3953 lines
110 KiB
JSON
3953 lines
110 KiB
JSON
|
{
|
||
|
|
"best_global_step": null,
|
||
|
|
"best_metric": null,
|
||
|
|
"best_model_checkpoint": null,
|
||
|
|
"epoch": 7.0,
|
||
|
|
"eval_steps": 500,
|
||
|
|
"global_step": 1778,
|
||
|
|
"is_hyper_param_search": false,
|
||
|
|
"is_local_process_zero": true,
|
||
|
|
"is_world_process_zero": true,
|
||
|
|
"log_history": [
|
||
|
|
{
|
||
|
|
"epoch": 0.01972386587771203,
|
||
|
|
"grad_norm": 12.393668111885583,
|
||
|
|
"learning_rate": 8.98876404494382e-07,
|
||
|
|
"loss": 0.855,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.42231136560440063,
|
||
|
|
"step": 5,
|
||
|
|
"valid_targets_mean": 2780.5,
|
||
|
|
"valid_targets_min": 2286
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03944773175542406,
|
||
|
|
"grad_norm": 10.568611218520317,
|
||
|
|
"learning_rate": 2.02247191011236e-06,
|
||
|
|
"loss": 0.8485,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4118260145187378,
|
||
|
|
"step": 10,
|
||
|
|
"valid_targets_mean": 2595.2,
|
||
|
|
"valid_targets_min": 1086
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.05917159763313609,
|
||
|
|
"grad_norm": 7.157784986316959,
|
||
|
|
"learning_rate": 3.146067415730337e-06,
|
||
|
|
"loss": 0.7901,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4182024896144867,
|
||
|
|
"step": 15,
|
||
|
|
"valid_targets_mean": 2818.2,
|
||
|
|
"valid_targets_min": 2457
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07889546351084813,
|
||
|
|
"grad_norm": 4.408996433642804,
|
||
|
|
"learning_rate": 4.269662921348315e-06,
|
||
|
|
"loss": 0.7372,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.35063397884368896,
|
||
|
|
"step": 20,
|
||
|
|
"valid_targets_mean": 2696.0,
|
||
|
|
"valid_targets_min": 2461
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09861932938856016,
|
||
|
|
"grad_norm": 3.030932610406114,
|
||
|
|
"learning_rate": 5.393258426966292e-06,
|
||
|
|
"loss": 0.6841,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.33855897188186646,
|
||
|
|
"step": 25,
|
||
|
|
"valid_targets_mean": 2817.0,
|
||
|
|
"valid_targets_min": 2545
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11834319526627218,
|
||
|
|
"grad_norm": 2.091968197966212,
|
||
|
|
"learning_rate": 6.51685393258427e-06,
|
||
|
|
"loss": 0.6258,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2944120168685913,
|
||
|
|
"step": 30,
|
||
|
|
"valid_targets_mean": 2796.9,
|
||
|
|
"valid_targets_min": 2374
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13806706114398423,
|
||
|
|
"grad_norm": 1.5698709348893622,
|
||
|
|
"learning_rate": 7.640449438202247e-06,
|
||
|
|
"loss": 0.575,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28275173902511597,
|
||
|
|
"step": 35,
|
||
|
|
"valid_targets_mean": 2847.1,
|
||
|
|
"valid_targets_min": 2387
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15779092702169625,
|
||
|
|
"grad_norm": 1.3739634811879422,
|
||
|
|
"learning_rate": 8.764044943820226e-06,
|
||
|
|
"loss": 0.5306,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2477639764547348,
|
||
|
|
"step": 40,
|
||
|
|
"valid_targets_mean": 2631.8,
|
||
|
|
"valid_targets_min": 2119
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17751479289940827,
|
||
|
|
"grad_norm": 0.9974792692592687,
|
||
|
|
"learning_rate": 9.887640449438202e-06,
|
||
|
|
"loss": 0.4711,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23885783553123474,
|
||
|
|
"step": 45,
|
||
|
|
"valid_targets_mean": 3069.0,
|
||
|
|
"valid_targets_min": 2811
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19723865877712032,
|
||
|
|
"grad_norm": 0.9951495464010057,
|
||
|
|
"learning_rate": 1.101123595505618e-05,
|
||
|
|
"loss": 0.4641,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21858510375022888,
|
||
|
|
"step": 50,
|
||
|
|
"valid_targets_mean": 2701.8,
|
||
|
|
"valid_targets_min": 2360
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.21696252465483234,
|
||
|
|
"grad_norm": 0.853374801452012,
|
||
|
|
"learning_rate": 1.213483146067416e-05,
|
||
|
|
"loss": 0.4218,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2046160101890564,
|
||
|
|
"step": 55,
|
||
|
|
"valid_targets_mean": 2669.0,
|
||
|
|
"valid_targets_min": 2057
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23668639053254437,
|
||
|
|
"grad_norm": 0.8673849188540548,
|
||
|
|
"learning_rate": 1.3258426966292135e-05,
|
||
|
|
"loss": 0.409,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22626200318336487,
|
||
|
|
"step": 60,
|
||
|
|
"valid_targets_mean": 3023.6,
|
||
|
|
"valid_targets_min": 2792
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2564102564102564,
|
||
|
|
"grad_norm": 0.8853328907131118,
|
||
|
|
"learning_rate": 1.4382022471910113e-05,
|
||
|
|
"loss": 0.3929,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1955169439315796,
|
||
|
|
"step": 65,
|
||
|
|
"valid_targets_mean": 2625.6,
|
||
|
|
"valid_targets_min": 2278
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.27613412228796846,
|
||
|
|
"grad_norm": 0.8384791833352463,
|
||
|
|
"learning_rate": 1.5505617977528093e-05,
|
||
|
|
"loss": 0.365,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1843859851360321,
|
||
|
|
"step": 70,
|
||
|
|
"valid_targets_mean": 2425.4,
|
||
|
|
"valid_targets_min": 890
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2958579881656805,
|
||
|
|
"grad_norm": 0.7140157492337951,
|
||
|
|
"learning_rate": 1.662921348314607e-05,
|
||
|
|
"loss": 0.3623,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17334681749343872,
|
||
|
|
"step": 75,
|
||
|
|
"valid_targets_mean": 2629.9,
|
||
|
|
"valid_targets_min": 2318
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3155818540433925,
|
||
|
|
"grad_norm": 0.7211533502723859,
|
||
|
|
"learning_rate": 1.7752808988764045e-05,
|
||
|
|
"loss": 0.352,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19573208689689636,
|
||
|
|
"step": 80,
|
||
|
|
"valid_targets_mean": 2928.8,
|
||
|
|
"valid_targets_min": 2541
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.33530571992110453,
|
||
|
|
"grad_norm": 0.6986225390135399,
|
||
|
|
"learning_rate": 1.8876404494382024e-05,
|
||
|
|
"loss": 0.345,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17175813019275665,
|
||
|
|
"step": 85,
|
||
|
|
"valid_targets_mean": 2744.9,
|
||
|
|
"valid_targets_min": 2278
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.35502958579881655,
|
||
|
|
"grad_norm": 0.7237528497028296,
|
||
|
|
"learning_rate": 2e-05,
|
||
|
|
"loss": 0.3452,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18763425946235657,
|
||
|
|
"step": 90,
|
||
|
|
"valid_targets_mean": 2874.9,
|
||
|
|
"valid_targets_min": 2520
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3747534516765286,
|
||
|
|
"grad_norm": 0.7792654371055137,
|
||
|
|
"learning_rate": 2.1123595505617976e-05,
|
||
|
|
"loss": 0.3375,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15737923979759216,
|
||
|
|
"step": 95,
|
||
|
|
"valid_targets_mean": 2686.2,
|
||
|
|
"valid_targets_min": 2518
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.39447731755424065,
|
||
|
|
"grad_norm": 0.7749501994818205,
|
||
|
|
"learning_rate": 2.2247191011235958e-05,
|
||
|
|
"loss": 0.329,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18453183770179749,
|
||
|
|
"step": 100,
|
||
|
|
"valid_targets_mean": 2518.8,
|
||
|
|
"valid_targets_min": 758
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.41420118343195267,
|
||
|
|
"grad_norm": 0.8022966405400229,
|
||
|
|
"learning_rate": 2.3370786516853933e-05,
|
||
|
|
"loss": 0.3276,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1502269208431244,
|
||
|
|
"step": 105,
|
||
|
|
"valid_targets_mean": 2681.0,
|
||
|
|
"valid_targets_min": 2455
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4339250493096647,
|
||
|
|
"grad_norm": 0.7798414089275748,
|
||
|
|
"learning_rate": 2.4494382022471914e-05,
|
||
|
|
"loss": 0.3193,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16151049733161926,
|
||
|
|
"step": 110,
|
||
|
|
"valid_targets_mean": 2455.6,
|
||
|
|
"valid_targets_min": 629
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4536489151873767,
|
||
|
|
"grad_norm": 0.7394131887198511,
|
||
|
|
"learning_rate": 2.561797752808989e-05,
|
||
|
|
"loss": 0.3191,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1483708620071411,
|
||
|
|
"step": 115,
|
||
|
|
"valid_targets_mean": 2713.2,
|
||
|
|
"valid_targets_min": 2357
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.47337278106508873,
|
||
|
|
"grad_norm": 0.7209483483643438,
|
||
|
|
"learning_rate": 2.6741573033707867e-05,
|
||
|
|
"loss": 0.3172,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15206953883171082,
|
||
|
|
"step": 120,
|
||
|
|
"valid_targets_mean": 2721.6,
|
||
|
|
"valid_targets_min": 2167
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4930966469428008,
|
||
|
|
"grad_norm": 0.6751742169745769,
|
||
|
|
"learning_rate": 2.7865168539325845e-05,
|
||
|
|
"loss": 0.3129,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15052777528762817,
|
||
|
|
"step": 125,
|
||
|
|
"valid_targets_mean": 2621.8,
|
||
|
|
"valid_targets_min": 1920
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5128205128205128,
|
||
|
|
"grad_norm": 0.7960222284377589,
|
||
|
|
"learning_rate": 2.8988764044943823e-05,
|
||
|
|
"loss": 0.316,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16217561066150665,
|
||
|
|
"step": 130,
|
||
|
|
"valid_targets_mean": 2773.5,
|
||
|
|
"valid_targets_min": 2511
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5325443786982249,
|
||
|
|
"grad_norm": 0.7866288927965381,
|
||
|
|
"learning_rate": 3.0112359550561798e-05,
|
||
|
|
"loss": 0.3114,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16387704014778137,
|
||
|
|
"step": 135,
|
||
|
|
"valid_targets_mean": 2720.6,
|
||
|
|
"valid_targets_min": 2360
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5522682445759369,
|
||
|
|
"grad_norm": 0.7176058157182661,
|
||
|
|
"learning_rate": 3.123595505617978e-05,
|
||
|
|
"loss": 0.3181,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1524210423231125,
|
||
|
|
"step": 140,
|
||
|
|
"valid_targets_mean": 2728.1,
|
||
|
|
"valid_targets_min": 2534
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.571992110453649,
|
||
|
|
"grad_norm": 0.7113552439834236,
|
||
|
|
"learning_rate": 3.235955056179776e-05,
|
||
|
|
"loss": 0.2947,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1595868170261383,
|
||
|
|
"step": 145,
|
||
|
|
"valid_targets_mean": 2730.5,
|
||
|
|
"valid_targets_min": 2476
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.591715976331361,
|
||
|
|
"grad_norm": 0.7744977015455972,
|
||
|
|
"learning_rate": 3.3483146067415736e-05,
|
||
|
|
"loss": 0.2905,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1445167511701584,
|
||
|
|
"step": 150,
|
||
|
|
"valid_targets_mean": 2774.1,
|
||
|
|
"valid_targets_min": 2541
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.611439842209073,
|
||
|
|
"grad_norm": 0.7571681383641046,
|
||
|
|
"learning_rate": 3.4606741573033714e-05,
|
||
|
|
"loss": 0.2919,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16327649354934692,
|
||
|
|
"step": 155,
|
||
|
|
"valid_targets_mean": 2844.4,
|
||
|
|
"valid_targets_min": 2593
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.631163708086785,
|
||
|
|
"grad_norm": 0.7803287581910531,
|
||
|
|
"learning_rate": 3.5730337078651685e-05,
|
||
|
|
"loss": 0.306,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1442846804857254,
|
||
|
|
"step": 160,
|
||
|
|
"valid_targets_mean": 2570.9,
|
||
|
|
"valid_targets_min": 2145
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.650887573964497,
|
||
|
|
"grad_norm": 0.8394464526183439,
|
||
|
|
"learning_rate": 3.685393258426967e-05,
|
||
|
|
"loss": 0.2975,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15439441800117493,
|
||
|
|
"step": 165,
|
||
|
|
"valid_targets_mean": 2764.0,
|
||
|
|
"valid_targets_min": 2317
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6706114398422091,
|
||
|
|
"grad_norm": 0.7543433273299673,
|
||
|
|
"learning_rate": 3.797752808988764e-05,
|
||
|
|
"loss": 0.3027,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1542845368385315,
|
||
|
|
"step": 170,
|
||
|
|
"valid_targets_mean": 2810.1,
|
||
|
|
"valid_targets_min": 2719
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6903353057199211,
|
||
|
|
"grad_norm": 0.8937613139999523,
|
||
|
|
"learning_rate": 3.910112359550562e-05,
|
||
|
|
"loss": 0.3027,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14578217267990112,
|
||
|
|
"step": 175,
|
||
|
|
"valid_targets_mean": 2728.0,
|
||
|
|
"valid_targets_min": 2589
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7100591715976331,
|
||
|
|
"grad_norm": 0.7110187508256404,
|
||
|
|
"learning_rate": 3.999996144687019e-05,
|
||
|
|
"loss": 0.2877,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1280485987663269,
|
||
|
|
"step": 180,
|
||
|
|
"valid_targets_mean": 2443.9,
|
||
|
|
"valid_targets_min": 240
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7297830374753451,
|
||
|
|
"grad_norm": 0.7363870011556279,
|
||
|
|
"learning_rate": 3.9998612102933544e-05,
|
||
|
|
"loss": 0.2916,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14857065677642822,
|
||
|
|
"step": 185,
|
||
|
|
"valid_targets_mean": 2717.6,
|
||
|
|
"valid_targets_min": 2458
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7495069033530573,
|
||
|
|
"grad_norm": 0.6764273367127231,
|
||
|
|
"learning_rate": 3.999533525113843e-05,
|
||
|
|
"loss": 0.2937,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14113926887512207,
|
||
|
|
"step": 190,
|
||
|
|
"valid_targets_mean": 2722.1,
|
||
|
|
"valid_targets_min": 2345
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7692307692307693,
|
||
|
|
"grad_norm": 0.755744969056552,
|
||
|
|
"learning_rate": 3.9990131207314634e-05,
|
||
|
|
"loss": 0.2977,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14644595980644226,
|
||
|
|
"step": 195,
|
||
|
|
"valid_targets_mean": 2784.1,
|
||
|
|
"valid_targets_min": 2500
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7889546351084813,
|
||
|
|
"grad_norm": 0.7338663547329065,
|
||
|
|
"learning_rate": 3.998300047303874e-05,
|
||
|
|
"loss": 0.2814,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13639897108078003,
|
||
|
|
"step": 200,
|
||
|
|
"valid_targets_mean": 2771.5,
|
||
|
|
"valid_targets_min": 2168
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8086785009861933,
|
||
|
|
"grad_norm": 0.6553399865152889,
|
||
|
|
"learning_rate": 3.997394373558576e-05,
|
||
|
|
"loss": 0.2811,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1425131857395172,
|
||
|
|
"step": 205,
|
||
|
|
"valid_targets_mean": 2755.1,
|
||
|
|
"valid_targets_min": 2411
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8284023668639053,
|
||
|
|
"grad_norm": 0.7077939310572914,
|
||
|
|
"learning_rate": 3.9962961867862894e-05,
|
||
|
|
"loss": 0.2869,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14431126415729523,
|
||
|
|
"step": 210,
|
||
|
|
"valid_targets_mean": 2430.6,
|
||
|
|
"valid_targets_min": 228
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8481262327416174,
|
||
|
|
"grad_norm": 0.650932809539881,
|
||
|
|
"learning_rate": 3.995005592832541e-05,
|
||
|
|
"loss": 0.271,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12961333990097046,
|
||
|
|
"step": 215,
|
||
|
|
"valid_targets_mean": 2659.9,
|
||
|
|
"valid_targets_min": 2483
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8678500986193294,
|
||
|
|
"grad_norm": 0.6706417952121615,
|
||
|
|
"learning_rate": 3.993522716087462e-05,
|
||
|
|
"loss": 0.2969,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14225119352340698,
|
||
|
|
"step": 220,
|
||
|
|
"valid_targets_mean": 2807.0,
|
||
|
|
"valid_targets_min": 2394
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8875739644970414,
|
||
|
|
"grad_norm": 0.6308353785376156,
|
||
|
|
"learning_rate": 3.991847699473801e-05,
|
||
|
|
"loss": 0.2757,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15311525762081146,
|
||
|
|
"step": 225,
|
||
|
|
"valid_targets_mean": 2949.0,
|
||
|
|
"valid_targets_min": 2230
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9072978303747534,
|
||
|
|
"grad_norm": 0.6560017144328685,
|
||
|
|
"learning_rate": 3.989980704433144e-05,
|
||
|
|
"loss": 0.2824,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1423225998878479,
|
||
|
|
"step": 230,
|
||
|
|
"valid_targets_mean": 2768.4,
|
||
|
|
"valid_targets_min": 2443
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9270216962524654,
|
||
|
|
"grad_norm": 0.6901957285417993,
|
||
|
|
"learning_rate": 3.98792191091036e-05,
|
||
|
|
"loss": 0.2826,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1249314695596695,
|
||
|
|
"step": 235,
|
||
|
|
"valid_targets_mean": 2661.4,
|
||
|
|
"valid_targets_min": 2291
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9467455621301775,
|
||
|
|
"grad_norm": 0.7286556164294888,
|
||
|
|
"learning_rate": 3.9856715173362527e-05,
|
||
|
|
"loss": 0.2732,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13277488946914673,
|
||
|
|
"step": 240,
|
||
|
|
"valid_targets_mean": 2377.2,
|
||
|
|
"valid_targets_min": 221
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9664694280078896,
|
||
|
|
"grad_norm": 0.6287868667926823,
|
||
|
|
"learning_rate": 3.9832297406084386e-05,
|
||
|
|
"loss": 0.2839,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.138014018535614,
|
||
|
|
"step": 245,
|
||
|
|
"valid_targets_mean": 2786.1,
|
||
|
|
"valid_targets_min": 2591
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9861932938856016,
|
||
|
|
"grad_norm": 0.6720514507470646,
|
||
|
|
"learning_rate": 3.980596816070442e-05,
|
||
|
|
"loss": 0.2799,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1344606578350067,
|
||
|
|
"step": 250,
|
||
|
|
"valid_targets_mean": 2749.8,
|
||
|
|
"valid_targets_min": 2283
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0039447731755424,
|
||
|
|
"grad_norm": 0.6216735135471295,
|
||
|
|
"learning_rate": 3.97777299748901e-05,
|
||
|
|
"loss": 0.2848,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1504630446434021,
|
||
|
|
"step": 255,
|
||
|
|
"valid_targets_mean": 2742.8,
|
||
|
|
"valid_targets_min": 1646
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0236686390532543,
|
||
|
|
"grad_norm": 0.7023669128134461,
|
||
|
|
"learning_rate": 3.974758557029653e-05,
|
||
|
|
"loss": 0.2858,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1299905627965927,
|
||
|
|
"step": 260,
|
||
|
|
"valid_targets_mean": 2617.8,
|
||
|
|
"valid_targets_min": 1887
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0433925049309665,
|
||
|
|
"grad_norm": 0.627595278839174,
|
||
|
|
"learning_rate": 3.971553785230418e-05,
|
||
|
|
"loss": 0.2767,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13176056742668152,
|
||
|
|
"step": 265,
|
||
|
|
"valid_targets_mean": 2869.8,
|
||
|
|
"valid_targets_min": 2674
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0631163708086786,
|
||
|
|
"grad_norm": 0.6391262041866768,
|
||
|
|
"learning_rate": 3.968158990973881e-05,
|
||
|
|
"loss": 0.273,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13228456676006317,
|
||
|
|
"step": 270,
|
||
|
|
"valid_targets_mean": 2691.6,
|
||
|
|
"valid_targets_min": 2353
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0828402366863905,
|
||
|
|
"grad_norm": 0.6018288963476707,
|
||
|
|
"learning_rate": 3.964574501457378e-05,
|
||
|
|
"loss": 0.2679,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13728728890419006,
|
||
|
|
"step": 275,
|
||
|
|
"valid_targets_mean": 2838.5,
|
||
|
|
"valid_targets_min": 2392
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1025641025641026,
|
||
|
|
"grad_norm": 0.7097189471016262,
|
||
|
|
"learning_rate": 3.960800662161469e-05,
|
||
|
|
"loss": 0.2701,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14349806308746338,
|
||
|
|
"step": 280,
|
||
|
|
"valid_targets_mean": 2759.0,
|
||
|
|
"valid_targets_min": 2301
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1222879684418146,
|
||
|
|
"grad_norm": 0.7107508454473763,
|
||
|
|
"learning_rate": 3.9568378368166406e-05,
|
||
|
|
"loss": 0.2671,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12640471756458282,
|
||
|
|
"step": 285,
|
||
|
|
"valid_targets_mean": 2651.2,
|
||
|
|
"valid_targets_min": 2458
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1420118343195267,
|
||
|
|
"grad_norm": 0.6821928728392587,
|
||
|
|
"learning_rate": 3.952686407368247e-05,
|
||
|
|
"loss": 0.2664,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15707197785377502,
|
||
|
|
"step": 290,
|
||
|
|
"valid_targets_mean": 2943.6,
|
||
|
|
"valid_targets_min": 2680
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1617357001972386,
|
||
|
|
"grad_norm": 0.6379847458458735,
|
||
|
|
"learning_rate": 3.948346773939699e-05,
|
||
|
|
"loss": 0.2827,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1453888714313507,
|
||
|
|
"step": 295,
|
||
|
|
"valid_targets_mean": 2814.9,
|
||
|
|
"valid_targets_min": 2416
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1814595660749507,
|
||
|
|
"grad_norm": 0.6377131340239722,
|
||
|
|
"learning_rate": 3.943819354793899e-05,
|
||
|
|
"loss": 0.2644,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1433432400226593,
|
||
|
|
"step": 300,
|
||
|
|
"valid_targets_mean": 2875.2,
|
||
|
|
"valid_targets_min": 2594
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2011834319526626,
|
||
|
|
"grad_norm": 0.6190195081369322,
|
||
|
|
"learning_rate": 3.9391045862929275e-05,
|
||
|
|
"loss": 0.2634,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1330021172761917,
|
||
|
|
"step": 305,
|
||
|
|
"valid_targets_mean": 2894.2,
|
||
|
|
"valid_targets_min": 2462
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2209072978303748,
|
||
|
|
"grad_norm": 0.6879916952105728,
|
||
|
|
"learning_rate": 3.934202922855985e-05,
|
||
|
|
"loss": 0.2665,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12800386548042297,
|
||
|
|
"step": 310,
|
||
|
|
"valid_targets_mean": 2512.1,
|
||
|
|
"valid_targets_min": 899
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2406311637080867,
|
||
|
|
"grad_norm": 0.6740962784039973,
|
||
|
|
"learning_rate": 3.9291148369155964e-05,
|
||
|
|
"loss": 0.2636,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1355883628129959,
|
||
|
|
"step": 315,
|
||
|
|
"valid_targets_mean": 2702.9,
|
||
|
|
"valid_targets_min": 2145
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2603550295857988,
|
||
|
|
"grad_norm": 0.630679211289767,
|
||
|
|
"learning_rate": 3.9238408188720745e-05,
|
||
|
|
"loss": 0.2697,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11943496763706207,
|
||
|
|
"step": 320,
|
||
|
|
"valid_targets_mean": 2745.5,
|
||
|
|
"valid_targets_min": 2205
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.280078895463511,
|
||
|
|
"grad_norm": 0.6849545453855597,
|
||
|
|
"learning_rate": 3.918381377046255e-05,
|
||
|
|
"loss": 0.2671,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11861124634742737,
|
||
|
|
"step": 325,
|
||
|
|
"valid_targets_mean": 2480.1,
|
||
|
|
"valid_targets_min": 913
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2998027613412229,
|
||
|
|
"grad_norm": 0.7034620049207961,
|
||
|
|
"learning_rate": 3.9127370376305045e-05,
|
||
|
|
"loss": 0.2647,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1283608376979828,
|
||
|
|
"step": 330,
|
||
|
|
"valid_targets_mean": 2704.1,
|
||
|
|
"valid_targets_min": 2510
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.319526627218935,
|
||
|
|
"grad_norm": 0.7393438012601564,
|
||
|
|
"learning_rate": 3.906908344638002e-05,
|
||
|
|
"loss": 0.2691,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12500716745853424,
|
||
|
|
"step": 335,
|
||
|
|
"valid_targets_mean": 2659.1,
|
||
|
|
"valid_targets_min": 2330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.339250493096647,
|
||
|
|
"grad_norm": 0.6742648960134628,
|
||
|
|
"learning_rate": 3.900895859850313e-05,
|
||
|
|
"loss": 0.2606,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12485789507627487,
|
||
|
|
"step": 340,
|
||
|
|
"valid_targets_mean": 2801.2,
|
||
|
|
"valid_targets_min": 2541
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.358974358974359,
|
||
|
|
"grad_norm": 0.6387775997430102,
|
||
|
|
"learning_rate": 3.8947001627632326e-05,
|
||
|
|
"loss": 0.2617,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12518510222434998,
|
||
|
|
"step": 345,
|
||
|
|
"valid_targets_mean": 2819.5,
|
||
|
|
"valid_targets_min": 2344
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.378698224852071,
|
||
|
|
"grad_norm": 0.8383430820622142,
|
||
|
|
"learning_rate": 3.888321850530943e-05,
|
||
|
|
"loss": 0.2723,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16160714626312256,
|
||
|
|
"step": 350,
|
||
|
|
"valid_targets_mean": 3243.1,
|
||
|
|
"valid_targets_min": 2351
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.398422090729783,
|
||
|
|
"grad_norm": 0.7672261457406931,
|
||
|
|
"learning_rate": 3.8817615379084514e-05,
|
||
|
|
"loss": 0.2639,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1340160369873047,
|
||
|
|
"step": 355,
|
||
|
|
"valid_targets_mean": 2834.5,
|
||
|
|
"valid_targets_min": 2638
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.418145956607495,
|
||
|
|
"grad_norm": 0.6046114376107636,
|
||
|
|
"learning_rate": 3.875019857192343e-05,
|
||
|
|
"loss": 0.2661,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12131619453430176,
|
||
|
|
"step": 360,
|
||
|
|
"valid_targets_mean": 2686.1,
|
||
|
|
"valid_targets_min": 1920
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4378698224852071,
|
||
|
|
"grad_norm": 0.7190305689143975,
|
||
|
|
"learning_rate": 3.8680974581598375e-05,
|
||
|
|
"loss": 0.2713,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17544794082641602,
|
||
|
|
"step": 365,
|
||
|
|
"valid_targets_mean": 3006.9,
|
||
|
|
"valid_targets_min": 2644
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4575936883629192,
|
||
|
|
"grad_norm": 0.6812402702902784,
|
||
|
|
"learning_rate": 3.860995008006161e-05,
|
||
|
|
"loss": 0.2798,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13226604461669922,
|
||
|
|
"step": 370,
|
||
|
|
"valid_targets_mean": 2752.2,
|
||
|
|
"valid_targets_min": 2382
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4773175542406312,
|
||
|
|
"grad_norm": 0.6200588811967641,
|
||
|
|
"learning_rate": 3.853713191280242e-05,
|
||
|
|
"loss": 0.2575,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12865135073661804,
|
||
|
|
"step": 375,
|
||
|
|
"valid_targets_mean": 2844.6,
|
||
|
|
"valid_targets_min": 2613
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.497041420118343,
|
||
|
|
"grad_norm": 1.5786547210414617,
|
||
|
|
"learning_rate": 3.846252709818733e-05,
|
||
|
|
"loss": 0.2563,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1177363470196724,
|
||
|
|
"step": 380,
|
||
|
|
"valid_targets_mean": 2440.1,
|
||
|
|
"valid_targets_min": 236
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5167652859960552,
|
||
|
|
"grad_norm": 0.6742498855124531,
|
||
|
|
"learning_rate": 3.8386142826783645e-05,
|
||
|
|
"loss": 0.2643,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13844269514083862,
|
||
|
|
"step": 385,
|
||
|
|
"valid_targets_mean": 2915.5,
|
||
|
|
"valid_targets_min": 2398
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5364891518737673,
|
||
|
|
"grad_norm": 0.6142760611695112,
|
||
|
|
"learning_rate": 3.830798646066642e-05,
|
||
|
|
"loss": 0.2619,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12407414615154266,
|
||
|
|
"step": 390,
|
||
|
|
"valid_targets_mean": 2409.5,
|
||
|
|
"valid_targets_min": 228
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5562130177514792,
|
||
|
|
"grad_norm": 0.5934316542899878,
|
||
|
|
"learning_rate": 3.8228065532708905e-05,
|
||
|
|
"loss": 0.265,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11504469811916351,
|
||
|
|
"step": 395,
|
||
|
|
"valid_targets_mean": 2382.8,
|
||
|
|
"valid_targets_min": 553
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5759368836291914,
|
||
|
|
"grad_norm": 0.5839005281410029,
|
||
|
|
"learning_rate": 3.814638774585648e-05,
|
||
|
|
"loss": 0.2565,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12977270781993866,
|
||
|
|
"step": 400,
|
||
|
|
"valid_targets_mean": 2765.8,
|
||
|
|
"valid_targets_min": 2210
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5956607495069033,
|
||
|
|
"grad_norm": 0.6163101497779825,
|
||
|
|
"learning_rate": 3.8062960972384223e-05,
|
||
|
|
"loss": 0.2552,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12079758942127228,
|
||
|
|
"step": 405,
|
||
|
|
"valid_targets_mean": 2386.1,
|
||
|
|
"valid_targets_min": 1067
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6153846153846154,
|
||
|
|
"grad_norm": 0.6175553569803202,
|
||
|
|
"learning_rate": 3.797779325313822e-05,
|
||
|
|
"loss": 0.259,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1276577264070511,
|
||
|
|
"step": 410,
|
||
|
|
"valid_targets_mean": 2598.8,
|
||
|
|
"valid_targets_min": 910
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6351084812623276,
|
||
|
|
"grad_norm": 0.6102981011446369,
|
||
|
|
"learning_rate": 3.78908927967605e-05,
|
||
|
|
"loss": 0.261,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14956462383270264,
|
||
|
|
"step": 415,
|
||
|
|
"valid_targets_mean": 2336.2,
|
||
|
|
"valid_targets_min": 620
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6548323471400395,
|
||
|
|
"grad_norm": 0.6569786250810341,
|
||
|
|
"learning_rate": 3.780226797889794e-05,
|
||
|
|
"loss": 0.2551,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13197916746139526,
|
||
|
|
"step": 420,
|
||
|
|
"valid_targets_mean": 2484.6,
|
||
|
|
"valid_targets_min": 235
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6745562130177514,
|
||
|
|
"grad_norm": 0.6137456740692272,
|
||
|
|
"learning_rate": 3.7711927341394916e-05,
|
||
|
|
"loss": 0.2623,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13518665730953217,
|
||
|
|
"step": 425,
|
||
|
|
"valid_targets_mean": 2763.5,
|
||
|
|
"valid_targets_min": 2315
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6942800788954635,
|
||
|
|
"grad_norm": 0.5716523540798796,
|
||
|
|
"learning_rate": 3.761987959147012e-05,
|
||
|
|
"loss": 0.2659,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12936533987522125,
|
||
|
|
"step": 430,
|
||
|
|
"valid_targets_mean": 2739.1,
|
||
|
|
"valid_targets_min": 2433
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7140039447731756,
|
||
|
|
"grad_norm": 0.6673338732342154,
|
||
|
|
"learning_rate": 3.7526133600877275e-05,
|
||
|
|
"loss": 0.2624,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13679540157318115,
|
||
|
|
"step": 435,
|
||
|
|
"valid_targets_mean": 2684.2,
|
||
|
|
"valid_targets_min": 2340
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7337278106508875,
|
||
|
|
"grad_norm": 0.654222096054993,
|
||
|
|
"learning_rate": 3.743069840505006e-05,
|
||
|
|
"loss": 0.2561,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12495963275432587,
|
||
|
|
"step": 440,
|
||
|
|
"valid_targets_mean": 2685.6,
|
||
|
|
"valid_targets_min": 2258
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7534516765285995,
|
||
|
|
"grad_norm": 0.5959377297017336,
|
||
|
|
"learning_rate": 3.733358320223128e-05,
|
||
|
|
"loss": 0.2555,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1379990577697754,
|
||
|
|
"step": 445,
|
||
|
|
"valid_targets_mean": 2889.8,
|
||
|
|
"valid_targets_min": 2521
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7731755424063116,
|
||
|
|
"grad_norm": 0.5974680334880117,
|
||
|
|
"learning_rate": 3.723479735258633e-05,
|
||
|
|
"loss": 0.2527,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1232835054397583,
|
||
|
|
"step": 450,
|
||
|
|
"valid_targets_mean": 2724.5,
|
||
|
|
"valid_targets_min": 2502
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7928994082840237,
|
||
|
|
"grad_norm": 0.598871857199513,
|
||
|
|
"learning_rate": 3.7134350377301e-05,
|
||
|
|
"loss": 0.2542,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1274440735578537,
|
||
|
|
"step": 455,
|
||
|
|
"valid_targets_mean": 2839.9,
|
||
|
|
"valid_targets_min": 2570
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8126232741617359,
|
||
|
|
"grad_norm": 0.54931006542639,
|
||
|
|
"learning_rate": 3.703225195766382e-05,
|
||
|
|
"loss": 0.2688,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13848382234573364,
|
||
|
|
"step": 460,
|
||
|
|
"valid_targets_mean": 2837.4,
|
||
|
|
"valid_targets_min": 2544
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8323471400394478,
|
||
|
|
"grad_norm": 0.6190858755325419,
|
||
|
|
"learning_rate": 3.692851193413299e-05,
|
||
|
|
"loss": 0.2556,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1246803030371666,
|
||
|
|
"step": 465,
|
||
|
|
"valid_targets_mean": 2657.6,
|
||
|
|
"valid_targets_min": 2297
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8520710059171597,
|
||
|
|
"grad_norm": 0.638850692251179,
|
||
|
|
"learning_rate": 3.682314030538788e-05,
|
||
|
|
"loss": 0.2518,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12336134910583496,
|
||
|
|
"step": 470,
|
||
|
|
"valid_targets_mean": 2636.2,
|
||
|
|
"valid_targets_min": 2443
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8717948717948718,
|
||
|
|
"grad_norm": 0.5693768334335082,
|
||
|
|
"learning_rate": 3.671614722736541e-05,
|
||
|
|
"loss": 0.2529,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13880418241024017,
|
||
|
|
"step": 475,
|
||
|
|
"valid_targets_mean": 2795.6,
|
||
|
|
"valid_targets_min": 2472
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.891518737672584,
|
||
|
|
"grad_norm": 0.5630668072866708,
|
||
|
|
"learning_rate": 3.6607543012281106e-05,
|
||
|
|
"loss": 0.251,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12563151121139526,
|
||
|
|
"step": 480,
|
||
|
|
"valid_targets_mean": 2829.8,
|
||
|
|
"valid_targets_min": 2406
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9112426035502958,
|
||
|
|
"grad_norm": 0.5844725715654421,
|
||
|
|
"learning_rate": 3.649733812763527e-05,
|
||
|
|
"loss": 0.2598,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12985971570014954,
|
||
|
|
"step": 485,
|
||
|
|
"valid_targets_mean": 2756.2,
|
||
|
|
"valid_targets_min": 2315
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9309664694280078,
|
||
|
|
"grad_norm": 0.5899966158069249,
|
||
|
|
"learning_rate": 3.638554319520406e-05,
|
||
|
|
"loss": 0.2648,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11524569988250732,
|
||
|
|
"step": 490,
|
||
|
|
"valid_targets_mean": 2704.6,
|
||
|
|
"valid_targets_min": 2488
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.95069033530572,
|
||
|
|
"grad_norm": 0.5293855483312215,
|
||
|
|
"learning_rate": 3.627216899001575e-05,
|
||
|
|
"loss": 0.2617,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12094192951917648,
|
||
|
|
"step": 495,
|
||
|
|
"valid_targets_mean": 2867.8,
|
||
|
|
"valid_targets_min": 2659
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.970414201183432,
|
||
|
|
"grad_norm": 0.5729667238870303,
|
||
|
|
"learning_rate": 3.6157226439312186e-05,
|
||
|
|
"loss": 0.2519,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12284328043460846,
|
||
|
|
"step": 500,
|
||
|
|
"valid_targets_mean": 2739.4,
|
||
|
|
"valid_targets_min": 2399
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.990138067061144,
|
||
|
|
"grad_norm": 0.6992688916817632,
|
||
|
|
"learning_rate": 3.604072662149567e-05,
|
||
|
|
"loss": 0.261,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1423102617263794,
|
||
|
|
"step": 505,
|
||
|
|
"valid_targets_mean": 2806.8,
|
||
|
|
"valid_targets_min": 2405
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.007889546351085,
|
||
|
|
"grad_norm": 0.6221341048350661,
|
||
|
|
"learning_rate": 3.5922680765061096e-05,
|
||
|
|
"loss": 0.2585,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12077319622039795,
|
||
|
|
"step": 510,
|
||
|
|
"valid_targets_mean": 2711.0,
|
||
|
|
"valid_targets_min": 2457
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.027613412228797,
|
||
|
|
"grad_norm": 0.6672089218604331,
|
||
|
|
"learning_rate": 3.580310024751381e-05,
|
||
|
|
"loss": 0.2458,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10570112615823746,
|
||
|
|
"step": 515,
|
||
|
|
"valid_targets_mean": 2262.9,
|
||
|
|
"valid_targets_min": 243
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0473372781065087,
|
||
|
|
"grad_norm": 0.5945800695837705,
|
||
|
|
"learning_rate": 3.568199659427298e-05,
|
||
|
|
"loss": 0.2454,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12192805111408234,
|
||
|
|
"step": 520,
|
||
|
|
"valid_targets_mean": 2743.1,
|
||
|
|
"valid_targets_min": 2329
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.067061143984221,
|
||
|
|
"grad_norm": 0.6274599334545877,
|
||
|
|
"learning_rate": 3.555938147756077e-05,
|
||
|
|
"loss": 0.2447,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11958570778369904,
|
||
|
|
"step": 525,
|
||
|
|
"valid_targets_mean": 2665.5,
|
||
|
|
"valid_targets_min": 2330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.086785009861933,
|
||
|
|
"grad_norm": 0.6385125746161686,
|
||
|
|
"learning_rate": 3.543526671527733e-05,
|
||
|
|
"loss": 0.2428,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11740340292453766,
|
||
|
|
"step": 530,
|
||
|
|
"valid_targets_mean": 2532.0,
|
||
|
|
"valid_targets_min": 758
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.106508875739645,
|
||
|
|
"grad_norm": 0.6831770036970266,
|
||
|
|
"learning_rate": 3.530966426986177e-05,
|
||
|
|
"loss": 0.24,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12677040696144104,
|
||
|
|
"step": 535,
|
||
|
|
"valid_targets_mean": 2766.4,
|
||
|
|
"valid_targets_min": 2546
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.126232741617357,
|
||
|
|
"grad_norm": 0.6794683854521593,
|
||
|
|
"learning_rate": 3.51825862471392e-05,
|
||
|
|
"loss": 0.2463,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13011988997459412,
|
||
|
|
"step": 540,
|
||
|
|
"valid_targets_mean": 2542.5,
|
||
|
|
"valid_targets_min": 726
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.145956607495069,
|
||
|
|
"grad_norm": 0.6063040383627043,
|
||
|
|
"learning_rate": 3.505404489515394e-05,
|
||
|
|
"loss": 0.2469,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12662534415721893,
|
||
|
|
"step": 545,
|
||
|
|
"valid_targets_mean": 2856.4,
|
||
|
|
"valid_targets_min": 2676
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.165680473372781,
|
||
|
|
"grad_norm": 0.5649605070606484,
|
||
|
|
"learning_rate": 3.492405260298905e-05,
|
||
|
|
"loss": 0.2438,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11776059120893478,
|
||
|
|
"step": 550,
|
||
|
|
"valid_targets_mean": 2867.5,
|
||
|
|
"valid_targets_min": 2628
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.185404339250493,
|
||
|
|
"grad_norm": 1.764427015727064,
|
||
|
|
"learning_rate": 3.47926218995722e-05,
|
||
|
|
"loss": 0.2426,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12325219810009003,
|
||
|
|
"step": 555,
|
||
|
|
"valid_targets_mean": 2648.9,
|
||
|
|
"valid_targets_min": 2295
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2051282051282053,
|
||
|
|
"grad_norm": 0.5735181697722106,
|
||
|
|
"learning_rate": 3.465976545246813e-05,
|
||
|
|
"loss": 0.2395,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12521851062774658,
|
||
|
|
"step": 560,
|
||
|
|
"valid_targets_mean": 2876.0,
|
||
|
|
"valid_targets_min": 2703
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.224852071005917,
|
||
|
|
"grad_norm": 0.560705297460117,
|
||
|
|
"learning_rate": 3.4525496066657735e-05,
|
||
|
|
"loss": 0.2452,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12072977423667908,
|
||
|
|
"step": 565,
|
||
|
|
"valid_targets_mean": 2887.6,
|
||
|
|
"valid_targets_min": 2711
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.244575936883629,
|
||
|
|
"grad_norm": 0.6102638099785221,
|
||
|
|
"learning_rate": 3.438982668330388e-05,
|
||
|
|
"loss": 0.249,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11241171509027481,
|
||
|
|
"step": 570,
|
||
|
|
"valid_targets_mean": 2786.4,
|
||
|
|
"valid_targets_min": 2563
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2642998027613412,
|
||
|
|
"grad_norm": 0.6049612103403902,
|
||
|
|
"learning_rate": 3.425277037850411e-05,
|
||
|
|
"loss": 0.2425,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12193508446216583,
|
||
|
|
"step": 575,
|
||
|
|
"valid_targets_mean": 2775.0,
|
||
|
|
"valid_targets_min": 2621
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2840236686390534,
|
||
|
|
"grad_norm": 0.6022071640874221,
|
||
|
|
"learning_rate": 3.411434036203035e-05,
|
||
|
|
"loss": 0.2432,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13640540838241577,
|
||
|
|
"step": 580,
|
||
|
|
"valid_targets_mean": 2792.4,
|
||
|
|
"valid_targets_min": 2348
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3037475345167655,
|
||
|
|
"grad_norm": 0.5810753725582615,
|
||
|
|
"learning_rate": 3.397454997605569e-05,
|
||
|
|
"loss": 0.2552,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13802054524421692,
|
||
|
|
"step": 585,
|
||
|
|
"valid_targets_mean": 2932.6,
|
||
|
|
"valid_targets_min": 2535
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.323471400394477,
|
||
|
|
"grad_norm": 0.7404569214228531,
|
||
|
|
"learning_rate": 3.38334126938685e-05,
|
||
|
|
"loss": 0.2494,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12563802301883698,
|
||
|
|
"step": 590,
|
||
|
|
"valid_targets_mean": 2703.0,
|
||
|
|
"valid_targets_min": 2317
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3431952662721893,
|
||
|
|
"grad_norm": 0.6250161744687328,
|
||
|
|
"learning_rate": 3.369094211857378e-05,
|
||
|
|
"loss": 0.2524,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1233794167637825,
|
||
|
|
"step": 595,
|
||
|
|
"valid_targets_mean": 2784.2,
|
||
|
|
"valid_targets_min": 2482
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3629191321499015,
|
||
|
|
"grad_norm": 0.638961484209998,
|
||
|
|
"learning_rate": 3.354715198178211e-05,
|
||
|
|
"loss": 0.2347,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.127358078956604,
|
||
|
|
"step": 600,
|
||
|
|
"valid_targets_mean": 2927.0,
|
||
|
|
"valid_targets_min": 2697
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3826429980276136,
|
||
|
|
"grad_norm": 0.6122052748591963,
|
||
|
|
"learning_rate": 3.3402056142286156e-05,
|
||
|
|
"loss": 0.2522,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16112154722213745,
|
||
|
|
"step": 605,
|
||
|
|
"valid_targets_mean": 3429.1,
|
||
|
|
"valid_targets_min": 2534
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4023668639053253,
|
||
|
|
"grad_norm": 0.5734915614532267,
|
||
|
|
"learning_rate": 3.3255668584724916e-05,
|
||
|
|
"loss": 0.2431,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12195591628551483,
|
||
|
|
"step": 610,
|
||
|
|
"valid_targets_mean": 2600.1,
|
||
|
|
"valid_targets_min": 840
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4220907297830374,
|
||
|
|
"grad_norm": 0.5442610781514553,
|
||
|
|
"learning_rate": 3.310800341823588e-05,
|
||
|
|
"loss": 0.2425,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11694778501987457,
|
||
|
|
"step": 615,
|
||
|
|
"valid_targets_mean": 2716.8,
|
||
|
|
"valid_targets_min": 2239
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4418145956607495,
|
||
|
|
"grad_norm": 0.6551616501559627,
|
||
|
|
"learning_rate": 3.2959074875095125e-05,
|
||
|
|
"loss": 0.2394,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11789815872907639,
|
||
|
|
"step": 620,
|
||
|
|
"valid_targets_mean": 2736.6,
|
||
|
|
"valid_targets_min": 2187
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4615384615384617,
|
||
|
|
"grad_norm": 0.6044403330185528,
|
||
|
|
"learning_rate": 3.280889730934562e-05,
|
||
|
|
"loss": 0.2472,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12093187868595123,
|
||
|
|
"step": 625,
|
||
|
|
"valid_targets_mean": 2658.4,
|
||
|
|
"valid_targets_min": 2195
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4812623274161734,
|
||
|
|
"grad_norm": 0.5411528292248211,
|
||
|
|
"learning_rate": 3.265748519541372e-05,
|
||
|
|
"loss": 0.2382,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11785396933555603,
|
||
|
|
"step": 630,
|
||
|
|
"valid_targets_mean": 2707.8,
|
||
|
|
"valid_targets_min": 2418
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5009861932938855,
|
||
|
|
"grad_norm": 0.6966260160611261,
|
||
|
|
"learning_rate": 3.250485312671411e-05,
|
||
|
|
"loss": 0.2441,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12316495925188065,
|
||
|
|
"step": 635,
|
||
|
|
"valid_targets_mean": 2821.5,
|
||
|
|
"valid_targets_min": 2447
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5207100591715976,
|
||
|
|
"grad_norm": 0.6021365969005681,
|
||
|
|
"learning_rate": 3.2351015814243235e-05,
|
||
|
|
"loss": 0.2472,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12455820292234421,
|
||
|
|
"step": 640,
|
||
|
|
"valid_targets_mean": 2708.1,
|
||
|
|
"valid_targets_min": 2504
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5404339250493098,
|
||
|
|
"grad_norm": 0.5729979937442556,
|
||
|
|
"learning_rate": 3.219598808516148e-05,
|
||
|
|
"loss": 0.2422,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13334402441978455,
|
||
|
|
"step": 645,
|
||
|
|
"valid_targets_mean": 2879.1,
|
||
|
|
"valid_targets_min": 2543
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.560157790927022,
|
||
|
|
"grad_norm": 0.5598659209311649,
|
||
|
|
"learning_rate": 3.203978488136403e-05,
|
||
|
|
"loss": 0.2446,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11669357120990753,
|
||
|
|
"step": 650,
|
||
|
|
"valid_targets_mean": 2730.8,
|
||
|
|
"valid_targets_min": 2505
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5798816568047336,
|
||
|
|
"grad_norm": 0.7723773132305327,
|
||
|
|
"learning_rate": 3.188242125804078e-05,
|
||
|
|
"loss": 0.2387,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12836016714572906,
|
||
|
|
"step": 655,
|
||
|
|
"valid_targets_mean": 2736.9,
|
||
|
|
"valid_targets_min": 2502
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5996055226824457,
|
||
|
|
"grad_norm": 0.5730554174107936,
|
||
|
|
"learning_rate": 3.1723912382225267e-05,
|
||
|
|
"loss": 0.241,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11414071917533875,
|
||
|
|
"step": 660,
|
||
|
|
"valid_targets_mean": 2759.1,
|
||
|
|
"valid_targets_min": 2511
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.619329388560158,
|
||
|
|
"grad_norm": 0.6367242357202733,
|
||
|
|
"learning_rate": 3.156427353133286e-05,
|
||
|
|
"loss": 0.238,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13788752257823944,
|
||
|
|
"step": 665,
|
||
|
|
"valid_targets_mean": 2787.8,
|
||
|
|
"valid_targets_min": 2457
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.63905325443787,
|
||
|
|
"grad_norm": 0.6057637105210123,
|
||
|
|
"learning_rate": 3.140352009168828e-05,
|
||
|
|
"loss": 0.2448,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11477711796760559,
|
||
|
|
"step": 670,
|
||
|
|
"valid_targets_mean": 2766.8,
|
||
|
|
"valid_targets_min": 2214
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.658777120315582,
|
||
|
|
"grad_norm": 0.5906650366547789,
|
||
|
|
"learning_rate": 3.124166755704261e-05,
|
||
|
|
"loss": 0.2351,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11703641712665558,
|
||
|
|
"step": 675,
|
||
|
|
"valid_targets_mean": 2617.5,
|
||
|
|
"valid_targets_min": 2250
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.678500986193294,
|
||
|
|
"grad_norm": 0.5869264967459172,
|
||
|
|
"learning_rate": 3.1078731527080023e-05,
|
||
|
|
"loss": 0.251,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1244022399187088,
|
||
|
|
"step": 680,
|
||
|
|
"valid_targets_mean": 2780.6,
|
||
|
|
"valid_targets_min": 2341
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.698224852071006,
|
||
|
|
"grad_norm": 0.571135382606071,
|
||
|
|
"learning_rate": 3.09147277059142e-05,
|
||
|
|
"loss": 0.251,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11044244468212128,
|
||
|
|
"step": 685,
|
||
|
|
"valid_targets_mean": 2754.4,
|
||
|
|
"valid_targets_min": 2476
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.717948717948718,
|
||
|
|
"grad_norm": 0.5469209277435115,
|
||
|
|
"learning_rate": 3.074967190057478e-05,
|
||
|
|
"loss": 0.2392,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1059972494840622,
|
||
|
|
"step": 690,
|
||
|
|
"valid_targets_mean": 2745.6,
|
||
|
|
"valid_targets_min": 2344
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7376725838264298,
|
||
|
|
"grad_norm": 0.6079541029025164,
|
||
|
|
"learning_rate": 3.058358001948381e-05,
|
||
|
|
"loss": 0.2501,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1203327625989914,
|
||
|
|
"step": 695,
|
||
|
|
"valid_targets_mean": 2431.0,
|
||
|
|
"valid_targets_min": 234
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.757396449704142,
|
||
|
|
"grad_norm": 0.5391166254307808,
|
||
|
|
"learning_rate": 3.0416468070922504e-05,
|
||
|
|
"loss": 0.2396,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12614591419696808,
|
||
|
|
"step": 700,
|
||
|
|
"valid_targets_mean": 2981.0,
|
||
|
|
"valid_targets_min": 2517
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.777120315581854,
|
||
|
|
"grad_norm": 0.5817257471159339,
|
||
|
|
"learning_rate": 3.0248352161488267e-05,
|
||
|
|
"loss": 0.2343,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1196541041135788,
|
||
|
|
"step": 705,
|
||
|
|
"valid_targets_mean": 2630.2,
|
||
|
|
"valid_targets_min": 1887
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.796844181459566,
|
||
|
|
"grad_norm": 0.6261176080441442,
|
||
|
|
"learning_rate": 3.007924849454235e-05,
|
||
|
|
"loss": 0.2439,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1341407597064972,
|
||
|
|
"step": 710,
|
||
|
|
"valid_targets_mean": 2759.4,
|
||
|
|
"valid_targets_min": 2341
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8165680473372783,
|
||
|
|
"grad_norm": 0.605185365072888,
|
||
|
|
"learning_rate": 2.9909173368648154e-05,
|
||
|
|
"loss": 0.2367,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1208367869257927,
|
||
|
|
"step": 715,
|
||
|
|
"valid_targets_mean": 2754.5,
|
||
|
|
"valid_targets_min": 2155
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.83629191321499,
|
||
|
|
"grad_norm": 0.5493689799378015,
|
||
|
|
"learning_rate": 2.9738143176000287e-05,
|
||
|
|
"loss": 0.2516,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11383502185344696,
|
||
|
|
"step": 720,
|
||
|
|
"valid_targets_mean": 2710.9,
|
||
|
|
"valid_targets_min": 2368
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.856015779092702,
|
||
|
|
"grad_norm": 0.5299969146379165,
|
||
|
|
"learning_rate": 2.9566174400844692e-05,
|
||
|
|
"loss": 0.2587,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11179862171411514,
|
||
|
|
"step": 725,
|
||
|
|
"valid_targets_mean": 2444.0,
|
||
|
|
"valid_targets_min": 886
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8757396449704142,
|
||
|
|
"grad_norm": 0.5965890530251023,
|
||
|
|
"learning_rate": 2.9393283617889846e-05,
|
||
|
|
"loss": 0.2449,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13717195391654968,
|
||
|
|
"step": 730,
|
||
|
|
"valid_targets_mean": 2874.1,
|
||
|
|
"valid_targets_min": 2640
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8954635108481264,
|
||
|
|
"grad_norm": 0.5335421932313925,
|
||
|
|
"learning_rate": 2.921948749070925e-05,
|
||
|
|
"loss": 0.2386,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10603675246238708,
|
||
|
|
"step": 735,
|
||
|
|
"valid_targets_mean": 2866.2,
|
||
|
|
"valid_targets_min": 2405
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9151873767258385,
|
||
|
|
"grad_norm": 0.5057273245036429,
|
||
|
|
"learning_rate": 2.9044802770135375e-05,
|
||
|
|
"loss": 0.241,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12216819822788239,
|
||
|
|
"step": 740,
|
||
|
|
"valid_targets_mean": 2815.2,
|
||
|
|
"valid_targets_min": 2625
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.93491124260355,
|
||
|
|
"grad_norm": 0.7197990148862006,
|
||
|
|
"learning_rate": 2.886924629264517e-05,
|
||
|
|
"loss": 0.2385,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1158512681722641,
|
||
|
|
"step": 745,
|
||
|
|
"valid_targets_mean": 2808.5,
|
||
|
|
"valid_targets_min": 2715
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9546351084812623,
|
||
|
|
"grad_norm": 0.5905281842524923,
|
||
|
|
"learning_rate": 2.8692834978737328e-05,
|
||
|
|
"loss": 0.2317,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11869832873344421,
|
||
|
|
"step": 750,
|
||
|
|
"valid_targets_mean": 2660.6,
|
||
|
|
"valid_targets_min": 2202
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9743589743589745,
|
||
|
|
"grad_norm": 0.563217611962283,
|
||
|
|
"learning_rate": 2.8515585831301456e-05,
|
||
|
|
"loss": 0.2539,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13342517614364624,
|
||
|
|
"step": 755,
|
||
|
|
"valid_targets_mean": 2922.2,
|
||
|
|
"valid_targets_min": 2548
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.994082840236686,
|
||
|
|
"grad_norm": 0.5452748406173739,
|
||
|
|
"learning_rate": 2.83375159339793e-05,
|
||
|
|
"loss": 0.2424,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10088188201189041,
|
||
|
|
"step": 760,
|
||
|
|
"valid_targets_mean": 2390.6,
|
||
|
|
"valid_targets_min": 509
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0118343195266273,
|
||
|
|
"grad_norm": 0.5394878651172428,
|
||
|
|
"learning_rate": 2.8158642449518186e-05,
|
||
|
|
"loss": 0.2358,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11956919729709625,
|
||
|
|
"step": 765,
|
||
|
|
"valid_targets_mean": 2932.8,
|
||
|
|
"valid_targets_min": 1763
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0315581854043394,
|
||
|
|
"grad_norm": 0.5770540952719748,
|
||
|
|
"learning_rate": 2.797898261811685e-05,
|
||
|
|
"loss": 0.2283,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11923639476299286,
|
||
|
|
"step": 770,
|
||
|
|
"valid_targets_mean": 2627.2,
|
||
|
|
"valid_targets_min": 1887
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.051282051282051,
|
||
|
|
"grad_norm": 0.5658145425694887,
|
||
|
|
"learning_rate": 2.7798553755763768e-05,
|
||
|
|
"loss": 0.2274,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11584222316741943,
|
||
|
|
"step": 775,
|
||
|
|
"valid_targets_mean": 2777.5,
|
||
|
|
"valid_targets_min": 2284
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0710059171597632,
|
||
|
|
"grad_norm": 0.5974928741375357,
|
||
|
|
"learning_rate": 2.7617373252568237e-05,
|
||
|
|
"loss": 0.2293,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11117218434810638,
|
||
|
|
"step": 780,
|
||
|
|
"valid_targets_mean": 2722.1,
|
||
|
|
"valid_targets_min": 2495
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0907297830374754,
|
||
|
|
"grad_norm": 0.5709461459187191,
|
||
|
|
"learning_rate": 2.7435458571084247e-05,
|
||
|
|
"loss": 0.2276,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10236341506242752,
|
||
|
|
"step": 785,
|
||
|
|
"valid_targets_mean": 2633.5,
|
||
|
|
"valid_targets_min": 2179
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1104536489151875,
|
||
|
|
"grad_norm": 0.6035740944098462,
|
||
|
|
"learning_rate": 2.725282724462743e-05,
|
||
|
|
"loss": 0.2243,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13170495629310608,
|
||
|
|
"step": 790,
|
||
|
|
"valid_targets_mean": 2718.8,
|
||
|
|
"valid_targets_min": 2407
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1301775147928996,
|
||
|
|
"grad_norm": 0.5313416799785263,
|
||
|
|
"learning_rate": 2.7069496875585145e-05,
|
||
|
|
"loss": 0.2306,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11041152477264404,
|
||
|
|
"step": 795,
|
||
|
|
"valid_targets_mean": 2704.9,
|
||
|
|
"valid_targets_min": 2231
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1499013806706113,
|
||
|
|
"grad_norm": 0.533414697398838,
|
||
|
|
"learning_rate": 2.688548513371994e-05,
|
||
|
|
"loss": 0.2308,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11932270228862762,
|
||
|
|
"step": 800,
|
||
|
|
"valid_targets_mean": 2894.5,
|
||
|
|
"valid_targets_min": 2295
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1696252465483234,
|
||
|
|
"grad_norm": 0.5705734439631395,
|
||
|
|
"learning_rate": 2.670080975446648e-05,
|
||
|
|
"loss": 0.2294,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10984625667333603,
|
||
|
|
"step": 805,
|
||
|
|
"valid_targets_mean": 2716.4,
|
||
|
|
"valid_targets_min": 2142
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1893491124260356,
|
||
|
|
"grad_norm": 0.56902879031506,
|
||
|
|
"learning_rate": 2.6515488537222198e-05,
|
||
|
|
"loss": 0.2246,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10160450637340546,
|
||
|
|
"step": 810,
|
||
|
|
"valid_targets_mean": 2734.4,
|
||
|
|
"valid_targets_min": 2207
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2090729783037477,
|
||
|
|
"grad_norm": 0.5452617893499204,
|
||
|
|
"learning_rate": 2.6329539343631725e-05,
|
||
|
|
"loss": 0.2249,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11202013492584229,
|
||
|
|
"step": 815,
|
||
|
|
"valid_targets_mean": 2996.8,
|
||
|
|
"valid_targets_min": 2679
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2287968441814594,
|
||
|
|
"grad_norm": 0.5972693217641168,
|
||
|
|
"learning_rate": 2.614298009586536e-05,
|
||
|
|
"loss": 0.2345,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11843322962522507,
|
||
|
|
"step": 820,
|
||
|
|
"valid_targets_mean": 2804.8,
|
||
|
|
"valid_targets_min": 2498
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2485207100591715,
|
||
|
|
"grad_norm": 0.5564823912843192,
|
||
|
|
"learning_rate": 2.595582877489171e-05,
|
||
|
|
"loss": 0.2322,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1069445013999939,
|
||
|
|
"step": 825,
|
||
|
|
"valid_targets_mean": 2702.0,
|
||
|
|
"valid_targets_min": 2318
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2682445759368837,
|
||
|
|
"grad_norm": 0.5784524393606165,
|
||
|
|
"learning_rate": 2.57681034187446e-05,
|
||
|
|
"loss": 0.2253,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11675705760717392,
|
||
|
|
"step": 830,
|
||
|
|
"valid_targets_mean": 2659.9,
|
||
|
|
"valid_targets_min": 525
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.287968441814596,
|
||
|
|
"grad_norm": 0.6417570973894795,
|
||
|
|
"learning_rate": 2.557982212078459e-05,
|
||
|
|
"loss": 0.2255,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10863907635211945,
|
||
|
|
"step": 835,
|
||
|
|
"valid_targets_mean": 2474.2,
|
||
|
|
"valid_targets_min": 894
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3076923076923075,
|
||
|
|
"grad_norm": 0.667522967334762,
|
||
|
|
"learning_rate": 2.5391003027955045e-05,
|
||
|
|
"loss": 0.2336,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1336827427148819,
|
||
|
|
"step": 840,
|
||
|
|
"valid_targets_mean": 2550.1,
|
||
|
|
"valid_targets_min": 756
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3274161735700196,
|
||
|
|
"grad_norm": 0.5859709200520877,
|
||
|
|
"learning_rate": 2.5201664339033138e-05,
|
||
|
|
"loss": 0.2356,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13315144181251526,
|
||
|
|
"step": 845,
|
||
|
|
"valid_targets_mean": 3100.6,
|
||
|
|
"valid_targets_min": 2699
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3471400394477318,
|
||
|
|
"grad_norm": 0.5746308464571264,
|
||
|
|
"learning_rate": 2.501182430287578e-05,
|
||
|
|
"loss": 0.2301,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11553173512220383,
|
||
|
|
"step": 850,
|
||
|
|
"valid_targets_mean": 2766.9,
|
||
|
|
"valid_targets_min": 2422
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.366863905325444,
|
||
|
|
"grad_norm": 0.5946484815216688,
|
||
|
|
"learning_rate": 2.4821501216660778e-05,
|
||
|
|
"loss": 0.2354,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13826578855514526,
|
||
|
|
"step": 855,
|
||
|
|
"valid_targets_mean": 2807.9,
|
||
|
|
"valid_targets_min": 2689
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.386587771203156,
|
||
|
|
"grad_norm": 0.6027994155218336,
|
||
|
|
"learning_rate": 2.4630713424123315e-05,
|
||
|
|
"loss": 0.2349,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11804911494255066,
|
||
|
|
"step": 860,
|
||
|
|
"valid_targets_mean": 2401.2,
|
||
|
|
"valid_targets_min": 706
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4063116370808677,
|
||
|
|
"grad_norm": 0.5950329300684662,
|
||
|
|
"learning_rate": 2.443947931378792e-05,
|
||
|
|
"loss": 0.2352,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11512573808431625,
|
||
|
|
"step": 865,
|
||
|
|
"valid_targets_mean": 2642.9,
|
||
|
|
"valid_targets_min": 1039
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.42603550295858,
|
||
|
|
"grad_norm": 0.7102236443711927,
|
||
|
|
"learning_rate": 2.4247817317196188e-05,
|
||
|
|
"loss": 0.2259,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09590330719947815,
|
||
|
|
"step": 870,
|
||
|
|
"valid_targets_mean": 2622.8,
|
||
|
|
"valid_targets_min": 2524
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.445759368836292,
|
||
|
|
"grad_norm": 0.5808040263882486,
|
||
|
|
"learning_rate": 2.405574590713025e-05,
|
||
|
|
"loss": 0.2315,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12733367085456848,
|
||
|
|
"step": 875,
|
||
|
|
"valid_targets_mean": 2933.6,
|
||
|
|
"valid_targets_min": 2742
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.465483234714004,
|
||
|
|
"grad_norm": 0.6149695489912174,
|
||
|
|
"learning_rate": 2.3863283595832387e-05,
|
||
|
|
"loss": 0.2377,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11119816452264786,
|
||
|
|
"step": 880,
|
||
|
|
"valid_targets_mean": 2720.4,
|
||
|
|
"valid_targets_min": 2397
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.485207100591716,
|
||
|
|
"grad_norm": 0.5407787836144836,
|
||
|
|
"learning_rate": 2.3670448933220732e-05,
|
||
|
|
"loss": 0.2314,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10149241983890533,
|
||
|
|
"step": 885,
|
||
|
|
"valid_targets_mean": 2615.0,
|
||
|
|
"valid_targets_min": 2286
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.504930966469428,
|
||
|
|
"grad_norm": 0.7003202120500566,
|
||
|
|
"learning_rate": 2.3477260505101427e-05,
|
||
|
|
"loss": 0.2392,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10941976308822632,
|
||
|
|
"step": 890,
|
||
|
|
"valid_targets_mean": 2556.2,
|
||
|
|
"valid_targets_min": 814
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.52465483234714,
|
||
|
|
"grad_norm": 0.6008857424340447,
|
||
|
|
"learning_rate": 2.328373693137726e-05,
|
||
|
|
"loss": 0.2393,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1187116801738739,
|
||
|
|
"step": 895,
|
||
|
|
"valid_targets_mean": 2609.9,
|
||
|
|
"valid_targets_min": 758
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.544378698224852,
|
||
|
|
"grad_norm": 0.6175869706655848,
|
||
|
|
"learning_rate": 2.3089896864253066e-05,
|
||
|
|
"loss": 0.233,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1316232681274414,
|
||
|
|
"step": 900,
|
||
|
|
"valid_targets_mean": 2775.1,
|
||
|
|
"valid_targets_min": 2506
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.564102564102564,
|
||
|
|
"grad_norm": 0.5772647985645465,
|
||
|
|
"learning_rate": 2.289575898643796e-05,
|
||
|
|
"loss": 0.2285,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12441572546958923,
|
||
|
|
"step": 905,
|
||
|
|
"valid_targets_mean": 2935.1,
|
||
|
|
"valid_targets_min": 2615
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.583826429980276,
|
||
|
|
"grad_norm": 0.5814530167069869,
|
||
|
|
"learning_rate": 2.270134200934466e-05,
|
||
|
|
"loss": 0.227,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1309787929058075,
|
||
|
|
"step": 910,
|
||
|
|
"valid_targets_mean": 2857.5,
|
||
|
|
"valid_targets_min": 2659
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.603550295857988,
|
||
|
|
"grad_norm": 0.5577133089768663,
|
||
|
|
"learning_rate": 2.2506664671286087e-05,
|
||
|
|
"loss": 0.2304,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11217945069074631,
|
||
|
|
"step": 915,
|
||
|
|
"valid_targets_mean": 2990.5,
|
||
|
|
"valid_targets_min": 2547
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6232741617357003,
|
||
|
|
"grad_norm": 0.5309025109571197,
|
||
|
|
"learning_rate": 2.2311745735669258e-05,
|
||
|
|
"loss": 0.2307,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11643368005752563,
|
||
|
|
"step": 920,
|
||
|
|
"valid_targets_mean": 2863.0,
|
||
|
|
"valid_targets_min": 2676
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6429980276134124,
|
||
|
|
"grad_norm": 0.5488354592542424,
|
||
|
|
"learning_rate": 2.2116603989186895e-05,
|
||
|
|
"loss": 0.2261,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.124149389564991,
|
||
|
|
"step": 925,
|
||
|
|
"valid_targets_mean": 2867.2,
|
||
|
|
"valid_targets_min": 2698
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.662721893491124,
|
||
|
|
"grad_norm": 0.595545887411219,
|
||
|
|
"learning_rate": 2.192125824000667e-05,
|
||
|
|
"loss": 0.2301,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11036589741706848,
|
||
|
|
"step": 930,
|
||
|
|
"valid_targets_mean": 2588.4,
|
||
|
|
"valid_targets_min": 2072
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6824457593688362,
|
||
|
|
"grad_norm": 0.6085924926330472,
|
||
|
|
"learning_rate": 2.1725727315958473e-05,
|
||
|
|
"loss": 0.223,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1208345964550972,
|
||
|
|
"step": 935,
|
||
|
|
"valid_targets_mean": 2770.0,
|
||
|
|
"valid_targets_min": 1850
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7021696252465484,
|
||
|
|
"grad_norm": 0.5575830245548056,
|
||
|
|
"learning_rate": 2.1530030062719724e-05,
|
||
|
|
"loss": 0.2291,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1100175529718399,
|
||
|
|
"step": 940,
|
||
|
|
"valid_targets_mean": 2444.2,
|
||
|
|
"valid_targets_min": 757
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7218934911242605,
|
||
|
|
"grad_norm": 0.5658214429441051,
|
||
|
|
"learning_rate": 2.1334185341999024e-05,
|
||
|
|
"loss": 0.23,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1086779534816742,
|
||
|
|
"step": 945,
|
||
|
|
"valid_targets_mean": 2806.1,
|
||
|
|
"valid_targets_min": 2470
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7416173570019726,
|
||
|
|
"grad_norm": 0.5491431411931478,
|
||
|
|
"learning_rate": 2.1138212029718158e-05,
|
||
|
|
"loss": 0.2239,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10670199245214462,
|
||
|
|
"step": 950,
|
||
|
|
"valid_targets_mean": 2740.6,
|
||
|
|
"valid_targets_min": 2453
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7613412228796843,
|
||
|
|
"grad_norm": 0.5621868558259059,
|
||
|
|
"learning_rate": 2.0942129014192854e-05,
|
||
|
|
"loss": 0.2318,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10864588618278503,
|
||
|
|
"step": 955,
|
||
|
|
"valid_targets_mean": 2582.4,
|
||
|
|
"valid_targets_min": 2480
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7810650887573964,
|
||
|
|
"grad_norm": 0.5642374192537544,
|
||
|
|
"learning_rate": 2.0745955194312276e-05,
|
||
|
|
"loss": 0.2269,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1140412837266922,
|
||
|
|
"step": 960,
|
||
|
|
"valid_targets_mean": 2733.8,
|
||
|
|
"valid_targets_min": 2397
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8007889546351086,
|
||
|
|
"grad_norm": 0.5767487410761784,
|
||
|
|
"learning_rate": 2.054970947771747e-05,
|
||
|
|
"loss": 0.2226,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11271451413631439,
|
||
|
|
"step": 965,
|
||
|
|
"valid_targets_mean": 3052.0,
|
||
|
|
"valid_targets_min": 2602
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8205128205128203,
|
||
|
|
"grad_norm": 0.5717678692463621,
|
||
|
|
"learning_rate": 2.0353410778979076e-05,
|
||
|
|
"loss": 0.2323,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11516307294368744,
|
||
|
|
"step": 970,
|
||
|
|
"valid_targets_mean": 2667.8,
|
||
|
|
"valid_targets_min": 2119
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8402366863905324,
|
||
|
|
"grad_norm": 0.6127135957946624,
|
||
|
|
"learning_rate": 2.0157078017774228e-05,
|
||
|
|
"loss": 0.2304,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1051267459988594,
|
||
|
|
"step": 975,
|
||
|
|
"valid_targets_mean": 2522.0,
|
||
|
|
"valid_targets_min": 220
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8599605522682445,
|
||
|
|
"grad_norm": 0.597642983394958,
|
||
|
|
"learning_rate": 1.99607301170631e-05,
|
||
|
|
"loss": 0.2267,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10642991960048676,
|
||
|
|
"step": 980,
|
||
|
|
"valid_targets_mean": 2665.5,
|
||
|
|
"valid_targets_min": 2341
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8796844181459567,
|
||
|
|
"grad_norm": 0.5617264390638574,
|
||
|
|
"learning_rate": 1.9764386001265015e-05,
|
||
|
|
"loss": 0.226,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10942518711090088,
|
||
|
|
"step": 985,
|
||
|
|
"valid_targets_mean": 2875.9,
|
||
|
|
"valid_targets_min": 2629
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.899408284023669,
|
||
|
|
"grad_norm": 0.5356173495359969,
|
||
|
|
"learning_rate": 1.956806459443453e-05,
|
||
|
|
"loss": 0.2238,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10982692241668701,
|
||
|
|
"step": 990,
|
||
|
|
"valid_targets_mean": 2755.8,
|
||
|
|
"valid_targets_min": 2377
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9191321499013805,
|
||
|
|
"grad_norm": 0.5714046702194722,
|
||
|
|
"learning_rate": 1.9371784818437436e-05,
|
||
|
|
"loss": 0.2322,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10803057998418808,
|
||
|
|
"step": 995,
|
||
|
|
"valid_targets_mean": 2653.9,
|
||
|
|
"valid_targets_min": 2391
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9388560157790926,
|
||
|
|
"grad_norm": 0.5552313785354733,
|
||
|
|
"learning_rate": 1.9175565591127073e-05,
|
||
|
|
"loss": 0.2264,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09931053966283798,
|
||
|
|
"step": 1000,
|
||
|
|
"valid_targets_mean": 2664.4,
|
||
|
|
"valid_targets_min": 2423
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9585798816568047,
|
||
|
|
"grad_norm": 0.552388613358886,
|
||
|
|
"learning_rate": 1.897942582452097e-05,
|
||
|
|
"loss": 0.2236,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11966521292924881,
|
||
|
|
"step": 1005,
|
||
|
|
"valid_targets_mean": 2755.4,
|
||
|
|
"valid_targets_min": 2535
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.978303747534517,
|
||
|
|
"grad_norm": 0.5438989000536992,
|
||
|
|
"learning_rate": 1.8783384422978066e-05,
|
||
|
|
"loss": 0.2296,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11552880704402924,
|
||
|
|
"step": 1010,
|
||
|
|
"valid_targets_mean": 2545.1,
|
||
|
|
"valid_targets_min": 1086
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.998027613412229,
|
||
|
|
"grad_norm": 0.5679634484160855,
|
||
|
|
"learning_rate": 1.8587460281376673e-05,
|
||
|
|
"loss": 0.2276,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10043665766716003,
|
||
|
|
"step": 1015,
|
||
|
|
"valid_targets_mean": 2407.0,
|
||
|
|
"valid_targets_min": 216
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.01577909270217,
|
||
|
|
"grad_norm": 0.5426153959281234,
|
||
|
|
"learning_rate": 1.8391672283293333e-05,
|
||
|
|
"loss": 0.2155,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10445930808782578,
|
||
|
|
"step": 1020,
|
||
|
|
"valid_targets_mean": 2756.6,
|
||
|
|
"valid_targets_min": 2598
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.035502958579881,
|
||
|
|
"grad_norm": 0.5634657480381071,
|
||
|
|
"learning_rate": 1.8196039299182818e-05,
|
||
|
|
"loss": 0.2192,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10721538215875626,
|
||
|
|
"step": 1025,
|
||
|
|
"valid_targets_mean": 2681.6,
|
||
|
|
"valid_targets_min": 2294
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.055226824457594,
|
||
|
|
"grad_norm": 0.5476429174978606,
|
||
|
|
"learning_rate": 1.8000580184559315e-05,
|
||
|
|
"loss": 0.2131,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09898987412452698,
|
||
|
|
"step": 1030,
|
||
|
|
"valid_targets_mean": 2721.5,
|
||
|
|
"valid_targets_min": 2469
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.074950690335306,
|
||
|
|
"grad_norm": 0.5782117145150883,
|
||
|
|
"learning_rate": 1.7805313778179095e-05,
|
||
|
|
"loss": 0.2107,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11142166703939438,
|
||
|
|
"step": 1035,
|
||
|
|
"valid_targets_mean": 2787.2,
|
||
|
|
"valid_targets_min": 2329
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.094674556213017,
|
||
|
|
"grad_norm": 0.6134021763569458,
|
||
|
|
"learning_rate": 1.7610258900224843e-05,
|
||
|
|
"loss": 0.2119,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10456886142492294,
|
||
|
|
"step": 1040,
|
||
|
|
"valid_targets_mean": 2760.5,
|
||
|
|
"valid_targets_min": 2385
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.11439842209073,
|
||
|
|
"grad_norm": 0.622905765719603,
|
||
|
|
"learning_rate": 1.741543435049165e-05,
|
||
|
|
"loss": 0.225,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11684265732765198,
|
||
|
|
"step": 1045,
|
||
|
|
"valid_targets_mean": 2781.8,
|
||
|
|
"valid_targets_min": 2403
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.134122287968442,
|
||
|
|
"grad_norm": 0.6688674230147973,
|
||
|
|
"learning_rate": 1.7220858906575126e-05,
|
||
|
|
"loss": 0.2297,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11058314144611359,
|
||
|
|
"step": 1050,
|
||
|
|
"valid_targets_mean": 2770.9,
|
||
|
|
"valid_targets_min": 2534
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.153846153846154,
|
||
|
|
"grad_norm": 0.5446637396644046,
|
||
|
|
"learning_rate": 1.702655132206154e-05,
|
||
|
|
"loss": 0.2153,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11232861876487732,
|
||
|
|
"step": 1055,
|
||
|
|
"valid_targets_mean": 2966.9,
|
||
|
|
"valid_targets_min": 2223
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.173570019723866,
|
||
|
|
"grad_norm": 0.5956928016618825,
|
||
|
|
"learning_rate": 1.6832530324720303e-05,
|
||
|
|
"loss": 0.2193,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12880636751651764,
|
||
|
|
"step": 1060,
|
||
|
|
"valid_targets_mean": 3357.0,
|
||
|
|
"valid_targets_min": 2118
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.193293885601578,
|
||
|
|
"grad_norm": 0.6213038274408813,
|
||
|
|
"learning_rate": 1.6638814614698965e-05,
|
||
|
|
"loss": 0.2188,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10869607329368591,
|
||
|
|
"step": 1065,
|
||
|
|
"valid_targets_mean": 2715.6,
|
||
|
|
"valid_targets_min": 2260
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.21301775147929,
|
||
|
|
"grad_norm": 0.5668384250549943,
|
||
|
|
"learning_rate": 1.6445422862720845e-05,
|
||
|
|
"loss": 0.2223,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10896855592727661,
|
||
|
|
"step": 1070,
|
||
|
|
"valid_targets_mean": 2847.9,
|
||
|
|
"valid_targets_min": 2600
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.232741617357002,
|
||
|
|
"grad_norm": 0.5434577739696346,
|
||
|
|
"learning_rate": 1.6252373708285505e-05,
|
||
|
|
"loss": 0.2104,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10187087953090668,
|
||
|
|
"step": 1075,
|
||
|
|
"valid_targets_mean": 2699.6,
|
||
|
|
"valid_targets_min": 2157
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.252465483234714,
|
||
|
|
"grad_norm": 0.5820415379961984,
|
||
|
|
"learning_rate": 1.6059685757872274e-05,
|
||
|
|
"loss": 0.2195,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11114576458930969,
|
||
|
|
"step": 1080,
|
||
|
|
"valid_targets_mean": 2726.9,
|
||
|
|
"valid_targets_min": 2444
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.272189349112426,
|
||
|
|
"grad_norm": 0.6232615254857719,
|
||
|
|
"learning_rate": 1.5867377583146836e-05,
|
||
|
|
"loss": 0.213,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10000680387020111,
|
||
|
|
"step": 1085,
|
||
|
|
"valid_targets_mean": 2861.9,
|
||
|
|
"valid_targets_min": 2296
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.291913214990138,
|
||
|
|
"grad_norm": 0.5807685631942436,
|
||
|
|
"learning_rate": 1.567546771917135e-05,
|
||
|
|
"loss": 0.2219,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11514294147491455,
|
||
|
|
"step": 1090,
|
||
|
|
"valid_targets_mean": 2758.8,
|
||
|
|
"valid_targets_min": 2661
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.31163708086785,
|
||
|
|
"grad_norm": 0.5352274047273502,
|
||
|
|
"learning_rate": 1.548397466261793e-05,
|
||
|
|
"loss": 0.2208,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10898380726575851,
|
||
|
|
"step": 1095,
|
||
|
|
"valid_targets_mean": 2732.8,
|
||
|
|
"valid_targets_min": 2328
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.331360946745562,
|
||
|
|
"grad_norm": 0.5935659443449228,
|
||
|
|
"learning_rate": 1.529291686998592e-05,
|
||
|
|
"loss": 0.2083,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1072545200586319,
|
||
|
|
"step": 1100,
|
||
|
|
"valid_targets_mean": 2808.1,
|
||
|
|
"valid_targets_min": 2471
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.351084812623274,
|
||
|
|
"grad_norm": 0.5635594927744844,
|
||
|
|
"learning_rate": 1.5102312755823053e-05,
|
||
|
|
"loss": 0.2126,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10902154445648193,
|
||
|
|
"step": 1105,
|
||
|
|
"valid_targets_mean": 2893.1,
|
||
|
|
"valid_targets_min": 1957
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.370808678500986,
|
||
|
|
"grad_norm": 0.5310059492766126,
|
||
|
|
"learning_rate": 1.4912180690950545e-05,
|
||
|
|
"loss": 0.2171,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09953179955482483,
|
||
|
|
"step": 1110,
|
||
|
|
"valid_targets_mean": 2812.8,
|
||
|
|
"valid_targets_min": 2435
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.390532544378698,
|
||
|
|
"grad_norm": 0.5749906425164959,
|
||
|
|
"learning_rate": 1.4722539000692548e-05,
|
||
|
|
"loss": 0.2252,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11796174943447113,
|
||
|
|
"step": 1115,
|
||
|
|
"valid_targets_mean": 2920.9,
|
||
|
|
"valid_targets_min": 561
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.410256410256411,
|
||
|
|
"grad_norm": 0.5546525639240448,
|
||
|
|
"learning_rate": 1.4533405963109867e-05,
|
||
|
|
"loss": 0.219,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10996496677398682,
|
||
|
|
"step": 1120,
|
||
|
|
"valid_targets_mean": 2703.8,
|
||
|
|
"valid_targets_min": 2384
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.429980276134122,
|
||
|
|
"grad_norm": 0.5704980326160944,
|
||
|
|
"learning_rate": 1.434479980723833e-05,
|
||
|
|
"loss": 0.2151,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10902923345565796,
|
||
|
|
"step": 1125,
|
||
|
|
"valid_targets_mean": 2485.2,
|
||
|
|
"valid_targets_min": 578
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.449704142011834,
|
||
|
|
"grad_norm": 0.5628501251388729,
|
||
|
|
"learning_rate": 1.4156738711331793e-05,
|
||
|
|
"loss": 0.22,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.100568488240242,
|
||
|
|
"step": 1130,
|
||
|
|
"valid_targets_mean": 2738.8,
|
||
|
|
"valid_targets_min": 1967
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.4694280078895465,
|
||
|
|
"grad_norm": 0.5724943267627032,
|
||
|
|
"learning_rate": 1.3969240801110088e-05,
|
||
|
|
"loss": 0.2184,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10781192034482956,
|
||
|
|
"step": 1135,
|
||
|
|
"valid_targets_mean": 2603.8,
|
||
|
|
"valid_targets_min": 678
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.489151873767258,
|
||
|
|
"grad_norm": 0.5400127434553933,
|
||
|
|
"learning_rate": 1.3782324148012061e-05,
|
||
|
|
"loss": 0.2061,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09783756732940674,
|
||
|
|
"step": 1140,
|
||
|
|
"valid_targets_mean": 2661.2,
|
||
|
|
"valid_targets_min": 2140
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.508875739644971,
|
||
|
|
"grad_norm": 0.5721781012732859,
|
||
|
|
"learning_rate": 1.3596006767453766e-05,
|
||
|
|
"loss": 0.2097,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10570011287927628,
|
||
|
|
"step": 1145,
|
||
|
|
"valid_targets_mean": 2768.6,
|
||
|
|
"valid_targets_min": 2411
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.5285996055226825,
|
||
|
|
"grad_norm": 0.5797709017154841,
|
||
|
|
"learning_rate": 1.3410306617092134e-05,
|
||
|
|
"loss": 0.2208,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1008017361164093,
|
||
|
|
"step": 1150,
|
||
|
|
"valid_targets_mean": 2708.9,
|
||
|
|
"valid_targets_min": 2605
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.548323471400394,
|
||
|
|
"grad_norm": 0.5807397420856328,
|
||
|
|
"learning_rate": 1.3225241595094173e-05,
|
||
|
|
"loss": 0.2194,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11709171533584595,
|
||
|
|
"step": 1155,
|
||
|
|
"valid_targets_mean": 2829.8,
|
||
|
|
"valid_targets_min": 2349
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.568047337278107,
|
||
|
|
"grad_norm": 0.5934600141097778,
|
||
|
|
"learning_rate": 1.3040829538411876e-05,
|
||
|
|
"loss": 0.2252,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11049273610115051,
|
||
|
|
"step": 1160,
|
||
|
|
"valid_targets_mean": 2799.0,
|
||
|
|
"valid_targets_min": 2481
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.587771203155818,
|
||
|
|
"grad_norm": 0.586504574090296,
|
||
|
|
"learning_rate": 1.2857088221063099e-05,
|
||
|
|
"loss": 0.2233,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11564389616250992,
|
||
|
|
"step": 1165,
|
||
|
|
"valid_targets_mean": 2571.1,
|
||
|
|
"valid_targets_min": 1070
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.607495069033531,
|
||
|
|
"grad_norm": 0.568713254434343,
|
||
|
|
"learning_rate": 1.2674035352418425e-05,
|
||
|
|
"loss": 0.2157,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10769792646169662,
|
||
|
|
"step": 1170,
|
||
|
|
"valid_targets_mean": 2746.2,
|
||
|
|
"valid_targets_min": 2230
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.627218934911243,
|
||
|
|
"grad_norm": 0.5883749097009137,
|
||
|
|
"learning_rate": 1.2491688575494337e-05,
|
||
|
|
"loss": 0.2181,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12988248467445374,
|
||
|
|
"step": 1175,
|
||
|
|
"valid_targets_mean": 2826.0,
|
||
|
|
"valid_targets_min": 2538
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.646942800788954,
|
||
|
|
"grad_norm": 0.5352791870570185,
|
||
|
|
"learning_rate": 1.231006546525273e-05,
|
||
|
|
"loss": 0.2184,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11645448207855225,
|
||
|
|
"step": 1180,
|
||
|
|
"valid_targets_mean": 2672.8,
|
||
|
|
"valid_targets_min": 1588
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.666666666666667,
|
||
|
|
"grad_norm": 0.6016777133894143,
|
||
|
|
"learning_rate": 1.2129183526906971e-05,
|
||
|
|
"loss": 0.2193,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11840753257274628,
|
||
|
|
"step": 1185,
|
||
|
|
"valid_targets_mean": 2780.2,
|
||
|
|
"valid_targets_min": 2288
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.686390532544379,
|
||
|
|
"grad_norm": 0.5530573238739942,
|
||
|
|
"learning_rate": 1.1949060194234775e-05,
|
||
|
|
"loss": 0.2176,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10779322683811188,
|
||
|
|
"step": 1190,
|
||
|
|
"valid_targets_mean": 2718.9,
|
||
|
|
"valid_targets_min": 2539
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.706114398422091,
|
||
|
|
"grad_norm": 0.5939209518186922,
|
||
|
|
"learning_rate": 1.1769712827897825e-05,
|
||
|
|
"loss": 0.2158,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10417772829532623,
|
||
|
|
"step": 1195,
|
||
|
|
"valid_targets_mean": 2733.8,
|
||
|
|
"valid_targets_min": 2276
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.725838264299803,
|
||
|
|
"grad_norm": 0.5847155608042839,
|
||
|
|
"learning_rate": 1.159115871376858e-05,
|
||
|
|
"loss": 0.2259,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11785180121660233,
|
||
|
|
"step": 1200,
|
||
|
|
"valid_targets_mean": 2744.1,
|
||
|
|
"valid_targets_min": 2445
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.745562130177515,
|
||
|
|
"grad_norm": 0.6079085062980414,
|
||
|
|
"learning_rate": 1.1413415061264205e-05,
|
||
|
|
"loss": 0.2151,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11582626402378082,
|
||
|
|
"step": 1205,
|
||
|
|
"valid_targets_mean": 2838.0,
|
||
|
|
"valid_targets_min": 2476
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.765285996055227,
|
||
|
|
"grad_norm": 0.5697903343992566,
|
||
|
|
"learning_rate": 1.1236499001687853e-05,
|
||
|
|
"loss": 0.2149,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10990774631500244,
|
||
|
|
"step": 1210,
|
||
|
|
"valid_targets_mean": 2896.2,
|
||
|
|
"valid_targets_min": 2517
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.785009861932939,
|
||
|
|
"grad_norm": 0.5985509055684338,
|
||
|
|
"learning_rate": 1.106042758657758e-05,
|
||
|
|
"loss": 0.2255,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11037649214267731,
|
||
|
|
"step": 1215,
|
||
|
|
"valid_targets_mean": 2780.1,
|
||
|
|
"valid_targets_min": 2299
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.804733727810651,
|
||
|
|
"grad_norm": 0.5987488992714199,
|
||
|
|
"learning_rate": 1.0885217786062837e-05,
|
||
|
|
"loss": 0.2142,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10688644647598267,
|
||
|
|
"step": 1220,
|
||
|
|
"valid_targets_mean": 2576.0,
|
||
|
|
"valid_targets_min": 2250
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.824457593688363,
|
||
|
|
"grad_norm": 0.5671645764737574,
|
||
|
|
"learning_rate": 1.0710886487228868e-05,
|
||
|
|
"loss": 0.2196,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10301928222179413,
|
||
|
|
"step": 1225,
|
||
|
|
"valid_targets_mean": 2858.1,
|
||
|
|
"valid_targets_min": 2575
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.844181459566075,
|
||
|
|
"grad_norm": 0.6260211012831314,
|
||
|
|
"learning_rate": 1.053745049248911e-05,
|
||
|
|
"loss": 0.2187,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10844558477401733,
|
||
|
|
"step": 1230,
|
||
|
|
"valid_targets_mean": 2625.0,
|
||
|
|
"valid_targets_min": 2074
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.8639053254437865,
|
||
|
|
"grad_norm": 0.5349751245601823,
|
||
|
|
"learning_rate": 1.0364926517965693e-05,
|
||
|
|
"loss": 0.2169,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10615304112434387,
|
||
|
|
"step": 1235,
|
||
|
|
"valid_targets_mean": 2692.9,
|
||
|
|
"valid_targets_min": 2443
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.883629191321499,
|
||
|
|
"grad_norm": 0.5454640864865432,
|
||
|
|
"learning_rate": 1.0193331191878388e-05,
|
||
|
|
"loss": 0.2271,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10507901012897491,
|
||
|
|
"step": 1240,
|
||
|
|
"valid_targets_mean": 2721.0,
|
||
|
|
"valid_targets_min": 2253
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.903353057199211,
|
||
|
|
"grad_norm": 0.5258954357514323,
|
||
|
|
"learning_rate": 1.0022681052941856e-05,
|
||
|
|
"loss": 0.2081,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09482446312904358,
|
||
|
|
"step": 1245,
|
||
|
|
"valid_targets_mean": 2661.8,
|
||
|
|
"valid_targets_min": 2301
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.923076923076923,
|
||
|
|
"grad_norm": 0.5970187543130571,
|
||
|
|
"learning_rate": 9.852992548771674e-06,
|
||
|
|
"loss": 0.2158,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10532170534133911,
|
||
|
|
"step": 1250,
|
||
|
|
"valid_targets_mean": 2923.8,
|
||
|
|
"valid_targets_min": 2557
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.942800788954635,
|
||
|
|
"grad_norm": 0.546286887729732,
|
||
|
|
"learning_rate": 9.684282034299053e-06,
|
||
|
|
"loss": 0.2174,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11268171668052673,
|
||
|
|
"step": 1255,
|
||
|
|
"valid_targets_mean": 2851.2,
|
||
|
|
"valid_targets_min": 2562
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.962524654832347,
|
||
|
|
"grad_norm": 0.5313090900813038,
|
||
|
|
"learning_rate": 9.516565770194523e-06,
|
||
|
|
"loss": 0.2173,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12095682322978973,
|
||
|
|
"step": 1260,
|
||
|
|
"valid_targets_mean": 3048.6,
|
||
|
|
"valid_targets_min": 2697
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.982248520710059,
|
||
|
|
"grad_norm": 0.6383130672021013,
|
||
|
|
"learning_rate": 9.349859921300704e-06,
|
||
|
|
"loss": 0.2138,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1049884483218193,
|
||
|
|
"step": 1265,
|
||
|
|
"valid_targets_mean": 2421.4,
|
||
|
|
"valid_targets_min": 811
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.0,
|
||
|
|
"grad_norm": 0.7635980492714537,
|
||
|
|
"learning_rate": 9.184180555074258e-06,
|
||
|
|
"loss": 0.2158,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21023963391780853,
|
||
|
|
"step": 1270,
|
||
|
|
"valid_targets_mean": 2632.4,
|
||
|
|
"valid_targets_min": 2420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.019723865877712,
|
||
|
|
"grad_norm": 0.6103692647848238,
|
||
|
|
"learning_rate": 9.019543640037363e-06,
|
||
|
|
"loss": 0.203,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10494416952133179,
|
||
|
|
"step": 1275,
|
||
|
|
"valid_targets_mean": 2897.0,
|
||
|
|
"valid_targets_min": 2628
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.039447731755424,
|
||
|
|
"grad_norm": 0.5803129887135982,
|
||
|
|
"learning_rate": 8.855965044238554e-06,
|
||
|
|
"loss": 0.2054,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0986366719007492,
|
||
|
|
"step": 1280,
|
||
|
|
"valid_targets_mean": 2766.9,
|
||
|
|
"valid_targets_min": 2588
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.059171597633136,
|
||
|
|
"grad_norm": 1.1089769014163477,
|
||
|
|
"learning_rate": 8.693460533723346e-06,
|
||
|
|
"loss": 0.2057,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1042357087135315,
|
||
|
|
"step": 1285,
|
||
|
|
"valid_targets_mean": 2862.5,
|
||
|
|
"valid_targets_min": 2253
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.0788954635108485,
|
||
|
|
"grad_norm": 0.5821716259572812,
|
||
|
|
"learning_rate": 8.532045771014693e-06,
|
||
|
|
"loss": 0.2028,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09927749633789062,
|
||
|
|
"step": 1290,
|
||
|
|
"valid_targets_mean": 2633.0,
|
||
|
|
"valid_targets_min": 2235
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.09861932938856,
|
||
|
|
"grad_norm": 0.5817689083770436,
|
||
|
|
"learning_rate": 8.37173631360339e-06,
|
||
|
|
"loss": 0.2084,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09932143986225128,
|
||
|
|
"step": 1295,
|
||
|
|
"valid_targets_mean": 2691.6,
|
||
|
|
"valid_targets_min": 2218
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.118343195266272,
|
||
|
|
"grad_norm": 0.5871296454692377,
|
||
|
|
"learning_rate": 8.212547612448595e-06,
|
||
|
|
"loss": 0.215,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10292601585388184,
|
||
|
|
"step": 1300,
|
||
|
|
"valid_targets_mean": 2495.6,
|
||
|
|
"valid_targets_min": 825
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.1380670611439845,
|
||
|
|
"grad_norm": 0.5784378464338366,
|
||
|
|
"learning_rate": 8.054495010488658e-06,
|
||
|
|
"loss": 0.2095,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10565447807312012,
|
||
|
|
"step": 1305,
|
||
|
|
"valid_targets_mean": 2827.6,
|
||
|
|
"valid_targets_min": 2543
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.157790927021696,
|
||
|
|
"grad_norm": 0.5654231764723436,
|
||
|
|
"learning_rate": 7.897593741162316e-06,
|
||
|
|
"loss": 0.2069,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10796831548213959,
|
||
|
|
"step": 1310,
|
||
|
|
"valid_targets_mean": 2792.1,
|
||
|
|
"valid_targets_min": 2258
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.177514792899408,
|
||
|
|
"grad_norm": 0.582761712788643,
|
||
|
|
"learning_rate": 7.741858926940475e-06,
|
||
|
|
"loss": 0.21,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10760889202356339,
|
||
|
|
"step": 1315,
|
||
|
|
"valid_targets_mean": 2839.0,
|
||
|
|
"valid_targets_min": 2595
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.19723865877712,
|
||
|
|
"grad_norm": 0.5658957308641317,
|
||
|
|
"learning_rate": 7.587305577868658e-06,
|
||
|
|
"loss": 0.2136,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11251413077116013,
|
||
|
|
"step": 1320,
|
||
|
|
"valid_targets_mean": 3097.5,
|
||
|
|
"valid_targets_min": 2300
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.216962524654832,
|
||
|
|
"grad_norm": 0.5395588278198526,
|
||
|
|
"learning_rate": 7.433948590120326e-06,
|
||
|
|
"loss": 0.2021,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10617254674434662,
|
||
|
|
"step": 1325,
|
||
|
|
"valid_targets_mean": 2761.4,
|
||
|
|
"valid_targets_min": 2475
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.236686390532545,
|
||
|
|
"grad_norm": 0.7280144483582203,
|
||
|
|
"learning_rate": 7.281802744561166e-06,
|
||
|
|
"loss": 0.2056,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10058635473251343,
|
||
|
|
"step": 1330,
|
||
|
|
"valid_targets_mean": 2711.1,
|
||
|
|
"valid_targets_min": 2521
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.256410256410256,
|
||
|
|
"grad_norm": 0.5573107835228384,
|
||
|
|
"learning_rate": 7.130882705324422e-06,
|
||
|
|
"loss": 0.2071,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10338783264160156,
|
||
|
|
"step": 1335,
|
||
|
|
"valid_targets_mean": 2623.1,
|
||
|
|
"valid_targets_min": 2069
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.276134122287968,
|
||
|
|
"grad_norm": 0.6053050045260399,
|
||
|
|
"learning_rate": 6.9812030183976e-06,
|
||
|
|
"loss": 0.2118,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10017573833465576,
|
||
|
|
"step": 1340,
|
||
|
|
"valid_targets_mean": 2779.1,
|
||
|
|
"valid_targets_min": 2500
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.295857988165681,
|
||
|
|
"grad_norm": 0.6040806744708079,
|
||
|
|
"learning_rate": 6.832778110220457e-06,
|
||
|
|
"loss": 0.2038,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10504000633955002,
|
||
|
|
"step": 1345,
|
||
|
|
"valid_targets_mean": 2810.1,
|
||
|
|
"valid_targets_min": 2459
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.315581854043392,
|
||
|
|
"grad_norm": 0.5902708046124696,
|
||
|
|
"learning_rate": 6.685622286294571e-06,
|
||
|
|
"loss": 0.2094,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10878709703683853,
|
||
|
|
"step": 1350,
|
||
|
|
"valid_targets_mean": 2719.6,
|
||
|
|
"valid_targets_min": 2064
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.335305719921105,
|
||
|
|
"grad_norm": 0.5899191230039799,
|
||
|
|
"learning_rate": 6.539749729804539e-06,
|
||
|
|
"loss": 0.2221,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11667156219482422,
|
||
|
|
"step": 1355,
|
||
|
|
"valid_targets_mean": 2746.2,
|
||
|
|
"valid_targets_min": 2325
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.355029585798817,
|
||
|
|
"grad_norm": 0.6347784277564364,
|
||
|
|
"learning_rate": 6.395174500250949e-06,
|
||
|
|
"loss": 0.208,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10981135070323944,
|
||
|
|
"step": 1360,
|
||
|
|
"valid_targets_mean": 2435.2,
|
||
|
|
"valid_targets_min": 611
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.374753451676528,
|
||
|
|
"grad_norm": 0.5628006993854829,
|
||
|
|
"learning_rate": 6.251910532095349e-06,
|
||
|
|
"loss": 0.2137,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10524934530258179,
|
||
|
|
"step": 1365,
|
||
|
|
"valid_targets_mean": 2659.2,
|
||
|
|
"valid_targets_min": 2148
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.394477317554241,
|
||
|
|
"grad_norm": 0.5981055577012766,
|
||
|
|
"learning_rate": 6.109971633417169e-06,
|
||
|
|
"loss": 0.2002,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12043331563472748,
|
||
|
|
"step": 1370,
|
||
|
|
"valid_targets_mean": 2762.6,
|
||
|
|
"valid_targets_min": 2392
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.414201183431953,
|
||
|
|
"grad_norm": 0.6007106811089689,
|
||
|
|
"learning_rate": 5.969371484582887e-06,
|
||
|
|
"loss": 0.2134,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10356853902339935,
|
||
|
|
"step": 1375,
|
||
|
|
"valid_targets_mean": 2468.5,
|
||
|
|
"valid_targets_min": 894
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.433925049309664,
|
||
|
|
"grad_norm": 0.571896554602496,
|
||
|
|
"learning_rate": 5.830123636927485e-06,
|
||
|
|
"loss": 0.2076,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10309138894081116,
|
||
|
|
"step": 1380,
|
||
|
|
"valid_targets_mean": 2495.9,
|
||
|
|
"valid_targets_min": 1067
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.453648915187377,
|
||
|
|
"grad_norm": 0.5674241658203291,
|
||
|
|
"learning_rate": 5.692241511448342e-06,
|
||
|
|
"loss": 0.2101,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08628429472446442,
|
||
|
|
"step": 1385,
|
||
|
|
"valid_targets_mean": 2511.6,
|
||
|
|
"valid_targets_min": 2242
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.4733727810650885,
|
||
|
|
"grad_norm": 0.5760369835813244,
|
||
|
|
"learning_rate": 5.555738397511699e-06,
|
||
|
|
"loss": 0.2133,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10998523235321045,
|
||
|
|
"step": 1390,
|
||
|
|
"valid_targets_mean": 2738.8,
|
||
|
|
"valid_targets_min": 2286
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.493096646942801,
|
||
|
|
"grad_norm": 0.5451989554041636,
|
||
|
|
"learning_rate": 5.4206274515717735e-06,
|
||
|
|
"loss": 0.2063,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09533542394638062,
|
||
|
|
"step": 1395,
|
||
|
|
"valid_targets_mean": 2659.6,
|
||
|
|
"valid_targets_min": 1561
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.512820512820513,
|
||
|
|
"grad_norm": 0.6440049530646881,
|
||
|
|
"learning_rate": 5.286921695902749e-06,
|
||
|
|
"loss": 0.2064,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09890514612197876,
|
||
|
|
"step": 1400,
|
||
|
|
"valid_targets_mean": 2307.2,
|
||
|
|
"valid_targets_min": 220
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.5325443786982245,
|
||
|
|
"grad_norm": 0.5672665440499342,
|
||
|
|
"learning_rate": 5.154634017343662e-06,
|
||
|
|
"loss": 0.2072,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10666382312774658,
|
||
|
|
"step": 1405,
|
||
|
|
"valid_targets_mean": 2901.0,
|
||
|
|
"valid_targets_min": 2427
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.552268244575937,
|
||
|
|
"grad_norm": 0.6083828575559277,
|
||
|
|
"learning_rate": 5.023777166056294e-06,
|
||
|
|
"loss": 0.2012,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11845268309116364,
|
||
|
|
"step": 1410,
|
||
|
|
"valid_targets_mean": 3099.5,
|
||
|
|
"valid_targets_min": 2592
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.571992110453649,
|
||
|
|
"grad_norm": 0.6109777681121973,
|
||
|
|
"learning_rate": 4.89436375429633e-06,
|
||
|
|
"loss": 0.202,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1107400581240654,
|
||
|
|
"step": 1415,
|
||
|
|
"valid_targets_mean": 2694.2,
|
||
|
|
"valid_targets_min": 2406
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.591715976331361,
|
||
|
|
"grad_norm": 0.6095815088053648,
|
||
|
|
"learning_rate": 4.766406255197751e-06,
|
||
|
|
"loss": 0.2091,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10473256558179855,
|
||
|
|
"step": 1420,
|
||
|
|
"valid_targets_mean": 2812.0,
|
||
|
|
"valid_targets_min": 2432
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.611439842209073,
|
||
|
|
"grad_norm": 0.5502160971702288,
|
||
|
|
"learning_rate": 4.639917001570644e-06,
|
||
|
|
"loss": 0.2105,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10024566948413849,
|
||
|
|
"step": 1425,
|
||
|
|
"valid_targets_mean": 2830.1,
|
||
|
|
"valid_targets_min": 2560
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.631163708086785,
|
||
|
|
"grad_norm": 0.5671734042589966,
|
||
|
|
"learning_rate": 4.51490818471255e-06,
|
||
|
|
"loss": 0.2087,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10085520148277283,
|
||
|
|
"step": 1430,
|
||
|
|
"valid_targets_mean": 2738.9,
|
||
|
|
"valid_targets_min": 2366
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.650887573964497,
|
||
|
|
"grad_norm": 0.6354154416970742,
|
||
|
|
"learning_rate": 4.391391853233404e-06,
|
||
|
|
"loss": 0.207,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09480464458465576,
|
||
|
|
"step": 1435,
|
||
|
|
"valid_targets_mean": 2387.6,
|
||
|
|
"valid_targets_min": 1086
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.670611439842209,
|
||
|
|
"grad_norm": 1.624896684734375,
|
||
|
|
"learning_rate": 4.269379911894336e-06,
|
||
|
|
"loss": 0.203,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0924990102648735,
|
||
|
|
"step": 1440,
|
||
|
|
"valid_targets_mean": 2728.1,
|
||
|
|
"valid_targets_min": 2244
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.6903353057199215,
|
||
|
|
"grad_norm": 0.591799498745929,
|
||
|
|
"learning_rate": 4.148884120460186e-06,
|
||
|
|
"loss": 0.2063,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1067214086651802,
|
||
|
|
"step": 1445,
|
||
|
|
"valid_targets_mean": 2714.5,
|
||
|
|
"valid_targets_min": 2388
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.710059171597633,
|
||
|
|
"grad_norm": 0.6369703532795925,
|
||
|
|
"learning_rate": 4.029916092566131e-06,
|
||
|
|
"loss": 0.2152,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11508543789386749,
|
||
|
|
"step": 1450,
|
||
|
|
"valid_targets_mean": 2516.5,
|
||
|
|
"valid_targets_min": 228
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.729783037475345,
|
||
|
|
"grad_norm": 0.5665887839217926,
|
||
|
|
"learning_rate": 3.91248729459831e-06,
|
||
|
|
"loss": 0.2033,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10095715522766113,
|
||
|
|
"step": 1455,
|
||
|
|
"valid_targets_mean": 2829.0,
|
||
|
|
"valid_targets_min": 2571
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.7495069033530575,
|
||
|
|
"grad_norm": 0.5782323903552244,
|
||
|
|
"learning_rate": 3.796609044588686e-06,
|
||
|
|
"loss": 0.2123,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10557662695646286,
|
||
|
|
"step": 1460,
|
||
|
|
"valid_targets_mean": 2774.0,
|
||
|
|
"valid_targets_min": 2657
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.769230769230769,
|
||
|
|
"grad_norm": 0.5792917141890924,
|
||
|
|
"learning_rate": 3.682292511124179e-06,
|
||
|
|
"loss": 0.2088,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11298931390047073,
|
||
|
|
"step": 1465,
|
||
|
|
"valid_targets_mean": 2848.8,
|
||
|
|
"valid_targets_min": 2366
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.788954635108482,
|
||
|
|
"grad_norm": 0.5895891791953276,
|
||
|
|
"learning_rate": 3.569548712270201e-06,
|
||
|
|
"loss": 0.2113,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10991150140762329,
|
||
|
|
"step": 1470,
|
||
|
|
"valid_targets_mean": 2726.2,
|
||
|
|
"valid_targets_min": 2034
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.808678500986193,
|
||
|
|
"grad_norm": 0.5515929409487375,
|
||
|
|
"learning_rate": 3.4583885145087613e-06,
|
||
|
|
"loss": 0.2213,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08973747491836548,
|
||
|
|
"step": 1475,
|
||
|
|
"valid_targets_mean": 2546.9,
|
||
|
|
"valid_targets_min": 2351
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.828402366863905,
|
||
|
|
"grad_norm": 0.7720608994591613,
|
||
|
|
"learning_rate": 3.348822631691082e-06,
|
||
|
|
"loss": 0.2133,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10680578649044037,
|
||
|
|
"step": 1480,
|
||
|
|
"valid_targets_mean": 2669.2,
|
||
|
|
"valid_targets_min": 2286
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.848126232741618,
|
||
|
|
"grad_norm": 0.5610109026525962,
|
||
|
|
"learning_rate": 3.240861624004983e-06,
|
||
|
|
"loss": 0.2113,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09859801083803177,
|
||
|
|
"step": 1485,
|
||
|
|
"valid_targets_mean": 2831.5,
|
||
|
|
"valid_targets_min": 2330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.867850098619329,
|
||
|
|
"grad_norm": 0.6069598593380445,
|
||
|
|
"learning_rate": 3.1345158969570933e-06,
|
||
|
|
"loss": 0.2085,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11143651604652405,
|
||
|
|
"step": 1490,
|
||
|
|
"valid_targets_mean": 2849.5,
|
||
|
|
"valid_targets_min": 2634
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.887573964497041,
|
||
|
|
"grad_norm": 0.6854844752708854,
|
||
|
|
"learning_rate": 3.0297957003699284e-06,
|
||
|
|
"loss": 0.2038,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09344105422496796,
|
||
|
|
"step": 1495,
|
||
|
|
"valid_targets_mean": 2737.9,
|
||
|
|
"valid_targets_min": 2524
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.907297830374754,
|
||
|
|
"grad_norm": 0.563041823558611,
|
||
|
|
"learning_rate": 2.926711127393993e-06,
|
||
|
|
"loss": 0.209,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11634556949138641,
|
||
|
|
"step": 1500,
|
||
|
|
"valid_targets_mean": 2692.0,
|
||
|
|
"valid_targets_min": 2084
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.927021696252465,
|
||
|
|
"grad_norm": 0.6177687719276024,
|
||
|
|
"learning_rate": 2.8252721135349892e-06,
|
||
|
|
"loss": 0.2071,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1214359849691391,
|
||
|
|
"step": 1505,
|
||
|
|
"valid_targets_mean": 2682.8,
|
||
|
|
"valid_targets_min": 900
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.946745562130177,
|
||
|
|
"grad_norm": 0.8090664295620404,
|
||
|
|
"learning_rate": 2.7254884356961976e-06,
|
||
|
|
"loss": 0.1996,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10884906351566315,
|
||
|
|
"step": 1510,
|
||
|
|
"valid_targets_mean": 2914.5,
|
||
|
|
"valid_targets_min": 2571
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.96646942800789,
|
||
|
|
"grad_norm": 0.6019392112644878,
|
||
|
|
"learning_rate": 2.6273697112361786e-06,
|
||
|
|
"loss": 0.2141,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11398676037788391,
|
||
|
|
"step": 1515,
|
||
|
|
"valid_targets_mean": 2455.0,
|
||
|
|
"valid_targets_min": 561
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.986193293885601,
|
||
|
|
"grad_norm": 0.5857073865386839,
|
||
|
|
"learning_rate": 2.5309253970418056e-06,
|
||
|
|
"loss": 0.2078,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11495976150035858,
|
||
|
|
"step": 1520,
|
||
|
|
"valid_targets_mean": 2876.9,
|
||
|
|
"valid_targets_min": 2412
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.003944773175542,
|
||
|
|
"grad_norm": 0.5719391096198033,
|
||
|
|
"learning_rate": 2.436164788616815e-06,
|
||
|
|
"loss": 0.2088,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10084079951047897,
|
||
|
|
"step": 1525,
|
||
|
|
"valid_targets_mean": 2712.9,
|
||
|
|
"valid_targets_min": 2286
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.023668639053255,
|
||
|
|
"grad_norm": 0.5426754389539966,
|
||
|
|
"learning_rate": 2.3430970191858873e-06,
|
||
|
|
"loss": 0.2059,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10399748384952545,
|
||
|
|
"step": 1530,
|
||
|
|
"valid_targets_mean": 2944.1,
|
||
|
|
"valid_targets_min": 2608
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.043392504930966,
|
||
|
|
"grad_norm": 0.5756797370773796,
|
||
|
|
"learning_rate": 2.2517310588143372e-06,
|
||
|
|
"loss": 0.2012,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10515573620796204,
|
||
|
|
"step": 1535,
|
||
|
|
"valid_targets_mean": 2881.6,
|
||
|
|
"valid_targets_min": 2697
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.063116370808679,
|
||
|
|
"grad_norm": 0.5952975569669119,
|
||
|
|
"learning_rate": 2.1620757135435875e-06,
|
||
|
|
"loss": 0.1984,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.107521191239357,
|
||
|
|
"step": 1540,
|
||
|
|
"valid_targets_mean": 2770.9,
|
||
|
|
"valid_targets_min": 2389
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.0828402366863905,
|
||
|
|
"grad_norm": 0.6046549679374532,
|
||
|
|
"learning_rate": 2.0741396245424263e-06,
|
||
|
|
"loss": 0.2027,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10322698950767517,
|
||
|
|
"step": 1545,
|
||
|
|
"valid_targets_mean": 2662.4,
|
||
|
|
"valid_targets_min": 2069
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.102564102564102,
|
||
|
|
"grad_norm": 0.5839425979169349,
|
||
|
|
"learning_rate": 1.9879312672741326e-06,
|
||
|
|
"loss": 0.207,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10858365893363953,
|
||
|
|
"step": 1550,
|
||
|
|
"valid_targets_mean": 2773.8,
|
||
|
|
"valid_targets_min": 2388
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.122287968441815,
|
||
|
|
"grad_norm": 0.5540668550502379,
|
||
|
|
"learning_rate": 1.903458950679613e-06,
|
||
|
|
"loss": 0.2051,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08865194022655487,
|
||
|
|
"step": 1555,
|
||
|
|
"valid_targets_mean": 2421.5,
|
||
|
|
"valid_targets_min": 224
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.1420118343195265,
|
||
|
|
"grad_norm": 0.5802375846674172,
|
||
|
|
"learning_rate": 1.820730816376548e-06,
|
||
|
|
"loss": 0.2074,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1142999529838562,
|
||
|
|
"step": 1560,
|
||
|
|
"valid_targets_mean": 2959.4,
|
||
|
|
"valid_targets_min": 2741
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.161735700197239,
|
||
|
|
"grad_norm": 0.5609881212176958,
|
||
|
|
"learning_rate": 1.7397548378747142e-06,
|
||
|
|
"loss": 0.2032,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1133670061826706,
|
||
|
|
"step": 1565,
|
||
|
|
"valid_targets_mean": 2941.5,
|
||
|
|
"valid_targets_min": 2522
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.181459566074951,
|
||
|
|
"grad_norm": 0.5650589343177295,
|
||
|
|
"learning_rate": 1.660538819807458e-06,
|
||
|
|
"loss": 0.2028,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09255445748567581,
|
||
|
|
"step": 1570,
|
||
|
|
"valid_targets_mean": 2634.8,
|
||
|
|
"valid_targets_min": 2009
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.201183431952662,
|
||
|
|
"grad_norm": 0.6899481169726128,
|
||
|
|
"learning_rate": 1.5830903971794765e-06,
|
||
|
|
"loss": 0.2045,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08773065358400345,
|
||
|
|
"step": 1575,
|
||
|
|
"valid_targets_mean": 2602.8,
|
||
|
|
"valid_targets_min": 2231
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.220907297830375,
|
||
|
|
"grad_norm": 0.5955601907980482,
|
||
|
|
"learning_rate": 1.5074170346309357e-06,
|
||
|
|
"loss": 0.2018,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09971213340759277,
|
||
|
|
"step": 1580,
|
||
|
|
"valid_targets_mean": 2628.8,
|
||
|
|
"valid_targets_min": 1850
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.240631163708087,
|
||
|
|
"grad_norm": 0.5976276287320689,
|
||
|
|
"learning_rate": 1.4335260257180262e-06,
|
||
|
|
"loss": 0.1996,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09339369833469391,
|
||
|
|
"step": 1585,
|
||
|
|
"valid_targets_mean": 2603.9,
|
||
|
|
"valid_targets_min": 2334
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.260355029585799,
|
||
|
|
"grad_norm": 0.5820063373583885,
|
||
|
|
"learning_rate": 1.3614244922099796e-06,
|
||
|
|
"loss": 0.2042,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09349517524242401,
|
||
|
|
"step": 1590,
|
||
|
|
"valid_targets_mean": 2527.1,
|
||
|
|
"valid_targets_min": 680
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.280078895463511,
|
||
|
|
"grad_norm": 0.5838820394012124,
|
||
|
|
"learning_rate": 1.2911193834026548e-06,
|
||
|
|
"loss": 0.2069,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09804850071668625,
|
||
|
|
"step": 1595,
|
||
|
|
"valid_targets_mean": 2784.9,
|
||
|
|
"valid_targets_min": 2403
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.299802761341223,
|
||
|
|
"grad_norm": 0.5675885842106687,
|
||
|
|
"learning_rate": 1.222617475448773e-06,
|
||
|
|
"loss": 0.1987,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08638069033622742,
|
||
|
|
"step": 1600,
|
||
|
|
"valid_targets_mean": 2639.1,
|
||
|
|
"valid_targets_min": 2470
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.319526627218935,
|
||
|
|
"grad_norm": 0.559816589058538,
|
||
|
|
"learning_rate": 1.1559253707048046e-06,
|
||
|
|
"loss": 0.2056,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10538795590400696,
|
||
|
|
"step": 1605,
|
||
|
|
"valid_targets_mean": 2713.5,
|
||
|
|
"valid_targets_min": 2202
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.339250493096647,
|
||
|
|
"grad_norm": 0.5750374705659049,
|
||
|
|
"learning_rate": 1.0910494970946073e-06,
|
||
|
|
"loss": 0.1965,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10088108479976654,
|
||
|
|
"step": 1610,
|
||
|
|
"valid_targets_mean": 2860.8,
|
||
|
|
"valid_targets_min": 2637
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.358974358974359,
|
||
|
|
"grad_norm": 0.5859680641427915,
|
||
|
|
"learning_rate": 1.027996107489908e-06,
|
||
|
|
"loss": 0.2018,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09206578135490417,
|
||
|
|
"step": 1615,
|
||
|
|
"valid_targets_mean": 2715.0,
|
||
|
|
"valid_targets_min": 2357
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.378698224852071,
|
||
|
|
"grad_norm": 0.57609224383997,
|
||
|
|
"learning_rate": 9.667712791076345e-07,
|
||
|
|
"loss": 0.2011,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.107093945145607,
|
||
|
|
"step": 1620,
|
||
|
|
"valid_targets_mean": 2827.8,
|
||
|
|
"valid_targets_min": 2361
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.398422090729783,
|
||
|
|
"grad_norm": 0.6063345195868304,
|
||
|
|
"learning_rate": 9.073809129241784e-07,
|
||
|
|
"loss": 0.2035,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1011471301317215,
|
||
|
|
"step": 1625,
|
||
|
|
"valid_targets_mean": 2627.9,
|
||
|
|
"valid_targets_min": 2070
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.418145956607495,
|
||
|
|
"grad_norm": 0.5477705710958664,
|
||
|
|
"learning_rate": 8.498307331066446e-07,
|
||
|
|
"loss": 0.2078,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10595261305570602,
|
||
|
|
"step": 1630,
|
||
|
|
"valid_targets_mean": 2764.4,
|
||
|
|
"valid_targets_min": 2560
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.437869822485207,
|
||
|
|
"grad_norm": 0.56207113011245,
|
||
|
|
"learning_rate": 7.941262864611387e-07,
|
||
|
|
"loss": 0.2014,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10150217264890671,
|
||
|
|
"step": 1635,
|
||
|
|
"valid_targets_mean": 2806.6,
|
||
|
|
"valid_targets_min": 2434
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.457593688362919,
|
||
|
|
"grad_norm": 0.5455741628807624,
|
||
|
|
"learning_rate": 7.402729418981769e-07,
|
||
|
|
"loss": 0.1961,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10178756713867188,
|
||
|
|
"step": 1640,
|
||
|
|
"valid_targets_mean": 2897.0,
|
||
|
|
"valid_targets_min": 2191
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.477317554240631,
|
||
|
|
"grad_norm": 0.5772883606123289,
|
||
|
|
"learning_rate": 6.882758899151886e-07,
|
||
|
|
"loss": 0.205,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09741091728210449,
|
||
|
|
"step": 1645,
|
||
|
|
"valid_targets_mean": 2811.6,
|
||
|
|
"valid_targets_min": 2464
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.497041420118343,
|
||
|
|
"grad_norm": 0.6275799312094796,
|
||
|
|
"learning_rate": 6.381401420962729e-07,
|
||
|
|
"loss": 0.2123,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08363485336303711,
|
||
|
|
"step": 1650,
|
||
|
|
"valid_targets_mean": 2621.0,
|
||
|
|
"valid_targets_min": 760
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.516765285996055,
|
||
|
|
"grad_norm": 0.6196319394254477,
|
||
|
|
"learning_rate": 5.898705306291508e-07,
|
||
|
|
"loss": 0.2074,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11036994308233261,
|
||
|
|
"step": 1655,
|
||
|
|
"valid_targets_mean": 2649.8,
|
||
|
|
"valid_targets_min": 2088
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.536489151873767,
|
||
|
|
"grad_norm": 0.5689210214939261,
|
||
|
|
"learning_rate": 5.434717078394447e-07,
|
||
|
|
"loss": 0.197,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09583219140768051,
|
||
|
|
"step": 1660,
|
||
|
|
"valid_targets_mean": 2811.6,
|
||
|
|
"valid_targets_min": 2487
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.556213017751479,
|
||
|
|
"grad_norm": 0.5646888458685277,
|
||
|
|
"learning_rate": 4.98948145742264e-07,
|
||
|
|
"loss": 0.1963,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08888030052185059,
|
||
|
|
"step": 1665,
|
||
|
|
"valid_targets_mean": 2753.8,
|
||
|
|
"valid_targets_min": 2487
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.575936883629192,
|
||
|
|
"grad_norm": 0.6034968843356804,
|
||
|
|
"learning_rate": 4.563041356111919e-07,
|
||
|
|
"loss": 0.2058,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08866098523139954,
|
||
|
|
"step": 1670,
|
||
|
|
"valid_targets_mean": 2696.8,
|
||
|
|
"valid_targets_min": 2581
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.595660749506903,
|
||
|
|
"grad_norm": 0.583847942388752,
|
||
|
|
"learning_rate": 4.155437875646828e-07,
|
||
|
|
"loss": 0.2045,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09415605664253235,
|
||
|
|
"step": 1675,
|
||
|
|
"valid_targets_mean": 2638.1,
|
||
|
|
"valid_targets_min": 2186
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.615384615384615,
|
||
|
|
"grad_norm": 0.5641061341462922,
|
||
|
|
"learning_rate": 3.7667103016992167e-07,
|
||
|
|
"loss": 0.206,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10915391147136688,
|
||
|
|
"step": 1680,
|
||
|
|
"valid_targets_mean": 2781.5,
|
||
|
|
"valid_targets_min": 2317
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.6351084812623276,
|
||
|
|
"grad_norm": 0.5990355601693913,
|
||
|
|
"learning_rate": 3.396896100641689e-07,
|
||
|
|
"loss": 0.2045,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1137113869190216,
|
||
|
|
"step": 1685,
|
||
|
|
"valid_targets_mean": 2603.4,
|
||
|
|
"valid_targets_min": 654
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.654832347140039,
|
||
|
|
"grad_norm": 0.5869862096354198,
|
||
|
|
"learning_rate": 3.0460309159366486e-07,
|
||
|
|
"loss": 0.2048,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10970069468021393,
|
||
|
|
"step": 1690,
|
||
|
|
"valid_targets_mean": 2766.0,
|
||
|
|
"valid_targets_min": 2278
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.674556213017752,
|
||
|
|
"grad_norm": 0.5926803043898645,
|
||
|
|
"learning_rate": 2.714148564700914e-07,
|
||
|
|
"loss": 0.2032,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09587039053440094,
|
||
|
|
"step": 1695,
|
||
|
|
"valid_targets_mean": 2492.2,
|
||
|
|
"valid_targets_min": 1737
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.6942800788954635,
|
||
|
|
"grad_norm": 0.5703254815335106,
|
||
|
|
"learning_rate": 2.401281034446212e-07,
|
||
|
|
"loss": 0.2022,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09007685631513596,
|
||
|
|
"step": 1700,
|
||
|
|
"valid_targets_mean": 2765.0,
|
||
|
|
"valid_targets_min": 2308
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.714003944773175,
|
||
|
|
"grad_norm": 0.5804134950472436,
|
||
|
|
"learning_rate": 2.107458479996316e-07,
|
||
|
|
"loss": 0.2012,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1073688194155693,
|
||
|
|
"step": 1705,
|
||
|
|
"valid_targets_mean": 2768.9,
|
||
|
|
"valid_targets_min": 2364
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.733727810650888,
|
||
|
|
"grad_norm": 0.5692009879043477,
|
||
|
|
"learning_rate": 1.8327092205805641e-07,
|
||
|
|
"loss": 0.2102,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09917967766523361,
|
||
|
|
"step": 1710,
|
||
|
|
"valid_targets_mean": 2800.0,
|
||
|
|
"valid_targets_min": 2432
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.7534516765285995,
|
||
|
|
"grad_norm": 0.5872382846025632,
|
||
|
|
"learning_rate": 1.577059737104447e-07,
|
||
|
|
"loss": 0.2093,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09657121449708939,
|
||
|
|
"step": 1715,
|
||
|
|
"valid_targets_mean": 2389.9,
|
||
|
|
"valid_targets_min": 509
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.773175542406312,
|
||
|
|
"grad_norm": 0.5542915172907693,
|
||
|
|
"learning_rate": 1.3405346695972265e-07,
|
||
|
|
"loss": 0.2012,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11030033975839615,
|
||
|
|
"step": 1720,
|
||
|
|
"valid_targets_mean": 2841.0,
|
||
|
|
"valid_targets_min": 2487
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.792899408284024,
|
||
|
|
"grad_norm": 0.6499077256942439,
|
||
|
|
"learning_rate": 1.1231568148372562e-07,
|
||
|
|
"loss": 0.2079,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10377727448940277,
|
||
|
|
"step": 1725,
|
||
|
|
"valid_targets_mean": 2773.2,
|
||
|
|
"valid_targets_min": 2571
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.812623274161735,
|
||
|
|
"grad_norm": 0.5979296909728743,
|
||
|
|
"learning_rate": 9.249471241545849e-08,
|
||
|
|
"loss": 0.2076,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10622979700565338,
|
||
|
|
"step": 1730,
|
||
|
|
"valid_targets_mean": 2625.0,
|
||
|
|
"valid_targets_min": 2140
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.832347140039448,
|
||
|
|
"grad_norm": 0.5505553156591817,
|
||
|
|
"learning_rate": 7.459247014117488e-08,
|
||
|
|
"loss": 0.204,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1144736185669899,
|
||
|
|
"step": 1735,
|
||
|
|
"valid_targets_mean": 2928.1,
|
||
|
|
"valid_targets_min": 2672
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.85207100591716,
|
||
|
|
"grad_norm": 0.6117935169652354,
|
||
|
|
"learning_rate": 5.861068011624449e-08,
|
||
|
|
"loss": 0.1979,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09437566995620728,
|
||
|
|
"step": 1740,
|
||
|
|
"valid_targets_mean": 2947.5,
|
||
|
|
"valid_targets_min": 2778
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.871794871794872,
|
||
|
|
"grad_norm": 0.5975320973333355,
|
||
|
|
"learning_rate": 4.4550882698857214e-08,
|
||
|
|
"loss": 0.2078,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1117662638425827,
|
||
|
|
"step": 1745,
|
||
|
|
"valid_targets_mean": 2751.0,
|
||
|
|
"valid_targets_min": 2074
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.891518737672584,
|
||
|
|
"grad_norm": 0.5874689123345994,
|
||
|
|
"learning_rate": 3.241443300154856e-08,
|
||
|
|
"loss": 0.2082,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09812627732753754,
|
||
|
|
"step": 1750,
|
||
|
|
"valid_targets_mean": 2506.1,
|
||
|
|
"valid_targets_min": 650
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.911242603550296,
|
||
|
|
"grad_norm": 0.5447364614407185,
|
||
|
|
"learning_rate": 2.220250076060193e-08,
|
||
|
|
"loss": 0.2038,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09424471110105515,
|
||
|
|
"step": 1755,
|
||
|
|
"valid_targets_mean": 2789.4,
|
||
|
|
"valid_targets_min": 2278
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.930966469428008,
|
||
|
|
"grad_norm": 0.5830877135242066,
|
||
|
|
"learning_rate": 1.3916070223298772e-08,
|
||
|
|
"loss": 0.2015,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10720057040452957,
|
||
|
|
"step": 1760,
|
||
|
|
"valid_targets_mean": 2683.1,
|
||
|
|
"valid_targets_min": 2465
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.95069033530572,
|
||
|
|
"grad_norm": 0.5748949285054279,
|
||
|
|
"learning_rate": 7.55594005306337e-09,
|
||
|
|
"loss": 0.2045,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1001519188284874,
|
||
|
|
"step": 1765,
|
||
|
|
"valid_targets_mean": 2860.1,
|
||
|
|
"valid_targets_min": 2482
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.970414201183432,
|
||
|
|
"grad_norm": 0.5736146068163355,
|
||
|
|
"learning_rate": 3.122723252477755e-09,
|
||
|
|
"loss": 0.2031,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10364716500043869,
|
||
|
|
"step": 1770,
|
||
|
|
"valid_targets_mean": 2770.8,
|
||
|
|
"valid_targets_min": 2503
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.990138067061144,
|
||
|
|
"grad_norm": 0.6141168538060929,
|
||
|
|
"learning_rate": 6.168471042067303e-10,
|
||
|
|
"loss": 0.2022,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11252442002296448,
|
||
|
|
"step": 1775,
|
||
|
|
"valid_targets_mean": 2815.0,
|
||
|
|
"valid_targets_min": 2205
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 7.0,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21355924010276794,
|
||
|
|
"step": 1778,
|
||
|
|
"total_flos": 6.986449832713912e+17,
|
||
|
|
"train_loss": 0.24971561980327953,
|
||
|
|
"train_runtime": 13514.0358,
|
||
|
|
"train_samples_per_second": 2.1,
|
||
|
|
"train_steps_per_second": 0.132,
|
||
|
|
"valid_targets_mean": 2789.8,
|
||
|
|
"valid_targets_min": 2416
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"logging_steps": 5,
|
||
|
|
"max_steps": 1778,
|
||
|
|
"num_input_tokens_seen": 0,
|
||
|
|
"num_train_epochs": 7,
|
||
|
|
"save_steps": 1500,
|
||
|
|
"stateful_callbacks": {
|
||
|
|
"TrainerControl": {
|
||
|
|
"args": {
|
||
|
|
"should_epoch_stop": false,
|
||
|
|
"should_evaluate": false,
|
||
|
|
"should_log": false,
|
||
|
|
"should_save": true,
|
||
|
|
"should_training_stop": true
|
||
|
|
},
|
||
|
|
"attributes": {}
|
||
|
|
}
|
||
|
|
},
|
||
|
|
"total_flos": 6.986449832713912e+17,
|
||
|
|
"train_batch_size": 1,
|
||
|
|
"trial_name": null,
|
||
|
|
"trial_params": null
|
||
|
|
}
|