"step": 805,
|
||
|
|
"valid_targets_mean": 2682.2,
|
||
|
|
"valid_targets_min": 902
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8517350157728707,
|
||
|
|
"grad_norm": 0.5942973141911478,
|
||
|
|
"learning_rate": 3.9943795684105104e-05,
|
||
|
|
"loss": 0.1675,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16043150424957275,
|
||
|
|
"step": 810,
|
||
|
|
"valid_targets_mean": 2266.4,
|
||
|
|
"valid_targets_min": 711
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8569926393270242,
|
||
|
|
"grad_norm": 0.5128367451048222,
|
||
|
|
"learning_rate": 3.993979860699403e-05,
|
||
|
|
"loss": 0.1752,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17938151955604553,
|
||
|
|
"step": 815,
|
||
|
|
"valid_targets_mean": 2735.4,
|
||
|
|
"valid_targets_min": 833
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8622502628811777,
|
||
|
|
"grad_norm": 0.46964310991514857,
|
||
|
|
"learning_rate": 3.993566445381641e-05,
|
||
|
|
"loss": 0.168,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13383260369300842,
|
||
|
|
"step": 820,
|
||
|
|
"valid_targets_mean": 2790.2,
|
||
|
|
"valid_targets_min": 446
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8675078864353313,
|
||
|
|
"grad_norm": 0.583921992918043,
|
||
|
|
"learning_rate": 3.9931393252992454e-05,
|
||
|
|
"loss": 0.167,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17725703120231628,
|
||
|
|
"step": 825,
|
||
|
|
"valid_targets_mean": 2180.7,
|
||
|
|
"valid_targets_min": 683
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8727655099894848,
|
||
|
|
"grad_norm": 0.5399592464700252,
|
||
|
|
"learning_rate": 3.992698503388453e-05,
|
||
|
|
"loss": 0.1783,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15243329107761383,
|
||
|
|
"step": 830,
|
||
|
|
"valid_targets_mean": 2000.4,
|
||
|
|
"valid_targets_min": 518
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8780231335436383,
|
||
|
|
"grad_norm": 0.5122009262637606,
|
||
|
|
"learning_rate": 3.992243982679691e-05,
|
||
|
|
"loss": 0.1645,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15500515699386597,
|
||
|
|
"step": 835,
|
||
|
|
"valid_targets_mean": 2338.6,
|
||
|
|
"valid_targets_min": 664
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8832807570977917,
|
||
|
|
"grad_norm": 0.5358477857493344,
|
||
|
|
"learning_rate": 3.991775766297562e-05,
|
||
|
|
"loss": 0.1729,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1721762865781784,
|
||
|
|
"step": 840,
|
||
|
|
"valid_targets_mean": 2463.5,
|
||
|
|
"valid_targets_min": 521
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8885383806519453,
|
||
|
|
"grad_norm": 0.5022921936433923,
|
||
|
|
"learning_rate": 3.991293857460815e-05,
|
||
|
|
"loss": 0.1758,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1488204300403595,
|
||
|
|
"step": 845,
|
||
|
|
"valid_targets_mean": 2454.8,
|
||
|
|
"valid_targets_min": 747
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8937960042060988,
|
||
|
|
"grad_norm": 0.588270886479337,
|
||
|
|
"learning_rate": 3.9907982594823326e-05,
|
||
|
|
"loss": 0.1665,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14433225989341736,
|
||
|
|
"step": 850,
|
||
|
|
"valid_targets_mean": 1942.8,
|
||
|
|
"valid_targets_min": 337
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8990536277602523,
|
||
|
|
"grad_norm": 0.518841925555911,
|
||
|
|
"learning_rate": 3.9902889757691e-05,
|
||
|
|
"loss": 0.1616,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15561543405056,
|
||
|
|
"step": 855,
|
||
|
|
"valid_targets_mean": 2479.9,
|
||
|
|
"valid_targets_min": 449
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9043112513144059,
|
||
|
|
"grad_norm": 0.5274949838957608,
|
||
|
|
"learning_rate": 3.9897660098221866e-05,
|
||
|
|
"loss": 0.1774,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1939384639263153,
|
||
|
|
"step": 860,
|
||
|
|
"valid_targets_mean": 2797.2,
|
||
|
|
"valid_targets_min": 1127
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9095688748685594,
|
||
|
|
"grad_norm": 0.7442818550986994,
|
||
|
|
"learning_rate": 3.98922936523672e-05,
|
||
|
|
"loss": 0.1609,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.153502956032753,
|
||
|
|
"step": 865,
|
||
|
|
"valid_targets_mean": 2642.8,
|
||
|
|
"valid_targets_min": 748
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9148264984227129,
|
||
|
|
"grad_norm": 0.5286740197393716,
|
||
|
|
"learning_rate": 3.9886790457018604e-05,
|
||
|
|
"loss": 0.1538,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14489783346652985,
|
||
|
|
"step": 870,
|
||
|
|
"valid_targets_mean": 2130.8,
|
||
|
|
"valid_targets_min": 554
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9200841219768665,
|
||
|
|
"grad_norm": 0.4874203031064565,
|
||
|
|
"learning_rate": 3.9881150550007776e-05,
|
||
|
|
"loss": 0.1642,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15502655506134033,
|
||
|
|
"step": 875,
|
||
|
|
"valid_targets_mean": 2639.6,
|
||
|
|
"valid_targets_min": 746
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.92534174553102,
|
||
|
|
"grad_norm": 0.611548653068185,
|
||
|
|
"learning_rate": 3.987537397010624e-05,
|
||
|
|
"loss": 0.1672,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18344169855117798,
|
||
|
|
"step": 880,
|
||
|
|
"valid_targets_mean": 1815.5,
|
||
|
|
"valid_targets_min": 803
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9305993690851735,
|
||
|
|
"grad_norm": 0.5628452882147774,
|
||
|
|
"learning_rate": 3.9869460757025064e-05,
|
||
|
|
"loss": 0.1566,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15735852718353271,
|
||
|
|
"step": 885,
|
||
|
|
"valid_targets_mean": 2084.2,
|
||
|
|
"valid_targets_min": 677
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.935856992639327,
|
||
|
|
"grad_norm": 0.4493789868053273,
|
||
|
|
"learning_rate": 3.9863410951414616e-05,
|
||
|
|
"loss": 0.1664,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14666572213172913,
|
||
|
|
"step": 890,
|
||
|
|
"valid_targets_mean": 2944.2,
|
||
|
|
"valid_targets_min": 855
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9411146161934806,
|
||
|
|
"grad_norm": 0.6393599680687345,
|
||
|
|
"learning_rate": 3.985722459486425e-05,
|
||
|
|
"loss": 0.1732,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19685781002044678,
|
||
|
|
"step": 895,
|
||
|
|
"valid_targets_mean": 1945.8,
|
||
|
|
"valid_targets_min": 639
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9463722397476341,
|
||
|
|
"grad_norm": 0.5480847359216666,
|
||
|
|
"learning_rate": 3.985090172990206e-05,
|
||
|
|
"loss": 0.1607,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16915547847747803,
|
||
|
|
"step": 900,
|
||
|
|
"valid_targets_mean": 2248.5,
|
||
|
|
"valid_targets_min": 684
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9516298633017876,
|
||
|
|
"grad_norm": 0.5777861645548802,
|
||
|
|
"learning_rate": 3.984444239999455e-05,
|
||
|
|
"loss": 0.1585,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16395241022109985,
|
||
|
|
"step": 905,
|
||
|
|
"valid_targets_mean": 1847.9,
|
||
|
|
"valid_targets_min": 566
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9568874868559412,
|
||
|
|
"grad_norm": 0.5940526813080879,
|
||
|
|
"learning_rate": 3.9837846649546354e-05,
|
||
|
|
"loss": 0.1719,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18966373801231384,
|
||
|
|
"step": 910,
|
||
|
|
"valid_targets_mean": 2282.3,
|
||
|
|
"valid_targets_min": 549
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9621451104100947,
|
||
|
|
"grad_norm": 0.5056453529696999,
|
||
|
|
"learning_rate": 3.9831114523899945e-05,
|
||
|
|
"loss": 0.1626,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1332794725894928,
|
||
|
|
"step": 915,
|
||
|
|
"valid_targets_mean": 2242.7,
|
||
|
|
"valid_targets_min": 539
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9674027339642481,
|
||
|
|
"grad_norm": 0.5567868529662247,
|
||
|
|
"learning_rate": 3.982424606933529e-05,
|
||
|
|
"loss": 0.1515,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15924754738807678,
|
||
|
|
"step": 920,
|
||
|
|
"valid_targets_mean": 1903.2,
|
||
|
|
"valid_targets_min": 486
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9726603575184016,
|
||
|
|
"grad_norm": 0.5025855838183128,
|
||
|
|
"learning_rate": 3.981724133306954e-05,
|
||
|
|
"loss": 0.1801,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16654324531555176,
|
||
|
|
"step": 925,
|
||
|
|
"valid_targets_mean": 2608.4,
|
||
|
|
"valid_targets_min": 847
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9779179810725552,
|
||
|
|
"grad_norm": 0.5579940702696247,
|
||
|
|
"learning_rate": 3.981010036325674e-05,
|
||
|
|
"loss": 0.1705,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15462610125541687,
|
||
|
|
"step": 930,
|
||
|
|
"valid_targets_mean": 1866.8,
|
||
|
|
"valid_targets_min": 526
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9831756046267087,
|
||
|
|
"grad_norm": 0.6330246846048638,
|
||
|
|
"learning_rate": 3.980282320898746e-05,
|
||
|
|
"loss": 0.1713,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14142312109470367,
|
||
|
|
"step": 935,
|
||
|
|
"valid_targets_mean": 2298.9,
|
||
|
|
"valid_targets_min": 802
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9884332281808622,
|
||
|
|
"grad_norm": 0.6222570958228409,
|
||
|
|
"learning_rate": 3.9795409920288456e-05,
|
||
|
|
"loss": 0.1771,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2011018991470337,
|
||
|
|
"step": 940,
|
||
|
|
"valid_targets_mean": 2405.1,
|
||
|
|
"valid_targets_min": 667
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9936908517350158,
|
||
|
|
"grad_norm": 0.5613259849145827,
|
||
|
|
"learning_rate": 3.978786054812236e-05,
|
||
|
|
"loss": 0.1582,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15669333934783936,
|
||
|
|
"step": 945,
|
||
|
|
"valid_targets_mean": 2029.0,
|
||
|
|
"valid_targets_min": 641
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9989484752891693,
|
||
|
|
"grad_norm": 0.5545292089417356,
|
||
|
|
"learning_rate": 3.9780175144387304e-05,
|
||
|
|
"loss": 0.1795,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18618620932102203,
|
||
|
|
"step": 950,
|
||
|
|
"valid_targets_mean": 2303.4,
|
||
|
|
"valid_targets_min": 751
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"logging_steps": 5,
|
||
|
|
"max_steps": 6657,
|
||
|
|
"num_input_tokens_seen": 0,
|
||
|
|
"num_train_epochs": 7,
|
||
|
|
"save_steps": 950,
|
||
|
|
"stateful_callbacks": {
|
||
|
|
"TrainerControl": {
|
||
|
|
"args": {
|
||
|
|
"should_epoch_stop": false,
|
||
|
|
"should_evaluate": false,
|
||
|
|
"should_log": false,
|
||
|
|
"should_save": true,
|
||
|
|
"should_training_stop": false
|
||
|
|
},
|
||
|
|
"attributes": {}
|
||
|
|
}
|
||
|
|
},
|
||
|
|
"total_flos": 204013831258112.0,
|
||
|
|
"train_batch_size": 1,
|
||
|
|
"trial_name": null,
|
||
|
|
"trial_params": null
|
||
|
|
}
|