9673 lines
258 KiB
JSON
9673 lines
258 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 7.0,
|
|
"eval_steps": 500,
|
|
"global_step": 4375,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.008,
|
|
"grad_norm": 19.031151891499,
|
|
"learning_rate": 3.6529680365296803e-07,
|
|
"loss": 0.7961,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.7910360097885132,
|
|
"step": 5,
|
|
"valid_targets_mean": 4487.0,
|
|
"valid_targets_min": 1038
|
|
},
|
|
{
|
|
"epoch": 0.016,
|
|
"grad_norm": 21.6574681809385,
|
|
"learning_rate": 8.219178082191781e-07,
|
|
"loss": 0.7641,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.7702529430389404,
|
|
"step": 10,
|
|
"valid_targets_mean": 4367.7,
|
|
"valid_targets_min": 2663
|
|
},
|
|
{
|
|
"epoch": 0.024,
|
|
"grad_norm": 18.61045320669567,
|
|
"learning_rate": 1.278538812785388e-06,
|
|
"loss": 0.7824,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.7942593693733215,
|
|
"step": 15,
|
|
"valid_targets_mean": 4549.8,
|
|
"valid_targets_min": 2530
|
|
},
|
|
{
|
|
"epoch": 0.032,
|
|
"grad_norm": 11.93015286495903,
|
|
"learning_rate": 1.7351598173515982e-06,
|
|
"loss": 0.7381,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.6816747188568115,
|
|
"step": 20,
|
|
"valid_targets_mean": 4554.2,
|
|
"valid_targets_min": 3046
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"grad_norm": 7.816459561620871,
|
|
"learning_rate": 2.191780821917808e-06,
|
|
"loss": 0.6619,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.6112380027770996,
|
|
"step": 25,
|
|
"valid_targets_mean": 4564.2,
|
|
"valid_targets_min": 3095
|
|
},
|
|
{
|
|
"epoch": 0.048,
|
|
"grad_norm": 3.950077327924629,
|
|
"learning_rate": 2.6484018264840183e-06,
|
|
"loss": 0.6374,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.6019740700721741,
|
|
"step": 30,
|
|
"valid_targets_mean": 4283.3,
|
|
"valid_targets_min": 2488
|
|
},
|
|
{
|
|
"epoch": 0.056,
|
|
"grad_norm": 2.443252710123367,
|
|
"learning_rate": 3.1050228310502285e-06,
|
|
"loss": 0.6055,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.5975817441940308,
|
|
"step": 35,
|
|
"valid_targets_mean": 4490.8,
|
|
"valid_targets_min": 3930
|
|
},
|
|
{
|
|
"epoch": 0.064,
|
|
"grad_norm": 1.631585869506433,
|
|
"learning_rate": 3.5616438356164386e-06,
|
|
"loss": 0.5597,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.5116957426071167,
|
|
"step": 40,
|
|
"valid_targets_mean": 4318.2,
|
|
"valid_targets_min": 2830
|
|
},
|
|
{
|
|
"epoch": 0.072,
|
|
"grad_norm": 1.5506415328901595,
|
|
"learning_rate": 4.018264840182649e-06,
|
|
"loss": 0.5229,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.5576603412628174,
|
|
"step": 45,
|
|
"valid_targets_mean": 4384.7,
|
|
"valid_targets_min": 3403
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"grad_norm": 1.1381426130117782,
|
|
"learning_rate": 4.4748858447488585e-06,
|
|
"loss": 0.5246,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4876716732978821,
|
|
"step": 50,
|
|
"valid_targets_mean": 4455.8,
|
|
"valid_targets_min": 2173
|
|
},
|
|
{
|
|
"epoch": 0.088,
|
|
"grad_norm": 0.975184316506151,
|
|
"learning_rate": 4.931506849315069e-06,
|
|
"loss": 0.4944,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4977094531059265,
|
|
"step": 55,
|
|
"valid_targets_mean": 4330.4,
|
|
"valid_targets_min": 2312
|
|
},
|
|
{
|
|
"epoch": 0.096,
|
|
"grad_norm": 0.857052809532433,
|
|
"learning_rate": 5.388127853881279e-06,
|
|
"loss": 0.5122,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.47971585392951965,
|
|
"step": 60,
|
|
"valid_targets_mean": 4388.4,
|
|
"valid_targets_min": 3205
|
|
},
|
|
{
|
|
"epoch": 0.104,
|
|
"grad_norm": 0.7171828372457968,
|
|
"learning_rate": 5.8447488584474885e-06,
|
|
"loss": 0.4684,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4935060441493988,
|
|
"step": 65,
|
|
"valid_targets_mean": 4365.2,
|
|
"valid_targets_min": 2169
|
|
},
|
|
{
|
|
"epoch": 0.112,
|
|
"grad_norm": 0.7194282877448056,
|
|
"learning_rate": 6.301369863013699e-06,
|
|
"loss": 0.4711,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.44747573137283325,
|
|
"step": 70,
|
|
"valid_targets_mean": 4520.1,
|
|
"valid_targets_min": 1768
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"grad_norm": 0.6300856706374344,
|
|
"learning_rate": 6.757990867579909e-06,
|
|
"loss": 0.4637,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.43163609504699707,
|
|
"step": 75,
|
|
"valid_targets_mean": 4633.4,
|
|
"valid_targets_min": 3923
|
|
},
|
|
{
|
|
"epoch": 0.128,
|
|
"grad_norm": 0.6756319994329208,
|
|
"learning_rate": 7.214611872146119e-06,
|
|
"loss": 0.4603,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4730415642261505,
|
|
"step": 80,
|
|
"valid_targets_mean": 4653.2,
|
|
"valid_targets_min": 3695
|
|
},
|
|
{
|
|
"epoch": 0.136,
|
|
"grad_norm": 0.6158232570056567,
|
|
"learning_rate": 7.671232876712329e-06,
|
|
"loss": 0.4589,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.428544819355011,
|
|
"step": 85,
|
|
"valid_targets_mean": 4531.2,
|
|
"valid_targets_min": 3376
|
|
},
|
|
{
|
|
"epoch": 0.144,
|
|
"grad_norm": 0.6611931364348244,
|
|
"learning_rate": 8.127853881278539e-06,
|
|
"loss": 0.4437,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.47876185178756714,
|
|
"step": 90,
|
|
"valid_targets_mean": 4528.2,
|
|
"valid_targets_min": 2911
|
|
},
|
|
{
|
|
"epoch": 0.152,
|
|
"grad_norm": 0.6159409916683618,
|
|
"learning_rate": 8.584474885844748e-06,
|
|
"loss": 0.4181,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4253089427947998,
|
|
"step": 95,
|
|
"valid_targets_mean": 4572.4,
|
|
"valid_targets_min": 3384
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"grad_norm": 0.6268508736944963,
|
|
"learning_rate": 9.04109589041096e-06,
|
|
"loss": 0.4286,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4430253803730011,
|
|
"step": 100,
|
|
"valid_targets_mean": 4519.5,
|
|
"valid_targets_min": 2326
|
|
},
|
|
{
|
|
"epoch": 0.168,
|
|
"grad_norm": 0.6081597860628081,
|
|
"learning_rate": 9.49771689497717e-06,
|
|
"loss": 0.4221,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4477927088737488,
|
|
"step": 105,
|
|
"valid_targets_mean": 4434.9,
|
|
"valid_targets_min": 2269
|
|
},
|
|
{
|
|
"epoch": 0.176,
|
|
"grad_norm": 0.6353985087340455,
|
|
"learning_rate": 9.95433789954338e-06,
|
|
"loss": 0.4187,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4329849183559418,
|
|
"step": 110,
|
|
"valid_targets_mean": 3999.9,
|
|
"valid_targets_min": 796
|
|
},
|
|
{
|
|
"epoch": 0.184,
|
|
"grad_norm": 0.6300850304742718,
|
|
"learning_rate": 1.0410958904109589e-05,
|
|
"loss": 0.419,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4073316752910614,
|
|
"step": 115,
|
|
"valid_targets_mean": 4400.6,
|
|
"valid_targets_min": 796
|
|
},
|
|
{
|
|
"epoch": 0.192,
|
|
"grad_norm": 0.5814246738643313,
|
|
"learning_rate": 1.08675799086758e-05,
|
|
"loss": 0.3884,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4161399006843567,
|
|
"step": 120,
|
|
"valid_targets_mean": 4406.8,
|
|
"valid_targets_min": 3712
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"grad_norm": 0.6359662723468273,
|
|
"learning_rate": 1.132420091324201e-05,
|
|
"loss": 0.389,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3953659236431122,
|
|
"step": 125,
|
|
"valid_targets_mean": 4140.9,
|
|
"valid_targets_min": 1768
|
|
},
|
|
{
|
|
"epoch": 0.208,
|
|
"grad_norm": 0.7772173219868077,
|
|
"learning_rate": 1.178082191780822e-05,
|
|
"loss": 0.3726,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.35869935154914856,
|
|
"step": 130,
|
|
"valid_targets_mean": 4183.6,
|
|
"valid_targets_min": 383
|
|
},
|
|
{
|
|
"epoch": 0.216,
|
|
"grad_norm": 0.5736605722405609,
|
|
"learning_rate": 1.223744292237443e-05,
|
|
"loss": 0.3898,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3817899823188782,
|
|
"step": 135,
|
|
"valid_targets_mean": 4541.3,
|
|
"valid_targets_min": 1663
|
|
},
|
|
{
|
|
"epoch": 0.224,
|
|
"grad_norm": 0.5642972952397545,
|
|
"learning_rate": 1.2694063926940641e-05,
|
|
"loss": 0.3887,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4149807095527649,
|
|
"step": 140,
|
|
"valid_targets_mean": 4649.8,
|
|
"valid_targets_min": 3948
|
|
},
|
|
{
|
|
"epoch": 0.232,
|
|
"grad_norm": 0.5471745969193615,
|
|
"learning_rate": 1.3150684931506849e-05,
|
|
"loss": 0.3832,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.36989864706993103,
|
|
"step": 145,
|
|
"valid_targets_mean": 4529.8,
|
|
"valid_targets_min": 1064
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"grad_norm": 0.5700885316867746,
|
|
"learning_rate": 1.360730593607306e-05,
|
|
"loss": 0.3611,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.33095479011535645,
|
|
"step": 150,
|
|
"valid_targets_mean": 4254.2,
|
|
"valid_targets_min": 1997
|
|
},
|
|
{
|
|
"epoch": 0.248,
|
|
"grad_norm": 0.6433049088279205,
|
|
"learning_rate": 1.406392694063927e-05,
|
|
"loss": 0.3865,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.42229828238487244,
|
|
"step": 155,
|
|
"valid_targets_mean": 4425.5,
|
|
"valid_targets_min": 1895
|
|
},
|
|
{
|
|
"epoch": 0.256,
|
|
"grad_norm": 0.5874719820319791,
|
|
"learning_rate": 1.4520547945205482e-05,
|
|
"loss": 0.3792,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.40710052847862244,
|
|
"step": 160,
|
|
"valid_targets_mean": 4567.0,
|
|
"valid_targets_min": 2277
|
|
},
|
|
{
|
|
"epoch": 0.264,
|
|
"grad_norm": 0.6225518498722266,
|
|
"learning_rate": 1.497716894977169e-05,
|
|
"loss": 0.3599,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.36803311109542847,
|
|
"step": 165,
|
|
"valid_targets_mean": 4571.4,
|
|
"valid_targets_min": 3765
|
|
},
|
|
{
|
|
"epoch": 0.272,
|
|
"grad_norm": 0.7978098671354809,
|
|
"learning_rate": 1.54337899543379e-05,
|
|
"loss": 0.3545,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3737052083015442,
|
|
"step": 170,
|
|
"valid_targets_mean": 4441.5,
|
|
"valid_targets_min": 2312
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"grad_norm": 0.7675053036135476,
|
|
"learning_rate": 1.589041095890411e-05,
|
|
"loss": 0.3757,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3923543095588684,
|
|
"step": 175,
|
|
"valid_targets_mean": 4220.6,
|
|
"valid_targets_min": 2033
|
|
},
|
|
{
|
|
"epoch": 0.288,
|
|
"grad_norm": 0.5774795964410557,
|
|
"learning_rate": 1.634703196347032e-05,
|
|
"loss": 0.3717,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.36189931631088257,
|
|
"step": 180,
|
|
"valid_targets_mean": 4623.9,
|
|
"valid_targets_min": 3787
|
|
},
|
|
{
|
|
"epoch": 0.296,
|
|
"grad_norm": 0.5613666644314715,
|
|
"learning_rate": 1.680365296803653e-05,
|
|
"loss": 0.3544,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.34768062829971313,
|
|
"step": 185,
|
|
"valid_targets_mean": 4565.7,
|
|
"valid_targets_min": 3368
|
|
},
|
|
{
|
|
"epoch": 0.304,
|
|
"grad_norm": 0.5787376459699338,
|
|
"learning_rate": 1.726027397260274e-05,
|
|
"loss": 0.3703,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3307119905948639,
|
|
"step": 190,
|
|
"valid_targets_mean": 4595.4,
|
|
"valid_targets_min": 3929
|
|
},
|
|
{
|
|
"epoch": 0.312,
|
|
"grad_norm": 0.5729822168949152,
|
|
"learning_rate": 1.771689497716895e-05,
|
|
"loss": 0.3695,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4279634654521942,
|
|
"step": 195,
|
|
"valid_targets_mean": 4838.6,
|
|
"valid_targets_min": 3695
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"grad_norm": 0.6699784452847818,
|
|
"learning_rate": 1.8173515981735163e-05,
|
|
"loss": 0.3655,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.42103224992752075,
|
|
"step": 200,
|
|
"valid_targets_mean": 4678.8,
|
|
"valid_targets_min": 2854
|
|
},
|
|
{
|
|
"epoch": 0.328,
|
|
"grad_norm": 0.6275564052364697,
|
|
"learning_rate": 1.863013698630137e-05,
|
|
"loss": 0.3539,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3345382511615753,
|
|
"step": 205,
|
|
"valid_targets_mean": 4436.8,
|
|
"valid_targets_min": 3621
|
|
},
|
|
{
|
|
"epoch": 0.336,
|
|
"grad_norm": 0.6180767934611819,
|
|
"learning_rate": 1.9086757990867582e-05,
|
|
"loss": 0.3642,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.33631786704063416,
|
|
"step": 210,
|
|
"valid_targets_mean": 4265.1,
|
|
"valid_targets_min": 2046
|
|
},
|
|
{
|
|
"epoch": 0.344,
|
|
"grad_norm": 0.6219326121732796,
|
|
"learning_rate": 1.954337899543379e-05,
|
|
"loss": 0.3627,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.37705540657043457,
|
|
"step": 215,
|
|
"valid_targets_mean": 4559.0,
|
|
"valid_targets_min": 2801
|
|
},
|
|
{
|
|
"epoch": 0.352,
|
|
"grad_norm": 0.6360576172350985,
|
|
"learning_rate": 2e-05,
|
|
"loss": 0.3332,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3344608247280121,
|
|
"step": 220,
|
|
"valid_targets_mean": 4612.6,
|
|
"valid_targets_min": 4138
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"grad_norm": 0.5472644977889617,
|
|
"learning_rate": 2.045662100456621e-05,
|
|
"loss": 0.3422,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3478066027164459,
|
|
"step": 225,
|
|
"valid_targets_mean": 4670.4,
|
|
"valid_targets_min": 3974
|
|
},
|
|
{
|
|
"epoch": 0.368,
|
|
"grad_norm": 0.9728294909204477,
|
|
"learning_rate": 2.0913242009132424e-05,
|
|
"loss": 0.3534,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.325618714094162,
|
|
"step": 230,
|
|
"valid_targets_mean": 4659.1,
|
|
"valid_targets_min": 3801
|
|
},
|
|
{
|
|
"epoch": 0.376,
|
|
"grad_norm": 0.6437091034095461,
|
|
"learning_rate": 2.1369863013698632e-05,
|
|
"loss": 0.3449,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3716578781604767,
|
|
"step": 235,
|
|
"valid_targets_mean": 4690.6,
|
|
"valid_targets_min": 3383
|
|
},
|
|
{
|
|
"epoch": 0.384,
|
|
"grad_norm": 0.6209195479018089,
|
|
"learning_rate": 2.182648401826484e-05,
|
|
"loss": 0.342,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.334907203912735,
|
|
"step": 240,
|
|
"valid_targets_mean": 4297.1,
|
|
"valid_targets_min": 709
|
|
},
|
|
{
|
|
"epoch": 0.392,
|
|
"grad_norm": 0.590723380369078,
|
|
"learning_rate": 2.2283105022831052e-05,
|
|
"loss": 0.3338,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3314967751502991,
|
|
"step": 245,
|
|
"valid_targets_mean": 4626.4,
|
|
"valid_targets_min": 2840
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"grad_norm": 0.5716169353633297,
|
|
"learning_rate": 2.2739726027397263e-05,
|
|
"loss": 0.349,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.34584173560142517,
|
|
"step": 250,
|
|
"valid_targets_mean": 4492.6,
|
|
"valid_targets_min": 1833
|
|
},
|
|
{
|
|
"epoch": 0.408,
|
|
"grad_norm": 0.5871290163737611,
|
|
"learning_rate": 2.3196347031963475e-05,
|
|
"loss": 0.3549,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3542795777320862,
|
|
"step": 255,
|
|
"valid_targets_mean": 4626.6,
|
|
"valid_targets_min": 3854
|
|
},
|
|
{
|
|
"epoch": 0.416,
|
|
"grad_norm": 0.6224674086258932,
|
|
"learning_rate": 2.3652968036529683e-05,
|
|
"loss": 0.3495,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.39130306243896484,
|
|
"step": 260,
|
|
"valid_targets_mean": 4820.4,
|
|
"valid_targets_min": 3789
|
|
},
|
|
{
|
|
"epoch": 0.424,
|
|
"grad_norm": 0.6289721051221905,
|
|
"learning_rate": 2.410958904109589e-05,
|
|
"loss": 0.3791,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.34234046936035156,
|
|
"step": 265,
|
|
"valid_targets_mean": 4702.2,
|
|
"valid_targets_min": 2514
|
|
},
|
|
{
|
|
"epoch": 0.432,
|
|
"grad_norm": 0.6259005624473737,
|
|
"learning_rate": 2.4566210045662106e-05,
|
|
"loss": 0.3476,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.32900434732437134,
|
|
"step": 270,
|
|
"valid_targets_mean": 4448.3,
|
|
"valid_targets_min": 3499
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"grad_norm": 0.5859961453144736,
|
|
"learning_rate": 2.5022831050228314e-05,
|
|
"loss": 0.3347,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3317798376083374,
|
|
"step": 275,
|
|
"valid_targets_mean": 4466.5,
|
|
"valid_targets_min": 2198
|
|
},
|
|
{
|
|
"epoch": 0.448,
|
|
"grad_norm": 0.5789991411893366,
|
|
"learning_rate": 2.547945205479452e-05,
|
|
"loss": 0.3411,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.32014352083206177,
|
|
"step": 280,
|
|
"valid_targets_mean": 4520.6,
|
|
"valid_targets_min": 2863
|
|
},
|
|
{
|
|
"epoch": 0.456,
|
|
"grad_norm": 0.5551965619238391,
|
|
"learning_rate": 2.593607305936073e-05,
|
|
"loss": 0.3399,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3320614993572235,
|
|
"step": 285,
|
|
"valid_targets_mean": 4417.2,
|
|
"valid_targets_min": 892
|
|
},
|
|
{
|
|
"epoch": 0.464,
|
|
"grad_norm": 0.6208316311126465,
|
|
"learning_rate": 2.6392694063926944e-05,
|
|
"loss": 0.3181,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.27898383140563965,
|
|
"step": 290,
|
|
"valid_targets_mean": 3889.0,
|
|
"valid_targets_min": 1611
|
|
},
|
|
{
|
|
"epoch": 0.472,
|
|
"grad_norm": 0.6365414944420186,
|
|
"learning_rate": 2.6849315068493153e-05,
|
|
"loss": 0.3318,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3356870710849762,
|
|
"step": 295,
|
|
"valid_targets_mean": 4355.2,
|
|
"valid_targets_min": 2252
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"grad_norm": 0.539147346125662,
|
|
"learning_rate": 2.7305936073059364e-05,
|
|
"loss": 0.3278,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.33671659231185913,
|
|
"step": 300,
|
|
"valid_targets_mean": 4805.1,
|
|
"valid_targets_min": 3814
|
|
},
|
|
{
|
|
"epoch": 0.488,
|
|
"grad_norm": 0.6334954822911109,
|
|
"learning_rate": 2.7762557077625572e-05,
|
|
"loss": 0.3328,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3289376497268677,
|
|
"step": 305,
|
|
"valid_targets_mean": 4255.5,
|
|
"valid_targets_min": 1585
|
|
},
|
|
{
|
|
"epoch": 0.496,
|
|
"grad_norm": 0.7542831396571542,
|
|
"learning_rate": 2.8219178082191783e-05,
|
|
"loss": 0.3292,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3443715274333954,
|
|
"step": 310,
|
|
"valid_targets_mean": 3974.0,
|
|
"valid_targets_min": 2111
|
|
},
|
|
{
|
|
"epoch": 0.504,
|
|
"grad_norm": 0.5878703128982131,
|
|
"learning_rate": 2.8675799086757995e-05,
|
|
"loss": 0.343,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.32190585136413574,
|
|
"step": 315,
|
|
"valid_targets_mean": 4510.7,
|
|
"valid_targets_min": 441
|
|
},
|
|
{
|
|
"epoch": 0.512,
|
|
"grad_norm": 0.6066330295738465,
|
|
"learning_rate": 2.9132420091324203e-05,
|
|
"loss": 0.3285,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3280946910381317,
|
|
"step": 320,
|
|
"valid_targets_mean": 4445.6,
|
|
"valid_targets_min": 1915
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"grad_norm": 0.6297276217201598,
|
|
"learning_rate": 2.958904109589041e-05,
|
|
"loss": 0.3356,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.31801527738571167,
|
|
"step": 325,
|
|
"valid_targets_mean": 4493.2,
|
|
"valid_targets_min": 2640
|
|
},
|
|
{
|
|
"epoch": 0.528,
|
|
"grad_norm": 0.5743054332736389,
|
|
"learning_rate": 3.0045662100456626e-05,
|
|
"loss": 0.3335,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3286020755767822,
|
|
"step": 330,
|
|
"valid_targets_mean": 4546.7,
|
|
"valid_targets_min": 3534
|
|
},
|
|
{
|
|
"epoch": 0.536,
|
|
"grad_norm": 0.6069662574889537,
|
|
"learning_rate": 3.0502283105022834e-05,
|
|
"loss": 0.3206,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3192424178123474,
|
|
"step": 335,
|
|
"valid_targets_mean": 4450.7,
|
|
"valid_targets_min": 2294
|
|
},
|
|
{
|
|
"epoch": 0.544,
|
|
"grad_norm": 0.7201520083591066,
|
|
"learning_rate": 3.0958904109589045e-05,
|
|
"loss": 0.3157,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2819150388240814,
|
|
"step": 340,
|
|
"valid_targets_mean": 4452.2,
|
|
"valid_targets_min": 3342
|
|
},
|
|
{
|
|
"epoch": 0.552,
|
|
"grad_norm": 0.5888311213879448,
|
|
"learning_rate": 3.141552511415525e-05,
|
|
"loss": 0.3282,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3358209729194641,
|
|
"step": 345,
|
|
"valid_targets_mean": 4580.2,
|
|
"valid_targets_min": 1533
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"grad_norm": 0.5663970602029285,
|
|
"learning_rate": 3.187214611872147e-05,
|
|
"loss": 0.3197,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.30763545632362366,
|
|
"step": 350,
|
|
"valid_targets_mean": 4550.8,
|
|
"valid_targets_min": 2908
|
|
},
|
|
{
|
|
"epoch": 0.568,
|
|
"grad_norm": 0.5868056060268125,
|
|
"learning_rate": 3.2328767123287676e-05,
|
|
"loss": 0.3465,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.33071446418762207,
|
|
"step": 355,
|
|
"valid_targets_mean": 4396.8,
|
|
"valid_targets_min": 2326
|
|
},
|
|
{
|
|
"epoch": 0.576,
|
|
"grad_norm": 0.5471882345116006,
|
|
"learning_rate": 3.2785388127853884e-05,
|
|
"loss": 0.308,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2997967004776001,
|
|
"step": 360,
|
|
"valid_targets_mean": 4723.9,
|
|
"valid_targets_min": 3180
|
|
},
|
|
{
|
|
"epoch": 0.584,
|
|
"grad_norm": 0.6086498123837335,
|
|
"learning_rate": 3.324200913242009e-05,
|
|
"loss": 0.3244,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.32957857847213745,
|
|
"step": 365,
|
|
"valid_targets_mean": 4539.7,
|
|
"valid_targets_min": 3168
|
|
},
|
|
{
|
|
"epoch": 0.592,
|
|
"grad_norm": 0.6352215584546738,
|
|
"learning_rate": 3.369863013698631e-05,
|
|
"loss": 0.3122,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.331981897354126,
|
|
"step": 370,
|
|
"valid_targets_mean": 4238.0,
|
|
"valid_targets_min": 970
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"grad_norm": 0.6489890003044103,
|
|
"learning_rate": 3.4155251141552515e-05,
|
|
"loss": 0.3148,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3532622456550598,
|
|
"step": 375,
|
|
"valid_targets_mean": 4598.5,
|
|
"valid_targets_min": 1875
|
|
},
|
|
{
|
|
"epoch": 0.608,
|
|
"grad_norm": 0.5784443324879636,
|
|
"learning_rate": 3.461187214611872e-05,
|
|
"loss": 0.3126,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3216385245323181,
|
|
"step": 380,
|
|
"valid_targets_mean": 4732.9,
|
|
"valid_targets_min": 3488
|
|
},
|
|
{
|
|
"epoch": 0.616,
|
|
"grad_norm": 0.5650016305435781,
|
|
"learning_rate": 3.506849315068493e-05,
|
|
"loss": 0.3253,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3246556520462036,
|
|
"step": 385,
|
|
"valid_targets_mean": 4726.4,
|
|
"valid_targets_min": 3726
|
|
},
|
|
{
|
|
"epoch": 0.624,
|
|
"grad_norm": 0.5688162085720793,
|
|
"learning_rate": 3.5525114155251146e-05,
|
|
"loss": 0.323,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.29329437017440796,
|
|
"step": 390,
|
|
"valid_targets_mean": 4398.9,
|
|
"valid_targets_min": 763
|
|
},
|
|
{
|
|
"epoch": 0.632,
|
|
"grad_norm": 0.5506423758378163,
|
|
"learning_rate": 3.5981735159817354e-05,
|
|
"loss": 0.3066,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28978243470191956,
|
|
"step": 395,
|
|
"valid_targets_mean": 4518.4,
|
|
"valid_targets_min": 3122
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"grad_norm": 0.5980025520191776,
|
|
"learning_rate": 3.643835616438356e-05,
|
|
"loss": 0.3212,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3178234398365021,
|
|
"step": 400,
|
|
"valid_targets_mean": 4729.4,
|
|
"valid_targets_min": 3537
|
|
},
|
|
{
|
|
"epoch": 0.648,
|
|
"grad_norm": 0.6183269091866439,
|
|
"learning_rate": 3.689497716894977e-05,
|
|
"loss": 0.315,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.32655179500579834,
|
|
"step": 405,
|
|
"valid_targets_mean": 4250.4,
|
|
"valid_targets_min": 2313
|
|
},
|
|
{
|
|
"epoch": 0.656,
|
|
"grad_norm": 0.5736290319522528,
|
|
"learning_rate": 3.7351598173515985e-05,
|
|
"loss": 0.3303,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3350864052772522,
|
|
"step": 410,
|
|
"valid_targets_mean": 4481.8,
|
|
"valid_targets_min": 2856
|
|
},
|
|
{
|
|
"epoch": 0.664,
|
|
"grad_norm": 0.599408395456047,
|
|
"learning_rate": 3.780821917808219e-05,
|
|
"loss": 0.3274,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.30340027809143066,
|
|
"step": 415,
|
|
"valid_targets_mean": 4503.7,
|
|
"valid_targets_min": 2780
|
|
},
|
|
{
|
|
"epoch": 0.672,
|
|
"grad_norm": 0.6306926969456081,
|
|
"learning_rate": 3.82648401826484e-05,
|
|
"loss": 0.3273,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3494199812412262,
|
|
"step": 420,
|
|
"valid_targets_mean": 4386.2,
|
|
"valid_targets_min": 2466
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"grad_norm": 0.5574754443008153,
|
|
"learning_rate": 3.8721461187214615e-05,
|
|
"loss": 0.3232,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3581283390522003,
|
|
"step": 425,
|
|
"valid_targets_mean": 4931.9,
|
|
"valid_targets_min": 4100
|
|
},
|
|
{
|
|
"epoch": 0.688,
|
|
"grad_norm": 0.5838443565520162,
|
|
"learning_rate": 3.9178082191780823e-05,
|
|
"loss": 0.3149,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.30937063694000244,
|
|
"step": 430,
|
|
"valid_targets_mean": 4445.9,
|
|
"valid_targets_min": 3257
|
|
},
|
|
{
|
|
"epoch": 0.696,
|
|
"grad_norm": 0.5648000277925969,
|
|
"learning_rate": 3.963470319634704e-05,
|
|
"loss": 0.3262,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3287450075149536,
|
|
"step": 435,
|
|
"valid_targets_mean": 4592.1,
|
|
"valid_targets_min": 3169
|
|
},
|
|
{
|
|
"epoch": 0.704,
|
|
"grad_norm": 0.5618585545103345,
|
|
"learning_rate": 3.99999936325009e-05,
|
|
"loss": 0.3162,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3002682626247406,
|
|
"step": 440,
|
|
"valid_targets_mean": 4453.6,
|
|
"valid_targets_min": 2306
|
|
},
|
|
{
|
|
"epoch": 0.712,
|
|
"grad_norm": 0.654707085899903,
|
|
"learning_rate": 3.9999770770457856e-05,
|
|
"loss": 0.3173,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.32616621255874634,
|
|
"step": 445,
|
|
"valid_targets_mean": 4888.4,
|
|
"valid_targets_min": 4011
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"grad_norm": 0.621528685184827,
|
|
"learning_rate": 3.9999229537513936e-05,
|
|
"loss": 0.2958,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3119966685771942,
|
|
"step": 450,
|
|
"valid_targets_mean": 4072.1,
|
|
"valid_targets_min": 926
|
|
},
|
|
{
|
|
"epoch": 0.728,
|
|
"grad_norm": 0.5439358249918728,
|
|
"learning_rate": 3.999836994228487e-05,
|
|
"loss": 0.3173,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3221753239631653,
|
|
"step": 455,
|
|
"valid_targets_mean": 4740.8,
|
|
"valid_targets_min": 3384
|
|
},
|
|
{
|
|
"epoch": 0.736,
|
|
"grad_norm": 0.5905848448789972,
|
|
"learning_rate": 3.999719199845432e-05,
|
|
"loss": 0.3284,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.30545833706855774,
|
|
"step": 460,
|
|
"valid_targets_mean": 4637.1,
|
|
"valid_targets_min": 3815
|
|
},
|
|
{
|
|
"epoch": 0.744,
|
|
"grad_norm": 0.5744939640039528,
|
|
"learning_rate": 3.999569572477366e-05,
|
|
"loss": 0.328,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.30196478962898254,
|
|
"step": 465,
|
|
"valid_targets_mean": 4571.3,
|
|
"valid_targets_min": 3052
|
|
},
|
|
{
|
|
"epoch": 0.752,
|
|
"grad_norm": 0.6156847707177951,
|
|
"learning_rate": 3.999388114506166e-05,
|
|
"loss": 0.3004,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3023531138896942,
|
|
"step": 470,
|
|
"valid_targets_mean": 4149.2,
|
|
"valid_targets_min": 926
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"grad_norm": 0.8170452216658209,
|
|
"learning_rate": 3.999174828820413e-05,
|
|
"loss": 0.3127,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.33499377965927124,
|
|
"step": 475,
|
|
"valid_targets_mean": 4773.4,
|
|
"valid_targets_min": 2652
|
|
},
|
|
{
|
|
"epoch": 0.768,
|
|
"grad_norm": 0.586438456660857,
|
|
"learning_rate": 3.998929718815341e-05,
|
|
"loss": 0.3135,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.311206579208374,
|
|
"step": 480,
|
|
"valid_targets_mean": 4698.1,
|
|
"valid_targets_min": 3467
|
|
},
|
|
{
|
|
"epoch": 0.776,
|
|
"grad_norm": 0.5933287538419926,
|
|
"learning_rate": 3.998652788392792e-05,
|
|
"loss": 0.3058,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2803855538368225,
|
|
"step": 485,
|
|
"valid_targets_mean": 4599.8,
|
|
"valid_targets_min": 2530
|
|
},
|
|
{
|
|
"epoch": 0.784,
|
|
"grad_norm": 0.597125952004893,
|
|
"learning_rate": 3.9983440419611445e-05,
|
|
"loss": 0.3032,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3011007606983185,
|
|
"step": 490,
|
|
"valid_targets_mean": 4269.1,
|
|
"valid_targets_min": 383
|
|
},
|
|
{
|
|
"epoch": 0.792,
|
|
"grad_norm": 0.5728831131888756,
|
|
"learning_rate": 3.9980034844352494e-05,
|
|
"loss": 0.2891,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.32238519191741943,
|
|
"step": 495,
|
|
"valid_targets_mean": 4705.0,
|
|
"valid_targets_min": 1594
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"grad_norm": 0.5731003315797196,
|
|
"learning_rate": 3.9976311212363495e-05,
|
|
"loss": 0.3116,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3044120669364929,
|
|
"step": 500,
|
|
"valid_targets_mean": 4542.1,
|
|
"valid_targets_min": 3672
|
|
},
|
|
{
|
|
"epoch": 0.808,
|
|
"grad_norm": 0.6001777146059886,
|
|
"learning_rate": 3.997226958291992e-05,
|
|
"loss": 0.3066,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.32697594165802,
|
|
"step": 505,
|
|
"valid_targets_mean": 4743.6,
|
|
"valid_targets_min": 2614
|
|
},
|
|
{
|
|
"epoch": 0.816,
|
|
"grad_norm": 0.5459030885288996,
|
|
"learning_rate": 3.996791002035937e-05,
|
|
"loss": 0.2963,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.30391639471054077,
|
|
"step": 510,
|
|
"valid_targets_mean": 4635.4,
|
|
"valid_targets_min": 3497
|
|
},
|
|
{
|
|
"epoch": 0.824,
|
|
"grad_norm": 0.5752233429531604,
|
|
"learning_rate": 3.996323259408055e-05,
|
|
"loss": 0.3072,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.30659013986587524,
|
|
"step": 515,
|
|
"valid_targets_mean": 4300.2,
|
|
"valid_targets_min": 1280
|
|
},
|
|
{
|
|
"epoch": 0.832,
|
|
"grad_norm": 0.561660686069106,
|
|
"learning_rate": 3.995823737854211e-05,
|
|
"loss": 0.3182,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3016003966331482,
|
|
"step": 520,
|
|
"valid_targets_mean": 4217.9,
|
|
"valid_targets_min": 2589
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"grad_norm": 0.5687483639146141,
|
|
"learning_rate": 3.9952924453261534e-05,
|
|
"loss": 0.306,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.31667256355285645,
|
|
"step": 525,
|
|
"valid_targets_mean": 4767.9,
|
|
"valid_targets_min": 4027
|
|
},
|
|
{
|
|
"epoch": 0.848,
|
|
"grad_norm": 0.568259528725315,
|
|
"learning_rate": 3.994729390281384e-05,
|
|
"loss": 0.3122,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3195444941520691,
|
|
"step": 530,
|
|
"valid_targets_mean": 4604.6,
|
|
"valid_targets_min": 2868
|
|
},
|
|
{
|
|
"epoch": 0.856,
|
|
"grad_norm": 0.5251910416408466,
|
|
"learning_rate": 3.994134581683021e-05,
|
|
"loss": 0.3073,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2904384136199951,
|
|
"step": 535,
|
|
"valid_targets_mean": 4685.7,
|
|
"valid_targets_min": 3888
|
|
},
|
|
{
|
|
"epoch": 0.864,
|
|
"grad_norm": 0.5940647785739603,
|
|
"learning_rate": 3.9935080289996626e-05,
|
|
"loss": 0.2999,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2823079824447632,
|
|
"step": 540,
|
|
"valid_targets_mean": 3859.9,
|
|
"valid_targets_min": 1753
|
|
},
|
|
{
|
|
"epoch": 0.872,
|
|
"grad_norm": 0.5772620624935296,
|
|
"learning_rate": 3.992849742205228e-05,
|
|
"loss": 0.3099,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2962970733642578,
|
|
"step": 545,
|
|
"valid_targets_mean": 4450.5,
|
|
"valid_targets_min": 2589
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"grad_norm": 0.5557098764699615,
|
|
"learning_rate": 3.9921597317788065e-05,
|
|
"loss": 0.3121,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3154350519180298,
|
|
"step": 550,
|
|
"valid_targets_mean": 4523.3,
|
|
"valid_targets_min": 1221
|
|
},
|
|
{
|
|
"epoch": 0.888,
|
|
"grad_norm": 0.5717904352901896,
|
|
"learning_rate": 3.991438008704486e-05,
|
|
"loss": 0.3244,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3106825351715088,
|
|
"step": 555,
|
|
"valid_targets_mean": 4614.3,
|
|
"valid_targets_min": 3471
|
|
},
|
|
{
|
|
"epoch": 0.896,
|
|
"grad_norm": 0.5414085791937698,
|
|
"learning_rate": 3.990684584471179e-05,
|
|
"loss": 0.2955,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28477743268013,
|
|
"step": 560,
|
|
"valid_targets_mean": 4634.1,
|
|
"valid_targets_min": 3176
|
|
},
|
|
{
|
|
"epoch": 0.904,
|
|
"grad_norm": 1.6688754590722357,
|
|
"learning_rate": 3.989899471072441e-05,
|
|
"loss": 0.2925,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2855817675590515,
|
|
"step": 565,
|
|
"valid_targets_mean": 4488.7,
|
|
"valid_targets_min": 2810
|
|
},
|
|
{
|
|
"epoch": 0.912,
|
|
"grad_norm": 0.5747955786408678,
|
|
"learning_rate": 3.9890826810062784e-05,
|
|
"loss": 0.3157,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.34171152114868164,
|
|
"step": 570,
|
|
"valid_targets_mean": 4768.1,
|
|
"valid_targets_min": 3716
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"grad_norm": 0.5955568983858046,
|
|
"learning_rate": 3.988234227274949e-05,
|
|
"loss": 0.2986,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.30315259099006653,
|
|
"step": 575,
|
|
"valid_targets_mean": 4347.4,
|
|
"valid_targets_min": 2182
|
|
},
|
|
{
|
|
"epoch": 0.928,
|
|
"grad_norm": 0.5594189753307507,
|
|
"learning_rate": 3.987354123384757e-05,
|
|
"loss": 0.3087,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3178068995475769,
|
|
"step": 580,
|
|
"valid_targets_mean": 4585.1,
|
|
"valid_targets_min": 3059
|
|
},
|
|
{
|
|
"epoch": 0.936,
|
|
"grad_norm": 0.5420489920592901,
|
|
"learning_rate": 3.9864423833458364e-05,
|
|
"loss": 0.2992,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3149924874305725,
|
|
"step": 585,
|
|
"valid_targets_mean": 4880.1,
|
|
"valid_targets_min": 4001
|
|
},
|
|
{
|
|
"epoch": 0.944,
|
|
"grad_norm": 0.5456965274351622,
|
|
"learning_rate": 3.9854990216719285e-05,
|
|
"loss": 0.2977,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28150367736816406,
|
|
"step": 590,
|
|
"valid_targets_mean": 4218.4,
|
|
"valid_targets_min": 1857
|
|
},
|
|
{
|
|
"epoch": 0.952,
|
|
"grad_norm": 0.5273814323601681,
|
|
"learning_rate": 3.98452405338015e-05,
|
|
"loss": 0.2855,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3029375672340393,
|
|
"step": 595,
|
|
"valid_targets_mean": 4481.4,
|
|
"valid_targets_min": 2503
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"grad_norm": 0.5536122651929943,
|
|
"learning_rate": 3.983517493990756e-05,
|
|
"loss": 0.2989,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.309809148311615,
|
|
"step": 600,
|
|
"valid_targets_mean": 4803.5,
|
|
"valid_targets_min": 3375
|
|
},
|
|
{
|
|
"epoch": 0.968,
|
|
"grad_norm": 0.5351703829937166,
|
|
"learning_rate": 3.982479359526892e-05,
|
|
"loss": 0.31,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3151604235172272,
|
|
"step": 605,
|
|
"valid_targets_mean": 4595.8,
|
|
"valid_targets_min": 3547
|
|
},
|
|
{
|
|
"epoch": 0.976,
|
|
"grad_norm": 0.5614631886925842,
|
|
"learning_rate": 3.981409666514336e-05,
|
|
"loss": 0.3017,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3181597888469696,
|
|
"step": 610,
|
|
"valid_targets_mean": 4773.3,
|
|
"valid_targets_min": 3011
|
|
},
|
|
{
|
|
"epoch": 0.984,
|
|
"grad_norm": 0.5156378502498516,
|
|
"learning_rate": 3.98030843198124e-05,
|
|
"loss": 0.2934,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2909044325351715,
|
|
"step": 615,
|
|
"valid_targets_mean": 4532.9,
|
|
"valid_targets_min": 3526
|
|
},
|
|
{
|
|
"epoch": 0.992,
|
|
"grad_norm": 0.5986192147321959,
|
|
"learning_rate": 3.979175673457858e-05,
|
|
"loss": 0.3052,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3034946322441101,
|
|
"step": 620,
|
|
"valid_targets_mean": 4490.9,
|
|
"valid_targets_min": 3406
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"grad_norm": 0.5233908842486453,
|
|
"learning_rate": 3.9780114089762616e-05,
|
|
"loss": 0.3079,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.30426180362701416,
|
|
"step": 625,
|
|
"valid_targets_mean": 4718.0,
|
|
"valid_targets_min": 3444
|
|
},
|
|
{
|
|
"epoch": 1.008,
|
|
"grad_norm": 0.5645680725275544,
|
|
"learning_rate": 3.976815657070062e-05,
|
|
"loss": 0.2694,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.281880259513855,
|
|
"step": 630,
|
|
"valid_targets_mean": 4390.5,
|
|
"valid_targets_min": 709
|
|
},
|
|
{
|
|
"epoch": 1.016,
|
|
"grad_norm": 0.5502011647236665,
|
|
"learning_rate": 3.975588436774107e-05,
|
|
"loss": 0.2887,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3047659993171692,
|
|
"step": 635,
|
|
"valid_targets_mean": 4613.4,
|
|
"valid_targets_min": 4000
|
|
},
|
|
{
|
|
"epoch": 1.024,
|
|
"grad_norm": 0.5966911038805575,
|
|
"learning_rate": 3.9743297676241826e-05,
|
|
"loss": 0.271,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28511953353881836,
|
|
"step": 640,
|
|
"valid_targets_mean": 4531.2,
|
|
"valid_targets_min": 3552
|
|
},
|
|
{
|
|
"epoch": 1.032,
|
|
"grad_norm": 0.5454673936341027,
|
|
"learning_rate": 3.9730396696566994e-05,
|
|
"loss": 0.2957,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2789442539215088,
|
|
"step": 645,
|
|
"valid_targets_mean": 4686.8,
|
|
"valid_targets_min": 2878
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"grad_norm": 0.5135827661900791,
|
|
"learning_rate": 3.971718163408375e-05,
|
|
"loss": 0.2756,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2611273527145386,
|
|
"step": 650,
|
|
"valid_targets_mean": 4509.1,
|
|
"valid_targets_min": 3754
|
|
},
|
|
{
|
|
"epoch": 1.048,
|
|
"grad_norm": 0.5401995972585367,
|
|
"learning_rate": 3.9703652699159093e-05,
|
|
"loss": 0.2791,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.27915990352630615,
|
|
"step": 655,
|
|
"valid_targets_mean": 4485.5,
|
|
"valid_targets_min": 2413
|
|
},
|
|
{
|
|
"epoch": 1.056,
|
|
"grad_norm": 0.5320943287838347,
|
|
"learning_rate": 3.9689810107156425e-05,
|
|
"loss": 0.274,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2547999322414398,
|
|
"step": 660,
|
|
"valid_targets_mean": 4294.6,
|
|
"valid_targets_min": 2173
|
|
},
|
|
{
|
|
"epoch": 1.064,
|
|
"grad_norm": 0.5283725024572096,
|
|
"learning_rate": 3.967565407843222e-05,
|
|
"loss": 0.2728,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25041675567626953,
|
|
"step": 665,
|
|
"valid_targets_mean": 4504.6,
|
|
"valid_targets_min": 3384
|
|
},
|
|
{
|
|
"epoch": 1.072,
|
|
"grad_norm": 0.562096771511752,
|
|
"learning_rate": 3.966118483833242e-05,
|
|
"loss": 0.2692,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.27407389879226685,
|
|
"step": 670,
|
|
"valid_targets_mean": 4674.1,
|
|
"valid_targets_min": 3960
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"grad_norm": 0.5394993986723041,
|
|
"learning_rate": 3.964640261718893e-05,
|
|
"loss": 0.2896,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2855004072189331,
|
|
"step": 675,
|
|
"valid_targets_mean": 4661.2,
|
|
"valid_targets_min": 3719
|
|
},
|
|
{
|
|
"epoch": 1.088,
|
|
"grad_norm": 0.5401374059383072,
|
|
"learning_rate": 3.963130765031589e-05,
|
|
"loss": 0.2755,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2641168236732483,
|
|
"step": 680,
|
|
"valid_targets_mean": 4500.5,
|
|
"valid_targets_min": 3426
|
|
},
|
|
{
|
|
"epoch": 1.096,
|
|
"grad_norm": 0.5317935782375285,
|
|
"learning_rate": 3.961590017800598e-05,
|
|
"loss": 0.285,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.27292317152023315,
|
|
"step": 685,
|
|
"valid_targets_mean": 4728.1,
|
|
"valid_targets_min": 2807
|
|
},
|
|
{
|
|
"epoch": 1.104,
|
|
"grad_norm": 0.6034207229797436,
|
|
"learning_rate": 3.960018044552653e-05,
|
|
"loss": 0.2689,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.30950093269348145,
|
|
"step": 690,
|
|
"valid_targets_mean": 4601.5,
|
|
"valid_targets_min": 3169
|
|
},
|
|
{
|
|
"epoch": 1.112,
|
|
"grad_norm": 0.5429319705626644,
|
|
"learning_rate": 3.9584148703115704e-05,
|
|
"loss": 0.2902,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25388938188552856,
|
|
"step": 695,
|
|
"valid_targets_mean": 4004.9,
|
|
"valid_targets_min": 1616
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"grad_norm": 0.5403726835846989,
|
|
"learning_rate": 3.956780520597842e-05,
|
|
"loss": 0.2921,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.27692919969558716,
|
|
"step": 700,
|
|
"valid_targets_mean": 4528.4,
|
|
"valid_targets_min": 2514
|
|
},
|
|
{
|
|
"epoch": 1.1280000000000001,
|
|
"grad_norm": 0.5774008754630514,
|
|
"learning_rate": 3.955115021428236e-05,
|
|
"loss": 0.3015,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2934579849243164,
|
|
"step": 705,
|
|
"valid_targets_mean": 4486.9,
|
|
"valid_targets_min": 3135
|
|
},
|
|
{
|
|
"epoch": 1.1360000000000001,
|
|
"grad_norm": 0.5522113177828276,
|
|
"learning_rate": 3.95341839931538e-05,
|
|
"loss": 0.2822,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28133076429367065,
|
|
"step": 710,
|
|
"valid_targets_mean": 4621.4,
|
|
"valid_targets_min": 3518
|
|
},
|
|
{
|
|
"epoch": 1.144,
|
|
"grad_norm": 0.5311310209496468,
|
|
"learning_rate": 3.95169068126734e-05,
|
|
"loss": 0.2753,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2587697505950928,
|
|
"step": 715,
|
|
"valid_targets_mean": 4629.9,
|
|
"valid_targets_min": 2910
|
|
},
|
|
{
|
|
"epoch": 1.152,
|
|
"grad_norm": 0.5925796710026722,
|
|
"learning_rate": 3.949931894787187e-05,
|
|
"loss": 0.27,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.29487308859825134,
|
|
"step": 720,
|
|
"valid_targets_mean": 4709.8,
|
|
"valid_targets_min": 2781
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"grad_norm": 0.5376507854122078,
|
|
"learning_rate": 3.948142067872565e-05,
|
|
"loss": 0.2746,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26245051622390747,
|
|
"step": 725,
|
|
"valid_targets_mean": 4419.6,
|
|
"valid_targets_min": 2952
|
|
},
|
|
{
|
|
"epoch": 1.168,
|
|
"grad_norm": 0.5689768065183267,
|
|
"learning_rate": 3.946321229015241e-05,
|
|
"loss": 0.2673,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2497783601284027,
|
|
"step": 730,
|
|
"valid_targets_mean": 4462.8,
|
|
"valid_targets_min": 2908
|
|
},
|
|
{
|
|
"epoch": 1.176,
|
|
"grad_norm": 0.5540679265530649,
|
|
"learning_rate": 3.944469407200652e-05,
|
|
"loss": 0.2891,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2979431748390198,
|
|
"step": 735,
|
|
"valid_targets_mean": 4964.1,
|
|
"valid_targets_min": 4052
|
|
},
|
|
{
|
|
"epoch": 1.184,
|
|
"grad_norm": 0.5888750072503279,
|
|
"learning_rate": 3.942586631907444e-05,
|
|
"loss": 0.2767,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28923922777175903,
|
|
"step": 740,
|
|
"valid_targets_mean": 4241.9,
|
|
"valid_targets_min": 2429
|
|
},
|
|
{
|
|
"epoch": 1.192,
|
|
"grad_norm": 0.5979023153814111,
|
|
"learning_rate": 3.9406729331070054e-05,
|
|
"loss": 0.2987,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.29468610882759094,
|
|
"step": 745,
|
|
"valid_targets_mean": 4337.9,
|
|
"valid_targets_min": 533
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"grad_norm": 0.5066412890955553,
|
|
"learning_rate": 3.938728341262985e-05,
|
|
"loss": 0.2525,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26115360856056213,
|
|
"step": 750,
|
|
"valid_targets_mean": 4708.9,
|
|
"valid_targets_min": 3449
|
|
},
|
|
{
|
|
"epoch": 1.208,
|
|
"grad_norm": 0.5155537386678077,
|
|
"learning_rate": 3.936752887330812e-05,
|
|
"loss": 0.2603,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2582944929599762,
|
|
"step": 755,
|
|
"valid_targets_mean": 4538.9,
|
|
"valid_targets_min": 3094
|
|
},
|
|
{
|
|
"epoch": 1.216,
|
|
"grad_norm": 0.5592405383275144,
|
|
"learning_rate": 3.9347466027571975e-05,
|
|
"loss": 0.2838,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26747220754623413,
|
|
"step": 760,
|
|
"valid_targets_mean": 4562.4,
|
|
"valid_targets_min": 2792
|
|
},
|
|
{
|
|
"epoch": 1.224,
|
|
"grad_norm": 0.6500622720548477,
|
|
"learning_rate": 3.932709519479639e-05,
|
|
"loss": 0.2799,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.29019367694854736,
|
|
"step": 765,
|
|
"valid_targets_mean": 4189.2,
|
|
"valid_targets_min": 1507
|
|
},
|
|
{
|
|
"epoch": 1.232,
|
|
"grad_norm": 0.5268945638812501,
|
|
"learning_rate": 3.930641669925911e-05,
|
|
"loss": 0.2746,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26922884583473206,
|
|
"step": 770,
|
|
"valid_targets_mean": 4824.9,
|
|
"valid_targets_min": 3854
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"grad_norm": 0.5403309618817408,
|
|
"learning_rate": 3.928543087013546e-05,
|
|
"loss": 0.2792,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2874947190284729,
|
|
"step": 775,
|
|
"valid_targets_mean": 4573.6,
|
|
"valid_targets_min": 3663
|
|
},
|
|
{
|
|
"epoch": 1.248,
|
|
"grad_norm": 0.5275108467631447,
|
|
"learning_rate": 3.926413804149315e-05,
|
|
"loss": 0.2796,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.29545095562934875,
|
|
"step": 780,
|
|
"valid_targets_mean": 4732.9,
|
|
"valid_targets_min": 2422
|
|
},
|
|
{
|
|
"epoch": 1.256,
|
|
"grad_norm": 0.620370817705532,
|
|
"learning_rate": 3.9242538552286894e-05,
|
|
"loss": 0.2913,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.31897711753845215,
|
|
"step": 785,
|
|
"valid_targets_mean": 4154.6,
|
|
"valid_targets_min": 1925
|
|
},
|
|
{
|
|
"epoch": 1.264,
|
|
"grad_norm": 0.6060237664998573,
|
|
"learning_rate": 3.9220632746353096e-05,
|
|
"loss": 0.2675,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25031667947769165,
|
|
"step": 790,
|
|
"valid_targets_mean": 4378.0,
|
|
"valid_targets_min": 912
|
|
},
|
|
{
|
|
"epoch": 1.272,
|
|
"grad_norm": 0.6051115609734747,
|
|
"learning_rate": 3.91984209724043e-05,
|
|
"loss": 0.2831,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2894507050514221,
|
|
"step": 795,
|
|
"valid_targets_mean": 4397.8,
|
|
"valid_targets_min": 1266
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"grad_norm": 0.5564017968436465,
|
|
"learning_rate": 3.917590358402369e-05,
|
|
"loss": 0.2726,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2730398178100586,
|
|
"step": 800,
|
|
"valid_targets_mean": 4577.2,
|
|
"valid_targets_min": 4031
|
|
},
|
|
{
|
|
"epoch": 1.288,
|
|
"grad_norm": 0.5443408511490985,
|
|
"learning_rate": 3.915308093965943e-05,
|
|
"loss": 0.271,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.27382051944732666,
|
|
"step": 805,
|
|
"valid_targets_mean": 4574.5,
|
|
"valid_targets_min": 3532
|
|
},
|
|
{
|
|
"epoch": 1.296,
|
|
"grad_norm": 0.5910835524136591,
|
|
"learning_rate": 3.9129953402618976e-05,
|
|
"loss": 0.2606,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2739621698856354,
|
|
"step": 810,
|
|
"valid_targets_mean": 4574.0,
|
|
"valid_targets_min": 2597
|
|
},
|
|
{
|
|
"epoch": 1.304,
|
|
"grad_norm": 0.5731499424919185,
|
|
"learning_rate": 3.91065213410633e-05,
|
|
"loss": 0.2743,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2773831784725189,
|
|
"step": 815,
|
|
"valid_targets_mean": 4488.7,
|
|
"valid_targets_min": 2450
|
|
},
|
|
{
|
|
"epoch": 1.312,
|
|
"grad_norm": 0.6397877359903824,
|
|
"learning_rate": 3.908278512800098e-05,
|
|
"loss": 0.264,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28440165519714355,
|
|
"step": 820,
|
|
"valid_targets_mean": 4444.6,
|
|
"valid_targets_min": 2911
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"grad_norm": 0.6098692570750375,
|
|
"learning_rate": 3.905874514128235e-05,
|
|
"loss": 0.277,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.30293822288513184,
|
|
"step": 825,
|
|
"valid_targets_mean": 4642.2,
|
|
"valid_targets_min": 3939
|
|
},
|
|
{
|
|
"epoch": 1.328,
|
|
"grad_norm": 0.5734684156489923,
|
|
"learning_rate": 3.903440176359338e-05,
|
|
"loss": 0.2712,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2627778649330139,
|
|
"step": 830,
|
|
"valid_targets_mean": 4295.7,
|
|
"valid_targets_min": 926
|
|
},
|
|
{
|
|
"epoch": 1.336,
|
|
"grad_norm": 0.5308135141821333,
|
|
"learning_rate": 3.90097553824497e-05,
|
|
"loss": 0.2643,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26772525906562805,
|
|
"step": 835,
|
|
"valid_targets_mean": 4366.0,
|
|
"valid_targets_min": 1585
|
|
},
|
|
{
|
|
"epoch": 1.3439999999999999,
|
|
"grad_norm": 0.5524603639814216,
|
|
"learning_rate": 3.8984806390190304e-05,
|
|
"loss": 0.2688,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2663770318031311,
|
|
"step": 840,
|
|
"valid_targets_mean": 4547.9,
|
|
"valid_targets_min": 2173
|
|
},
|
|
{
|
|
"epoch": 1.3519999999999999,
|
|
"grad_norm": 0.48622023181203317,
|
|
"learning_rate": 3.895955518397141e-05,
|
|
"loss": 0.2469,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2332693487405777,
|
|
"step": 845,
|
|
"valid_targets_mean": 4487.3,
|
|
"valid_targets_min": 3818
|
|
},
|
|
{
|
|
"epoch": 1.3599999999999999,
|
|
"grad_norm": 0.5922422218510031,
|
|
"learning_rate": 3.893400216576011e-05,
|
|
"loss": 0.2756,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.27962732315063477,
|
|
"step": 850,
|
|
"valid_targets_mean": 3954.2,
|
|
"valid_targets_min": 1507
|
|
},
|
|
{
|
|
"epoch": 1.3679999999999999,
|
|
"grad_norm": 0.643709626785372,
|
|
"learning_rate": 3.89081477423279e-05,
|
|
"loss": 0.2949,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.29329532384872437,
|
|
"step": 855,
|
|
"valid_targets_mean": 4409.8,
|
|
"valid_targets_min": 1756
|
|
},
|
|
{
|
|
"epoch": 1.376,
|
|
"grad_norm": 0.5539220969897855,
|
|
"learning_rate": 3.888199232524434e-05,
|
|
"loss": 0.2774,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2806379199028015,
|
|
"step": 860,
|
|
"valid_targets_mean": 4279.1,
|
|
"valid_targets_min": 2268
|
|
},
|
|
{
|
|
"epoch": 1.384,
|
|
"grad_norm": 0.7165484588962573,
|
|
"learning_rate": 3.8855536330870354e-05,
|
|
"loss": 0.2785,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2794676423072815,
|
|
"step": 865,
|
|
"valid_targets_mean": 4698.6,
|
|
"valid_targets_min": 3449
|
|
},
|
|
{
|
|
"epoch": 1.392,
|
|
"grad_norm": 0.6017584353070601,
|
|
"learning_rate": 3.882878018035173e-05,
|
|
"loss": 0.2615,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28225210309028625,
|
|
"step": 870,
|
|
"valid_targets_mean": 4364.3,
|
|
"valid_targets_min": 2594
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"grad_norm": 0.5917226238786201,
|
|
"learning_rate": 3.880172429961232e-05,
|
|
"loss": 0.2751,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28232043981552124,
|
|
"step": 875,
|
|
"valid_targets_mean": 4490.0,
|
|
"valid_targets_min": 3334
|
|
},
|
|
{
|
|
"epoch": 1.408,
|
|
"grad_norm": 0.5384371025690399,
|
|
"learning_rate": 3.877436911934733e-05,
|
|
"loss": 0.2622,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2497349977493286,
|
|
"step": 880,
|
|
"valid_targets_mean": 4265.1,
|
|
"valid_targets_min": 1276
|
|
},
|
|
{
|
|
"epoch": 1.416,
|
|
"grad_norm": 0.5496696893087772,
|
|
"learning_rate": 3.874671507501641e-05,
|
|
"loss": 0.2545,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2798706591129303,
|
|
"step": 885,
|
|
"valid_targets_mean": 4586.9,
|
|
"valid_targets_min": 3883
|
|
},
|
|
{
|
|
"epoch": 1.424,
|
|
"grad_norm": 0.5901771101802331,
|
|
"learning_rate": 3.871876260683677e-05,
|
|
"loss": 0.2596,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2610628008842468,
|
|
"step": 890,
|
|
"valid_targets_mean": 4225.6,
|
|
"valid_targets_min": 2558
|
|
},
|
|
{
|
|
"epoch": 1.432,
|
|
"grad_norm": 0.563510552372382,
|
|
"learning_rate": 3.869051215977612e-05,
|
|
"loss": 0.2671,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25134116411209106,
|
|
"step": 895,
|
|
"valid_targets_mean": 4316.4,
|
|
"valid_targets_min": 2589
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"grad_norm": 0.6999351399066672,
|
|
"learning_rate": 3.8661964183545634e-05,
|
|
"loss": 0.2574,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2672742009162903,
|
|
"step": 900,
|
|
"valid_targets_mean": 4490.4,
|
|
"valid_targets_min": 1502
|
|
},
|
|
{
|
|
"epoch": 1.448,
|
|
"grad_norm": 0.5713283886476633,
|
|
"learning_rate": 3.863311913259276e-05,
|
|
"loss": 0.2712,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26115670800209045,
|
|
"step": 905,
|
|
"valid_targets_mean": 4082.6,
|
|
"valid_targets_min": 2530
|
|
},
|
|
{
|
|
"epoch": 1.456,
|
|
"grad_norm": 0.6246550423508,
|
|
"learning_rate": 3.860397746609402e-05,
|
|
"loss": 0.2496,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23793759942054749,
|
|
"step": 910,
|
|
"valid_targets_mean": 4330.5,
|
|
"valid_targets_min": 1064
|
|
},
|
|
{
|
|
"epoch": 1.464,
|
|
"grad_norm": 0.5562914747823162,
|
|
"learning_rate": 3.857453964794764e-05,
|
|
"loss": 0.2718,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28893595933914185,
|
|
"step": 915,
|
|
"valid_targets_mean": 4702.5,
|
|
"valid_targets_min": 3681
|
|
},
|
|
{
|
|
"epoch": 1.472,
|
|
"grad_norm": 0.5889698255605341,
|
|
"learning_rate": 3.854480614676624e-05,
|
|
"loss": 0.2527,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25224125385284424,
|
|
"step": 920,
|
|
"valid_targets_mean": 4722.4,
|
|
"valid_targets_min": 3208
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"grad_norm": 0.5619840603369763,
|
|
"learning_rate": 3.851477743586932e-05,
|
|
"loss": 0.264,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2762981653213501,
|
|
"step": 925,
|
|
"valid_targets_mean": 4542.1,
|
|
"valid_targets_min": 2771
|
|
},
|
|
{
|
|
"epoch": 1.488,
|
|
"grad_norm": 0.6289238839468987,
|
|
"learning_rate": 3.8484453993275746e-05,
|
|
"loss": 0.2608,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2775125503540039,
|
|
"step": 930,
|
|
"valid_targets_mean": 4406.3,
|
|
"valid_targets_min": 1732
|
|
},
|
|
{
|
|
"epoch": 1.496,
|
|
"grad_norm": 0.5840983767511602,
|
|
"learning_rate": 3.8453836301696134e-05,
|
|
"loss": 0.2707,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.31399524211883545,
|
|
"step": 935,
|
|
"valid_targets_mean": 4701.6,
|
|
"valid_targets_min": 1611
|
|
},
|
|
{
|
|
"epoch": 1.504,
|
|
"grad_norm": 0.5453974195259924,
|
|
"learning_rate": 3.842292484852518e-05,
|
|
"loss": 0.2734,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26793837547302246,
|
|
"step": 940,
|
|
"valid_targets_mean": 4527.9,
|
|
"valid_targets_min": 3001
|
|
},
|
|
{
|
|
"epoch": 1.512,
|
|
"grad_norm": 0.6029438283984964,
|
|
"learning_rate": 3.8391720125833875e-05,
|
|
"loss": 0.2635,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2808925211429596,
|
|
"step": 945,
|
|
"valid_targets_mean": 4543.4,
|
|
"valid_targets_min": 2940
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"grad_norm": 0.5653252208412846,
|
|
"learning_rate": 3.83602226303617e-05,
|
|
"loss": 0.2741,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2835398316383362,
|
|
"step": 950,
|
|
"valid_targets_mean": 4288.4,
|
|
"valid_targets_min": 2632
|
|
},
|
|
{
|
|
"epoch": 1.528,
|
|
"grad_norm": 0.5740782374428135,
|
|
"learning_rate": 3.83284328635087e-05,
|
|
"loss": 0.2592,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28318852186203003,
|
|
"step": 955,
|
|
"valid_targets_mean": 4726.3,
|
|
"valid_targets_min": 3534
|
|
},
|
|
{
|
|
"epoch": 1.536,
|
|
"grad_norm": 0.5147390002096152,
|
|
"learning_rate": 3.829635133132751e-05,
|
|
"loss": 0.2665,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24325799942016602,
|
|
"step": 960,
|
|
"valid_targets_mean": 4775.6,
|
|
"valid_targets_min": 3947
|
|
},
|
|
{
|
|
"epoch": 1.544,
|
|
"grad_norm": 0.5477607619146632,
|
|
"learning_rate": 3.8263978544515304e-05,
|
|
"loss": 0.2644,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26991504430770874,
|
|
"step": 965,
|
|
"valid_targets_mean": 4752.8,
|
|
"valid_targets_min": 3817
|
|
},
|
|
{
|
|
"epoch": 1.552,
|
|
"grad_norm": 0.5585339627968549,
|
|
"learning_rate": 3.823131501840565e-05,
|
|
"loss": 0.2726,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28317582607269287,
|
|
"step": 970,
|
|
"valid_targets_mean": 4565.5,
|
|
"valid_targets_min": 3635
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"grad_norm": 0.5601335724855683,
|
|
"learning_rate": 3.819836127296032e-05,
|
|
"loss": 0.2548,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2538621425628662,
|
|
"step": 975,
|
|
"valid_targets_mean": 4221.9,
|
|
"valid_targets_min": 2527
|
|
},
|
|
{
|
|
"epoch": 1.568,
|
|
"grad_norm": 0.5451708690901231,
|
|
"learning_rate": 3.8165117832761016e-05,
|
|
"loss": 0.2645,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25721418857574463,
|
|
"step": 980,
|
|
"valid_targets_mean": 4633.4,
|
|
"valid_targets_min": 3967
|
|
},
|
|
{
|
|
"epoch": 1.576,
|
|
"grad_norm": 0.5776461712786491,
|
|
"learning_rate": 3.813158522700098e-05,
|
|
"loss": 0.266,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22501063346862793,
|
|
"step": 985,
|
|
"valid_targets_mean": 4457.2,
|
|
"valid_targets_min": 3046
|
|
},
|
|
{
|
|
"epoch": 1.584,
|
|
"grad_norm": 0.5321239076402918,
|
|
"learning_rate": 3.809776398947665e-05,
|
|
"loss": 0.2596,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26491671800613403,
|
|
"step": 990,
|
|
"valid_targets_mean": 4671.9,
|
|
"valid_targets_min": 3693
|
|
},
|
|
{
|
|
"epoch": 1.592,
|
|
"grad_norm": 0.6177159791971688,
|
|
"learning_rate": 3.806365465857908e-05,
|
|
"loss": 0.2562,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28787925839424133,
|
|
"step": 995,
|
|
"valid_targets_mean": 4414.7,
|
|
"valid_targets_min": 2342
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"grad_norm": 0.5446675255960364,
|
|
"learning_rate": 3.802925777728541e-05,
|
|
"loss": 0.2652,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23680800199508667,
|
|
"step": 1000,
|
|
"valid_targets_mean": 4274.9,
|
|
"valid_targets_min": 2566
|
|
},
|
|
{
|
|
"epoch": 1.608,
|
|
"grad_norm": 0.8232416542457747,
|
|
"learning_rate": 3.799457389315023e-05,
|
|
"loss": 0.2812,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2802053689956665,
|
|
"step": 1005,
|
|
"valid_targets_mean": 4192.4,
|
|
"valid_targets_min": 2644
|
|
},
|
|
{
|
|
"epoch": 1.616,
|
|
"grad_norm": 0.5430433445877927,
|
|
"learning_rate": 3.795960355829683e-05,
|
|
"loss": 0.2734,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2667580842971802,
|
|
"step": 1010,
|
|
"valid_targets_mean": 4712.8,
|
|
"valid_targets_min": 3281
|
|
},
|
|
{
|
|
"epoch": 1.624,
|
|
"grad_norm": 0.5645459332867225,
|
|
"learning_rate": 3.7924347329408444e-05,
|
|
"loss": 0.2717,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26407909393310547,
|
|
"step": 1015,
|
|
"valid_targets_mean": 4072.7,
|
|
"valid_targets_min": 934
|
|
},
|
|
{
|
|
"epoch": 1.6320000000000001,
|
|
"grad_norm": 0.5527220671547023,
|
|
"learning_rate": 3.788880576771937e-05,
|
|
"loss": 0.2501,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26285576820373535,
|
|
"step": 1020,
|
|
"valid_targets_mean": 4323.4,
|
|
"valid_targets_min": 383
|
|
},
|
|
{
|
|
"epoch": 1.6400000000000001,
|
|
"grad_norm": 0.546570041015299,
|
|
"learning_rate": 3.785297943900605e-05,
|
|
"loss": 0.265,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.29476824402809143,
|
|
"step": 1025,
|
|
"valid_targets_mean": 4714.8,
|
|
"valid_targets_min": 3180
|
|
},
|
|
{
|
|
"epoch": 1.6480000000000001,
|
|
"grad_norm": 0.5650303061621147,
|
|
"learning_rate": 3.7816868913578044e-05,
|
|
"loss": 0.2589,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2785050868988037,
|
|
"step": 1030,
|
|
"valid_targets_mean": 4323.9,
|
|
"valid_targets_min": 2326
|
|
},
|
|
{
|
|
"epoch": 1.6560000000000001,
|
|
"grad_norm": 0.5540344684978876,
|
|
"learning_rate": 3.778047476626897e-05,
|
|
"loss": 0.2594,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26538190245628357,
|
|
"step": 1035,
|
|
"valid_targets_mean": 4063.5,
|
|
"valid_targets_min": 2830
|
|
},
|
|
{
|
|
"epoch": 1.6640000000000001,
|
|
"grad_norm": 0.5448161540115258,
|
|
"learning_rate": 3.7743797576427335e-05,
|
|
"loss": 0.2421,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22615385055541992,
|
|
"step": 1040,
|
|
"valid_targets_mean": 4292.4,
|
|
"valid_targets_min": 2652
|
|
},
|
|
{
|
|
"epoch": 1.6720000000000002,
|
|
"grad_norm": 0.5452992120992455,
|
|
"learning_rate": 3.770683792790733e-05,
|
|
"loss": 0.2516,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2638772428035736,
|
|
"step": 1045,
|
|
"valid_targets_mean": 4576.2,
|
|
"valid_targets_min": 3502
|
|
},
|
|
{
|
|
"epoch": 1.6800000000000002,
|
|
"grad_norm": 0.5645675763072664,
|
|
"learning_rate": 3.766959640905954e-05,
|
|
"loss": 0.2603,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26780176162719727,
|
|
"step": 1050,
|
|
"valid_targets_mean": 4682.9,
|
|
"valid_targets_min": 3309
|
|
},
|
|
{
|
|
"epoch": 1.688,
|
|
"grad_norm": 0.5913240298197293,
|
|
"learning_rate": 3.763207361272153e-05,
|
|
"loss": 0.2606,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.27208465337753296,
|
|
"step": 1055,
|
|
"valid_targets_mean": 4461.3,
|
|
"valid_targets_min": 3217
|
|
},
|
|
{
|
|
"epoch": 1.696,
|
|
"grad_norm": 0.5310912018370454,
|
|
"learning_rate": 3.759427013620849e-05,
|
|
"loss": 0.2519,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2439044713973999,
|
|
"step": 1060,
|
|
"valid_targets_mean": 4545.6,
|
|
"valid_targets_min": 3753
|
|
},
|
|
{
|
|
"epoch": 1.704,
|
|
"grad_norm": 0.5442454084976005,
|
|
"learning_rate": 3.755618658130366e-05,
|
|
"loss": 0.2533,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22730106115341187,
|
|
"step": 1065,
|
|
"valid_targets_mean": 4587.8,
|
|
"valid_targets_min": 2988
|
|
},
|
|
{
|
|
"epoch": 1.712,
|
|
"grad_norm": 0.6249479734007123,
|
|
"learning_rate": 3.751782355424877e-05,
|
|
"loss": 0.2651,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2661554515361786,
|
|
"step": 1070,
|
|
"valid_targets_mean": 4075.0,
|
|
"valid_targets_min": 1221
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"grad_norm": 0.5594095396821931,
|
|
"learning_rate": 3.7479181665734395e-05,
|
|
"loss": 0.2575,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25086405873298645,
|
|
"step": 1075,
|
|
"valid_targets_mean": 4205.8,
|
|
"valid_targets_min": 796
|
|
},
|
|
{
|
|
"epoch": 1.728,
|
|
"grad_norm": 0.5359372331708528,
|
|
"learning_rate": 3.7440261530890213e-05,
|
|
"loss": 0.2507,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2717932164669037,
|
|
"step": 1080,
|
|
"valid_targets_mean": 4632.9,
|
|
"valid_targets_min": 1354
|
|
},
|
|
{
|
|
"epoch": 1.736,
|
|
"grad_norm": 0.5880204806209435,
|
|
"learning_rate": 3.740106376927527e-05,
|
|
"loss": 0.2669,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2813637852668762,
|
|
"step": 1085,
|
|
"valid_targets_mean": 4721.8,
|
|
"valid_targets_min": 2708
|
|
},
|
|
{
|
|
"epoch": 1.744,
|
|
"grad_norm": 0.6246292636354145,
|
|
"learning_rate": 3.7361589004868035e-05,
|
|
"loss": 0.257,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24439872801303864,
|
|
"step": 1090,
|
|
"valid_targets_mean": 4300.0,
|
|
"valid_targets_min": 3046
|
|
},
|
|
{
|
|
"epoch": 1.752,
|
|
"grad_norm": 0.5435672541179415,
|
|
"learning_rate": 3.7321837866056535e-05,
|
|
"loss": 0.25,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2435004711151123,
|
|
"step": 1095,
|
|
"valid_targets_mean": 4501.7,
|
|
"valid_targets_min": 3837
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"grad_norm": 0.5783303996107718,
|
|
"learning_rate": 3.728181098562831e-05,
|
|
"loss": 0.2522,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24512404203414917,
|
|
"step": 1100,
|
|
"valid_targets_mean": 4474.1,
|
|
"valid_targets_min": 3235
|
|
},
|
|
{
|
|
"epoch": 1.768,
|
|
"grad_norm": 0.6504547213798294,
|
|
"learning_rate": 3.7241509000760355e-05,
|
|
"loss": 0.2402,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.226511150598526,
|
|
"step": 1105,
|
|
"valid_targets_mean": 4137.1,
|
|
"valid_targets_min": 533
|
|
},
|
|
{
|
|
"epoch": 1.776,
|
|
"grad_norm": 0.5624265613083258,
|
|
"learning_rate": 3.720093255300899e-05,
|
|
"loss": 0.2543,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24739189445972443,
|
|
"step": 1110,
|
|
"valid_targets_mean": 4443.7,
|
|
"valid_targets_min": 3355
|
|
},
|
|
{
|
|
"epoch": 1.784,
|
|
"grad_norm": 0.5524103339705932,
|
|
"learning_rate": 3.7160082288299645e-05,
|
|
"loss": 0.2558,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2450713813304901,
|
|
"step": 1115,
|
|
"valid_targets_mean": 4423.1,
|
|
"valid_targets_min": 892
|
|
},
|
|
{
|
|
"epoch": 1.792,
|
|
"grad_norm": 0.560112598757554,
|
|
"learning_rate": 3.7118958856916534e-05,
|
|
"loss": 0.2546,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2667468190193176,
|
|
"step": 1120,
|
|
"valid_targets_mean": 4576.2,
|
|
"valid_targets_min": 3312
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"grad_norm": 0.56391540580291,
|
|
"learning_rate": 3.707756291349237e-05,
|
|
"loss": 0.2533,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23388142883777618,
|
|
"step": 1125,
|
|
"valid_targets_mean": 4231.3,
|
|
"valid_targets_min": 2503
|
|
},
|
|
{
|
|
"epoch": 1.808,
|
|
"grad_norm": 0.6705583659290554,
|
|
"learning_rate": 3.703589511699787e-05,
|
|
"loss": 0.2572,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24915233254432678,
|
|
"step": 1130,
|
|
"valid_targets_mean": 4158.8,
|
|
"valid_targets_min": 1585
|
|
},
|
|
{
|
|
"epoch": 1.8159999999999998,
|
|
"grad_norm": 0.5334190725316186,
|
|
"learning_rate": 3.6993956130731355e-05,
|
|
"loss": 0.2544,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26935499906539917,
|
|
"step": 1135,
|
|
"valid_targets_mean": 4735.7,
|
|
"valid_targets_min": 2868
|
|
},
|
|
{
|
|
"epoch": 1.8239999999999998,
|
|
"grad_norm": 0.5587526656592765,
|
|
"learning_rate": 3.6951746622308106e-05,
|
|
"loss": 0.2562,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2673352360725403,
|
|
"step": 1140,
|
|
"valid_targets_mean": 4128.7,
|
|
"valid_targets_min": 437
|
|
},
|
|
{
|
|
"epoch": 1.8319999999999999,
|
|
"grad_norm": 0.6000198800309797,
|
|
"learning_rate": 3.69092672636498e-05,
|
|
"loss": 0.2637,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26417607069015503,
|
|
"step": 1145,
|
|
"valid_targets_mean": 4819.9,
|
|
"valid_targets_min": 3980
|
|
},
|
|
{
|
|
"epoch": 1.8399999999999999,
|
|
"grad_norm": 1.0018845778078547,
|
|
"learning_rate": 3.686651873097375e-05,
|
|
"loss": 0.2515,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23706850409507751,
|
|
"step": 1150,
|
|
"valid_targets_mean": 4489.2,
|
|
"valid_targets_min": 2588
|
|
},
|
|
{
|
|
"epoch": 1.8479999999999999,
|
|
"grad_norm": 0.6054771563228303,
|
|
"learning_rate": 3.682350170478223e-05,
|
|
"loss": 0.2615,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24834175407886505,
|
|
"step": 1155,
|
|
"valid_targets_mean": 4525.9,
|
|
"valid_targets_min": 2918
|
|
},
|
|
{
|
|
"epoch": 1.8559999999999999,
|
|
"grad_norm": 0.6277848376697793,
|
|
"learning_rate": 3.678021686985153e-05,
|
|
"loss": 0.2494,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24348101019859314,
|
|
"step": 1160,
|
|
"valid_targets_mean": 4410.1,
|
|
"valid_targets_min": 3588
|
|
},
|
|
{
|
|
"epoch": 1.8639999999999999,
|
|
"grad_norm": 0.5840515707369159,
|
|
"learning_rate": 3.6736664915221144e-05,
|
|
"loss": 0.2489,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2422848790884018,
|
|
"step": 1165,
|
|
"valid_targets_mean": 4572.8,
|
|
"valid_targets_min": 3472
|
|
},
|
|
{
|
|
"epoch": 1.8719999999999999,
|
|
"grad_norm": 0.5503238054065828,
|
|
"learning_rate": 3.669284653418278e-05,
|
|
"loss": 0.2497,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2383195012807846,
|
|
"step": 1170,
|
|
"valid_targets_mean": 4797.6,
|
|
"valid_targets_min": 4136
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"grad_norm": 0.5109385304599156,
|
|
"learning_rate": 3.6648762424269306e-05,
|
|
"loss": 0.2513,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23124974966049194,
|
|
"step": 1175,
|
|
"valid_targets_mean": 4797.1,
|
|
"valid_targets_min": 3923
|
|
},
|
|
{
|
|
"epoch": 1.888,
|
|
"grad_norm": 0.5681843788769078,
|
|
"learning_rate": 3.660441328724365e-05,
|
|
"loss": 0.2415,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2655305564403534,
|
|
"step": 1180,
|
|
"valid_targets_mean": 4607.1,
|
|
"valid_targets_min": 3000
|
|
},
|
|
{
|
|
"epoch": 1.896,
|
|
"grad_norm": 0.6152564643722765,
|
|
"learning_rate": 3.655979982908764e-05,
|
|
"loss": 0.2543,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23537468910217285,
|
|
"step": 1185,
|
|
"valid_targets_mean": 4237.4,
|
|
"valid_targets_min": 3391
|
|
},
|
|
{
|
|
"epoch": 1.904,
|
|
"grad_norm": 0.5692146128948513,
|
|
"learning_rate": 3.6514922759990756e-05,
|
|
"loss": 0.2603,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2569218575954437,
|
|
"step": 1190,
|
|
"valid_targets_mean": 4222.9,
|
|
"valid_targets_min": 1753
|
|
},
|
|
{
|
|
"epoch": 1.912,
|
|
"grad_norm": 0.5165210689236093,
|
|
"learning_rate": 3.646978279433883e-05,
|
|
"loss": 0.2349,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20811985433101654,
|
|
"step": 1195,
|
|
"valid_targets_mean": 4657.8,
|
|
"valid_targets_min": 2910
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"grad_norm": 0.5515435488514879,
|
|
"learning_rate": 3.6424380650702685e-05,
|
|
"loss": 0.2528,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2342337965965271,
|
|
"step": 1200,
|
|
"valid_targets_mean": 4510.0,
|
|
"valid_targets_min": 3958
|
|
},
|
|
{
|
|
"epoch": 1.928,
|
|
"grad_norm": 0.6117191288823864,
|
|
"learning_rate": 3.637871705182667e-05,
|
|
"loss": 0.2398,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23031620681285858,
|
|
"step": 1205,
|
|
"valid_targets_mean": 4567.4,
|
|
"valid_targets_min": 3685
|
|
},
|
|
{
|
|
"epoch": 1.936,
|
|
"grad_norm": 0.5933744227233284,
|
|
"learning_rate": 3.633279272461717e-05,
|
|
"loss": 0.2531,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.27962976694107056,
|
|
"step": 1210,
|
|
"valid_targets_mean": 4395.9,
|
|
"valid_targets_min": 2657
|
|
},
|
|
{
|
|
"epoch": 1.944,
|
|
"grad_norm": 0.5624967272091034,
|
|
"learning_rate": 3.628660840013102e-05,
|
|
"loss": 0.2556,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23646654188632965,
|
|
"step": 1215,
|
|
"valid_targets_mean": 4179.0,
|
|
"valid_targets_min": 1367
|
|
},
|
|
{
|
|
"epoch": 1.952,
|
|
"grad_norm": 0.5663348573085811,
|
|
"learning_rate": 3.624016481356392e-05,
|
|
"loss": 0.259,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2401682585477829,
|
|
"step": 1220,
|
|
"valid_targets_mean": 4683.1,
|
|
"valid_targets_min": 2221
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"grad_norm": 0.5709404939982533,
|
|
"learning_rate": 3.619346270423866e-05,
|
|
"loss": 0.2447,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22117024660110474,
|
|
"step": 1225,
|
|
"valid_targets_mean": 4775.0,
|
|
"valid_targets_min": 3746
|
|
},
|
|
{
|
|
"epoch": 1.968,
|
|
"grad_norm": 0.5733088245129365,
|
|
"learning_rate": 3.6146502815593384e-05,
|
|
"loss": 0.2612,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2772558629512787,
|
|
"step": 1230,
|
|
"valid_targets_mean": 4618.6,
|
|
"valid_targets_min": 446
|
|
},
|
|
{
|
|
"epoch": 1.976,
|
|
"grad_norm": 0.5354336910437538,
|
|
"learning_rate": 3.609928589516977e-05,
|
|
"loss": 0.2575,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24273596704006195,
|
|
"step": 1235,
|
|
"valid_targets_mean": 4692.0,
|
|
"valid_targets_min": 3668
|
|
},
|
|
{
|
|
"epoch": 1.984,
|
|
"grad_norm": 0.538298311452803,
|
|
"learning_rate": 3.6051812694601114e-05,
|
|
"loss": 0.2411,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24419942498207092,
|
|
"step": 1240,
|
|
"valid_targets_mean": 4465.9,
|
|
"valid_targets_min": 2541
|
|
},
|
|
{
|
|
"epoch": 1.992,
|
|
"grad_norm": 0.5152690007154265,
|
|
"learning_rate": 3.6004083969600346e-05,
|
|
"loss": 0.2306,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.227997824549675,
|
|
"step": 1245,
|
|
"valid_targets_mean": 4827.6,
|
|
"valid_targets_min": 3779
|
|
},
|
|
{
|
|
"epoch": 2.0,
|
|
"grad_norm": 0.5501664179681109,
|
|
"learning_rate": 3.595610047994804e-05,
|
|
"loss": 0.2481,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25922834873199463,
|
|
"step": 1250,
|
|
"valid_targets_mean": 4440.4,
|
|
"valid_targets_min": 2379
|
|
},
|
|
{
|
|
"epoch": 2.008,
|
|
"grad_norm": 0.5582146595448779,
|
|
"learning_rate": 3.5907862989480285e-05,
|
|
"loss": 0.2163,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2133411467075348,
|
|
"step": 1255,
|
|
"valid_targets_mean": 4456.1,
|
|
"valid_targets_min": 2887
|
|
},
|
|
{
|
|
"epoch": 2.016,
|
|
"grad_norm": 0.5832136209121199,
|
|
"learning_rate": 3.585937226607656e-05,
|
|
"loss": 0.2177,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23650860786437988,
|
|
"step": 1260,
|
|
"valid_targets_mean": 4759.6,
|
|
"valid_targets_min": 3876
|
|
},
|
|
{
|
|
"epoch": 2.024,
|
|
"grad_norm": 0.6031189148926421,
|
|
"learning_rate": 3.5810629081647476e-05,
|
|
"loss": 0.2332,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24701716005802155,
|
|
"step": 1265,
|
|
"valid_targets_mean": 4589.8,
|
|
"valid_targets_min": 3205
|
|
},
|
|
{
|
|
"epoch": 2.032,
|
|
"grad_norm": 0.5617151239416491,
|
|
"learning_rate": 3.576163421212249e-05,
|
|
"loss": 0.2308,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24185265600681305,
|
|
"step": 1270,
|
|
"valid_targets_mean": 5003.8,
|
|
"valid_targets_min": 3710
|
|
},
|
|
{
|
|
"epoch": 2.04,
|
|
"grad_norm": 0.6269511081142009,
|
|
"learning_rate": 3.5712388437437576e-05,
|
|
"loss": 0.2215,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21043658256530762,
|
|
"step": 1275,
|
|
"valid_targets_mean": 3758.2,
|
|
"valid_targets_min": 441
|
|
},
|
|
{
|
|
"epoch": 2.048,
|
|
"grad_norm": 0.5946150451854865,
|
|
"learning_rate": 3.566289254152283e-05,
|
|
"loss": 0.2323,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22661244869232178,
|
|
"step": 1280,
|
|
"valid_targets_mean": 4625.9,
|
|
"valid_targets_min": 3242
|
|
},
|
|
{
|
|
"epoch": 2.056,
|
|
"grad_norm": 0.6364207693410725,
|
|
"learning_rate": 3.56131473122899e-05,
|
|
"loss": 0.2362,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23021018505096436,
|
|
"step": 1285,
|
|
"valid_targets_mean": 4578.1,
|
|
"valid_targets_min": 3798
|
|
},
|
|
{
|
|
"epoch": 2.064,
|
|
"grad_norm": 0.5751777337459273,
|
|
"learning_rate": 3.556315354161955e-05,
|
|
"loss": 0.2192,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19548694789409637,
|
|
"step": 1290,
|
|
"valid_targets_mean": 4301.4,
|
|
"valid_targets_min": 3502
|
|
},
|
|
{
|
|
"epoch": 2.072,
|
|
"grad_norm": 0.6691024998509097,
|
|
"learning_rate": 3.551291202534899e-05,
|
|
"loss": 0.2251,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2374153584241867,
|
|
"step": 1295,
|
|
"valid_targets_mean": 4386.8,
|
|
"valid_targets_min": 1043
|
|
},
|
|
{
|
|
"epoch": 2.08,
|
|
"grad_norm": 0.523908160573919,
|
|
"learning_rate": 3.546242356325922e-05,
|
|
"loss": 0.227,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1890295296907425,
|
|
"step": 1300,
|
|
"valid_targets_mean": 4714.5,
|
|
"valid_targets_min": 3610
|
|
},
|
|
{
|
|
"epoch": 2.088,
|
|
"grad_norm": 0.6236897317896717,
|
|
"learning_rate": 3.5411688959062323e-05,
|
|
"loss": 0.2159,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20969998836517334,
|
|
"step": 1305,
|
|
"valid_targets_mean": 4241.5,
|
|
"valid_targets_min": 912
|
|
},
|
|
{
|
|
"epoch": 2.096,
|
|
"grad_norm": 0.5825736082019941,
|
|
"learning_rate": 3.5360709020388625e-05,
|
|
"loss": 0.2178,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2224734127521515,
|
|
"step": 1310,
|
|
"valid_targets_mean": 4681.1,
|
|
"valid_targets_min": 2248
|
|
},
|
|
{
|
|
"epoch": 2.104,
|
|
"grad_norm": 0.6527714324210339,
|
|
"learning_rate": 3.530948455877388e-05,
|
|
"loss": 0.2294,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21715235710144043,
|
|
"step": 1315,
|
|
"valid_targets_mean": 4307.6,
|
|
"valid_targets_min": 1201
|
|
},
|
|
{
|
|
"epoch": 2.112,
|
|
"grad_norm": 0.6912181604104021,
|
|
"learning_rate": 3.525801638964634e-05,
|
|
"loss": 0.229,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2568296790122986,
|
|
"step": 1320,
|
|
"valid_targets_mean": 4816.1,
|
|
"valid_targets_min": 4028
|
|
},
|
|
{
|
|
"epoch": 2.12,
|
|
"grad_norm": 0.6523923553140656,
|
|
"learning_rate": 3.520630533231376e-05,
|
|
"loss": 0.2165,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21029867231845856,
|
|
"step": 1325,
|
|
"valid_targets_mean": 4628.9,
|
|
"valid_targets_min": 3449
|
|
},
|
|
{
|
|
"epoch": 2.128,
|
|
"grad_norm": 0.5762203347149927,
|
|
"learning_rate": 3.5154352209950376e-05,
|
|
"loss": 0.2245,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23122583329677582,
|
|
"step": 1330,
|
|
"valid_targets_mean": 4750.0,
|
|
"valid_targets_min": 2840
|
|
},
|
|
{
|
|
"epoch": 2.136,
|
|
"grad_norm": 0.5561466781055587,
|
|
"learning_rate": 3.510215784958376e-05,
|
|
"loss": 0.2122,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20478473603725433,
|
|
"step": 1335,
|
|
"valid_targets_mean": 4541.6,
|
|
"valid_targets_min": 2508
|
|
},
|
|
{
|
|
"epoch": 2.144,
|
|
"grad_norm": 0.5875253403761592,
|
|
"learning_rate": 3.5049723082081755e-05,
|
|
"loss": 0.2351,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24400833249092102,
|
|
"step": 1340,
|
|
"valid_targets_mean": 4531.4,
|
|
"valid_targets_min": 3384
|
|
},
|
|
{
|
|
"epoch": 2.152,
|
|
"grad_norm": 0.5545917744266169,
|
|
"learning_rate": 3.49970487421391e-05,
|
|
"loss": 0.2133,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19406016170978546,
|
|
"step": 1345,
|
|
"valid_targets_mean": 4566.6,
|
|
"valid_targets_min": 3716
|
|
},
|
|
{
|
|
"epoch": 2.16,
|
|
"grad_norm": 0.6154352205361276,
|
|
"learning_rate": 3.494413566826427e-05,
|
|
"loss": 0.2214,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22522397339344025,
|
|
"step": 1350,
|
|
"valid_targets_mean": 4323.7,
|
|
"valid_targets_min": 1594
|
|
},
|
|
{
|
|
"epoch": 2.168,
|
|
"grad_norm": 0.5723304524279321,
|
|
"learning_rate": 3.489098470276608e-05,
|
|
"loss": 0.2241,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21622180938720703,
|
|
"step": 1355,
|
|
"valid_targets_mean": 4652.8,
|
|
"valid_targets_min": 3421
|
|
},
|
|
{
|
|
"epoch": 2.176,
|
|
"grad_norm": 0.5602236779823958,
|
|
"learning_rate": 3.483759669174024e-05,
|
|
"loss": 0.2232,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23774118721485138,
|
|
"step": 1360,
|
|
"valid_targets_mean": 4575.8,
|
|
"valid_targets_min": 3694
|
|
},
|
|
{
|
|
"epoch": 2.184,
|
|
"grad_norm": 0.6501857971658852,
|
|
"learning_rate": 3.478397248505598e-05,
|
|
"loss": 0.2126,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2366846352815628,
|
|
"step": 1365,
|
|
"valid_targets_mean": 4061.2,
|
|
"valid_targets_min": 1616
|
|
},
|
|
{
|
|
"epoch": 2.192,
|
|
"grad_norm": 0.5957271107147547,
|
|
"learning_rate": 3.473011293634241e-05,
|
|
"loss": 0.2204,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20847156643867493,
|
|
"step": 1370,
|
|
"valid_targets_mean": 4440.6,
|
|
"valid_targets_min": 2878
|
|
},
|
|
{
|
|
"epoch": 2.2,
|
|
"grad_norm": 0.5931651328036286,
|
|
"learning_rate": 3.467601890297502e-05,
|
|
"loss": 0.2291,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24591544270515442,
|
|
"step": 1375,
|
|
"valid_targets_mean": 4635.2,
|
|
"valid_targets_min": 3625
|
|
},
|
|
{
|
|
"epoch": 2.208,
|
|
"grad_norm": 0.5822104956753541,
|
|
"learning_rate": 3.4621691246061976e-05,
|
|
"loss": 0.2104,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2121216356754303,
|
|
"step": 1380,
|
|
"valid_targets_mean": 4441.4,
|
|
"valid_targets_min": 2221
|
|
},
|
|
{
|
|
"epoch": 2.216,
|
|
"grad_norm": 0.5724289879845731,
|
|
"learning_rate": 3.456713083043046e-05,
|
|
"loss": 0.2136,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20560818910598755,
|
|
"step": 1385,
|
|
"valid_targets_mean": 4653.3,
|
|
"valid_targets_min": 4009
|
|
},
|
|
{
|
|
"epoch": 2.224,
|
|
"grad_norm": 0.6083489497060326,
|
|
"learning_rate": 3.451233852461285e-05,
|
|
"loss": 0.221,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22648689150810242,
|
|
"step": 1390,
|
|
"valid_targets_mean": 4486.5,
|
|
"valid_targets_min": 2221
|
|
},
|
|
{
|
|
"epoch": 2.232,
|
|
"grad_norm": 0.7044793738522117,
|
|
"learning_rate": 3.4457315200832935e-05,
|
|
"loss": 0.2223,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20412814617156982,
|
|
"step": 1395,
|
|
"valid_targets_mean": 4629.4,
|
|
"valid_targets_min": 4149
|
|
},
|
|
{
|
|
"epoch": 2.24,
|
|
"grad_norm": 0.675805084465064,
|
|
"learning_rate": 3.440206173499201e-05,
|
|
"loss": 0.2233,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.221166729927063,
|
|
"step": 1400,
|
|
"valid_targets_mean": 4029.9,
|
|
"valid_targets_min": 533
|
|
},
|
|
{
|
|
"epoch": 2.248,
|
|
"grad_norm": 0.6139637216380665,
|
|
"learning_rate": 3.4346579006654945e-05,
|
|
"loss": 0.2176,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20849201083183289,
|
|
"step": 1405,
|
|
"valid_targets_mean": 4401.9,
|
|
"valid_targets_min": 2550
|
|
},
|
|
{
|
|
"epoch": 2.2560000000000002,
|
|
"grad_norm": 0.584295202695642,
|
|
"learning_rate": 3.4290867899036166e-05,
|
|
"loss": 0.2237,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22592870891094208,
|
|
"step": 1410,
|
|
"valid_targets_mean": 4455.4,
|
|
"valid_targets_min": 1624
|
|
},
|
|
{
|
|
"epoch": 2.2640000000000002,
|
|
"grad_norm": 0.6106464865196701,
|
|
"learning_rate": 3.4234929298985614e-05,
|
|
"loss": 0.2132,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21731606125831604,
|
|
"step": 1415,
|
|
"valid_targets_mean": 4518.1,
|
|
"valid_targets_min": 2878
|
|
},
|
|
{
|
|
"epoch": 2.2720000000000002,
|
|
"grad_norm": 0.6122124015347076,
|
|
"learning_rate": 3.417876409697463e-05,
|
|
"loss": 0.2389,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21166938543319702,
|
|
"step": 1420,
|
|
"valid_targets_mean": 4538.2,
|
|
"valid_targets_min": 1875
|
|
},
|
|
{
|
|
"epoch": 2.2800000000000002,
|
|
"grad_norm": 0.5694359002193767,
|
|
"learning_rate": 3.412237318708175e-05,
|
|
"loss": 0.2223,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20801980793476105,
|
|
"step": 1425,
|
|
"valid_targets_mean": 4755.6,
|
|
"valid_targets_min": 4182
|
|
},
|
|
{
|
|
"epoch": 2.288,
|
|
"grad_norm": 0.5754851921698292,
|
|
"learning_rate": 3.4065757466978504e-05,
|
|
"loss": 0.2123,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17738023400306702,
|
|
"step": 1430,
|
|
"valid_targets_mean": 3904.9,
|
|
"valid_targets_min": 1533
|
|
},
|
|
{
|
|
"epoch": 2.296,
|
|
"grad_norm": 0.6142207630654527,
|
|
"learning_rate": 3.400891783791511e-05,
|
|
"loss": 0.2231,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21368850767612457,
|
|
"step": 1435,
|
|
"valid_targets_mean": 4456.6,
|
|
"valid_targets_min": 3156
|
|
},
|
|
{
|
|
"epoch": 2.304,
|
|
"grad_norm": 0.6087453093160878,
|
|
"learning_rate": 3.395185520470614e-05,
|
|
"loss": 0.2334,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2234778106212616,
|
|
"step": 1440,
|
|
"valid_targets_mean": 4194.8,
|
|
"valid_targets_min": 2003
|
|
},
|
|
{
|
|
"epoch": 2.312,
|
|
"grad_norm": 0.6390763078439579,
|
|
"learning_rate": 3.38945704757161e-05,
|
|
"loss": 0.217,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20924900472164154,
|
|
"step": 1445,
|
|
"valid_targets_mean": 4646.9,
|
|
"valid_targets_min": 4133
|
|
},
|
|
{
|
|
"epoch": 2.32,
|
|
"grad_norm": 0.5458562309340034,
|
|
"learning_rate": 3.383706456284498e-05,
|
|
"loss": 0.2127,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22154280543327332,
|
|
"step": 1450,
|
|
"valid_targets_mean": 4744.4,
|
|
"valid_targets_min": 2527
|
|
},
|
|
{
|
|
"epoch": 2.328,
|
|
"grad_norm": 0.5610168688707511,
|
|
"learning_rate": 3.377933838151374e-05,
|
|
"loss": 0.2183,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19419562816619873,
|
|
"step": 1455,
|
|
"valid_targets_mean": 4503.9,
|
|
"valid_targets_min": 2663
|
|
},
|
|
{
|
|
"epoch": 2.336,
|
|
"grad_norm": 0.6048378754904387,
|
|
"learning_rate": 3.3721392850649714e-05,
|
|
"loss": 0.2146,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19261109828948975,
|
|
"step": 1460,
|
|
"valid_targets_mean": 4673.2,
|
|
"valid_targets_min": 2663
|
|
},
|
|
{
|
|
"epoch": 2.344,
|
|
"grad_norm": 0.7708286198650987,
|
|
"learning_rate": 3.3663228892672034e-05,
|
|
"loss": 0.2202,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20756103098392487,
|
|
"step": 1465,
|
|
"valid_targets_mean": 4660.4,
|
|
"valid_targets_min": 4014
|
|
},
|
|
{
|
|
"epoch": 2.352,
|
|
"grad_norm": 0.5999742101817844,
|
|
"learning_rate": 3.36048474334769e-05,
|
|
"loss": 0.2287,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2102992832660675,
|
|
"step": 1470,
|
|
"valid_targets_mean": 4573.2,
|
|
"valid_targets_min": 3879
|
|
},
|
|
{
|
|
"epoch": 2.36,
|
|
"grad_norm": 0.5528159649898882,
|
|
"learning_rate": 3.3546249402422834e-05,
|
|
"loss": 0.2268,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19125226140022278,
|
|
"step": 1475,
|
|
"valid_targets_mean": 4687.9,
|
|
"valid_targets_min": 3621
|
|
},
|
|
{
|
|
"epoch": 2.368,
|
|
"grad_norm": 0.593007208345445,
|
|
"learning_rate": 3.3487435732315944e-05,
|
|
"loss": 0.2207,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21677760779857635,
|
|
"step": 1480,
|
|
"valid_targets_mean": 4440.5,
|
|
"valid_targets_min": 752
|
|
},
|
|
{
|
|
"epoch": 2.376,
|
|
"grad_norm": 0.6093062964033565,
|
|
"learning_rate": 3.342840735939501e-05,
|
|
"loss": 0.2247,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21120436489582062,
|
|
"step": 1485,
|
|
"valid_targets_mean": 4668.5,
|
|
"valid_targets_min": 3334
|
|
},
|
|
{
|
|
"epoch": 2.384,
|
|
"grad_norm": 0.6096665292799268,
|
|
"learning_rate": 3.33691652233166e-05,
|
|
"loss": 0.2079,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2017737329006195,
|
|
"step": 1490,
|
|
"valid_targets_mean": 4532.2,
|
|
"valid_targets_min": 3431
|
|
},
|
|
{
|
|
"epoch": 2.392,
|
|
"grad_norm": 0.6287681957914089,
|
|
"learning_rate": 3.330971026714016e-05,
|
|
"loss": 0.2261,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21089962124824524,
|
|
"step": 1495,
|
|
"valid_targets_mean": 4316.1,
|
|
"valid_targets_min": 2204
|
|
},
|
|
{
|
|
"epoch": 2.4,
|
|
"grad_norm": 0.5988690159439845,
|
|
"learning_rate": 3.325004343731292e-05,
|
|
"loss": 0.2177,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20130027830600739,
|
|
"step": 1500,
|
|
"valid_targets_mean": 4487.2,
|
|
"valid_targets_min": 2850
|
|
},
|
|
{
|
|
"epoch": 2.408,
|
|
"grad_norm": 0.622060571137543,
|
|
"learning_rate": 3.3190165683654885e-05,
|
|
"loss": 0.2263,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2422848343849182,
|
|
"step": 1505,
|
|
"valid_targets_mean": 4609.2,
|
|
"valid_targets_min": 2241
|
|
},
|
|
{
|
|
"epoch": 2.416,
|
|
"grad_norm": 0.6059809489279075,
|
|
"learning_rate": 3.31300779593437e-05,
|
|
"loss": 0.21,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21591085195541382,
|
|
"step": 1510,
|
|
"valid_targets_mean": 4314.2,
|
|
"valid_targets_min": 2326
|
|
},
|
|
{
|
|
"epoch": 2.424,
|
|
"grad_norm": 0.5919456096931015,
|
|
"learning_rate": 3.306978122089948e-05,
|
|
"loss": 0.2165,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2247156798839569,
|
|
"step": 1515,
|
|
"valid_targets_mean": 4672.9,
|
|
"valid_targets_min": 2754
|
|
},
|
|
{
|
|
"epoch": 2.432,
|
|
"grad_norm": 0.6205304379312415,
|
|
"learning_rate": 3.300927642816957e-05,
|
|
"loss": 0.2195,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23405715823173523,
|
|
"step": 1520,
|
|
"valid_targets_mean": 4451.4,
|
|
"valid_targets_min": 3697
|
|
},
|
|
{
|
|
"epoch": 2.44,
|
|
"grad_norm": 0.5865497862035505,
|
|
"learning_rate": 3.294856454431328e-05,
|
|
"loss": 0.206,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.200160950422287,
|
|
"step": 1525,
|
|
"valid_targets_mean": 4359.7,
|
|
"valid_targets_min": 3552
|
|
},
|
|
{
|
|
"epoch": 2.448,
|
|
"grad_norm": 0.7362852858098001,
|
|
"learning_rate": 3.288764653578653e-05,
|
|
"loss": 0.2253,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19714310765266418,
|
|
"step": 1530,
|
|
"valid_targets_mean": 4689.0,
|
|
"valid_targets_min": 3948
|
|
},
|
|
{
|
|
"epoch": 2.456,
|
|
"grad_norm": 0.6326535269418605,
|
|
"learning_rate": 3.2826523372326516e-05,
|
|
"loss": 0.2185,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2291913479566574,
|
|
"step": 1535,
|
|
"valid_targets_mean": 4435.3,
|
|
"valid_targets_min": 3527
|
|
},
|
|
{
|
|
"epoch": 2.464,
|
|
"grad_norm": 0.6462541377963396,
|
|
"learning_rate": 3.276519602693621e-05,
|
|
"loss": 0.2197,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2315208464860916,
|
|
"step": 1540,
|
|
"valid_targets_mean": 4310.9,
|
|
"valid_targets_min": 2926
|
|
},
|
|
{
|
|
"epoch": 2.472,
|
|
"grad_norm": 0.6142955882767627,
|
|
"learning_rate": 3.270366547586892e-05,
|
|
"loss": 0.2168,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2122419774532318,
|
|
"step": 1545,
|
|
"valid_targets_mean": 4815.6,
|
|
"valid_targets_min": 3566
|
|
},
|
|
{
|
|
"epoch": 2.48,
|
|
"grad_norm": 0.6234588696866334,
|
|
"learning_rate": 3.2641932698612715e-05,
|
|
"loss": 0.2155,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23945504426956177,
|
|
"step": 1550,
|
|
"valid_targets_mean": 4477.9,
|
|
"valid_targets_min": 2277
|
|
},
|
|
{
|
|
"epoch": 2.488,
|
|
"grad_norm": 0.6465714584755653,
|
|
"learning_rate": 3.2579998677874855e-05,
|
|
"loss": 0.2151,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24272841215133667,
|
|
"step": 1555,
|
|
"valid_targets_mean": 4330.6,
|
|
"valid_targets_min": 1064
|
|
},
|
|
{
|
|
"epoch": 2.496,
|
|
"grad_norm": 0.6658121536084254,
|
|
"learning_rate": 3.251786439956614e-05,
|
|
"loss": 0.2129,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20680904388427734,
|
|
"step": 1560,
|
|
"valid_targets_mean": 4540.8,
|
|
"valid_targets_min": 2173
|
|
},
|
|
{
|
|
"epoch": 2.504,
|
|
"grad_norm": 0.6294512444998034,
|
|
"learning_rate": 3.2455530852785206e-05,
|
|
"loss": 0.2064,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19804677367210388,
|
|
"step": 1565,
|
|
"valid_targets_mean": 4754.9,
|
|
"valid_targets_min": 4053
|
|
},
|
|
{
|
|
"epoch": 2.512,
|
|
"grad_norm": 0.6599328770433809,
|
|
"learning_rate": 3.239299902980281e-05,
|
|
"loss": 0.2156,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23814725875854492,
|
|
"step": 1570,
|
|
"valid_targets_mean": 4246.0,
|
|
"valid_targets_min": 2775
|
|
},
|
|
{
|
|
"epoch": 2.52,
|
|
"grad_norm": 0.6200616486403315,
|
|
"learning_rate": 3.2330269926046e-05,
|
|
"loss": 0.2205,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2262628823518753,
|
|
"step": 1575,
|
|
"valid_targets_mean": 4568.5,
|
|
"valid_targets_min": 2182
|
|
},
|
|
{
|
|
"epoch": 2.528,
|
|
"grad_norm": 0.5766621316929544,
|
|
"learning_rate": 3.2267344540082284e-05,
|
|
"loss": 0.1976,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19369341433048248,
|
|
"step": 1580,
|
|
"valid_targets_mean": 4488.2,
|
|
"valid_targets_min": 3647
|
|
},
|
|
{
|
|
"epoch": 2.536,
|
|
"grad_norm": 0.7579321808369266,
|
|
"learning_rate": 3.220422387360373e-05,
|
|
"loss": 0.2229,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2228652536869049,
|
|
"step": 1585,
|
|
"valid_targets_mean": 4433.4,
|
|
"valid_targets_min": 1991
|
|
},
|
|
{
|
|
"epoch": 2.544,
|
|
"grad_norm": 0.6422731585042506,
|
|
"learning_rate": 3.2140908931411026e-05,
|
|
"loss": 0.2333,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2688996195793152,
|
|
"step": 1590,
|
|
"valid_targets_mean": 4637.2,
|
|
"valid_targets_min": 3927
|
|
},
|
|
{
|
|
"epoch": 2.552,
|
|
"grad_norm": 0.6534548006358079,
|
|
"learning_rate": 3.207740072139748e-05,
|
|
"loss": 0.214,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21586550772190094,
|
|
"step": 1595,
|
|
"valid_targets_mean": 4739.6,
|
|
"valid_targets_min": 3566
|
|
},
|
|
{
|
|
"epoch": 2.56,
|
|
"grad_norm": 0.606704419999282,
|
|
"learning_rate": 3.2013700254532996e-05,
|
|
"loss": 0.2152,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2186005562543869,
|
|
"step": 1600,
|
|
"valid_targets_mean": 4480.3,
|
|
"valid_targets_min": 2597
|
|
},
|
|
{
|
|
"epoch": 2.568,
|
|
"grad_norm": 0.647425867556911,
|
|
"learning_rate": 3.194980854484794e-05,
|
|
"loss": 0.2008,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18927501142024994,
|
|
"step": 1605,
|
|
"valid_targets_mean": 4248.3,
|
|
"valid_targets_min": 1681
|
|
},
|
|
{
|
|
"epoch": 2.576,
|
|
"grad_norm": 0.9644034097501023,
|
|
"learning_rate": 3.188572660941702e-05,
|
|
"loss": 0.2226,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22622108459472656,
|
|
"step": 1610,
|
|
"valid_targets_mean": 4476.1,
|
|
"valid_targets_min": 3078
|
|
},
|
|
{
|
|
"epoch": 2.584,
|
|
"grad_norm": 0.6030828392576895,
|
|
"learning_rate": 3.182145546834311e-05,
|
|
"loss": 0.2041,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23037052154541016,
|
|
"step": 1615,
|
|
"valid_targets_mean": 4624.5,
|
|
"valid_targets_min": 3088
|
|
},
|
|
{
|
|
"epoch": 2.592,
|
|
"grad_norm": 0.7884053193340356,
|
|
"learning_rate": 3.1756996144740994e-05,
|
|
"loss": 0.2179,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22793933749198914,
|
|
"step": 1620,
|
|
"valid_targets_mean": 4616.6,
|
|
"valid_targets_min": 2941
|
|
},
|
|
{
|
|
"epoch": 2.6,
|
|
"grad_norm": 0.5555988422976395,
|
|
"learning_rate": 3.1692349664721074e-05,
|
|
"loss": 0.2191,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20210662484169006,
|
|
"step": 1625,
|
|
"valid_targets_mean": 4738.2,
|
|
"valid_targets_min": 3731
|
|
},
|
|
{
|
|
"epoch": 2.608,
|
|
"grad_norm": 0.6649512626521268,
|
|
"learning_rate": 3.1627517057373046e-05,
|
|
"loss": 0.1937,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20128515362739563,
|
|
"step": 1630,
|
|
"valid_targets_mean": 4447.6,
|
|
"valid_targets_min": 3265
|
|
},
|
|
{
|
|
"epoch": 2.616,
|
|
"grad_norm": 0.6046416684924685,
|
|
"learning_rate": 3.156249935474953e-05,
|
|
"loss": 0.2199,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18843211233615875,
|
|
"step": 1635,
|
|
"valid_targets_mean": 4420.7,
|
|
"valid_targets_min": 1753
|
|
},
|
|
{
|
|
"epoch": 2.624,
|
|
"grad_norm": 0.7090533200737186,
|
|
"learning_rate": 3.1497297591849614e-05,
|
|
"loss": 0.207,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20444434881210327,
|
|
"step": 1640,
|
|
"valid_targets_mean": 4114.6,
|
|
"valid_targets_min": 2655
|
|
},
|
|
{
|
|
"epoch": 2.632,
|
|
"grad_norm": 0.6190191091269116,
|
|
"learning_rate": 3.143191280660238e-05,
|
|
"loss": 0.2014,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2053518295288086,
|
|
"step": 1645,
|
|
"valid_targets_mean": 4372.4,
|
|
"valid_targets_min": 1767
|
|
},
|
|
{
|
|
"epoch": 2.64,
|
|
"grad_norm": 0.6567966641491492,
|
|
"learning_rate": 3.1366346039850424e-05,
|
|
"loss": 0.2087,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22472810745239258,
|
|
"step": 1650,
|
|
"valid_targets_mean": 4364.2,
|
|
"valid_targets_min": 796
|
|
},
|
|
{
|
|
"epoch": 2.648,
|
|
"grad_norm": 0.6192333887687917,
|
|
"learning_rate": 3.130059833533323e-05,
|
|
"loss": 0.2172,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22299574315547943,
|
|
"step": 1655,
|
|
"valid_targets_mean": 4873.0,
|
|
"valid_targets_min": 4230
|
|
},
|
|
{
|
|
"epoch": 2.656,
|
|
"grad_norm": 0.6255644759615485,
|
|
"learning_rate": 3.123467073967059e-05,
|
|
"loss": 0.2086,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19573351740837097,
|
|
"step": 1660,
|
|
"valid_targets_mean": 4427.1,
|
|
"valid_targets_min": 3592
|
|
},
|
|
{
|
|
"epoch": 2.664,
|
|
"grad_norm": 0.6584115333174764,
|
|
"learning_rate": 3.116856430234594e-05,
|
|
"loss": 0.22,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2267571985721588,
|
|
"step": 1665,
|
|
"valid_targets_mean": 4357.6,
|
|
"valid_targets_min": 2614
|
|
},
|
|
{
|
|
"epoch": 2.672,
|
|
"grad_norm": 0.6394802485006209,
|
|
"learning_rate": 3.110228007568963e-05,
|
|
"loss": 0.221,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2543630599975586,
|
|
"step": 1670,
|
|
"valid_targets_mean": 4593.7,
|
|
"valid_targets_min": 1611
|
|
},
|
|
{
|
|
"epoch": 2.68,
|
|
"grad_norm": 0.6412418251447299,
|
|
"learning_rate": 3.103581911486221e-05,
|
|
"loss": 0.2041,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21916908025741577,
|
|
"step": 1675,
|
|
"valid_targets_mean": 4290.6,
|
|
"valid_targets_min": 1062
|
|
},
|
|
{
|
|
"epoch": 2.6879999999999997,
|
|
"grad_norm": 0.6450519841464782,
|
|
"learning_rate": 3.0969182477837604e-05,
|
|
"loss": 0.2198,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2431817352771759,
|
|
"step": 1680,
|
|
"valid_targets_mean": 4538.1,
|
|
"valid_targets_min": 3634
|
|
},
|
|
{
|
|
"epoch": 2.6959999999999997,
|
|
"grad_norm": 0.6123387920479804,
|
|
"learning_rate": 3.090237122538628e-05,
|
|
"loss": 0.1919,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18903237581253052,
|
|
"step": 1685,
|
|
"valid_targets_mean": 4511.2,
|
|
"valid_targets_min": 2528
|
|
},
|
|
{
|
|
"epoch": 2.7039999999999997,
|
|
"grad_norm": 0.7110342449591243,
|
|
"learning_rate": 3.0835386421058345e-05,
|
|
"loss": 0.2189,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25073370337486267,
|
|
"step": 1690,
|
|
"valid_targets_mean": 4033.6,
|
|
"valid_targets_min": 2033
|
|
},
|
|
{
|
|
"epoch": 2.7119999999999997,
|
|
"grad_norm": 0.6596833436036363,
|
|
"learning_rate": 3.0768229131166664e-05,
|
|
"loss": 0.1998,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19766391813755035,
|
|
"step": 1695,
|
|
"valid_targets_mean": 4536.1,
|
|
"valid_targets_min": 2589
|
|
},
|
|
{
|
|
"epoch": 2.7199999999999998,
|
|
"grad_norm": 0.6248987774108561,
|
|
"learning_rate": 3.070090042476983e-05,
|
|
"loss": 0.2149,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22887420654296875,
|
|
"step": 1700,
|
|
"valid_targets_mean": 4439.7,
|
|
"valid_targets_min": 1991
|
|
},
|
|
{
|
|
"epoch": 2.7279999999999998,
|
|
"grad_norm": 0.6568348438451338,
|
|
"learning_rate": 3.063340137365517e-05,
|
|
"loss": 0.2149,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21074923872947693,
|
|
"step": 1705,
|
|
"valid_targets_mean": 4260.6,
|
|
"valid_targets_min": 1768
|
|
},
|
|
{
|
|
"epoch": 2.7359999999999998,
|
|
"grad_norm": 0.5795724823019238,
|
|
"learning_rate": 3.0565733052321674e-05,
|
|
"loss": 0.1979,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19304868578910828,
|
|
"step": 1710,
|
|
"valid_targets_mean": 4747.4,
|
|
"valid_targets_min": 3846
|
|
},
|
|
{
|
|
"epoch": 2.7439999999999998,
|
|
"grad_norm": 0.6363787249536077,
|
|
"learning_rate": 3.0497896537962924e-05,
|
|
"loss": 0.2203,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22860632836818695,
|
|
"step": 1715,
|
|
"valid_targets_mean": 4421.3,
|
|
"valid_targets_min": 3914
|
|
},
|
|
{
|
|
"epoch": 2.752,
|
|
"grad_norm": 0.6821762197311857,
|
|
"learning_rate": 3.042989291044991e-05,
|
|
"loss": 0.2038,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22558745741844177,
|
|
"step": 1720,
|
|
"valid_targets_mean": 4346.4,
|
|
"valid_targets_min": 2619
|
|
},
|
|
{
|
|
"epoch": 2.76,
|
|
"grad_norm": 0.5536107704495562,
|
|
"learning_rate": 3.036172325231383e-05,
|
|
"loss": 0.1948,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.172830730676651,
|
|
"step": 1725,
|
|
"valid_targets_mean": 4832.9,
|
|
"valid_targets_min": 4145
|
|
},
|
|
{
|
|
"epoch": 2.768,
|
|
"grad_norm": 0.6521174892153598,
|
|
"learning_rate": 3.0293388648728908e-05,
|
|
"loss": 0.1998,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23022960126399994,
|
|
"step": 1730,
|
|
"valid_targets_mean": 4539.1,
|
|
"valid_targets_min": 2270
|
|
},
|
|
{
|
|
"epoch": 2.776,
|
|
"grad_norm": 0.6617910863284133,
|
|
"learning_rate": 3.022489018749508e-05,
|
|
"loss": 0.2072,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22436785697937012,
|
|
"step": 1735,
|
|
"valid_targets_mean": 4532.1,
|
|
"valid_targets_min": 383
|
|
},
|
|
{
|
|
"epoch": 2.784,
|
|
"grad_norm": 0.6068556611892771,
|
|
"learning_rate": 3.015622895902068e-05,
|
|
"loss": 0.2021,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1872781366109848,
|
|
"step": 1740,
|
|
"valid_targets_mean": 4492.3,
|
|
"valid_targets_min": 2169
|
|
},
|
|
{
|
|
"epoch": 2.792,
|
|
"grad_norm": 0.6345642082611994,
|
|
"learning_rate": 3.008740605630508e-05,
|
|
"loss": 0.2101,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20982220768928528,
|
|
"step": 1745,
|
|
"valid_targets_mean": 4546.9,
|
|
"valid_targets_min": 3447
|
|
},
|
|
{
|
|
"epoch": 2.8,
|
|
"grad_norm": 0.6546895138635057,
|
|
"learning_rate": 3.0018422574921337e-05,
|
|
"loss": 0.2163,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22267594933509827,
|
|
"step": 1750,
|
|
"valid_targets_mean": 4662.2,
|
|
"valid_targets_min": 2255
|
|
},
|
|
{
|
|
"epoch": 2.808,
|
|
"grad_norm": 0.6779079921648047,
|
|
"learning_rate": 2.9949279612998673e-05,
|
|
"loss": 0.2152,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23363904654979706,
|
|
"step": 1755,
|
|
"valid_targets_mean": 4548.7,
|
|
"valid_targets_min": 1168
|
|
},
|
|
{
|
|
"epoch": 2.816,
|
|
"grad_norm": 0.6340281486831346,
|
|
"learning_rate": 2.9879978271205064e-05,
|
|
"loss": 0.2164,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21770212054252625,
|
|
"step": 1760,
|
|
"valid_targets_mean": 4367.6,
|
|
"valid_targets_min": 1997
|
|
},
|
|
{
|
|
"epoch": 2.824,
|
|
"grad_norm": 0.6469063628167159,
|
|
"learning_rate": 2.9810519652729692e-05,
|
|
"loss": 0.1997,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2050325572490692,
|
|
"step": 1765,
|
|
"valid_targets_mean": 4748.9,
|
|
"valid_targets_min": 3834
|
|
},
|
|
{
|
|
"epoch": 2.832,
|
|
"grad_norm": 0.5828593204318403,
|
|
"learning_rate": 2.9740904863265378e-05,
|
|
"loss": 0.1977,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18355226516723633,
|
|
"step": 1770,
|
|
"valid_targets_mean": 4357.1,
|
|
"valid_targets_min": 1624
|
|
},
|
|
{
|
|
"epoch": 2.84,
|
|
"grad_norm": 0.6197430062562591,
|
|
"learning_rate": 2.967113501099097e-05,
|
|
"loss": 0.203,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1888769268989563,
|
|
"step": 1775,
|
|
"valid_targets_mean": 4452.2,
|
|
"valid_targets_min": 2887
|
|
},
|
|
{
|
|
"epoch": 2.848,
|
|
"grad_norm": 0.6407473544939916,
|
|
"learning_rate": 2.9601211206553745e-05,
|
|
"loss": 0.1976,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20623719692230225,
|
|
"step": 1780,
|
|
"valid_targets_mean": 4672.6,
|
|
"valid_targets_min": 3635
|
|
},
|
|
{
|
|
"epoch": 2.856,
|
|
"grad_norm": 0.6203577411199663,
|
|
"learning_rate": 2.9531134563051686e-05,
|
|
"loss": 0.2079,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22331242263317108,
|
|
"step": 1785,
|
|
"valid_targets_mean": 4745.8,
|
|
"valid_targets_min": 2198
|
|
},
|
|
{
|
|
"epoch": 2.864,
|
|
"grad_norm": 0.6343693648150197,
|
|
"learning_rate": 2.946090619601579e-05,
|
|
"loss": 0.2145,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2414296567440033,
|
|
"step": 1790,
|
|
"valid_targets_mean": 4791.1,
|
|
"valid_targets_min": 2918
|
|
},
|
|
{
|
|
"epoch": 2.872,
|
|
"grad_norm": 0.5772783245412467,
|
|
"learning_rate": 2.9390527223392292e-05,
|
|
"loss": 0.2132,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17813526093959808,
|
|
"step": 1795,
|
|
"valid_targets_mean": 4548.1,
|
|
"valid_targets_min": 2614
|
|
},
|
|
{
|
|
"epoch": 2.88,
|
|
"grad_norm": 0.6699359487007163,
|
|
"learning_rate": 2.931999876552488e-05,
|
|
"loss": 0.2033,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23955471813678741,
|
|
"step": 1800,
|
|
"valid_targets_mean": 4557.4,
|
|
"valid_targets_min": 3844
|
|
},
|
|
{
|
|
"epoch": 2.888,
|
|
"grad_norm": 0.628988844937057,
|
|
"learning_rate": 2.9249321945136854e-05,
|
|
"loss": 0.1939,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20162507891654968,
|
|
"step": 1805,
|
|
"valid_targets_mean": 4572.4,
|
|
"valid_targets_min": 3484
|
|
},
|
|
{
|
|
"epoch": 2.896,
|
|
"grad_norm": 0.6226172415814413,
|
|
"learning_rate": 2.9178497887313257e-05,
|
|
"loss": 0.2063,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2108525037765503,
|
|
"step": 1810,
|
|
"valid_targets_mean": 4451.0,
|
|
"valid_targets_min": 3004
|
|
},
|
|
{
|
|
"epoch": 2.904,
|
|
"grad_norm": 0.6524697307141797,
|
|
"learning_rate": 2.9107527719482968e-05,
|
|
"loss": 0.2072,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20220604538917542,
|
|
"step": 1815,
|
|
"valid_targets_mean": 4271.7,
|
|
"valid_targets_min": 2528
|
|
},
|
|
{
|
|
"epoch": 2.912,
|
|
"grad_norm": 0.6594539990085777,
|
|
"learning_rate": 2.9036412571400747e-05,
|
|
"loss": 0.2217,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19932571053504944,
|
|
"step": 1820,
|
|
"valid_targets_mean": 4059.0,
|
|
"valid_targets_min": 383
|
|
},
|
|
{
|
|
"epoch": 2.92,
|
|
"grad_norm": 0.6322548674200522,
|
|
"learning_rate": 2.8965153575129255e-05,
|
|
"loss": 0.2011,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16368697583675385,
|
|
"step": 1825,
|
|
"valid_targets_mean": 4401.4,
|
|
"valid_targets_min": 2241
|
|
},
|
|
{
|
|
"epoch": 2.928,
|
|
"grad_norm": 0.6905951523296558,
|
|
"learning_rate": 2.8893751865021044e-05,
|
|
"loss": 0.1971,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18371345102787018,
|
|
"step": 1830,
|
|
"valid_targets_mean": 4574.6,
|
|
"valid_targets_min": 3451
|
|
},
|
|
{
|
|
"epoch": 2.936,
|
|
"grad_norm": 0.6760919692480537,
|
|
"learning_rate": 2.8822208577700473e-05,
|
|
"loss": 0.1841,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17449943721294403,
|
|
"step": 1835,
|
|
"valid_targets_mean": 4131.6,
|
|
"valid_targets_min": 1857
|
|
},
|
|
{
|
|
"epoch": 2.944,
|
|
"grad_norm": 0.6279966540948257,
|
|
"learning_rate": 2.8750524852045642e-05,
|
|
"loss": 0.1965,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1772339940071106,
|
|
"step": 1840,
|
|
"valid_targets_mean": 4386.8,
|
|
"valid_targets_min": 3844
|
|
},
|
|
{
|
|
"epoch": 2.952,
|
|
"grad_norm": 0.6494732136391637,
|
|
"learning_rate": 2.867870182917024e-05,
|
|
"loss": 0.2038,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21635648608207703,
|
|
"step": 1845,
|
|
"valid_targets_mean": 4605.3,
|
|
"valid_targets_min": 1800
|
|
},
|
|
{
|
|
"epoch": 2.96,
|
|
"grad_norm": 0.6338120349019097,
|
|
"learning_rate": 2.8606740652405394e-05,
|
|
"loss": 0.2099,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1874855011701584,
|
|
"step": 1850,
|
|
"valid_targets_mean": 4728.0,
|
|
"valid_targets_min": 3842
|
|
},
|
|
{
|
|
"epoch": 2.968,
|
|
"grad_norm": 0.863902047067183,
|
|
"learning_rate": 2.853464246728147e-05,
|
|
"loss": 0.196,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18281200528144836,
|
|
"step": 1855,
|
|
"valid_targets_mean": 4100.2,
|
|
"valid_targets_min": 1833
|
|
},
|
|
{
|
|
"epoch": 2.976,
|
|
"grad_norm": 0.6616021668640071,
|
|
"learning_rate": 2.846240842150984e-05,
|
|
"loss": 0.2029,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19817513227462769,
|
|
"step": 1860,
|
|
"valid_targets_mean": 4479.2,
|
|
"valid_targets_min": 3754
|
|
},
|
|
{
|
|
"epoch": 2.984,
|
|
"grad_norm": 0.6985557285952828,
|
|
"learning_rate": 2.839003966496458e-05,
|
|
"loss": 0.2076,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21524721384048462,
|
|
"step": 1865,
|
|
"valid_targets_mean": 4613.6,
|
|
"valid_targets_min": 3773
|
|
},
|
|
{
|
|
"epoch": 2.992,
|
|
"grad_norm": 0.6517695278599833,
|
|
"learning_rate": 2.8317537349664215e-05,
|
|
"loss": 0.193,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20121997594833374,
|
|
"step": 1870,
|
|
"valid_targets_mean": 4380.2,
|
|
"valid_targets_min": 1641
|
|
},
|
|
{
|
|
"epoch": 3.0,
|
|
"grad_norm": 0.6881518734463911,
|
|
"learning_rate": 2.824490262975334e-05,
|
|
"loss": 0.2041,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23973992466926575,
|
|
"step": 1875,
|
|
"valid_targets_mean": 4849.5,
|
|
"valid_targets_min": 3085
|
|
},
|
|
{
|
|
"epoch": 3.008,
|
|
"grad_norm": 0.6584117550757836,
|
|
"learning_rate": 2.817213666148427e-05,
|
|
"loss": 0.1712,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17237037420272827,
|
|
"step": 1880,
|
|
"valid_targets_mean": 4448.9,
|
|
"valid_targets_min": 2270
|
|
},
|
|
{
|
|
"epoch": 3.016,
|
|
"grad_norm": 0.6890669176698888,
|
|
"learning_rate": 2.809924060319862e-05,
|
|
"loss": 0.1811,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18438558280467987,
|
|
"step": 1885,
|
|
"valid_targets_mean": 4322.2,
|
|
"valid_targets_min": 934
|
|
},
|
|
{
|
|
"epoch": 3.024,
|
|
"grad_norm": 0.6823437139214746,
|
|
"learning_rate": 2.802621561530888e-05,
|
|
"loss": 0.18,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16331693530082703,
|
|
"step": 1890,
|
|
"valid_targets_mean": 4033.4,
|
|
"valid_targets_min": 674
|
|
},
|
|
{
|
|
"epoch": 3.032,
|
|
"grad_norm": 0.668321493919484,
|
|
"learning_rate": 2.7953062860279937e-05,
|
|
"loss": 0.1685,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1623179316520691,
|
|
"step": 1895,
|
|
"valid_targets_mean": 4485.8,
|
|
"valid_targets_min": 3816
|
|
},
|
|
{
|
|
"epoch": 3.04,
|
|
"grad_norm": 0.7235413257756489,
|
|
"learning_rate": 2.7879783502610557e-05,
|
|
"loss": 0.1686,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16372078657150269,
|
|
"step": 1900,
|
|
"valid_targets_mean": 4089.6,
|
|
"valid_targets_min": 437
|
|
},
|
|
{
|
|
"epoch": 3.048,
|
|
"grad_norm": 0.8073255006904592,
|
|
"learning_rate": 2.7806378708814875e-05,
|
|
"loss": 0.1868,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22582300007343292,
|
|
"step": 1905,
|
|
"valid_targets_mean": 4506.6,
|
|
"valid_targets_min": 2663
|
|
},
|
|
{
|
|
"epoch": 3.056,
|
|
"grad_norm": 0.6968734989303648,
|
|
"learning_rate": 2.773284964740379e-05,
|
|
"loss": 0.1731,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16369201242923737,
|
|
"step": 1910,
|
|
"valid_targets_mean": 4553.9,
|
|
"valid_targets_min": 1833
|
|
},
|
|
{
|
|
"epoch": 3.064,
|
|
"grad_norm": 0.7403391240459269,
|
|
"learning_rate": 2.7659197488866403e-05,
|
|
"loss": 0.1819,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19810575246810913,
|
|
"step": 1915,
|
|
"valid_targets_mean": 4497.6,
|
|
"valid_targets_min": 3540
|
|
},
|
|
{
|
|
"epoch": 3.072,
|
|
"grad_norm": 0.6897058721915196,
|
|
"learning_rate": 2.7585423405651347e-05,
|
|
"loss": 0.1738,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16929036378860474,
|
|
"step": 1920,
|
|
"valid_targets_mean": 4304.3,
|
|
"valid_targets_min": 2688
|
|
},
|
|
{
|
|
"epoch": 3.08,
|
|
"grad_norm": 0.7944465680025203,
|
|
"learning_rate": 2.7511528572148153e-05,
|
|
"loss": 0.1743,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17681191861629486,
|
|
"step": 1925,
|
|
"valid_targets_mean": 4383.9,
|
|
"valid_targets_min": 2461
|
|
},
|
|
{
|
|
"epoch": 3.088,
|
|
"grad_norm": 0.7832746059106002,
|
|
"learning_rate": 2.7437514164668536e-05,
|
|
"loss": 0.1765,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1997135877609253,
|
|
"step": 1930,
|
|
"valid_targets_mean": 4570.8,
|
|
"valid_targets_min": 796
|
|
},
|
|
{
|
|
"epoch": 3.096,
|
|
"grad_norm": 0.7654148393837111,
|
|
"learning_rate": 2.7363381361427692e-05,
|
|
"loss": 0.1634,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15489351749420166,
|
|
"step": 1935,
|
|
"valid_targets_mean": 4396.2,
|
|
"valid_targets_min": 3858
|
|
},
|
|
{
|
|
"epoch": 3.104,
|
|
"grad_norm": 0.6932894830539476,
|
|
"learning_rate": 2.72891313425255e-05,
|
|
"loss": 0.1561,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15014304220676422,
|
|
"step": 1940,
|
|
"valid_targets_mean": 4456.7,
|
|
"valid_targets_min": 2534
|
|
},
|
|
{
|
|
"epoch": 3.112,
|
|
"grad_norm": 0.6859012757456313,
|
|
"learning_rate": 2.7214765289927777e-05,
|
|
"loss": 0.167,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16869190335273743,
|
|
"step": 1945,
|
|
"valid_targets_mean": 4664.1,
|
|
"valid_targets_min": 3001
|
|
},
|
|
{
|
|
"epoch": 3.12,
|
|
"grad_norm": 0.6881051555093767,
|
|
"learning_rate": 2.714028438744746e-05,
|
|
"loss": 0.1772,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18348908424377441,
|
|
"step": 1950,
|
|
"valid_targets_mean": 4669.9,
|
|
"valid_targets_min": 4028
|
|
},
|
|
{
|
|
"epoch": 3.128,
|
|
"grad_norm": 0.676586342945555,
|
|
"learning_rate": 2.706568982072573e-05,
|
|
"loss": 0.1817,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1808232069015503,
|
|
"step": 1955,
|
|
"valid_targets_mean": 4791.5,
|
|
"valid_targets_min": 3502
|
|
},
|
|
{
|
|
"epoch": 3.136,
|
|
"grad_norm": 0.7544579622038925,
|
|
"learning_rate": 2.6990982777213174e-05,
|
|
"loss": 0.1812,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19533181190490723,
|
|
"step": 1960,
|
|
"valid_targets_mean": 4436.6,
|
|
"valid_targets_min": 3484
|
|
},
|
|
{
|
|
"epoch": 3.144,
|
|
"grad_norm": 0.7309166670903791,
|
|
"learning_rate": 2.691616444615085e-05,
|
|
"loss": 0.1727,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17915305495262146,
|
|
"step": 1965,
|
|
"valid_targets_mean": 4742.8,
|
|
"valid_targets_min": 3960
|
|
},
|
|
{
|
|
"epoch": 3.152,
|
|
"grad_norm": 0.7568747440851656,
|
|
"learning_rate": 2.6841236018551402e-05,
|
|
"loss": 0.1865,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1846478283405304,
|
|
"step": 1970,
|
|
"valid_targets_mean": 4104.7,
|
|
"valid_targets_min": 934
|
|
},
|
|
{
|
|
"epoch": 3.16,
|
|
"grad_norm": 0.7304634316097307,
|
|
"learning_rate": 2.6766198687180028e-05,
|
|
"loss": 0.1631,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1656237244606018,
|
|
"step": 1975,
|
|
"valid_targets_mean": 4378.9,
|
|
"valid_targets_min": 2342
|
|
},
|
|
{
|
|
"epoch": 3.168,
|
|
"grad_norm": 0.6835278918270478,
|
|
"learning_rate": 2.6691053646535564e-05,
|
|
"loss": 0.1774,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16619622707366943,
|
|
"step": 1980,
|
|
"valid_targets_mean": 4604.1,
|
|
"valid_targets_min": 3472
|
|
},
|
|
{
|
|
"epoch": 3.176,
|
|
"grad_norm": 0.7207154394413637,
|
|
"learning_rate": 2.6615802092831446e-05,
|
|
"loss": 0.158,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13958649337291718,
|
|
"step": 1985,
|
|
"valid_targets_mean": 4761.4,
|
|
"valid_targets_min": 2503
|
|
},
|
|
{
|
|
"epoch": 3.184,
|
|
"grad_norm": 0.7728396631110935,
|
|
"learning_rate": 2.6540445223976637e-05,
|
|
"loss": 0.1783,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17282788455486298,
|
|
"step": 1990,
|
|
"valid_targets_mean": 4188.0,
|
|
"valid_targets_min": 2268
|
|
},
|
|
{
|
|
"epoch": 3.192,
|
|
"grad_norm": 0.6948108926339397,
|
|
"learning_rate": 2.6464984239556602e-05,
|
|
"loss": 0.1665,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17473545670509338,
|
|
"step": 1995,
|
|
"valid_targets_mean": 4674.2,
|
|
"valid_targets_min": 3606
|
|
},
|
|
{
|
|
"epoch": 3.2,
|
|
"grad_norm": 0.7960284463673977,
|
|
"learning_rate": 2.63894203408142e-05,
|
|
"loss": 0.1819,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17627307772636414,
|
|
"step": 2000,
|
|
"valid_targets_mean": 4506.6,
|
|
"valid_targets_min": 3801
|
|
},
|
|
{
|
|
"epoch": 3.208,
|
|
"grad_norm": 0.6786645174486711,
|
|
"learning_rate": 2.6313754730630528e-05,
|
|
"loss": 0.1816,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1778252124786377,
|
|
"step": 2005,
|
|
"valid_targets_mean": 4630.9,
|
|
"valid_targets_min": 2854
|
|
},
|
|
{
|
|
"epoch": 3.216,
|
|
"grad_norm": 0.6700059153745351,
|
|
"learning_rate": 2.623798861350582e-05,
|
|
"loss": 0.1706,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14134109020233154,
|
|
"step": 2010,
|
|
"valid_targets_mean": 4646.8,
|
|
"valid_targets_min": 3927
|
|
},
|
|
{
|
|
"epoch": 3.224,
|
|
"grad_norm": 0.7425036888296542,
|
|
"learning_rate": 2.6162123195540247e-05,
|
|
"loss": 0.1761,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1791936606168747,
|
|
"step": 2015,
|
|
"valid_targets_mean": 4490.2,
|
|
"valid_targets_min": 2878
|
|
},
|
|
{
|
|
"epoch": 3.232,
|
|
"grad_norm": 0.740878305658572,
|
|
"learning_rate": 2.6086159684414726e-05,
|
|
"loss": 0.1824,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1859646737575531,
|
|
"step": 2020,
|
|
"valid_targets_mean": 4675.5,
|
|
"valid_targets_min": 3951
|
|
},
|
|
{
|
|
"epoch": 3.24,
|
|
"grad_norm": 0.7290197772281836,
|
|
"learning_rate": 2.6010099289371694e-05,
|
|
"loss": 0.1716,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13927796483039856,
|
|
"step": 2025,
|
|
"valid_targets_mean": 4235.9,
|
|
"valid_targets_min": 2708
|
|
},
|
|
{
|
|
"epoch": 3.248,
|
|
"grad_norm": 0.8034700391209433,
|
|
"learning_rate": 2.5933943221195844e-05,
|
|
"loss": 0.1749,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14626024663448334,
|
|
"step": 2030,
|
|
"valid_targets_mean": 3803.8,
|
|
"valid_targets_min": 1318
|
|
},
|
|
{
|
|
"epoch": 3.2560000000000002,
|
|
"grad_norm": 0.7900641366732813,
|
|
"learning_rate": 2.5857692692194884e-05,
|
|
"loss": 0.1728,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1817488968372345,
|
|
"step": 2035,
|
|
"valid_targets_mean": 4421.8,
|
|
"valid_targets_min": 534
|
|
},
|
|
{
|
|
"epoch": 3.2640000000000002,
|
|
"grad_norm": 0.7618614330827502,
|
|
"learning_rate": 2.5781348916180195e-05,
|
|
"loss": 0.1668,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14263778924942017,
|
|
"step": 2040,
|
|
"valid_targets_mean": 4314.8,
|
|
"valid_targets_min": 1623
|
|
},
|
|
{
|
|
"epoch": 3.2720000000000002,
|
|
"grad_norm": 0.7289268775227461,
|
|
"learning_rate": 2.570491310844755e-05,
|
|
"loss": 0.184,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17487305402755737,
|
|
"step": 2045,
|
|
"valid_targets_mean": 4583.6,
|
|
"valid_targets_min": 3048
|
|
},
|
|
{
|
|
"epoch": 3.2800000000000002,
|
|
"grad_norm": 0.7652645560494707,
|
|
"learning_rate": 2.562838648575774e-05,
|
|
"loss": 0.1698,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1700042486190796,
|
|
"step": 2050,
|
|
"valid_targets_mean": 4245.7,
|
|
"valid_targets_min": 1915
|
|
},
|
|
{
|
|
"epoch": 3.288,
|
|
"grad_norm": 0.7278549315311669,
|
|
"learning_rate": 2.5551770266317224e-05,
|
|
"loss": 0.1801,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17379678785800934,
|
|
"step": 2055,
|
|
"valid_targets_mean": 4315.6,
|
|
"valid_targets_min": 2450
|
|
},
|
|
{
|
|
"epoch": 3.296,
|
|
"grad_norm": 0.7685029417075684,
|
|
"learning_rate": 2.5475065669758713e-05,
|
|
"loss": 0.1643,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14998362958431244,
|
|
"step": 2060,
|
|
"valid_targets_mean": 3987.4,
|
|
"valid_targets_min": 785
|
|
},
|
|
{
|
|
"epoch": 3.304,
|
|
"grad_norm": 0.7407280792377957,
|
|
"learning_rate": 2.5398273917121786e-05,
|
|
"loss": 0.1689,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14474965631961823,
|
|
"step": 2065,
|
|
"valid_targets_mean": 4576.1,
|
|
"valid_targets_min": 3621
|
|
},
|
|
{
|
|
"epoch": 3.312,
|
|
"grad_norm": 0.7402276082403204,
|
|
"learning_rate": 2.532139623083342e-05,
|
|
"loss": 0.1657,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15302659571170807,
|
|
"step": 2070,
|
|
"valid_targets_mean": 4348.1,
|
|
"valid_targets_min": 2589
|
|
},
|
|
{
|
|
"epoch": 3.32,
|
|
"grad_norm": 0.9395610931697639,
|
|
"learning_rate": 2.5244433834688552e-05,
|
|
"loss": 0.1722,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18299683928489685,
|
|
"step": 2075,
|
|
"valid_targets_mean": 5006.2,
|
|
"valid_targets_min": 4091
|
|
},
|
|
{
|
|
"epoch": 3.328,
|
|
"grad_norm": 0.7364825525167654,
|
|
"learning_rate": 2.5167387953830602e-05,
|
|
"loss": 0.1603,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14602452516555786,
|
|
"step": 2080,
|
|
"valid_targets_mean": 4744.1,
|
|
"valid_targets_min": 3896
|
|
},
|
|
{
|
|
"epoch": 3.336,
|
|
"grad_norm": 0.7689569157469457,
|
|
"learning_rate": 2.5090259814731946e-05,
|
|
"loss": 0.1699,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18590813875198364,
|
|
"step": 2085,
|
|
"valid_targets_mean": 4374.1,
|
|
"valid_targets_min": 2248
|
|
},
|
|
{
|
|
"epoch": 3.344,
|
|
"grad_norm": 0.7021748547924481,
|
|
"learning_rate": 2.5013050645174414e-05,
|
|
"loss": 0.1654,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17635685205459595,
|
|
"step": 2090,
|
|
"valid_targets_mean": 4731.2,
|
|
"valid_targets_min": 3959
|
|
},
|
|
{
|
|
"epoch": 3.352,
|
|
"grad_norm": 1.0544312488099348,
|
|
"learning_rate": 2.4935761674229735e-05,
|
|
"loss": 0.1737,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.163224995136261,
|
|
"step": 2095,
|
|
"valid_targets_mean": 4576.5,
|
|
"valid_targets_min": 1585
|
|
},
|
|
{
|
|
"epoch": 3.36,
|
|
"grad_norm": 0.8144708611276341,
|
|
"learning_rate": 2.4858394132239982e-05,
|
|
"loss": 0.174,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1789839267730713,
|
|
"step": 2100,
|
|
"valid_targets_mean": 4341.3,
|
|
"valid_targets_min": 1180
|
|
},
|
|
{
|
|
"epoch": 3.368,
|
|
"grad_norm": 0.722218790000107,
|
|
"learning_rate": 2.4780949250797964e-05,
|
|
"loss": 0.1696,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1648513227701187,
|
|
"step": 2105,
|
|
"valid_targets_mean": 4532.6,
|
|
"valid_targets_min": 2277
|
|
},
|
|
{
|
|
"epoch": 3.376,
|
|
"grad_norm": 0.741842686744923,
|
|
"learning_rate": 2.4703428262727656e-05,
|
|
"loss": 0.168,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17283059656620026,
|
|
"step": 2110,
|
|
"valid_targets_mean": 4576.2,
|
|
"valid_targets_min": 2754
|
|
},
|
|
{
|
|
"epoch": 3.384,
|
|
"grad_norm": 0.7025042419473128,
|
|
"learning_rate": 2.4625832402064525e-05,
|
|
"loss": 0.1639,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16021236777305603,
|
|
"step": 2115,
|
|
"valid_targets_mean": 4334.1,
|
|
"valid_targets_min": 2655
|
|
},
|
|
{
|
|
"epoch": 3.392,
|
|
"grad_norm": 0.7595683887261002,
|
|
"learning_rate": 2.454816290403595e-05,
|
|
"loss": 0.1764,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18192517757415771,
|
|
"step": 2120,
|
|
"valid_targets_mean": 4268.2,
|
|
"valid_targets_min": 2252
|
|
},
|
|
{
|
|
"epoch": 3.4,
|
|
"grad_norm": 0.7275508475765697,
|
|
"learning_rate": 2.4470421005041492e-05,
|
|
"loss": 0.1684,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16349178552627563,
|
|
"step": 2125,
|
|
"valid_targets_mean": 4416.0,
|
|
"valid_targets_min": 1661
|
|
},
|
|
{
|
|
"epoch": 3.408,
|
|
"grad_norm": 0.6969523954593061,
|
|
"learning_rate": 2.4392607942633263e-05,
|
|
"loss": 0.1632,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14763638377189636,
|
|
"step": 2130,
|
|
"valid_targets_mean": 4729.2,
|
|
"valid_targets_min": 3854
|
|
},
|
|
{
|
|
"epoch": 3.416,
|
|
"grad_norm": 0.70535276338042,
|
|
"learning_rate": 2.43147249554962e-05,
|
|
"loss": 0.1603,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16657257080078125,
|
|
"step": 2135,
|
|
"valid_targets_mean": 4866.2,
|
|
"valid_targets_min": 3606
|
|
},
|
|
{
|
|
"epoch": 3.424,
|
|
"grad_norm": 0.7074803985483549,
|
|
"learning_rate": 2.423677328342835e-05,
|
|
"loss": 0.1648,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16990341246128082,
|
|
"step": 2140,
|
|
"valid_targets_mean": 4668.6,
|
|
"valid_targets_min": 1064
|
|
},
|
|
{
|
|
"epoch": 3.432,
|
|
"grad_norm": 0.757464915113217,
|
|
"learning_rate": 2.415875416732113e-05,
|
|
"loss": 0.1765,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1767418086528778,
|
|
"step": 2145,
|
|
"valid_targets_mean": 4716.2,
|
|
"valid_targets_min": 3784
|
|
},
|
|
{
|
|
"epoch": 3.44,
|
|
"grad_norm": 0.6954789896167358,
|
|
"learning_rate": 2.4080668849139603e-05,
|
|
"loss": 0.1521,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14761747419834137,
|
|
"step": 2150,
|
|
"valid_targets_mean": 4514.6,
|
|
"valid_targets_min": 1452
|
|
},
|
|
{
|
|
"epoch": 3.448,
|
|
"grad_norm": 0.803780345974481,
|
|
"learning_rate": 2.4002518571902665e-05,
|
|
"loss": 0.1706,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15911833941936493,
|
|
"step": 2155,
|
|
"valid_targets_mean": 4433.6,
|
|
"valid_targets_min": 1925
|
|
},
|
|
{
|
|
"epoch": 3.456,
|
|
"grad_norm": 0.7407615150228559,
|
|
"learning_rate": 2.392430457966328e-05,
|
|
"loss": 0.183,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17995017766952515,
|
|
"step": 2160,
|
|
"valid_targets_mean": 4415.8,
|
|
"valid_targets_min": 3078
|
|
},
|
|
{
|
|
"epoch": 3.464,
|
|
"grad_norm": 0.7811600173674533,
|
|
"learning_rate": 2.3846028117488686e-05,
|
|
"loss": 0.174,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18306660652160645,
|
|
"step": 2165,
|
|
"valid_targets_mean": 4687.6,
|
|
"valid_targets_min": 3566
|
|
},
|
|
{
|
|
"epoch": 3.472,
|
|
"grad_norm": 0.7501572368205324,
|
|
"learning_rate": 2.3767690431440533e-05,
|
|
"loss": 0.1626,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15919369459152222,
|
|
"step": 2170,
|
|
"valid_targets_mean": 4604.8,
|
|
"valid_targets_min": 3746
|
|
},
|
|
{
|
|
"epoch": 3.48,
|
|
"grad_norm": 0.7252255076307738,
|
|
"learning_rate": 2.368929276855512e-05,
|
|
"loss": 0.1548,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14515303075313568,
|
|
"step": 2175,
|
|
"valid_targets_mean": 4547.7,
|
|
"valid_targets_min": 2528
|
|
},
|
|
{
|
|
"epoch": 3.488,
|
|
"grad_norm": 0.7838586037998898,
|
|
"learning_rate": 2.361083637682347e-05,
|
|
"loss": 0.16,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18825480341911316,
|
|
"step": 2180,
|
|
"valid_targets_mean": 4567.8,
|
|
"valid_targets_min": 3368
|
|
},
|
|
{
|
|
"epoch": 3.496,
|
|
"grad_norm": 0.8586989889649361,
|
|
"learning_rate": 2.3532322505171502e-05,
|
|
"loss": 0.1769,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19984722137451172,
|
|
"step": 2185,
|
|
"valid_targets_mean": 4670.2,
|
|
"valid_targets_min": 2582
|
|
},
|
|
{
|
|
"epoch": 3.504,
|
|
"grad_norm": 0.7268917010994708,
|
|
"learning_rate": 2.3453752403440147e-05,
|
|
"loss": 0.1691,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15694065392017365,
|
|
"step": 2190,
|
|
"valid_targets_mean": 4421.4,
|
|
"valid_targets_min": 1062
|
|
},
|
|
{
|
|
"epoch": 3.512,
|
|
"grad_norm": 0.8766115165893474,
|
|
"learning_rate": 2.337512732236545e-05,
|
|
"loss": 0.1649,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19197137653827667,
|
|
"step": 2195,
|
|
"valid_targets_mean": 4506.7,
|
|
"valid_targets_min": 890
|
|
},
|
|
{
|
|
"epoch": 3.52,
|
|
"grad_norm": 0.7858554649871525,
|
|
"learning_rate": 2.3296448513558628e-05,
|
|
"loss": 0.1634,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1569642871618271,
|
|
"step": 2200,
|
|
"valid_targets_mean": 4779.6,
|
|
"valid_targets_min": 3375
|
|
},
|
|
{
|
|
"epoch": 3.528,
|
|
"grad_norm": 0.7679106628525494,
|
|
"learning_rate": 2.321771722948622e-05,
|
|
"loss": 0.1755,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17182579636573792,
|
|
"step": 2205,
|
|
"valid_targets_mean": 4802.2,
|
|
"valid_targets_min": 3911
|
|
},
|
|
{
|
|
"epoch": 3.536,
|
|
"grad_norm": 0.7548962571576605,
|
|
"learning_rate": 2.3138934723450074e-05,
|
|
"loss": 0.1568,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16351479291915894,
|
|
"step": 2210,
|
|
"valid_targets_mean": 4455.4,
|
|
"valid_targets_min": 1594
|
|
},
|
|
{
|
|
"epoch": 3.544,
|
|
"grad_norm": 0.7785215535689904,
|
|
"learning_rate": 2.306010224956744e-05,
|
|
"loss": 0.161,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17092761397361755,
|
|
"step": 2215,
|
|
"valid_targets_mean": 4679.8,
|
|
"valid_targets_min": 3712
|
|
},
|
|
{
|
|
"epoch": 3.552,
|
|
"grad_norm": 0.7265491098292198,
|
|
"learning_rate": 2.2981221062750986e-05,
|
|
"loss": 0.1597,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16401392221450806,
|
|
"step": 2220,
|
|
"valid_targets_mean": 4472.6,
|
|
"valid_targets_min": 2663
|
|
},
|
|
{
|
|
"epoch": 3.56,
|
|
"grad_norm": 0.7932816074419308,
|
|
"learning_rate": 2.290229241868882e-05,
|
|
"loss": 0.1559,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16041916608810425,
|
|
"step": 2225,
|
|
"valid_targets_mean": 4480.8,
|
|
"valid_targets_min": 3886
|
|
},
|
|
{
|
|
"epoch": 3.568,
|
|
"grad_norm": 0.7421382105594192,
|
|
"learning_rate": 2.282331757382454e-05,
|
|
"loss": 0.166,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16886839270591736,
|
|
"step": 2230,
|
|
"valid_targets_mean": 4509.6,
|
|
"valid_targets_min": 1038
|
|
},
|
|
{
|
|
"epoch": 3.576,
|
|
"grad_norm": 0.7757021722466771,
|
|
"learning_rate": 2.2744297785337155e-05,
|
|
"loss": 0.1697,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1749383807182312,
|
|
"step": 2235,
|
|
"valid_targets_mean": 4388.6,
|
|
"valid_targets_min": 2182
|
|
},
|
|
{
|
|
"epoch": 3.584,
|
|
"grad_norm": 0.8254813027207493,
|
|
"learning_rate": 2.2665234311121155e-05,
|
|
"loss": 0.1631,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17627069354057312,
|
|
"step": 2240,
|
|
"valid_targets_mean": 4076.2,
|
|
"valid_targets_min": 1356
|
|
},
|
|
{
|
|
"epoch": 3.592,
|
|
"grad_norm": 0.7925150407101634,
|
|
"learning_rate": 2.258612840976645e-05,
|
|
"loss": 0.1724,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17663386464118958,
|
|
"step": 2245,
|
|
"valid_targets_mean": 4397.8,
|
|
"valid_targets_min": 3403
|
|
},
|
|
{
|
|
"epoch": 3.6,
|
|
"grad_norm": 0.7272233011735565,
|
|
"learning_rate": 2.2506981340538315e-05,
|
|
"loss": 0.1638,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16130034625530243,
|
|
"step": 2250,
|
|
"valid_targets_mean": 4724.5,
|
|
"valid_targets_min": 3411
|
|
},
|
|
{
|
|
"epoch": 3.608,
|
|
"grad_norm": 0.723442035628587,
|
|
"learning_rate": 2.2427794363357384e-05,
|
|
"loss": 0.1623,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16438783705234528,
|
|
"step": 2255,
|
|
"valid_targets_mean": 4588.1,
|
|
"valid_targets_min": 2620
|
|
},
|
|
{
|
|
"epoch": 3.616,
|
|
"grad_norm": 0.7878600601397067,
|
|
"learning_rate": 2.2348568738779566e-05,
|
|
"loss": 0.1711,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1789105087518692,
|
|
"step": 2260,
|
|
"valid_targets_mean": 4355.9,
|
|
"valid_targets_min": 1921
|
|
},
|
|
{
|
|
"epoch": 3.624,
|
|
"grad_norm": 0.7538881464784375,
|
|
"learning_rate": 2.2269305727975993e-05,
|
|
"loss": 0.1605,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1658523976802826,
|
|
"step": 2265,
|
|
"valid_targets_mean": 4493.4,
|
|
"valid_targets_min": 2774
|
|
},
|
|
{
|
|
"epoch": 3.632,
|
|
"grad_norm": 0.7230235466408017,
|
|
"learning_rate": 2.2190006592712927e-05,
|
|
"loss": 0.1711,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16584663093090057,
|
|
"step": 2270,
|
|
"valid_targets_mean": 4457.9,
|
|
"valid_targets_min": 2411
|
|
},
|
|
{
|
|
"epoch": 3.64,
|
|
"grad_norm": 0.7662117989633142,
|
|
"learning_rate": 2.2110672595331698e-05,
|
|
"loss": 0.164,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17641490697860718,
|
|
"step": 2275,
|
|
"valid_targets_mean": 4517.8,
|
|
"valid_targets_min": 3288
|
|
},
|
|
{
|
|
"epoch": 3.648,
|
|
"grad_norm": 0.7174427457202215,
|
|
"learning_rate": 2.2031304998728587e-05,
|
|
"loss": 0.1679,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15006238222122192,
|
|
"step": 2280,
|
|
"valid_targets_mean": 4797.6,
|
|
"valid_targets_min": 3610
|
|
},
|
|
{
|
|
"epoch": 3.656,
|
|
"grad_norm": 0.7937523840871874,
|
|
"learning_rate": 2.1951905066334737e-05,
|
|
"loss": 0.1641,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15922710299491882,
|
|
"step": 2285,
|
|
"valid_targets_mean": 4593.2,
|
|
"valid_targets_min": 3355
|
|
},
|
|
{
|
|
"epoch": 3.664,
|
|
"grad_norm": 0.8616793875901728,
|
|
"learning_rate": 2.1872474062096046e-05,
|
|
"loss": 0.1575,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16093303263187408,
|
|
"step": 2290,
|
|
"valid_targets_mean": 4380.8,
|
|
"valid_targets_min": 2663
|
|
},
|
|
{
|
|
"epoch": 3.672,
|
|
"grad_norm": 0.7142384461092642,
|
|
"learning_rate": 2.179301325045301e-05,
|
|
"loss": 0.1531,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1537773609161377,
|
|
"step": 2295,
|
|
"valid_targets_mean": 4718.9,
|
|
"valid_targets_min": 3665
|
|
},
|
|
{
|
|
"epoch": 3.68,
|
|
"grad_norm": 0.7535742407062561,
|
|
"learning_rate": 2.1713523896320647e-05,
|
|
"loss": 0.1601,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1746906340122223,
|
|
"step": 2300,
|
|
"valid_targets_mean": 4653.4,
|
|
"valid_targets_min": 1803
|
|
},
|
|
{
|
|
"epoch": 3.6879999999999997,
|
|
"grad_norm": 0.7578754317219941,
|
|
"learning_rate": 2.163400726506832e-05,
|
|
"loss": 0.1513,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14377814531326294,
|
|
"step": 2305,
|
|
"valid_targets_mean": 4387.6,
|
|
"valid_targets_min": 2625
|
|
},
|
|
{
|
|
"epoch": 3.6959999999999997,
|
|
"grad_norm": 0.7240363866948977,
|
|
"learning_rate": 2.155446462249961e-05,
|
|
"loss": 0.1617,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1296854019165039,
|
|
"step": 2310,
|
|
"valid_targets_mean": 4542.4,
|
|
"valid_targets_min": 912
|
|
},
|
|
{
|
|
"epoch": 3.7039999999999997,
|
|
"grad_norm": 0.7158727967725006,
|
|
"learning_rate": 2.147489723483217e-05,
|
|
"loss": 0.1507,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13783171772956848,
|
|
"step": 2315,
|
|
"valid_targets_mean": 4190.2,
|
|
"valid_targets_min": 1767
|
|
},
|
|
{
|
|
"epoch": 3.7119999999999997,
|
|
"grad_norm": 1.0171494646151558,
|
|
"learning_rate": 2.139530636867757e-05,
|
|
"loss": 0.1706,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1726381480693817,
|
|
"step": 2320,
|
|
"valid_targets_mean": 4804.3,
|
|
"valid_targets_min": 3476
|
|
},
|
|
{
|
|
"epoch": 3.7199999999999998,
|
|
"grad_norm": 0.8111961243800389,
|
|
"learning_rate": 2.1315693291021114e-05,
|
|
"loss": 0.1663,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1637105643749237,
|
|
"step": 2325,
|
|
"valid_targets_mean": 4731.8,
|
|
"valid_targets_min": 1641
|
|
},
|
|
{
|
|
"epoch": 3.7279999999999998,
|
|
"grad_norm": 0.7444083417685533,
|
|
"learning_rate": 2.1236059269201686e-05,
|
|
"loss": 0.156,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15476304292678833,
|
|
"step": 2330,
|
|
"valid_targets_mean": 4758.3,
|
|
"valid_targets_min": 3449
|
|
},
|
|
{
|
|
"epoch": 3.7359999999999998,
|
|
"grad_norm": 0.7900722850776294,
|
|
"learning_rate": 2.1156405570891584e-05,
|
|
"loss": 0.1691,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17280042171478271,
|
|
"step": 2335,
|
|
"valid_targets_mean": 4230.6,
|
|
"valid_targets_min": 1266
|
|
},
|
|
{
|
|
"epoch": 3.7439999999999998,
|
|
"grad_norm": 0.8269768892231929,
|
|
"learning_rate": 2.1076733464076322e-05,
|
|
"loss": 0.1584,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17558909952640533,
|
|
"step": 2340,
|
|
"valid_targets_mean": 4404.8,
|
|
"valid_targets_min": 2911
|
|
},
|
|
{
|
|
"epoch": 3.752,
|
|
"grad_norm": 0.751654545486785,
|
|
"learning_rate": 2.0997044217034462e-05,
|
|
"loss": 0.168,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14578205347061157,
|
|
"step": 2345,
|
|
"valid_targets_mean": 4440.6,
|
|
"valid_targets_min": 3289
|
|
},
|
|
{
|
|
"epoch": 3.76,
|
|
"grad_norm": 0.7723475906233473,
|
|
"learning_rate": 2.0917339098317405e-05,
|
|
"loss": 0.1603,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16291284561157227,
|
|
"step": 2350,
|
|
"valid_targets_mean": 4502.1,
|
|
"valid_targets_min": 3820
|
|
},
|
|
{
|
|
"epoch": 3.768,
|
|
"grad_norm": 0.7848124558376547,
|
|
"learning_rate": 2.083761937672922e-05,
|
|
"loss": 0.1577,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14275816082954407,
|
|
"step": 2355,
|
|
"valid_targets_mean": 4400.0,
|
|
"valid_targets_min": 1756
|
|
},
|
|
{
|
|
"epoch": 3.776,
|
|
"grad_norm": 0.8411905648427629,
|
|
"learning_rate": 2.0757886321306433e-05,
|
|
"loss": 0.1534,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1524171531200409,
|
|
"step": 2360,
|
|
"valid_targets_mean": 4226.6,
|
|
"valid_targets_min": 1171
|
|
},
|
|
{
|
|
"epoch": 3.784,
|
|
"grad_norm": 0.8582196175410304,
|
|
"learning_rate": 2.0678141201297827e-05,
|
|
"loss": 0.1629,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15119555592536926,
|
|
"step": 2365,
|
|
"valid_targets_mean": 3712.5,
|
|
"valid_targets_min": 796
|
|
},
|
|
{
|
|
"epoch": 3.792,
|
|
"grad_norm": 0.7800338585211174,
|
|
"learning_rate": 2.059838528614423e-05,
|
|
"loss": 0.1606,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16101251542568207,
|
|
"step": 2370,
|
|
"valid_targets_mean": 4324.8,
|
|
"valid_targets_min": 1623
|
|
},
|
|
{
|
|
"epoch": 3.8,
|
|
"grad_norm": 0.752524731996287,
|
|
"learning_rate": 2.0518619845458322e-05,
|
|
"loss": 0.1577,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13612759113311768,
|
|
"step": 2375,
|
|
"valid_targets_mean": 4458.1,
|
|
"valid_targets_min": 2878
|
|
},
|
|
{
|
|
"epoch": 3.808,
|
|
"grad_norm": 0.7341983582346249,
|
|
"learning_rate": 2.0438846149004426e-05,
|
|
"loss": 0.1586,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14201219379901886,
|
|
"step": 2380,
|
|
"valid_targets_mean": 4641.8,
|
|
"valid_targets_min": 3838
|
|
},
|
|
{
|
|
"epoch": 3.816,
|
|
"grad_norm": 0.7676651757238573,
|
|
"learning_rate": 2.0359065466678268e-05,
|
|
"loss": 0.156,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17265881597995758,
|
|
"step": 2385,
|
|
"valid_targets_mean": 4540.4,
|
|
"valid_targets_min": 3393
|
|
},
|
|
{
|
|
"epoch": 3.824,
|
|
"grad_norm": 0.7888974281860628,
|
|
"learning_rate": 2.0279279068486795e-05,
|
|
"loss": 0.1444,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14057913422584534,
|
|
"step": 2390,
|
|
"valid_targets_mean": 4758.4,
|
|
"valid_targets_min": 3718
|
|
},
|
|
{
|
|
"epoch": 3.832,
|
|
"grad_norm": 0.7685339751364212,
|
|
"learning_rate": 2.019948822452794e-05,
|
|
"loss": 0.1441,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12885694205760956,
|
|
"step": 2395,
|
|
"valid_targets_mean": 4358.3,
|
|
"valid_targets_min": 2918
|
|
},
|
|
{
|
|
"epoch": 3.84,
|
|
"grad_norm": 0.7671697655841537,
|
|
"learning_rate": 2.0119694204970393e-05,
|
|
"loss": 0.1535,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13525399565696716,
|
|
"step": 2400,
|
|
"valid_targets_mean": 4513.4,
|
|
"valid_targets_min": 3146
|
|
},
|
|
{
|
|
"epoch": 3.848,
|
|
"grad_norm": 0.8118108915843892,
|
|
"learning_rate": 2.0039898280033414e-05,
|
|
"loss": 0.1625,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18759211897850037,
|
|
"step": 2405,
|
|
"valid_targets_mean": 4627.2,
|
|
"valid_targets_min": 2908
|
|
},
|
|
{
|
|
"epoch": 3.856,
|
|
"grad_norm": 0.8148011826371031,
|
|
"learning_rate": 1.9960101719966592e-05,
|
|
"loss": 0.1582,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.155015230178833,
|
|
"step": 2410,
|
|
"valid_targets_mean": 4692.4,
|
|
"valid_targets_min": 3488
|
|
},
|
|
{
|
|
"epoch": 3.864,
|
|
"grad_norm": 0.8027644472067336,
|
|
"learning_rate": 1.9880305795029617e-05,
|
|
"loss": 0.1559,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17668181657791138,
|
|
"step": 2415,
|
|
"valid_targets_mean": 4252.5,
|
|
"valid_targets_min": 1500
|
|
},
|
|
{
|
|
"epoch": 3.872,
|
|
"grad_norm": 0.8700584076582782,
|
|
"learning_rate": 1.980051177547207e-05,
|
|
"loss": 0.1577,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1771048903465271,
|
|
"step": 2420,
|
|
"valid_targets_mean": 4497.0,
|
|
"valid_targets_min": 1275
|
|
},
|
|
{
|
|
"epoch": 3.88,
|
|
"grad_norm": 0.9107260555156427,
|
|
"learning_rate": 1.9720720931513212e-05,
|
|
"loss": 0.1523,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17244121432304382,
|
|
"step": 2425,
|
|
"valid_targets_mean": 4524.6,
|
|
"valid_targets_min": 3671
|
|
},
|
|
{
|
|
"epoch": 3.888,
|
|
"grad_norm": 0.8091914178362604,
|
|
"learning_rate": 1.9640934533321735e-05,
|
|
"loss": 0.1605,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15282407402992249,
|
|
"step": 2430,
|
|
"valid_targets_mean": 4813.2,
|
|
"valid_targets_min": 3868
|
|
},
|
|
{
|
|
"epoch": 3.896,
|
|
"grad_norm": 0.8560961139521632,
|
|
"learning_rate": 1.9561153850995577e-05,
|
|
"loss": 0.1594,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15245641767978668,
|
|
"step": 2435,
|
|
"valid_targets_mean": 4436.8,
|
|
"valid_targets_min": 437
|
|
},
|
|
{
|
|
"epoch": 3.904,
|
|
"grad_norm": 0.7951858926958523,
|
|
"learning_rate": 1.948138015454168e-05,
|
|
"loss": 0.1695,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16454783082008362,
|
|
"step": 2440,
|
|
"valid_targets_mean": 4208.1,
|
|
"valid_targets_min": 709
|
|
},
|
|
{
|
|
"epoch": 3.912,
|
|
"grad_norm": 0.909964933441964,
|
|
"learning_rate": 1.9401614713855775e-05,
|
|
"loss": 0.1608,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17729684710502625,
|
|
"step": 2445,
|
|
"valid_targets_mean": 4387.7,
|
|
"valid_targets_min": 2819
|
|
},
|
|
{
|
|
"epoch": 3.92,
|
|
"grad_norm": 0.7316059109329837,
|
|
"learning_rate": 1.932185879870218e-05,
|
|
"loss": 0.1524,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1363392472267151,
|
|
"step": 2450,
|
|
"valid_targets_mean": 4578.8,
|
|
"valid_targets_min": 2747
|
|
},
|
|
{
|
|
"epoch": 3.928,
|
|
"grad_norm": 0.7471432830499678,
|
|
"learning_rate": 1.924211367869357e-05,
|
|
"loss": 0.1558,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1411188542842865,
|
|
"step": 2455,
|
|
"valid_targets_mean": 4605.3,
|
|
"valid_targets_min": 3395
|
|
},
|
|
{
|
|
"epoch": 3.936,
|
|
"grad_norm": 0.873155365002312,
|
|
"learning_rate": 1.9162380623270783e-05,
|
|
"loss": 0.1596,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16278159618377686,
|
|
"step": 2460,
|
|
"valid_targets_mean": 4184.2,
|
|
"valid_targets_min": 2198
|
|
},
|
|
{
|
|
"epoch": 3.944,
|
|
"grad_norm": 0.7759888623240812,
|
|
"learning_rate": 1.90826609016826e-05,
|
|
"loss": 0.1562,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12739640474319458,
|
|
"step": 2465,
|
|
"valid_targets_mean": 4289.1,
|
|
"valid_targets_min": 2514
|
|
},
|
|
{
|
|
"epoch": 3.952,
|
|
"grad_norm": 0.8237490054952596,
|
|
"learning_rate": 1.9002955782965548e-05,
|
|
"loss": 0.1508,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1336348056793213,
|
|
"step": 2470,
|
|
"valid_targets_mean": 4211.6,
|
|
"valid_targets_min": 2488
|
|
},
|
|
{
|
|
"epoch": 3.96,
|
|
"grad_norm": 0.8034580006910028,
|
|
"learning_rate": 1.8923266535923688e-05,
|
|
"loss": 0.1519,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15886086225509644,
|
|
"step": 2475,
|
|
"valid_targets_mean": 4826.2,
|
|
"valid_targets_min": 3788
|
|
},
|
|
{
|
|
"epoch": 3.968,
|
|
"grad_norm": 0.8346675545506835,
|
|
"learning_rate": 1.8843594429108426e-05,
|
|
"loss": 0.1545,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13874715566635132,
|
|
"step": 2480,
|
|
"valid_targets_mean": 4239.8,
|
|
"valid_targets_min": 533
|
|
},
|
|
{
|
|
"epoch": 3.976,
|
|
"grad_norm": 0.8622534996950038,
|
|
"learning_rate": 1.8763940730798324e-05,
|
|
"loss": 0.1418,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14953193068504333,
|
|
"step": 2485,
|
|
"valid_targets_mean": 4396.6,
|
|
"valid_targets_min": 2588
|
|
},
|
|
{
|
|
"epoch": 3.984,
|
|
"grad_norm": 0.8821912691867786,
|
|
"learning_rate": 1.8684306708978896e-05,
|
|
"loss": 0.1488,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13349349796772003,
|
|
"step": 2490,
|
|
"valid_targets_mean": 4648.9,
|
|
"valid_targets_min": 4046
|
|
},
|
|
{
|
|
"epoch": 3.992,
|
|
"grad_norm": 0.7256897200149186,
|
|
"learning_rate": 1.8604693631322433e-05,
|
|
"loss": 0.16,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16528138518333435,
|
|
"step": 2495,
|
|
"valid_targets_mean": 4877.3,
|
|
"valid_targets_min": 4175
|
|
},
|
|
{
|
|
"epoch": 4.0,
|
|
"grad_norm": 0.8223355433651999,
|
|
"learning_rate": 1.852510276516783e-05,
|
|
"loss": 0.156,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14207009971141815,
|
|
"step": 2500,
|
|
"valid_targets_mean": 4126.0,
|
|
"valid_targets_min": 1502
|
|
},
|
|
{
|
|
"epoch": 4.008,
|
|
"grad_norm": 0.8210415268759927,
|
|
"learning_rate": 1.8445535377500393e-05,
|
|
"loss": 0.1277,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11087614297866821,
|
|
"step": 2505,
|
|
"valid_targets_mean": 4454.9,
|
|
"valid_targets_min": 2306
|
|
},
|
|
{
|
|
"epoch": 4.016,
|
|
"grad_norm": 0.8787296957109687,
|
|
"learning_rate": 1.8365992734931686e-05,
|
|
"loss": 0.1326,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13519947230815887,
|
|
"step": 2510,
|
|
"valid_targets_mean": 5031.2,
|
|
"valid_targets_min": 3901
|
|
},
|
|
{
|
|
"epoch": 4.024,
|
|
"grad_norm": 0.8692322767488014,
|
|
"learning_rate": 1.8286476103679356e-05,
|
|
"loss": 0.1441,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1450907588005066,
|
|
"step": 2515,
|
|
"valid_targets_mean": 4487.4,
|
|
"valid_targets_min": 2644
|
|
},
|
|
{
|
|
"epoch": 4.032,
|
|
"grad_norm": 0.9885756443501106,
|
|
"learning_rate": 1.8206986749546992e-05,
|
|
"loss": 0.1311,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13911330699920654,
|
|
"step": 2520,
|
|
"valid_targets_mean": 4180.9,
|
|
"valid_targets_min": 926
|
|
},
|
|
{
|
|
"epoch": 4.04,
|
|
"grad_norm": 1.0302523927225058,
|
|
"learning_rate": 1.8127525937903957e-05,
|
|
"loss": 0.1365,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1420406997203827,
|
|
"step": 2525,
|
|
"valid_targets_mean": 4149.2,
|
|
"valid_targets_min": 383
|
|
},
|
|
{
|
|
"epoch": 4.048,
|
|
"grad_norm": 0.8881708947914299,
|
|
"learning_rate": 1.8048094933665262e-05,
|
|
"loss": 0.1262,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15089157223701477,
|
|
"step": 2530,
|
|
"valid_targets_mean": 4746.3,
|
|
"valid_targets_min": 3046
|
|
},
|
|
{
|
|
"epoch": 4.056,
|
|
"grad_norm": 0.9122016960128606,
|
|
"learning_rate": 1.7968695001271416e-05,
|
|
"loss": 0.1354,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13081161677837372,
|
|
"step": 2535,
|
|
"valid_targets_mean": 4357.5,
|
|
"valid_targets_min": 3095
|
|
},
|
|
{
|
|
"epoch": 4.064,
|
|
"grad_norm": 1.216067627308065,
|
|
"learning_rate": 1.7889327404668316e-05,
|
|
"loss": 0.1282,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12417822331190109,
|
|
"step": 2540,
|
|
"valid_targets_mean": 4257.1,
|
|
"valid_targets_min": 2111
|
|
},
|
|
{
|
|
"epoch": 4.072,
|
|
"grad_norm": 0.9018583318140123,
|
|
"learning_rate": 1.7809993407287083e-05,
|
|
"loss": 0.1301,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13237899541854858,
|
|
"step": 2545,
|
|
"valid_targets_mean": 4461.6,
|
|
"valid_targets_min": 2264
|
|
},
|
|
{
|
|
"epoch": 4.08,
|
|
"grad_norm": 0.8700299805620602,
|
|
"learning_rate": 1.7730694272024018e-05,
|
|
"loss": 0.1209,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12628108263015747,
|
|
"step": 2550,
|
|
"valid_targets_mean": 4434.8,
|
|
"valid_targets_min": 3017
|
|
},
|
|
{
|
|
"epoch": 4.088,
|
|
"grad_norm": 0.8143284490519167,
|
|
"learning_rate": 1.765143126122044e-05,
|
|
"loss": 0.134,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1355004459619522,
|
|
"step": 2555,
|
|
"valid_targets_mean": 4855.0,
|
|
"valid_targets_min": 3474
|
|
},
|
|
{
|
|
"epoch": 4.096,
|
|
"grad_norm": 0.8294635235096921,
|
|
"learning_rate": 1.7572205636642622e-05,
|
|
"loss": 0.1269,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11806997656822205,
|
|
"step": 2560,
|
|
"valid_targets_mean": 4702.6,
|
|
"valid_targets_min": 3709
|
|
},
|
|
{
|
|
"epoch": 4.104,
|
|
"grad_norm": 0.9444857905112815,
|
|
"learning_rate": 1.749301865946169e-05,
|
|
"loss": 0.1369,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14227548241615295,
|
|
"step": 2565,
|
|
"valid_targets_mean": 4586.6,
|
|
"valid_targets_min": 3088
|
|
},
|
|
{
|
|
"epoch": 4.112,
|
|
"grad_norm": 0.9114227848713481,
|
|
"learning_rate": 1.7413871590233557e-05,
|
|
"loss": 0.1264,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12511909008026123,
|
|
"step": 2570,
|
|
"valid_targets_mean": 4506.2,
|
|
"valid_targets_min": 2480
|
|
},
|
|
{
|
|
"epoch": 4.12,
|
|
"grad_norm": 0.9010182953808246,
|
|
"learning_rate": 1.7334765688878848e-05,
|
|
"loss": 0.1339,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15343663096427917,
|
|
"step": 2575,
|
|
"valid_targets_mean": 4486.4,
|
|
"valid_targets_min": 1502
|
|
},
|
|
{
|
|
"epoch": 4.128,
|
|
"grad_norm": 0.7995261725084607,
|
|
"learning_rate": 1.7255702214662852e-05,
|
|
"loss": 0.121,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11626158654689789,
|
|
"step": 2580,
|
|
"valid_targets_mean": 4378.1,
|
|
"valid_targets_min": 1691
|
|
},
|
|
{
|
|
"epoch": 4.136,
|
|
"grad_norm": 0.8327792061773963,
|
|
"learning_rate": 1.7176682426175468e-05,
|
|
"loss": 0.1336,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10958704352378845,
|
|
"step": 2585,
|
|
"valid_targets_mean": 4508.5,
|
|
"valid_targets_min": 3168
|
|
},
|
|
{
|
|
"epoch": 4.144,
|
|
"grad_norm": 0.899809546107282,
|
|
"learning_rate": 1.709770758131118e-05,
|
|
"loss": 0.1217,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13027742505073547,
|
|
"step": 2590,
|
|
"valid_targets_mean": 4718.6,
|
|
"valid_targets_min": 3749
|
|
},
|
|
{
|
|
"epoch": 4.152,
|
|
"grad_norm": 0.8784764171571812,
|
|
"learning_rate": 1.7018778937249017e-05,
|
|
"loss": 0.1319,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12138954550027847,
|
|
"step": 2595,
|
|
"valid_targets_mean": 4669.0,
|
|
"valid_targets_min": 3853
|
|
},
|
|
{
|
|
"epoch": 4.16,
|
|
"grad_norm": 0.8844569214781467,
|
|
"learning_rate": 1.6939897750432562e-05,
|
|
"loss": 0.1323,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12189703434705734,
|
|
"step": 2600,
|
|
"valid_targets_mean": 4236.5,
|
|
"valid_targets_min": 1452
|
|
},
|
|
{
|
|
"epoch": 4.168,
|
|
"grad_norm": 0.9002792638268112,
|
|
"learning_rate": 1.6861065276549933e-05,
|
|
"loss": 0.1286,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12161644548177719,
|
|
"step": 2605,
|
|
"valid_targets_mean": 4379.1,
|
|
"valid_targets_min": 3046
|
|
},
|
|
{
|
|
"epoch": 4.176,
|
|
"grad_norm": 0.8521485074093523,
|
|
"learning_rate": 1.6782282770513788e-05,
|
|
"loss": 0.1296,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13108961284160614,
|
|
"step": 2610,
|
|
"valid_targets_mean": 4523.9,
|
|
"valid_targets_min": 3610
|
|
},
|
|
{
|
|
"epoch": 4.184,
|
|
"grad_norm": 0.8387757232304655,
|
|
"learning_rate": 1.6703551486441382e-05,
|
|
"loss": 0.1313,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13820071518421173,
|
|
"step": 2615,
|
|
"valid_targets_mean": 4476.8,
|
|
"valid_targets_min": 2169
|
|
},
|
|
{
|
|
"epoch": 4.192,
|
|
"grad_norm": 1.0324616660443247,
|
|
"learning_rate": 1.6624872677634565e-05,
|
|
"loss": 0.1402,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14743241667747498,
|
|
"step": 2620,
|
|
"valid_targets_mean": 4496.7,
|
|
"valid_targets_min": 1800
|
|
},
|
|
{
|
|
"epoch": 4.2,
|
|
"grad_norm": 0.9729444207584627,
|
|
"learning_rate": 1.654624759655986e-05,
|
|
"loss": 0.1302,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13991251587867737,
|
|
"step": 2625,
|
|
"valid_targets_mean": 4681.2,
|
|
"valid_targets_min": 4058
|
|
},
|
|
{
|
|
"epoch": 4.208,
|
|
"grad_norm": 0.8702809379036429,
|
|
"learning_rate": 1.64676774948285e-05,
|
|
"loss": 0.1291,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13113875687122345,
|
|
"step": 2630,
|
|
"valid_targets_mean": 4496.2,
|
|
"valid_targets_min": 2754
|
|
},
|
|
{
|
|
"epoch": 4.216,
|
|
"grad_norm": 0.9526638810942808,
|
|
"learning_rate": 1.6389163623176536e-05,
|
|
"loss": 0.1246,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1293298602104187,
|
|
"step": 2635,
|
|
"valid_targets_mean": 4454.7,
|
|
"valid_targets_min": 3511
|
|
},
|
|
{
|
|
"epoch": 4.224,
|
|
"grad_norm": 0.8737913675338372,
|
|
"learning_rate": 1.6310707231444884e-05,
|
|
"loss": 0.1198,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12094807624816895,
|
|
"step": 2640,
|
|
"valid_targets_mean": 4319.8,
|
|
"valid_targets_min": 1275
|
|
},
|
|
{
|
|
"epoch": 4.232,
|
|
"grad_norm": 0.841005437053232,
|
|
"learning_rate": 1.623230956855947e-05,
|
|
"loss": 0.1284,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12554490566253662,
|
|
"step": 2645,
|
|
"valid_targets_mean": 4806.9,
|
|
"valid_targets_min": 3808
|
|
},
|
|
{
|
|
"epoch": 4.24,
|
|
"grad_norm": 0.8204309043498031,
|
|
"learning_rate": 1.6153971882511324e-05,
|
|
"loss": 0.1345,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13911524415016174,
|
|
"step": 2650,
|
|
"valid_targets_mean": 4688.9,
|
|
"valid_targets_min": 3523
|
|
},
|
|
{
|
|
"epoch": 4.248,
|
|
"grad_norm": 0.8498532250307234,
|
|
"learning_rate": 1.6075695420336724e-05,
|
|
"loss": 0.1283,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11888610571622849,
|
|
"step": 2655,
|
|
"valid_targets_mean": 4575.1,
|
|
"valid_targets_min": 889
|
|
},
|
|
{
|
|
"epoch": 4.256,
|
|
"grad_norm": 0.8826077361228213,
|
|
"learning_rate": 1.5997481428097338e-05,
|
|
"loss": 0.1313,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12786248326301575,
|
|
"step": 2660,
|
|
"valid_targets_mean": 4494.4,
|
|
"valid_targets_min": 2269
|
|
},
|
|
{
|
|
"epoch": 4.264,
|
|
"grad_norm": 0.8720631217941316,
|
|
"learning_rate": 1.5919331150860396e-05,
|
|
"loss": 0.1255,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1225660964846611,
|
|
"step": 2665,
|
|
"valid_targets_mean": 4423.9,
|
|
"valid_targets_min": 3531
|
|
},
|
|
{
|
|
"epoch": 4.272,
|
|
"grad_norm": 1.0442842907713012,
|
|
"learning_rate": 1.5841245832678873e-05,
|
|
"loss": 0.1376,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13624045252799988,
|
|
"step": 2670,
|
|
"valid_targets_mean": 4220.5,
|
|
"valid_targets_min": 1661
|
|
},
|
|
{
|
|
"epoch": 4.28,
|
|
"grad_norm": 0.9077795177306202,
|
|
"learning_rate": 1.576322671657166e-05,
|
|
"loss": 0.1261,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11935368180274963,
|
|
"step": 2675,
|
|
"valid_targets_mean": 4367.9,
|
|
"valid_targets_min": 2248
|
|
},
|
|
{
|
|
"epoch": 4.288,
|
|
"grad_norm": 0.9620729909784931,
|
|
"learning_rate": 1.5685275044503804e-05,
|
|
"loss": 0.1394,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13141000270843506,
|
|
"step": 2680,
|
|
"valid_targets_mean": 4731.5,
|
|
"valid_targets_min": 3694
|
|
},
|
|
{
|
|
"epoch": 4.296,
|
|
"grad_norm": 0.8569711911455616,
|
|
"learning_rate": 1.560739205736674e-05,
|
|
"loss": 0.1226,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11056534945964813,
|
|
"step": 2685,
|
|
"valid_targets_mean": 4451.7,
|
|
"valid_targets_min": 3168
|
|
},
|
|
{
|
|
"epoch": 4.304,
|
|
"grad_norm": 0.8229272877588962,
|
|
"learning_rate": 1.552957899495851e-05,
|
|
"loss": 0.1222,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1084807962179184,
|
|
"step": 2690,
|
|
"valid_targets_mean": 4650.6,
|
|
"valid_targets_min": 4020
|
|
},
|
|
{
|
|
"epoch": 4.312,
|
|
"grad_norm": 1.0026405404034844,
|
|
"learning_rate": 1.5451837095964054e-05,
|
|
"loss": 0.1293,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13712434470653534,
|
|
"step": 2695,
|
|
"valid_targets_mean": 4594.6,
|
|
"valid_targets_min": 2911
|
|
},
|
|
{
|
|
"epoch": 4.32,
|
|
"grad_norm": 0.8613013483172972,
|
|
"learning_rate": 1.5374167597935478e-05,
|
|
"loss": 0.1305,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1049533560872078,
|
|
"step": 2700,
|
|
"valid_targets_mean": 4459.2,
|
|
"valid_targets_min": 3686
|
|
},
|
|
{
|
|
"epoch": 4.328,
|
|
"grad_norm": 0.9135314528415804,
|
|
"learning_rate": 1.5296571737272354e-05,
|
|
"loss": 0.1368,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13919486105442047,
|
|
"step": 2705,
|
|
"valid_targets_mean": 4580.5,
|
|
"valid_targets_min": 1502
|
|
},
|
|
{
|
|
"epoch": 4.336,
|
|
"grad_norm": 0.8757763208865634,
|
|
"learning_rate": 1.5219050749202037e-05,
|
|
"loss": 0.1245,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.140527606010437,
|
|
"step": 2710,
|
|
"valid_targets_mean": 4674.4,
|
|
"valid_targets_min": 3342
|
|
},
|
|
{
|
|
"epoch": 4.344,
|
|
"grad_norm": 1.000696951787401,
|
|
"learning_rate": 1.5141605867760021e-05,
|
|
"loss": 0.1203,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13654690980911255,
|
|
"step": 2715,
|
|
"valid_targets_mean": 4507.5,
|
|
"valid_targets_min": 3000
|
|
},
|
|
{
|
|
"epoch": 4.352,
|
|
"grad_norm": 0.8671273350726071,
|
|
"learning_rate": 1.5064238325770267e-05,
|
|
"loss": 0.1309,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12745189666748047,
|
|
"step": 2720,
|
|
"valid_targets_mean": 4609.5,
|
|
"valid_targets_min": 3801
|
|
},
|
|
{
|
|
"epoch": 4.36,
|
|
"grad_norm": 0.8162186009156405,
|
|
"learning_rate": 1.498694935482559e-05,
|
|
"loss": 0.1234,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11081044375896454,
|
|
"step": 2725,
|
|
"valid_targets_mean": 4391.2,
|
|
"valid_targets_min": 3006
|
|
},
|
|
{
|
|
"epoch": 4.368,
|
|
"grad_norm": 0.8212202747287932,
|
|
"learning_rate": 1.4909740185268056e-05,
|
|
"loss": 0.1179,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11198791861534119,
|
|
"step": 2730,
|
|
"valid_targets_mean": 4525.9,
|
|
"valid_targets_min": 934
|
|
},
|
|
{
|
|
"epoch": 4.376,
|
|
"grad_norm": 0.8979093771568071,
|
|
"learning_rate": 1.4832612046169408e-05,
|
|
"loss": 0.1233,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13183970749378204,
|
|
"step": 2735,
|
|
"valid_targets_mean": 4780.4,
|
|
"valid_targets_min": 3696
|
|
},
|
|
{
|
|
"epoch": 4.384,
|
|
"grad_norm": 1.015097933976018,
|
|
"learning_rate": 1.4755566165311455e-05,
|
|
"loss": 0.1401,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1437522917985916,
|
|
"step": 2740,
|
|
"valid_targets_mean": 4539.9,
|
|
"valid_targets_min": 2781
|
|
},
|
|
{
|
|
"epoch": 4.392,
|
|
"grad_norm": 0.8420293607228045,
|
|
"learning_rate": 1.4678603769166591e-05,
|
|
"loss": 0.1253,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11503274738788605,
|
|
"step": 2745,
|
|
"valid_targets_mean": 4572.4,
|
|
"valid_targets_min": 3122
|
|
},
|
|
{
|
|
"epoch": 4.4,
|
|
"grad_norm": 0.8246502135189038,
|
|
"learning_rate": 1.4601726082878226e-05,
|
|
"loss": 0.1201,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10758624970912933,
|
|
"step": 2750,
|
|
"valid_targets_mean": 4305.6,
|
|
"valid_targets_min": 1500
|
|
},
|
|
{
|
|
"epoch": 4.408,
|
|
"grad_norm": 0.8466461327869255,
|
|
"learning_rate": 1.4524934330241292e-05,
|
|
"loss": 0.1252,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1151241809129715,
|
|
"step": 2755,
|
|
"valid_targets_mean": 4218.7,
|
|
"valid_targets_min": 1623
|
|
},
|
|
{
|
|
"epoch": 4.416,
|
|
"grad_norm": 1.0534560980630954,
|
|
"learning_rate": 1.4448229733682784e-05,
|
|
"loss": 0.1158,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10284535586833954,
|
|
"step": 2760,
|
|
"valid_targets_mean": 4488.2,
|
|
"valid_targets_min": 674
|
|
},
|
|
{
|
|
"epoch": 4.424,
|
|
"grad_norm": 0.9252345382987744,
|
|
"learning_rate": 1.4371613514242264e-05,
|
|
"loss": 0.1294,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14723145961761475,
|
|
"step": 2765,
|
|
"valid_targets_mean": 4472.1,
|
|
"valid_targets_min": 2222
|
|
},
|
|
{
|
|
"epoch": 4.432,
|
|
"grad_norm": 0.8872043428706476,
|
|
"learning_rate": 1.4295086891552457e-05,
|
|
"loss": 0.1259,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13603493571281433,
|
|
"step": 2770,
|
|
"valid_targets_mean": 4300.8,
|
|
"valid_targets_min": 2221
|
|
},
|
|
{
|
|
"epoch": 4.44,
|
|
"grad_norm": 0.9191173117551973,
|
|
"learning_rate": 1.4218651083819811e-05,
|
|
"loss": 0.1287,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1368560791015625,
|
|
"step": 2775,
|
|
"valid_targets_mean": 4612.9,
|
|
"valid_targets_min": 2541
|
|
},
|
|
{
|
|
"epoch": 4.448,
|
|
"grad_norm": 0.9688986589620144,
|
|
"learning_rate": 1.4142307307805125e-05,
|
|
"loss": 0.1379,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12872910499572754,
|
|
"step": 2780,
|
|
"valid_targets_mean": 4603.2,
|
|
"valid_targets_min": 3444
|
|
},
|
|
{
|
|
"epoch": 4.456,
|
|
"grad_norm": 1.022927393945829,
|
|
"learning_rate": 1.406605677880416e-05,
|
|
"loss": 0.1298,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12192612141370773,
|
|
"step": 2785,
|
|
"valid_targets_mean": 4423.8,
|
|
"valid_targets_min": 2527
|
|
},
|
|
{
|
|
"epoch": 4.464,
|
|
"grad_norm": 0.8932486528250327,
|
|
"learning_rate": 1.3989900710628313e-05,
|
|
"loss": 0.1218,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11987736076116562,
|
|
"step": 2790,
|
|
"valid_targets_mean": 4706.3,
|
|
"valid_targets_min": 2663
|
|
},
|
|
{
|
|
"epoch": 4.4719999999999995,
|
|
"grad_norm": 0.8786227250439019,
|
|
"learning_rate": 1.3913840315585279e-05,
|
|
"loss": 0.1168,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10237512737512589,
|
|
"step": 2795,
|
|
"valid_targets_mean": 4194.4,
|
|
"valid_targets_min": 1852
|
|
},
|
|
{
|
|
"epoch": 4.48,
|
|
"grad_norm": 0.9337658811945582,
|
|
"learning_rate": 1.3837876804459765e-05,
|
|
"loss": 0.128,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1305350363254547,
|
|
"step": 2800,
|
|
"valid_targets_mean": 4233.6,
|
|
"valid_targets_min": 709
|
|
},
|
|
{
|
|
"epoch": 4.4879999999999995,
|
|
"grad_norm": 0.9574630960859271,
|
|
"learning_rate": 1.3762011386494191e-05,
|
|
"loss": 0.1205,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11888418346643448,
|
|
"step": 2805,
|
|
"valid_targets_mean": 4455.6,
|
|
"valid_targets_min": 2566
|
|
},
|
|
{
|
|
"epoch": 4.496,
|
|
"grad_norm": 0.9014362447208852,
|
|
"learning_rate": 1.3686245269369485e-05,
|
|
"loss": 0.1186,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11498626321554184,
|
|
"step": 2810,
|
|
"valid_targets_mean": 4156.5,
|
|
"valid_targets_min": 1875
|
|
},
|
|
{
|
|
"epoch": 4.504,
|
|
"grad_norm": 1.0623775331231256,
|
|
"learning_rate": 1.3610579659185809e-05,
|
|
"loss": 0.122,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12701629102230072,
|
|
"step": 2815,
|
|
"valid_targets_mean": 4006.9,
|
|
"valid_targets_min": 796
|
|
},
|
|
{
|
|
"epoch": 4.5120000000000005,
|
|
"grad_norm": 0.8915639614153672,
|
|
"learning_rate": 1.35350157604434e-05,
|
|
"loss": 0.1116,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10582106560468674,
|
|
"step": 2820,
|
|
"valid_targets_mean": 4422.6,
|
|
"valid_targets_min": 1875
|
|
},
|
|
{
|
|
"epoch": 4.52,
|
|
"grad_norm": 2.6407820943775135,
|
|
"learning_rate": 1.345955477602337e-05,
|
|
"loss": 0.1199,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11786778271198273,
|
|
"step": 2825,
|
|
"valid_targets_mean": 4479.7,
|
|
"valid_targets_min": 3427
|
|
},
|
|
{
|
|
"epoch": 4.5280000000000005,
|
|
"grad_norm": 0.9820171671919866,
|
|
"learning_rate": 1.3384197907168561e-05,
|
|
"loss": 0.1288,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1197080910205841,
|
|
"step": 2830,
|
|
"valid_targets_mean": 4415.9,
|
|
"valid_targets_min": 3000
|
|
},
|
|
{
|
|
"epoch": 4.536,
|
|
"grad_norm": 0.9061857091875595,
|
|
"learning_rate": 1.3308946353464438e-05,
|
|
"loss": 0.1245,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10080514848232269,
|
|
"step": 2835,
|
|
"valid_targets_mean": 4698.4,
|
|
"valid_targets_min": 2850
|
|
},
|
|
{
|
|
"epoch": 4.5440000000000005,
|
|
"grad_norm": 0.8867463820049024,
|
|
"learning_rate": 1.3233801312819979e-05,
|
|
"loss": 0.1276,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10956710577011108,
|
|
"step": 2840,
|
|
"valid_targets_mean": 4777.8,
|
|
"valid_targets_min": 3048
|
|
},
|
|
{
|
|
"epoch": 4.552,
|
|
"grad_norm": 0.9922109849451005,
|
|
"learning_rate": 1.3158763981448606e-05,
|
|
"loss": 0.1118,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1070832908153534,
|
|
"step": 2845,
|
|
"valid_targets_mean": 4362.7,
|
|
"valid_targets_min": 970
|
|
},
|
|
{
|
|
"epoch": 4.5600000000000005,
|
|
"grad_norm": 1.102073820679673,
|
|
"learning_rate": 1.3083835553849148e-05,
|
|
"loss": 0.1164,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12305476516485214,
|
|
"step": 2850,
|
|
"valid_targets_mean": 4975.6,
|
|
"valid_targets_min": 3447
|
|
},
|
|
{
|
|
"epoch": 4.568,
|
|
"grad_norm": 0.8539481112653702,
|
|
"learning_rate": 1.3009017222786828e-05,
|
|
"loss": 0.1252,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12367376685142517,
|
|
"step": 2855,
|
|
"valid_targets_mean": 4561.4,
|
|
"valid_targets_min": 3308
|
|
},
|
|
{
|
|
"epoch": 4.576,
|
|
"grad_norm": 0.8891715986159021,
|
|
"learning_rate": 1.2934310179274269e-05,
|
|
"loss": 0.1167,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11885467171669006,
|
|
"step": 2860,
|
|
"valid_targets_mean": 4351.4,
|
|
"valid_targets_min": 2003
|
|
},
|
|
{
|
|
"epoch": 4.584,
|
|
"grad_norm": 0.8922758915804034,
|
|
"learning_rate": 1.2859715612552541e-05,
|
|
"loss": 0.1166,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10838243365287781,
|
|
"step": 2865,
|
|
"valid_targets_mean": 4373.2,
|
|
"valid_targets_min": 2093
|
|
},
|
|
{
|
|
"epoch": 4.592,
|
|
"grad_norm": 0.8071226000670852,
|
|
"learning_rate": 1.278523471007223e-05,
|
|
"loss": 0.1183,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09847372025251389,
|
|
"step": 2870,
|
|
"valid_targets_mean": 4509.2,
|
|
"valid_targets_min": 3704
|
|
},
|
|
{
|
|
"epoch": 4.6,
|
|
"grad_norm": 0.938106336761023,
|
|
"learning_rate": 1.271086865747451e-05,
|
|
"loss": 0.1277,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12543395161628723,
|
|
"step": 2875,
|
|
"valid_targets_mean": 4239.5,
|
|
"valid_targets_min": 2264
|
|
},
|
|
{
|
|
"epoch": 4.608,
|
|
"grad_norm": 0.9474488928691247,
|
|
"learning_rate": 1.2636618638572316e-05,
|
|
"loss": 0.1218,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12688979506492615,
|
|
"step": 2880,
|
|
"valid_targets_mean": 4475.5,
|
|
"valid_targets_min": 2941
|
|
},
|
|
{
|
|
"epoch": 4.616,
|
|
"grad_norm": 0.988137030539896,
|
|
"learning_rate": 1.2562485835331466e-05,
|
|
"loss": 0.1292,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13637849688529968,
|
|
"step": 2885,
|
|
"valid_targets_mean": 4295.7,
|
|
"valid_targets_min": 2242
|
|
},
|
|
{
|
|
"epoch": 4.624,
|
|
"grad_norm": 0.9530852873066519,
|
|
"learning_rate": 1.2488471427851852e-05,
|
|
"loss": 0.1191,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11682432889938354,
|
|
"step": 2890,
|
|
"valid_targets_mean": 4366.5,
|
|
"valid_targets_min": 2771
|
|
},
|
|
{
|
|
"epoch": 4.632,
|
|
"grad_norm": 0.8889419684691198,
|
|
"learning_rate": 1.241457659434866e-05,
|
|
"loss": 0.1274,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10850934684276581,
|
|
"step": 2895,
|
|
"valid_targets_mean": 4562.9,
|
|
"valid_targets_min": 2268
|
|
},
|
|
{
|
|
"epoch": 4.64,
|
|
"grad_norm": 0.912220778345041,
|
|
"learning_rate": 1.2340802511133605e-05,
|
|
"loss": 0.1217,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11009922623634338,
|
|
"step": 2900,
|
|
"valid_targets_mean": 4677.8,
|
|
"valid_targets_min": 3707
|
|
},
|
|
{
|
|
"epoch": 4.648,
|
|
"grad_norm": 0.8724837106567461,
|
|
"learning_rate": 1.2267150352596216e-05,
|
|
"loss": 0.1174,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10858947038650513,
|
|
"step": 2905,
|
|
"valid_targets_mean": 4507.5,
|
|
"valid_targets_min": 3619
|
|
},
|
|
{
|
|
"epoch": 4.656,
|
|
"grad_norm": 0.8659109477996534,
|
|
"learning_rate": 1.2193621291185132e-05,
|
|
"loss": 0.122,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12659163773059845,
|
|
"step": 2910,
|
|
"valid_targets_mean": 4229.7,
|
|
"valid_targets_min": 1997
|
|
},
|
|
{
|
|
"epoch": 4.664,
|
|
"grad_norm": 0.9280204858098339,
|
|
"learning_rate": 1.2120216497389446e-05,
|
|
"loss": 0.121,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10974974185228348,
|
|
"step": 2915,
|
|
"valid_targets_mean": 4770.1,
|
|
"valid_targets_min": 2608
|
|
},
|
|
{
|
|
"epoch": 4.672,
|
|
"grad_norm": 0.9711744349412768,
|
|
"learning_rate": 1.2046937139720068e-05,
|
|
"loss": 0.1225,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12705110013484955,
|
|
"step": 2920,
|
|
"valid_targets_mean": 4487.7,
|
|
"valid_targets_min": 2472
|
|
},
|
|
{
|
|
"epoch": 4.68,
|
|
"grad_norm": 0.884793838324572,
|
|
"learning_rate": 1.1973784384691121e-05,
|
|
"loss": 0.1239,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1327436864376068,
|
|
"step": 2925,
|
|
"valid_targets_mean": 4665.1,
|
|
"valid_targets_min": 3199
|
|
},
|
|
{
|
|
"epoch": 4.688,
|
|
"grad_norm": 0.9552787115315434,
|
|
"learning_rate": 1.1900759396801382e-05,
|
|
"loss": 0.1257,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13108131289482117,
|
|
"step": 2930,
|
|
"valid_targets_mean": 4540.8,
|
|
"valid_targets_min": 2450
|
|
},
|
|
{
|
|
"epoch": 4.696,
|
|
"grad_norm": 1.0566777508862333,
|
|
"learning_rate": 1.1827863338515741e-05,
|
|
"loss": 0.1212,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12033368647098541,
|
|
"step": 2935,
|
|
"valid_targets_mean": 4728.9,
|
|
"valid_targets_min": 1448
|
|
},
|
|
{
|
|
"epoch": 4.704,
|
|
"grad_norm": 1.025847285433629,
|
|
"learning_rate": 1.1755097370246669e-05,
|
|
"loss": 0.1237,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12928824126720428,
|
|
"step": 2940,
|
|
"valid_targets_mean": 4205.5,
|
|
"valid_targets_min": 1839
|
|
},
|
|
{
|
|
"epoch": 4.712,
|
|
"grad_norm": 1.0168841320844981,
|
|
"learning_rate": 1.1682462650335791e-05,
|
|
"loss": 0.1094,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10925668478012085,
|
|
"step": 2945,
|
|
"valid_targets_mean": 4396.7,
|
|
"valid_targets_min": 934
|
|
},
|
|
{
|
|
"epoch": 4.72,
|
|
"grad_norm": 0.9330806808769733,
|
|
"learning_rate": 1.1609960335035423e-05,
|
|
"loss": 0.1153,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10928137600421906,
|
|
"step": 2950,
|
|
"valid_targets_mean": 4567.1,
|
|
"valid_targets_min": 3085
|
|
},
|
|
{
|
|
"epoch": 4.728,
|
|
"grad_norm": 1.1218529032321893,
|
|
"learning_rate": 1.1537591578490165e-05,
|
|
"loss": 0.1349,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14489446580410004,
|
|
"step": 2955,
|
|
"valid_targets_mean": 4496.2,
|
|
"valid_targets_min": 3365
|
|
},
|
|
{
|
|
"epoch": 4.736,
|
|
"grad_norm": 0.9695867505363509,
|
|
"learning_rate": 1.146535753271853e-05,
|
|
"loss": 0.1205,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14045220613479614,
|
|
"step": 2960,
|
|
"valid_targets_mean": 4572.9,
|
|
"valid_targets_min": 3403
|
|
},
|
|
{
|
|
"epoch": 4.744,
|
|
"grad_norm": 0.8743440117551271,
|
|
"learning_rate": 1.139325934759461e-05,
|
|
"loss": 0.1232,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1123942881822586,
|
|
"step": 2965,
|
|
"valid_targets_mean": 4200.6,
|
|
"valid_targets_min": 1280
|
|
},
|
|
{
|
|
"epoch": 4.752,
|
|
"grad_norm": 0.9576088926180587,
|
|
"learning_rate": 1.1321298170829768e-05,
|
|
"loss": 0.1166,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12689080834388733,
|
|
"step": 2970,
|
|
"valid_targets_mean": 4559.9,
|
|
"valid_targets_min": 3941
|
|
},
|
|
{
|
|
"epoch": 4.76,
|
|
"grad_norm": 0.9585316195440083,
|
|
"learning_rate": 1.1249475147954363e-05,
|
|
"loss": 0.1178,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13638204336166382,
|
|
"step": 2975,
|
|
"valid_targets_mean": 4378.1,
|
|
"valid_targets_min": 2604
|
|
},
|
|
{
|
|
"epoch": 4.768,
|
|
"grad_norm": 1.0812214352933986,
|
|
"learning_rate": 1.1177791422299528e-05,
|
|
"loss": 0.1252,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13600561022758484,
|
|
"step": 2980,
|
|
"valid_targets_mean": 4581.5,
|
|
"valid_targets_min": 2908
|
|
},
|
|
{
|
|
"epoch": 4.776,
|
|
"grad_norm": 0.8830470435294472,
|
|
"learning_rate": 1.1106248134978959e-05,
|
|
"loss": 0.1185,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12618562579154968,
|
|
"step": 2985,
|
|
"valid_targets_mean": 4087.4,
|
|
"valid_targets_min": 1276
|
|
},
|
|
{
|
|
"epoch": 4.784,
|
|
"grad_norm": 0.8742380932469804,
|
|
"learning_rate": 1.1034846424870744e-05,
|
|
"loss": 0.107,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09562106430530548,
|
|
"step": 2990,
|
|
"valid_targets_mean": 4638.6,
|
|
"valid_targets_min": 3421
|
|
},
|
|
{
|
|
"epoch": 4.792,
|
|
"grad_norm": 1.2269050404703068,
|
|
"learning_rate": 1.0963587428599256e-05,
|
|
"loss": 0.1308,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16860944032669067,
|
|
"step": 2995,
|
|
"valid_targets_mean": 4368.0,
|
|
"valid_targets_min": 2647
|
|
},
|
|
{
|
|
"epoch": 4.8,
|
|
"grad_norm": 0.9829312494021705,
|
|
"learning_rate": 1.089247228051704e-05,
|
|
"loss": 0.1207,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14158979058265686,
|
|
"step": 3000,
|
|
"valid_targets_mean": 4453.6,
|
|
"valid_targets_min": 1500
|
|
},
|
|
{
|
|
"epoch": 4.808,
|
|
"grad_norm": 1.0034742075418455,
|
|
"learning_rate": 1.0821502112686753e-05,
|
|
"loss": 0.116,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10320864617824554,
|
|
"step": 3005,
|
|
"valid_targets_mean": 4592.9,
|
|
"valid_targets_min": 1180
|
|
},
|
|
{
|
|
"epoch": 4.816,
|
|
"grad_norm": 0.8698848202094884,
|
|
"learning_rate": 1.0750678054863158e-05,
|
|
"loss": 0.1225,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12271146476268768,
|
|
"step": 3010,
|
|
"valid_targets_mean": 4809.0,
|
|
"valid_targets_min": 2604
|
|
},
|
|
{
|
|
"epoch": 4.824,
|
|
"grad_norm": 0.9898558172858881,
|
|
"learning_rate": 1.0680001234475127e-05,
|
|
"loss": 0.1158,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11627490073442459,
|
|
"step": 3015,
|
|
"valid_targets_mean": 4430.2,
|
|
"valid_targets_min": 1623
|
|
},
|
|
{
|
|
"epoch": 4.832,
|
|
"grad_norm": 0.8690060731526007,
|
|
"learning_rate": 1.0609472776607715e-05,
|
|
"loss": 0.1109,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11362262070178986,
|
|
"step": 3020,
|
|
"valid_targets_mean": 4715.4,
|
|
"valid_targets_min": 3818
|
|
},
|
|
{
|
|
"epoch": 4.84,
|
|
"grad_norm": 1.0021596354916502,
|
|
"learning_rate": 1.0539093803984217e-05,
|
|
"loss": 0.1214,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12027707695960999,
|
|
"step": 3025,
|
|
"valid_targets_mean": 4274.6,
|
|
"valid_targets_min": 1833
|
|
},
|
|
{
|
|
"epoch": 4.848,
|
|
"grad_norm": 1.0058443346999417,
|
|
"learning_rate": 1.046886543694832e-05,
|
|
"loss": 0.1237,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11630566418170929,
|
|
"step": 3030,
|
|
"valid_targets_mean": 4027.4,
|
|
"valid_targets_min": 1855
|
|
},
|
|
{
|
|
"epoch": 4.856,
|
|
"grad_norm": 0.8065244905255626,
|
|
"learning_rate": 1.0398788793446263e-05,
|
|
"loss": 0.1178,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11266714334487915,
|
|
"step": 3035,
|
|
"valid_targets_mean": 4917.6,
|
|
"valid_targets_min": 4091
|
|
},
|
|
{
|
|
"epoch": 4.864,
|
|
"grad_norm": 0.8484087197502739,
|
|
"learning_rate": 1.0328864989009037e-05,
|
|
"loss": 0.1213,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10497960448265076,
|
|
"step": 3040,
|
|
"valid_targets_mean": 4671.1,
|
|
"valid_targets_min": 3563
|
|
},
|
|
{
|
|
"epoch": 4.872,
|
|
"grad_norm": 1.0107470848497029,
|
|
"learning_rate": 1.0259095136734634e-05,
|
|
"loss": 0.1228,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1646127700805664,
|
|
"step": 3045,
|
|
"valid_targets_mean": 4579.3,
|
|
"valid_targets_min": 1038
|
|
},
|
|
{
|
|
"epoch": 4.88,
|
|
"grad_norm": 1.0992434765146812,
|
|
"learning_rate": 1.0189480347270311e-05,
|
|
"loss": 0.1256,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14360107481479645,
|
|
"step": 3050,
|
|
"valid_targets_mean": 4535.1,
|
|
"valid_targets_min": 1915
|
|
},
|
|
{
|
|
"epoch": 4.888,
|
|
"grad_norm": 1.223058178551748,
|
|
"learning_rate": 1.0120021728794938e-05,
|
|
"loss": 0.1135,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11999551951885223,
|
|
"step": 3055,
|
|
"valid_targets_mean": 4789.8,
|
|
"valid_targets_min": 3726
|
|
},
|
|
{
|
|
"epoch": 4.896,
|
|
"grad_norm": 0.8885256605099695,
|
|
"learning_rate": 1.0050720387001334e-05,
|
|
"loss": 0.1187,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12425725162029266,
|
|
"step": 3060,
|
|
"valid_targets_mean": 4822.6,
|
|
"valid_targets_min": 3855
|
|
},
|
|
{
|
|
"epoch": 4.904,
|
|
"grad_norm": 0.9819369971237449,
|
|
"learning_rate": 9.981577425078672e-06,
|
|
"loss": 0.1257,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12856510281562805,
|
|
"step": 3065,
|
|
"valid_targets_mean": 4775.0,
|
|
"valid_targets_min": 3943
|
|
},
|
|
{
|
|
"epoch": 4.912,
|
|
"grad_norm": 0.9473558729264966,
|
|
"learning_rate": 9.912593943694924e-06,
|
|
"loss": 0.1096,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1181732639670372,
|
|
"step": 3070,
|
|
"valid_targets_mean": 4819.2,
|
|
"valid_targets_min": 4196
|
|
},
|
|
{
|
|
"epoch": 4.92,
|
|
"grad_norm": 1.0345669511086757,
|
|
"learning_rate": 9.843771040979328e-06,
|
|
"loss": 0.1207,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12042224407196045,
|
|
"step": 3075,
|
|
"valid_targets_mean": 4545.2,
|
|
"valid_targets_min": 2893
|
|
},
|
|
{
|
|
"epoch": 4.928,
|
|
"grad_norm": 0.900063118579883,
|
|
"learning_rate": 9.775109812504922e-06,
|
|
"loss": 0.1243,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11888638138771057,
|
|
"step": 3080,
|
|
"valid_targets_mean": 4631.7,
|
|
"valid_targets_min": 1855
|
|
},
|
|
{
|
|
"epoch": 4.936,
|
|
"grad_norm": 0.8848835727351212,
|
|
"learning_rate": 9.706611351271088e-06,
|
|
"loss": 0.1257,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09977660328149796,
|
|
"step": 3085,
|
|
"valid_targets_mean": 4701.0,
|
|
"valid_targets_min": 4107
|
|
},
|
|
{
|
|
"epoch": 4.944,
|
|
"grad_norm": 0.9028218147111922,
|
|
"learning_rate": 9.638276747686169e-06,
|
|
"loss": 0.1096,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1146727204322815,
|
|
"step": 3090,
|
|
"valid_targets_mean": 4827.4,
|
|
"valid_targets_min": 3836
|
|
},
|
|
{
|
|
"epoch": 4.952,
|
|
"grad_norm": 1.007029956734003,
|
|
"learning_rate": 9.570107089550091e-06,
|
|
"loss": 0.1217,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11671005934476852,
|
|
"step": 3095,
|
|
"valid_targets_mean": 4505.1,
|
|
"valid_targets_min": 2527
|
|
},
|
|
{
|
|
"epoch": 4.96,
|
|
"grad_norm": 0.9832776574776416,
|
|
"learning_rate": 9.502103462037074e-06,
|
|
"loss": 0.1158,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11825068295001984,
|
|
"step": 3100,
|
|
"valid_targets_mean": 4730.7,
|
|
"valid_targets_min": 4180
|
|
},
|
|
{
|
|
"epoch": 4.968,
|
|
"grad_norm": 0.9627688925323846,
|
|
"learning_rate": 9.434266947678326e-06,
|
|
"loss": 0.1097,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10409554094076157,
|
|
"step": 3105,
|
|
"valid_targets_mean": 4411.9,
|
|
"valid_targets_min": 3318
|
|
},
|
|
{
|
|
"epoch": 4.976,
|
|
"grad_norm": 0.971238302948729,
|
|
"learning_rate": 9.366598626344836e-06,
|
|
"loss": 0.1173,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12805861234664917,
|
|
"step": 3110,
|
|
"valid_targets_mean": 4427.3,
|
|
"valid_targets_min": 3243
|
|
},
|
|
{
|
|
"epoch": 4.984,
|
|
"grad_norm": 1.0255247622308403,
|
|
"learning_rate": 9.299099575230172e-06,
|
|
"loss": 0.1144,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12859445810317993,
|
|
"step": 3115,
|
|
"valid_targets_mean": 4488.8,
|
|
"valid_targets_min": 2893
|
|
},
|
|
{
|
|
"epoch": 4.992,
|
|
"grad_norm": 1.0558641294927158,
|
|
"learning_rate": 9.231770868833334e-06,
|
|
"loss": 0.1128,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10420358926057816,
|
|
"step": 3120,
|
|
"valid_targets_mean": 4235.7,
|
|
"valid_targets_min": 3599
|
|
},
|
|
{
|
|
"epoch": 5.0,
|
|
"grad_norm": 1.3915541744244435,
|
|
"learning_rate": 9.164613578941652e-06,
|
|
"loss": 0.1197,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11326853185892105,
|
|
"step": 3125,
|
|
"valid_targets_mean": 4434.3,
|
|
"valid_targets_min": 2940
|
|
},
|
|
{
|
|
"epoch": 5.008,
|
|
"grad_norm": 0.8563069329481541,
|
|
"learning_rate": 9.097628774613732e-06,
|
|
"loss": 0.0993,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10614681243896484,
|
|
"step": 3130,
|
|
"valid_targets_mean": 4696.1,
|
|
"valid_targets_min": 3421
|
|
},
|
|
{
|
|
"epoch": 5.016,
|
|
"grad_norm": 0.845376958180274,
|
|
"learning_rate": 9.030817522162403e-06,
|
|
"loss": 0.0926,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08138807117938995,
|
|
"step": 3135,
|
|
"valid_targets_mean": 4268.4,
|
|
"valid_targets_min": 2476
|
|
},
|
|
{
|
|
"epoch": 5.024,
|
|
"grad_norm": 0.9025465388150725,
|
|
"learning_rate": 8.964180885137797e-06,
|
|
"loss": 0.1008,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09270966053009033,
|
|
"step": 3140,
|
|
"valid_targets_mean": 4310.6,
|
|
"valid_targets_min": 1857
|
|
},
|
|
{
|
|
"epoch": 5.032,
|
|
"grad_norm": 0.9660939209894497,
|
|
"learning_rate": 8.897719924310375e-06,
|
|
"loss": 0.0933,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09960483014583588,
|
|
"step": 3145,
|
|
"valid_targets_mean": 4679.1,
|
|
"valid_targets_min": 3318
|
|
},
|
|
{
|
|
"epoch": 5.04,
|
|
"grad_norm": 0.9676424380810908,
|
|
"learning_rate": 8.831435697654068e-06,
|
|
"loss": 0.0879,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09373152256011963,
|
|
"step": 3150,
|
|
"valid_targets_mean": 4721.7,
|
|
"valid_targets_min": 4140
|
|
},
|
|
{
|
|
"epoch": 5.048,
|
|
"grad_norm": 0.9702672615298229,
|
|
"learning_rate": 8.765329260329413e-06,
|
|
"loss": 0.1033,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09621229767799377,
|
|
"step": 3155,
|
|
"valid_targets_mean": 4400.4,
|
|
"valid_targets_min": 3451
|
|
},
|
|
{
|
|
"epoch": 5.056,
|
|
"grad_norm": 0.9565132150477523,
|
|
"learning_rate": 8.699401664666774e-06,
|
|
"loss": 0.0967,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07502689957618713,
|
|
"step": 3160,
|
|
"valid_targets_mean": 4646.4,
|
|
"valid_targets_min": 3385
|
|
},
|
|
{
|
|
"epoch": 5.064,
|
|
"grad_norm": 0.9467154067484276,
|
|
"learning_rate": 8.633653960149579e-06,
|
|
"loss": 0.1004,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11934950947761536,
|
|
"step": 3165,
|
|
"valid_targets_mean": 4632.7,
|
|
"valid_targets_min": 3085
|
|
},
|
|
{
|
|
"epoch": 5.072,
|
|
"grad_norm": 1.0256072893015145,
|
|
"learning_rate": 8.56808719339762e-06,
|
|
"loss": 0.0926,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08504219353199005,
|
|
"step": 3170,
|
|
"valid_targets_mean": 4481.8,
|
|
"valid_targets_min": 2306
|
|
},
|
|
{
|
|
"epoch": 5.08,
|
|
"grad_norm": 0.9811366155787422,
|
|
"learning_rate": 8.502702408150391e-06,
|
|
"loss": 0.1064,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11032101511955261,
|
|
"step": 3175,
|
|
"valid_targets_mean": 4277.8,
|
|
"valid_targets_min": 2775
|
|
},
|
|
{
|
|
"epoch": 5.088,
|
|
"grad_norm": 1.0326895154388958,
|
|
"learning_rate": 8.43750064525047e-06,
|
|
"loss": 0.0993,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10859635472297668,
|
|
"step": 3180,
|
|
"valid_targets_mean": 4665.2,
|
|
"valid_targets_min": 1062
|
|
},
|
|
{
|
|
"epoch": 5.096,
|
|
"grad_norm": 1.1048807230739326,
|
|
"learning_rate": 8.372482942626952e-06,
|
|
"loss": 0.0929,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09459155797958374,
|
|
"step": 3185,
|
|
"valid_targets_mean": 4373.8,
|
|
"valid_targets_min": 2608
|
|
},
|
|
{
|
|
"epoch": 5.104,
|
|
"grad_norm": 0.9567198073688751,
|
|
"learning_rate": 8.307650335278927e-06,
|
|
"loss": 0.1072,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10010776668787003,
|
|
"step": 3190,
|
|
"valid_targets_mean": 3917.3,
|
|
"valid_targets_min": 912
|
|
},
|
|
{
|
|
"epoch": 5.112,
|
|
"grad_norm": 1.0445356772810388,
|
|
"learning_rate": 8.243003855259015e-06,
|
|
"loss": 0.111,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11553865671157837,
|
|
"step": 3195,
|
|
"valid_targets_mean": 4665.4,
|
|
"valid_targets_min": 3727
|
|
},
|
|
{
|
|
"epoch": 5.12,
|
|
"grad_norm": 0.9884150896781629,
|
|
"learning_rate": 8.178544531656897e-06,
|
|
"loss": 0.1085,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08703594654798508,
|
|
"step": 3200,
|
|
"valid_targets_mean": 4698.6,
|
|
"valid_targets_min": 2620
|
|
},
|
|
{
|
|
"epoch": 5.128,
|
|
"grad_norm": 0.9608154950720944,
|
|
"learning_rate": 8.11427339058299e-06,
|
|
"loss": 0.1123,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11205752193927765,
|
|
"step": 3205,
|
|
"valid_targets_mean": 4684.2,
|
|
"valid_targets_min": 3670
|
|
},
|
|
{
|
|
"epoch": 5.136,
|
|
"grad_norm": 0.9208900814977037,
|
|
"learning_rate": 8.050191455152072e-06,
|
|
"loss": 0.1032,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09493430703878403,
|
|
"step": 3210,
|
|
"valid_targets_mean": 4684.9,
|
|
"valid_targets_min": 4087
|
|
},
|
|
{
|
|
"epoch": 5.144,
|
|
"grad_norm": 1.0389486221005826,
|
|
"learning_rate": 7.986299745467013e-06,
|
|
"loss": 0.09,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09533935785293579,
|
|
"step": 3215,
|
|
"valid_targets_mean": 4703.8,
|
|
"valid_targets_min": 2908
|
|
},
|
|
{
|
|
"epoch": 5.152,
|
|
"grad_norm": 1.0260858732256064,
|
|
"learning_rate": 7.922599278602524e-06,
|
|
"loss": 0.1029,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08555421233177185,
|
|
"step": 3220,
|
|
"valid_targets_mean": 4381.8,
|
|
"valid_targets_min": 534
|
|
},
|
|
{
|
|
"epoch": 5.16,
|
|
"grad_norm": 1.0869170517101912,
|
|
"learning_rate": 7.859091068588987e-06,
|
|
"loss": 0.0921,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08201434463262558,
|
|
"step": 3225,
|
|
"valid_targets_mean": 4586.9,
|
|
"valid_targets_min": 3529
|
|
},
|
|
{
|
|
"epoch": 5.168,
|
|
"grad_norm": 1.0962303078223183,
|
|
"learning_rate": 7.795776126396284e-06,
|
|
"loss": 0.1103,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11936794966459274,
|
|
"step": 3230,
|
|
"valid_targets_mean": 4664.1,
|
|
"valid_targets_min": 3525
|
|
},
|
|
{
|
|
"epoch": 5.176,
|
|
"grad_norm": 1.5080967357355342,
|
|
"learning_rate": 7.732655459917726e-06,
|
|
"loss": 0.0963,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10037938505411148,
|
|
"step": 3235,
|
|
"valid_targets_mean": 4369.3,
|
|
"valid_targets_min": 2046
|
|
},
|
|
{
|
|
"epoch": 5.184,
|
|
"grad_norm": 0.9009102176331055,
|
|
"learning_rate": 7.669730073954005e-06,
|
|
"loss": 0.1094,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08869896084070206,
|
|
"step": 3240,
|
|
"valid_targets_mean": 4349.8,
|
|
"valid_targets_min": 2277
|
|
},
|
|
{
|
|
"epoch": 5.192,
|
|
"grad_norm": 0.9029354968762454,
|
|
"learning_rate": 7.607000970197194e-06,
|
|
"loss": 0.1071,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0901302695274353,
|
|
"step": 3245,
|
|
"valid_targets_mean": 4862.9,
|
|
"valid_targets_min": 3235
|
|
},
|
|
{
|
|
"epoch": 5.2,
|
|
"grad_norm": 1.021803260454566,
|
|
"learning_rate": 7.544469147214797e-06,
|
|
"loss": 0.0962,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10541929304599762,
|
|
"step": 3250,
|
|
"valid_targets_mean": 4292.1,
|
|
"valid_targets_min": 2810
|
|
},
|
|
{
|
|
"epoch": 5.208,
|
|
"grad_norm": 1.0532504692469233,
|
|
"learning_rate": 7.482135600433868e-06,
|
|
"loss": 0.1021,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10755719244480133,
|
|
"step": 3255,
|
|
"valid_targets_mean": 4112.4,
|
|
"valid_targets_min": 1624
|
|
},
|
|
{
|
|
"epoch": 5.216,
|
|
"grad_norm": 0.9135676112953338,
|
|
"learning_rate": 7.420001322125156e-06,
|
|
"loss": 0.0904,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08535128086805344,
|
|
"step": 3260,
|
|
"valid_targets_mean": 4495.6,
|
|
"valid_targets_min": 3502
|
|
},
|
|
{
|
|
"epoch": 5.224,
|
|
"grad_norm": 1.0321296347461633,
|
|
"learning_rate": 7.3580673013872946e-06,
|
|
"loss": 0.0974,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08713699132204056,
|
|
"step": 3265,
|
|
"valid_targets_mean": 4334.7,
|
|
"valid_targets_min": 2904
|
|
},
|
|
{
|
|
"epoch": 5.232,
|
|
"grad_norm": 1.0672062983053705,
|
|
"learning_rate": 7.2963345241310904e-06,
|
|
"loss": 0.1021,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1000540480017662,
|
|
"step": 3270,
|
|
"valid_targets_mean": 4453.2,
|
|
"valid_targets_min": 3277
|
|
},
|
|
{
|
|
"epoch": 5.24,
|
|
"grad_norm": 1.0648268301982886,
|
|
"learning_rate": 7.234803973063797e-06,
|
|
"loss": 0.1008,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11028389632701874,
|
|
"step": 3275,
|
|
"valid_targets_mean": 4438.5,
|
|
"valid_targets_min": 1852
|
|
},
|
|
{
|
|
"epoch": 5.248,
|
|
"grad_norm": 0.967344255783242,
|
|
"learning_rate": 7.173476627673492e-06,
|
|
"loss": 0.1037,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10425382107496262,
|
|
"step": 3280,
|
|
"valid_targets_mean": 4746.1,
|
|
"valid_targets_min": 3846
|
|
},
|
|
{
|
|
"epoch": 5.256,
|
|
"grad_norm": 0.9869044823926637,
|
|
"learning_rate": 7.112353464213477e-06,
|
|
"loss": 0.1061,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10941453278064728,
|
|
"step": 3285,
|
|
"valid_targets_mean": 4657.8,
|
|
"valid_targets_min": 2173
|
|
},
|
|
{
|
|
"epoch": 5.264,
|
|
"grad_norm": 0.9711559572722612,
|
|
"learning_rate": 7.051435455686735e-06,
|
|
"loss": 0.1046,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10553494095802307,
|
|
"step": 3290,
|
|
"valid_targets_mean": 4458.4,
|
|
"valid_targets_min": 1661
|
|
},
|
|
{
|
|
"epoch": 5.272,
|
|
"grad_norm": 0.9378400705195068,
|
|
"learning_rate": 6.990723571830438e-06,
|
|
"loss": 0.0963,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08980098366737366,
|
|
"step": 3295,
|
|
"valid_targets_mean": 4519.5,
|
|
"valid_targets_min": 970
|
|
},
|
|
{
|
|
"epoch": 5.28,
|
|
"grad_norm": 0.9651910972519293,
|
|
"learning_rate": 6.93021877910052e-06,
|
|
"loss": 0.0945,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07966376841068268,
|
|
"step": 3300,
|
|
"valid_targets_mean": 4408.7,
|
|
"valid_targets_min": 1171
|
|
},
|
|
{
|
|
"epoch": 5.288,
|
|
"grad_norm": 1.06319518956601,
|
|
"learning_rate": 6.8699220406562985e-06,
|
|
"loss": 0.0935,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1122937723994255,
|
|
"step": 3305,
|
|
"valid_targets_mean": 4716.1,
|
|
"valid_targets_min": 3946
|
|
},
|
|
{
|
|
"epoch": 5.296,
|
|
"grad_norm": 1.0298982959295162,
|
|
"learning_rate": 6.809834316345117e-06,
|
|
"loss": 0.0963,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10752555727958679,
|
|
"step": 3310,
|
|
"valid_targets_mean": 4501.9,
|
|
"valid_targets_min": 3822
|
|
},
|
|
{
|
|
"epoch": 5.304,
|
|
"grad_norm": 0.9593398528464043,
|
|
"learning_rate": 6.749956562687083e-06,
|
|
"loss": 0.1075,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10177193582057953,
|
|
"step": 3315,
|
|
"valid_targets_mean": 4480.3,
|
|
"valid_targets_min": 1502
|
|
},
|
|
{
|
|
"epoch": 5.312,
|
|
"grad_norm": 0.9893980191623689,
|
|
"learning_rate": 6.690289732859841e-06,
|
|
"loss": 0.0946,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11240974068641663,
|
|
"step": 3320,
|
|
"valid_targets_mean": 4504.6,
|
|
"valid_targets_min": 2270
|
|
},
|
|
{
|
|
"epoch": 5.32,
|
|
"grad_norm": 1.0080299837060387,
|
|
"learning_rate": 6.630834776683403e-06,
|
|
"loss": 0.0977,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10046304762363434,
|
|
"step": 3325,
|
|
"valid_targets_mean": 4313.6,
|
|
"valid_targets_min": 2326
|
|
},
|
|
{
|
|
"epoch": 5.328,
|
|
"grad_norm": 1.0841189053772304,
|
|
"learning_rate": 6.571592640605e-06,
|
|
"loss": 0.1071,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1049947589635849,
|
|
"step": 3330,
|
|
"valid_targets_mean": 4530.8,
|
|
"valid_targets_min": 1991
|
|
},
|
|
{
|
|
"epoch": 5.336,
|
|
"grad_norm": 1.0025426594634692,
|
|
"learning_rate": 6.512564267684061e-06,
|
|
"loss": 0.094,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09264200925827026,
|
|
"step": 3335,
|
|
"valid_targets_mean": 4782.6,
|
|
"valid_targets_min": 3912
|
|
},
|
|
{
|
|
"epoch": 5.344,
|
|
"grad_norm": 0.9992626977859915,
|
|
"learning_rate": 6.453750597577167e-06,
|
|
"loss": 0.0944,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09321926534175873,
|
|
"step": 3340,
|
|
"valid_targets_mean": 4581.4,
|
|
"valid_targets_min": 2389
|
|
},
|
|
{
|
|
"epoch": 5.352,
|
|
"grad_norm": 0.9979953954634814,
|
|
"learning_rate": 6.395152566523106e-06,
|
|
"loss": 0.1058,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10344021022319794,
|
|
"step": 3345,
|
|
"valid_targets_mean": 4848.6,
|
|
"valid_targets_min": 2608
|
|
},
|
|
{
|
|
"epoch": 5.36,
|
|
"grad_norm": 0.9272900098479718,
|
|
"learning_rate": 6.336771107327966e-06,
|
|
"loss": 0.097,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09526834636926651,
|
|
"step": 3350,
|
|
"valid_targets_mean": 4900.1,
|
|
"valid_targets_min": 3552
|
|
},
|
|
{
|
|
"epoch": 5.368,
|
|
"grad_norm": 1.0126519388132873,
|
|
"learning_rate": 6.278607149350289e-06,
|
|
"loss": 0.0986,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09526057541370392,
|
|
"step": 3355,
|
|
"valid_targets_mean": 4522.6,
|
|
"valid_targets_min": 3774
|
|
},
|
|
{
|
|
"epoch": 5.376,
|
|
"grad_norm": 1.1517106757423723,
|
|
"learning_rate": 6.220661618486268e-06,
|
|
"loss": 0.1043,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10477059334516525,
|
|
"step": 3360,
|
|
"valid_targets_mean": 4231.8,
|
|
"valid_targets_min": 1839
|
|
},
|
|
{
|
|
"epoch": 5.384,
|
|
"grad_norm": 1.1363453019296743,
|
|
"learning_rate": 6.162935437155024e-06,
|
|
"loss": 0.1019,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12003234773874283,
|
|
"step": 3365,
|
|
"valid_targets_mean": 4648.0,
|
|
"valid_targets_min": 3670
|
|
},
|
|
{
|
|
"epoch": 5.392,
|
|
"grad_norm": 0.9246807761311088,
|
|
"learning_rate": 6.105429524283901e-06,
|
|
"loss": 0.0937,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07637076824903488,
|
|
"step": 3370,
|
|
"valid_targets_mean": 4402.3,
|
|
"valid_targets_min": 1386
|
|
},
|
|
{
|
|
"epoch": 5.4,
|
|
"grad_norm": 1.0592886976295979,
|
|
"learning_rate": 6.04814479529386e-06,
|
|
"loss": 0.0927,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08888541162014008,
|
|
"step": 3375,
|
|
"valid_targets_mean": 4029.1,
|
|
"valid_targets_min": 1064
|
|
},
|
|
{
|
|
"epoch": 5.408,
|
|
"grad_norm": 1.009489901211204,
|
|
"learning_rate": 5.991082162084889e-06,
|
|
"loss": 0.0986,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09777955710887909,
|
|
"step": 3380,
|
|
"valid_targets_mean": 4409.1,
|
|
"valid_targets_min": 912
|
|
},
|
|
{
|
|
"epoch": 5.416,
|
|
"grad_norm": 0.9778310749751995,
|
|
"learning_rate": 5.934242533021499e-06,
|
|
"loss": 0.0946,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0967833623290062,
|
|
"step": 3385,
|
|
"valid_targets_mean": 4523.7,
|
|
"valid_targets_min": 3069
|
|
},
|
|
{
|
|
"epoch": 5.424,
|
|
"grad_norm": 1.1393219562727224,
|
|
"learning_rate": 5.877626812918258e-06,
|
|
"loss": 0.1066,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11012902855873108,
|
|
"step": 3390,
|
|
"valid_targets_mean": 4385.0,
|
|
"valid_targets_min": 2222
|
|
},
|
|
{
|
|
"epoch": 5.432,
|
|
"grad_norm": 0.9479435326982311,
|
|
"learning_rate": 5.821235903025378e-06,
|
|
"loss": 0.1046,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10226097702980042,
|
|
"step": 3395,
|
|
"valid_targets_mean": 4599.2,
|
|
"valid_targets_min": 3887
|
|
},
|
|
{
|
|
"epoch": 5.44,
|
|
"grad_norm": 0.8818753266270228,
|
|
"learning_rate": 5.765070701014391e-06,
|
|
"loss": 0.0805,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0755777582526207,
|
|
"step": 3400,
|
|
"valid_targets_mean": 4336.2,
|
|
"valid_targets_min": 2640
|
|
},
|
|
{
|
|
"epoch": 5.448,
|
|
"grad_norm": 1.115991667957166,
|
|
"learning_rate": 5.709132100963841e-06,
|
|
"loss": 0.0964,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10201806575059891,
|
|
"step": 3405,
|
|
"valid_targets_mean": 4365.1,
|
|
"valid_targets_min": 2926
|
|
},
|
|
{
|
|
"epoch": 5.456,
|
|
"grad_norm": 0.9268980232426772,
|
|
"learning_rate": 5.653420993345062e-06,
|
|
"loss": 0.094,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09449397027492523,
|
|
"step": 3410,
|
|
"valid_targets_mean": 4606.7,
|
|
"valid_targets_min": 3265
|
|
},
|
|
{
|
|
"epoch": 5.464,
|
|
"grad_norm": 1.1130510292580091,
|
|
"learning_rate": 5.597938265007994e-06,
|
|
"loss": 0.0994,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12211880832910538,
|
|
"step": 3415,
|
|
"valid_targets_mean": 4351.2,
|
|
"valid_targets_min": 2480
|
|
},
|
|
{
|
|
"epoch": 5.4719999999999995,
|
|
"grad_norm": 1.025459368559189,
|
|
"learning_rate": 5.542684799167069e-06,
|
|
"loss": 0.0978,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1003492921590805,
|
|
"step": 3420,
|
|
"valid_targets_mean": 4869.4,
|
|
"valid_targets_min": 4176
|
|
},
|
|
{
|
|
"epoch": 5.48,
|
|
"grad_norm": 0.9908968231609966,
|
|
"learning_rate": 5.487661475387152e-06,
|
|
"loss": 0.0907,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09571412950754166,
|
|
"step": 3425,
|
|
"valid_targets_mean": 4802.2,
|
|
"valid_targets_min": 3818
|
|
},
|
|
{
|
|
"epoch": 5.4879999999999995,
|
|
"grad_norm": 0.906981506515572,
|
|
"learning_rate": 5.432869169569541e-06,
|
|
"loss": 0.0982,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0993971973657608,
|
|
"step": 3430,
|
|
"valid_targets_mean": 4894.7,
|
|
"valid_targets_min": 4086
|
|
},
|
|
{
|
|
"epoch": 5.496,
|
|
"grad_norm": 1.0542740271535298,
|
|
"learning_rate": 5.378308753938024e-06,
|
|
"loss": 0.0956,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08996215462684631,
|
|
"step": 3435,
|
|
"valid_targets_mean": 4703.3,
|
|
"valid_targets_min": 3421
|
|
},
|
|
{
|
|
"epoch": 5.504,
|
|
"grad_norm": 0.9352581455307021,
|
|
"learning_rate": 5.323981097024986e-06,
|
|
"loss": 0.0955,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08220089972019196,
|
|
"step": 3440,
|
|
"valid_targets_mean": 4525.1,
|
|
"valid_targets_min": 1997
|
|
},
|
|
{
|
|
"epoch": 5.5120000000000005,
|
|
"grad_norm": 0.9718146573788276,
|
|
"learning_rate": 5.269887063657595e-06,
|
|
"loss": 0.0937,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09245629608631134,
|
|
"step": 3445,
|
|
"valid_targets_mean": 4281.8,
|
|
"valid_targets_min": 709
|
|
},
|
|
{
|
|
"epoch": 5.52,
|
|
"grad_norm": 0.9577985842250504,
|
|
"learning_rate": 5.216027514944027e-06,
|
|
"loss": 0.1012,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10035169124603271,
|
|
"step": 3450,
|
|
"valid_targets_mean": 4393.9,
|
|
"valid_targets_min": 3641
|
|
},
|
|
{
|
|
"epoch": 5.5280000000000005,
|
|
"grad_norm": 1.2124662085869782,
|
|
"learning_rate": 5.162403308259767e-06,
|
|
"loss": 0.0921,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10835231840610504,
|
|
"step": 3455,
|
|
"valid_targets_mean": 4203.6,
|
|
"valid_targets_min": 2632
|
|
},
|
|
{
|
|
"epoch": 5.536,
|
|
"grad_norm": 1.0502210119402213,
|
|
"learning_rate": 5.109015297233935e-06,
|
|
"loss": 0.0986,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.121584951877594,
|
|
"step": 3460,
|
|
"valid_targets_mean": 4442.4,
|
|
"valid_targets_min": 2326
|
|
},
|
|
{
|
|
"epoch": 5.5440000000000005,
|
|
"grad_norm": 0.7922061404155397,
|
|
"learning_rate": 5.055864331735736e-06,
|
|
"loss": 0.0921,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07856306433677673,
|
|
"step": 3465,
|
|
"valid_targets_mean": 4519.6,
|
|
"valid_targets_min": 2372
|
|
},
|
|
{
|
|
"epoch": 5.552,
|
|
"grad_norm": 1.0350376414344746,
|
|
"learning_rate": 5.002951257860909e-06,
|
|
"loss": 0.0942,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09397441148757935,
|
|
"step": 3470,
|
|
"valid_targets_mean": 4468.1,
|
|
"valid_targets_min": 2411
|
|
},
|
|
{
|
|
"epoch": 5.5600000000000005,
|
|
"grad_norm": 0.9494039639556755,
|
|
"learning_rate": 4.950276917918256e-06,
|
|
"loss": 0.0922,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09576588869094849,
|
|
"step": 3475,
|
|
"valid_targets_mean": 4490.8,
|
|
"valid_targets_min": 3095
|
|
},
|
|
{
|
|
"epoch": 5.568,
|
|
"grad_norm": 0.9318794190214731,
|
|
"learning_rate": 4.8978421504162385e-06,
|
|
"loss": 0.0937,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08758287876844406,
|
|
"step": 3480,
|
|
"valid_targets_mean": 4197.8,
|
|
"valid_targets_min": 1839
|
|
},
|
|
{
|
|
"epoch": 5.576,
|
|
"grad_norm": 0.9519758546210891,
|
|
"learning_rate": 4.845647790049634e-06,
|
|
"loss": 0.0982,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1009417250752449,
|
|
"step": 3485,
|
|
"valid_targets_mean": 4574.1,
|
|
"valid_targets_min": 3890
|
|
},
|
|
{
|
|
"epoch": 5.584,
|
|
"grad_norm": 1.0566022610527137,
|
|
"learning_rate": 4.793694667686244e-06,
|
|
"loss": 0.0924,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09925191104412079,
|
|
"step": 3490,
|
|
"valid_targets_mean": 4664.8,
|
|
"valid_targets_min": 796
|
|
},
|
|
{
|
|
"epoch": 5.592,
|
|
"grad_norm": 1.1698827287551596,
|
|
"learning_rate": 4.741983610353664e-06,
|
|
"loss": 0.0886,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09215767681598663,
|
|
"step": 3495,
|
|
"valid_targets_mean": 4623.9,
|
|
"valid_targets_min": 3186
|
|
},
|
|
{
|
|
"epoch": 5.6,
|
|
"grad_norm": 0.9698458596208371,
|
|
"learning_rate": 4.690515441226122e-06,
|
|
"loss": 0.0889,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0767895057797432,
|
|
"step": 3500,
|
|
"valid_targets_mean": 4576.2,
|
|
"valid_targets_min": 2411
|
|
},
|
|
{
|
|
"epoch": 5.608,
|
|
"grad_norm": 1.0951356983614102,
|
|
"learning_rate": 4.639290979611379e-06,
|
|
"loss": 0.0931,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09801951050758362,
|
|
"step": 3505,
|
|
"valid_targets_mean": 4519.5,
|
|
"valid_targets_min": 2541
|
|
},
|
|
{
|
|
"epoch": 5.616,
|
|
"grad_norm": 1.091954389328433,
|
|
"learning_rate": 4.588311040937683e-06,
|
|
"loss": 0.0944,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10716378688812256,
|
|
"step": 3510,
|
|
"valid_targets_mean": 4108.4,
|
|
"valid_targets_min": 2594
|
|
},
|
|
{
|
|
"epoch": 5.624,
|
|
"grad_norm": 0.9216084457200024,
|
|
"learning_rate": 4.537576436740783e-06,
|
|
"loss": 0.0918,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09752807766199112,
|
|
"step": 3515,
|
|
"valid_targets_mean": 4355.3,
|
|
"valid_targets_min": 2908
|
|
},
|
|
{
|
|
"epoch": 5.632,
|
|
"grad_norm": 0.9365013293550658,
|
|
"learning_rate": 4.487087974651016e-06,
|
|
"loss": 0.0927,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08773821592330933,
|
|
"step": 3520,
|
|
"valid_targets_mean": 4398.9,
|
|
"valid_targets_min": 1354
|
|
},
|
|
{
|
|
"epoch": 5.64,
|
|
"grad_norm": 1.0226555967780488,
|
|
"learning_rate": 4.436846458380455e-06,
|
|
"loss": 0.1,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10363728553056717,
|
|
"step": 3525,
|
|
"valid_targets_mean": 4653.2,
|
|
"valid_targets_min": 2550
|
|
},
|
|
{
|
|
"epoch": 5.648,
|
|
"grad_norm": 0.9812638515592906,
|
|
"learning_rate": 4.386852687710104e-06,
|
|
"loss": 0.091,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10121522098779678,
|
|
"step": 3530,
|
|
"valid_targets_mean": 4693.8,
|
|
"valid_targets_min": 3982
|
|
},
|
|
{
|
|
"epoch": 5.656,
|
|
"grad_norm": 0.8930549568298898,
|
|
"learning_rate": 4.337107458477177e-06,
|
|
"loss": 0.106,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09476905316114426,
|
|
"step": 3535,
|
|
"valid_targets_mean": 4608.4,
|
|
"valid_targets_min": 1533
|
|
},
|
|
{
|
|
"epoch": 5.664,
|
|
"grad_norm": 0.9386757395468496,
|
|
"learning_rate": 4.287611562562422e-06,
|
|
"loss": 0.0918,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10383667051792145,
|
|
"step": 3540,
|
|
"valid_targets_mean": 4248.5,
|
|
"valid_targets_min": 1266
|
|
},
|
|
{
|
|
"epoch": 5.672,
|
|
"grad_norm": 0.9814950154994561,
|
|
"learning_rate": 4.238365787877516e-06,
|
|
"loss": 0.0923,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09031368792057037,
|
|
"step": 3545,
|
|
"valid_targets_mean": 4792.9,
|
|
"valid_targets_min": 3958
|
|
},
|
|
{
|
|
"epoch": 5.68,
|
|
"grad_norm": 0.7980989236671729,
|
|
"learning_rate": 4.189370918352531e-06,
|
|
"loss": 0.0827,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07505756616592407,
|
|
"step": 3550,
|
|
"valid_targets_mean": 4819.3,
|
|
"valid_targets_min": 4106
|
|
},
|
|
{
|
|
"epoch": 5.688,
|
|
"grad_norm": 0.9047323663097401,
|
|
"learning_rate": 4.140627733923439e-06,
|
|
"loss": 0.0917,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07517348229885101,
|
|
"step": 3555,
|
|
"valid_targets_mean": 4539.6,
|
|
"valid_targets_min": 3832
|
|
},
|
|
{
|
|
"epoch": 5.696,
|
|
"grad_norm": 0.9565573885511617,
|
|
"learning_rate": 4.092137010519712e-06,
|
|
"loss": 0.1008,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10186438262462616,
|
|
"step": 3560,
|
|
"valid_targets_mean": 4527.6,
|
|
"valid_targets_min": 2854
|
|
},
|
|
{
|
|
"epoch": 5.704,
|
|
"grad_norm": 1.1048993345884468,
|
|
"learning_rate": 4.043899520051964e-06,
|
|
"loss": 0.101,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12873142957687378,
|
|
"step": 3565,
|
|
"valid_targets_mean": 4839.4,
|
|
"valid_targets_min": 3773
|
|
},
|
|
{
|
|
"epoch": 5.712,
|
|
"grad_norm": 1.1147062718920944,
|
|
"learning_rate": 3.995916030399658e-06,
|
|
"loss": 0.1014,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10289431363344193,
|
|
"step": 3570,
|
|
"valid_targets_mean": 4418.1,
|
|
"valid_targets_min": 2003
|
|
},
|
|
{
|
|
"epoch": 5.72,
|
|
"grad_norm": 1.0692444366260967,
|
|
"learning_rate": 3.948187305398892e-06,
|
|
"loss": 0.0997,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.101779505610466,
|
|
"step": 3575,
|
|
"valid_targets_mean": 4921.0,
|
|
"valid_targets_min": 4126
|
|
},
|
|
{
|
|
"epoch": 5.728,
|
|
"grad_norm": 1.031295734601785,
|
|
"learning_rate": 3.90071410483023e-06,
|
|
"loss": 0.0896,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08697981387376785,
|
|
"step": 3580,
|
|
"valid_targets_mean": 4229.8,
|
|
"valid_targets_min": 2619
|
|
},
|
|
{
|
|
"epoch": 5.736,
|
|
"grad_norm": 1.1631339310406366,
|
|
"learning_rate": 3.853497184406623e-06,
|
|
"loss": 0.0924,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09651214629411697,
|
|
"step": 3585,
|
|
"valid_targets_mean": 3801.9,
|
|
"valid_targets_min": 889
|
|
},
|
|
{
|
|
"epoch": 5.744,
|
|
"grad_norm": 0.9940222501659088,
|
|
"learning_rate": 3.80653729576135e-06,
|
|
"loss": 0.0881,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08622810989618301,
|
|
"step": 3590,
|
|
"valid_targets_mean": 4448.9,
|
|
"valid_targets_min": 3450
|
|
},
|
|
{
|
|
"epoch": 5.752,
|
|
"grad_norm": 0.9665497859677963,
|
|
"learning_rate": 3.7598351864360872e-06,
|
|
"loss": 0.0984,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09384353458881378,
|
|
"step": 3595,
|
|
"valid_targets_mean": 4188.3,
|
|
"valid_targets_min": 3059
|
|
},
|
|
{
|
|
"epoch": 5.76,
|
|
"grad_norm": 1.0868981049413289,
|
|
"learning_rate": 3.713391599868985e-06,
|
|
"loss": 0.0911,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09497397392988205,
|
|
"step": 3600,
|
|
"valid_targets_mean": 4594.2,
|
|
"valid_targets_min": 2003
|
|
},
|
|
{
|
|
"epoch": 5.768,
|
|
"grad_norm": 1.0521287467884775,
|
|
"learning_rate": 3.6672072753828424e-06,
|
|
"loss": 0.0905,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09298671782016754,
|
|
"step": 3605,
|
|
"valid_targets_mean": 4389.2,
|
|
"valid_targets_min": 3525
|
|
},
|
|
{
|
|
"epoch": 5.776,
|
|
"grad_norm": 0.8877810745990372,
|
|
"learning_rate": 3.6212829481733368e-06,
|
|
"loss": 0.099,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08877648413181305,
|
|
"step": 3610,
|
|
"valid_targets_mean": 4835.7,
|
|
"valid_targets_min": 4271
|
|
},
|
|
{
|
|
"epoch": 5.784,
|
|
"grad_norm": 0.9850684980810048,
|
|
"learning_rate": 3.575619349297317e-06,
|
|
"loss": 0.0832,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08490557223558426,
|
|
"step": 3615,
|
|
"valid_targets_mean": 4727.8,
|
|
"valid_targets_min": 3694
|
|
},
|
|
{
|
|
"epoch": 5.792,
|
|
"grad_norm": 1.0965879572150923,
|
|
"learning_rate": 3.5302172056611682e-06,
|
|
"loss": 0.0983,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10039995610713959,
|
|
"step": 3620,
|
|
"valid_targets_mean": 4422.6,
|
|
"valid_targets_min": 1800
|
|
},
|
|
{
|
|
"epoch": 5.8,
|
|
"grad_norm": 0.8845773586602016,
|
|
"learning_rate": 3.485077240009247e-06,
|
|
"loss": 0.0977,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07733885943889618,
|
|
"step": 3625,
|
|
"valid_targets_mean": 4354.7,
|
|
"valid_targets_min": 3488
|
|
},
|
|
{
|
|
"epoch": 5.808,
|
|
"grad_norm": 1.0425257659496143,
|
|
"learning_rate": 3.4402001709123643e-06,
|
|
"loss": 0.093,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09721488505601883,
|
|
"step": 3630,
|
|
"valid_targets_mean": 4644.7,
|
|
"valid_targets_min": 3599
|
|
},
|
|
{
|
|
"epoch": 5.816,
|
|
"grad_norm": 1.0140631579380122,
|
|
"learning_rate": 3.3955867127563515e-06,
|
|
"loss": 0.0974,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09973962604999542,
|
|
"step": 3635,
|
|
"valid_targets_mean": 4518.1,
|
|
"valid_targets_min": 3449
|
|
},
|
|
{
|
|
"epoch": 5.824,
|
|
"grad_norm": 1.0993766606042434,
|
|
"learning_rate": 3.351237575730695e-06,
|
|
"loss": 0.0923,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08937374502420425,
|
|
"step": 3640,
|
|
"valid_targets_mean": 4234.1,
|
|
"valid_targets_min": 1168
|
|
},
|
|
{
|
|
"epoch": 5.832,
|
|
"grad_norm": 1.193730065434566,
|
|
"learning_rate": 3.307153465817219e-06,
|
|
"loss": 0.0992,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09890756011009216,
|
|
"step": 3645,
|
|
"valid_targets_mean": 4628.4,
|
|
"valid_targets_min": 2807
|
|
},
|
|
{
|
|
"epoch": 5.84,
|
|
"grad_norm": 0.9886911200702848,
|
|
"learning_rate": 3.263335084778856e-06,
|
|
"loss": 0.0914,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10946749895811081,
|
|
"step": 3650,
|
|
"valid_targets_mean": 4763.6,
|
|
"valid_targets_min": 3547
|
|
},
|
|
{
|
|
"epoch": 5.848,
|
|
"grad_norm": 0.9122211171419388,
|
|
"learning_rate": 3.2197831301484816e-06,
|
|
"loss": 0.0888,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09050600230693817,
|
|
"step": 3655,
|
|
"valid_targets_mean": 4873.1,
|
|
"valid_targets_min": 1623
|
|
},
|
|
{
|
|
"epoch": 5.856,
|
|
"grad_norm": 1.0105083382222246,
|
|
"learning_rate": 3.1764982952177805e-06,
|
|
"loss": 0.094,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07956604659557343,
|
|
"step": 3660,
|
|
"valid_targets_mean": 4539.8,
|
|
"valid_targets_min": 3383
|
|
},
|
|
{
|
|
"epoch": 5.864,
|
|
"grad_norm": 1.0895687048650797,
|
|
"learning_rate": 3.1334812690262507e-06,
|
|
"loss": 0.1,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09592334181070328,
|
|
"step": 3665,
|
|
"valid_targets_mean": 4105.1,
|
|
"valid_targets_min": 1623
|
|
},
|
|
{
|
|
"epoch": 5.872,
|
|
"grad_norm": 0.934850128382635,
|
|
"learning_rate": 3.0907327363502084e-06,
|
|
"loss": 0.0947,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07556062191724777,
|
|
"step": 3670,
|
|
"valid_targets_mean": 4620.1,
|
|
"valid_targets_min": 3046
|
|
},
|
|
{
|
|
"epoch": 5.88,
|
|
"grad_norm": 0.9690343293101364,
|
|
"learning_rate": 3.0482533776918987e-06,
|
|
"loss": 0.0916,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08898059278726578,
|
|
"step": 3675,
|
|
"valid_targets_mean": 4447.6,
|
|
"valid_targets_min": 1367
|
|
},
|
|
{
|
|
"epoch": 5.888,
|
|
"grad_norm": 1.6473360206578607,
|
|
"learning_rate": 3.0060438692686533e-06,
|
|
"loss": 0.0958,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09584605693817139,
|
|
"step": 3680,
|
|
"valid_targets_mean": 4835.7,
|
|
"valid_targets_min": 3928
|
|
},
|
|
{
|
|
"epoch": 5.896,
|
|
"grad_norm": 1.072737015924698,
|
|
"learning_rate": 2.964104883002139e-06,
|
|
"loss": 0.0944,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10440277308225632,
|
|
"step": 3685,
|
|
"valid_targets_mean": 4387.6,
|
|
"valid_targets_min": 2264
|
|
},
|
|
{
|
|
"epoch": 5.904,
|
|
"grad_norm": 0.951904533527667,
|
|
"learning_rate": 2.9224370865076457e-06,
|
|
"loss": 0.0931,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08394080400466919,
|
|
"step": 3690,
|
|
"valid_targets_mean": 4404.2,
|
|
"valid_targets_min": 1925
|
|
},
|
|
{
|
|
"epoch": 5.912,
|
|
"grad_norm": 1.0175178696785432,
|
|
"learning_rate": 2.8810411430834716e-06,
|
|
"loss": 0.0926,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08819282054901123,
|
|
"step": 3695,
|
|
"valid_targets_mean": 4633.1,
|
|
"valid_targets_min": 3665
|
|
},
|
|
{
|
|
"epoch": 5.92,
|
|
"grad_norm": 0.8968020587831622,
|
|
"learning_rate": 2.8399177117003595e-06,
|
|
"loss": 0.0948,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09017691016197205,
|
|
"step": 3700,
|
|
"valid_targets_mean": 4717.8,
|
|
"valid_targets_min": 3411
|
|
},
|
|
{
|
|
"epoch": 5.928,
|
|
"grad_norm": 0.9036576387424691,
|
|
"learning_rate": 2.7990674469910085e-06,
|
|
"loss": 0.0945,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08503346145153046,
|
|
"step": 3705,
|
|
"valid_targets_mean": 4439.2,
|
|
"valid_targets_min": 2597
|
|
},
|
|
{
|
|
"epoch": 5.936,
|
|
"grad_norm": 0.95286029704266,
|
|
"learning_rate": 2.7584909992396515e-06,
|
|
"loss": 0.0867,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07618775963783264,
|
|
"step": 3710,
|
|
"valid_targets_mean": 4343.2,
|
|
"valid_targets_min": 2708
|
|
},
|
|
{
|
|
"epoch": 5.944,
|
|
"grad_norm": 1.0366480558294295,
|
|
"learning_rate": 2.7181890143716995e-06,
|
|
"loss": 0.0963,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.106142558157444,
|
|
"step": 3715,
|
|
"valid_targets_mean": 4751.8,
|
|
"valid_targets_min": 2604
|
|
},
|
|
{
|
|
"epoch": 5.952,
|
|
"grad_norm": 1.158935742630901,
|
|
"learning_rate": 2.6781621339434717e-06,
|
|
"loss": 0.1002,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11273472011089325,
|
|
"step": 3720,
|
|
"valid_targets_mean": 4728.0,
|
|
"valid_targets_min": 3960
|
|
},
|
|
{
|
|
"epoch": 5.96,
|
|
"grad_norm": 0.9936556826471483,
|
|
"learning_rate": 2.638410995131966e-06,
|
|
"loss": 0.0858,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08023962378501892,
|
|
"step": 3725,
|
|
"valid_targets_mean": 4607.8,
|
|
"valid_targets_min": 2688
|
|
},
|
|
{
|
|
"epoch": 5.968,
|
|
"grad_norm": 0.8914909898206873,
|
|
"learning_rate": 2.5989362307247313e-06,
|
|
"loss": 0.089,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08502813428640366,
|
|
"step": 3730,
|
|
"valid_targets_mean": 4804.0,
|
|
"valid_targets_min": 4042
|
|
},
|
|
{
|
|
"epoch": 5.976,
|
|
"grad_norm": 0.9702519688302588,
|
|
"learning_rate": 2.5597384691097847e-06,
|
|
"loss": 0.0881,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08771882206201553,
|
|
"step": 3735,
|
|
"valid_targets_mean": 4662.7,
|
|
"valid_targets_min": 1507
|
|
},
|
|
{
|
|
"epoch": 5.984,
|
|
"grad_norm": 0.9988559244748249,
|
|
"learning_rate": 2.520818334265611e-06,
|
|
"loss": 0.0964,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08962014317512512,
|
|
"step": 3740,
|
|
"valid_targets_mean": 4364.3,
|
|
"valid_targets_min": 3391
|
|
},
|
|
{
|
|
"epoch": 5.992,
|
|
"grad_norm": 1.0150253256495827,
|
|
"learning_rate": 2.482176445751232e-06,
|
|
"loss": 0.0982,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1042180210351944,
|
|
"step": 3745,
|
|
"valid_targets_mean": 4628.1,
|
|
"valid_targets_min": 3379
|
|
},
|
|
{
|
|
"epoch": 6.0,
|
|
"grad_norm": 0.9466667834456809,
|
|
"learning_rate": 2.4438134186963415e-06,
|
|
"loss": 0.0983,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08395225554704666,
|
|
"step": 3750,
|
|
"valid_targets_mean": 4533.2,
|
|
"valid_targets_min": 2988
|
|
},
|
|
{
|
|
"epoch": 6.008,
|
|
"grad_norm": 0.977075557175262,
|
|
"learning_rate": 2.4057298637915105e-06,
|
|
"loss": 0.0876,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08931310474872589,
|
|
"step": 3755,
|
|
"valid_targets_mean": 4423.5,
|
|
"valid_targets_min": 3531
|
|
},
|
|
{
|
|
"epoch": 6.016,
|
|
"grad_norm": 1.0296256847941045,
|
|
"learning_rate": 2.3679263872784717e-06,
|
|
"loss": 0.0823,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08132696151733398,
|
|
"step": 3760,
|
|
"valid_targets_mean": 4504.8,
|
|
"valid_targets_min": 2506
|
|
},
|
|
{
|
|
"epoch": 6.024,
|
|
"grad_norm": 1.351986257844334,
|
|
"learning_rate": 2.330403590940471e-06,
|
|
"loss": 0.081,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10204030573368073,
|
|
"step": 3765,
|
|
"valid_targets_mean": 4626.7,
|
|
"valid_targets_min": 3004
|
|
},
|
|
{
|
|
"epoch": 6.032,
|
|
"grad_norm": 1.033847005975066,
|
|
"learning_rate": 2.2931620720926717e-06,
|
|
"loss": 0.0878,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08778373897075653,
|
|
"step": 3770,
|
|
"valid_targets_mean": 4495.7,
|
|
"valid_targets_min": 2422
|
|
},
|
|
{
|
|
"epoch": 6.04,
|
|
"grad_norm": 0.8944381017307728,
|
|
"learning_rate": 2.256202423572669e-06,
|
|
"loss": 0.0889,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09564901143312454,
|
|
"step": 3775,
|
|
"valid_targets_mean": 4696.7,
|
|
"valid_targets_min": 3430
|
|
},
|
|
{
|
|
"epoch": 6.048,
|
|
"grad_norm": 0.9153472829822796,
|
|
"learning_rate": 2.219525233731035e-06,
|
|
"loss": 0.0784,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07495303452014923,
|
|
"step": 3780,
|
|
"valid_targets_mean": 4601.4,
|
|
"valid_targets_min": 2255
|
|
},
|
|
{
|
|
"epoch": 6.056,
|
|
"grad_norm": 0.963750805776716,
|
|
"learning_rate": 2.183131086421961e-06,
|
|
"loss": 0.0899,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09702135622501373,
|
|
"step": 3785,
|
|
"valid_targets_mean": 4566.1,
|
|
"valid_targets_min": 2774
|
|
},
|
|
{
|
|
"epoch": 6.064,
|
|
"grad_norm": 0.8537793040971933,
|
|
"learning_rate": 2.1470205609939533e-06,
|
|
"loss": 0.0779,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07274477183818817,
|
|
"step": 3790,
|
|
"valid_targets_mean": 4466.4,
|
|
"valid_targets_min": 3814
|
|
},
|
|
{
|
|
"epoch": 6.072,
|
|
"grad_norm": 1.0727551287385353,
|
|
"learning_rate": 2.1111942322806335e-06,
|
|
"loss": 0.0902,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09366320818662643,
|
|
"step": 3795,
|
|
"valid_targets_mean": 4439.1,
|
|
"valid_targets_min": 3246
|
|
},
|
|
{
|
|
"epoch": 6.08,
|
|
"grad_norm": 1.1131175388704868,
|
|
"learning_rate": 2.0756526705915635e-06,
|
|
"loss": 0.0836,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09015049040317535,
|
|
"step": 3800,
|
|
"valid_targets_mean": 4142.1,
|
|
"valid_targets_min": 2248
|
|
},
|
|
{
|
|
"epoch": 6.088,
|
|
"grad_norm": 0.8517340286487266,
|
|
"learning_rate": 2.0403964417031764e-06,
|
|
"loss": 0.0886,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07115870714187622,
|
|
"step": 3805,
|
|
"valid_targets_mean": 4471.5,
|
|
"valid_targets_min": 2878
|
|
},
|
|
{
|
|
"epoch": 6.096,
|
|
"grad_norm": 1.000338997380535,
|
|
"learning_rate": 2.0054261068497773e-06,
|
|
"loss": 0.084,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08141026645898819,
|
|
"step": 3810,
|
|
"valid_targets_mean": 4390.2,
|
|
"valid_targets_min": 2926
|
|
},
|
|
{
|
|
"epoch": 6.104,
|
|
"grad_norm": 0.8858946092804942,
|
|
"learning_rate": 1.9707422227145922e-06,
|
|
"loss": 0.0778,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07216652482748032,
|
|
"step": 3815,
|
|
"valid_targets_mean": 4512.2,
|
|
"valid_targets_min": 2534
|
|
},
|
|
{
|
|
"epoch": 6.112,
|
|
"grad_norm": 0.9363531713816285,
|
|
"learning_rate": 1.936345341420924e-06,
|
|
"loss": 0.0803,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08437725901603699,
|
|
"step": 3820,
|
|
"valid_targets_mean": 4416.3,
|
|
"valid_targets_min": 1594
|
|
},
|
|
{
|
|
"epoch": 6.12,
|
|
"grad_norm": 0.9405946711453356,
|
|
"learning_rate": 1.9022360105233507e-06,
|
|
"loss": 0.0888,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08383543789386749,
|
|
"step": 3825,
|
|
"valid_targets_mean": 4343.9,
|
|
"valid_targets_min": 3169
|
|
},
|
|
{
|
|
"epoch": 6.128,
|
|
"grad_norm": 0.9655258447300491,
|
|
"learning_rate": 1.8684147729990188e-06,
|
|
"loss": 0.0844,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07891859859228134,
|
|
"step": 3830,
|
|
"valid_targets_mean": 4467.3,
|
|
"valid_targets_min": 2268
|
|
},
|
|
{
|
|
"epoch": 6.136,
|
|
"grad_norm": 0.9345291553188987,
|
|
"learning_rate": 1.8348821672389893e-06,
|
|
"loss": 0.0826,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06917843222618103,
|
|
"step": 3835,
|
|
"valid_targets_mean": 4605.4,
|
|
"valid_targets_min": 3642
|
|
},
|
|
{
|
|
"epoch": 6.144,
|
|
"grad_norm": 0.9792711793543556,
|
|
"learning_rate": 1.8016387270396784e-06,
|
|
"loss": 0.078,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08932572603225708,
|
|
"step": 3840,
|
|
"valid_targets_mean": 4478.0,
|
|
"valid_targets_min": 2608
|
|
},
|
|
{
|
|
"epoch": 6.152,
|
|
"grad_norm": 1.0733497208308675,
|
|
"learning_rate": 1.7686849815943486e-06,
|
|
"loss": 0.0844,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09008129686117172,
|
|
"step": 3845,
|
|
"valid_targets_mean": 4398.5,
|
|
"valid_targets_min": 2780
|
|
},
|
|
{
|
|
"epoch": 6.16,
|
|
"grad_norm": 0.8987695720385824,
|
|
"learning_rate": 1.7360214554847e-06,
|
|
"loss": 0.0851,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07093490660190582,
|
|
"step": 3850,
|
|
"valid_targets_mean": 4597.2,
|
|
"valid_targets_min": 2003
|
|
},
|
|
{
|
|
"epoch": 6.168,
|
|
"grad_norm": 0.9484898182449262,
|
|
"learning_rate": 1.703648668672495e-06,
|
|
"loss": 0.0856,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07835480570793152,
|
|
"step": 3855,
|
|
"valid_targets_mean": 4718.4,
|
|
"valid_targets_min": 3391
|
|
},
|
|
{
|
|
"epoch": 6.176,
|
|
"grad_norm": 0.8993444591406407,
|
|
"learning_rate": 1.6715671364913077e-06,
|
|
"loss": 0.0842,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07994028925895691,
|
|
"step": 3860,
|
|
"valid_targets_mean": 4624.1,
|
|
"valid_targets_min": 3754
|
|
},
|
|
{
|
|
"epoch": 6.184,
|
|
"grad_norm": 0.8444013155253536,
|
|
"learning_rate": 1.6397773696383091e-06,
|
|
"loss": 0.0814,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07129291445016861,
|
|
"step": 3865,
|
|
"valid_targets_mean": 4751.8,
|
|
"valid_targets_min": 3832
|
|
},
|
|
{
|
|
"epoch": 6.192,
|
|
"grad_norm": 0.975104098046281,
|
|
"learning_rate": 1.6082798741661321e-06,
|
|
"loss": 0.0787,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08080554753541946,
|
|
"step": 3870,
|
|
"valid_targets_mean": 4725.4,
|
|
"valid_targets_min": 2810
|
|
},
|
|
{
|
|
"epoch": 6.2,
|
|
"grad_norm": 0.9786745855466956,
|
|
"learning_rate": 1.5770751514748273e-06,
|
|
"loss": 0.086,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08651769161224365,
|
|
"step": 3875,
|
|
"valid_targets_mean": 4703.6,
|
|
"valid_targets_min": 1857
|
|
},
|
|
{
|
|
"epoch": 6.208,
|
|
"grad_norm": 1.03110527594668,
|
|
"learning_rate": 1.5461636983038686e-06,
|
|
"loss": 0.0932,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09436431527137756,
|
|
"step": 3880,
|
|
"valid_targets_mean": 4432.4,
|
|
"valid_targets_min": 2312
|
|
},
|
|
{
|
|
"epoch": 6.216,
|
|
"grad_norm": 1.1471586737466828,
|
|
"learning_rate": 1.5155460067242578e-06,
|
|
"loss": 0.0876,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10137753188610077,
|
|
"step": 3885,
|
|
"valid_targets_mean": 3824.6,
|
|
"valid_targets_min": 1062
|
|
},
|
|
{
|
|
"epoch": 6.224,
|
|
"grad_norm": 0.9305173224026609,
|
|
"learning_rate": 1.4852225641306816e-06,
|
|
"loss": 0.0785,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08005651831626892,
|
|
"step": 3890,
|
|
"valid_targets_mean": 4387.0,
|
|
"valid_targets_min": 441
|
|
},
|
|
{
|
|
"epoch": 6.232,
|
|
"grad_norm": 0.8849865048374633,
|
|
"learning_rate": 1.4551938532337607e-06,
|
|
"loss": 0.077,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0619070790708065,
|
|
"step": 3895,
|
|
"valid_targets_mean": 4246.2,
|
|
"valid_targets_min": 2270
|
|
},
|
|
{
|
|
"epoch": 6.24,
|
|
"grad_norm": 0.9353360153089825,
|
|
"learning_rate": 1.4254603520523614e-06,
|
|
"loss": 0.0906,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09257589280605316,
|
|
"step": 3900,
|
|
"valid_targets_mean": 4676.6,
|
|
"valid_targets_min": 2472
|
|
},
|
|
{
|
|
"epoch": 6.248,
|
|
"grad_norm": 0.9659763870524511,
|
|
"learning_rate": 1.3960225339059875e-06,
|
|
"loss": 0.0843,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07981941848993301,
|
|
"step": 3905,
|
|
"valid_targets_mean": 4673.6,
|
|
"valid_targets_min": 3832
|
|
},
|
|
{
|
|
"epoch": 6.256,
|
|
"grad_norm": 1.0391038814941895,
|
|
"learning_rate": 1.3668808674072409e-06,
|
|
"loss": 0.0814,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08510047197341919,
|
|
"step": 3910,
|
|
"valid_targets_mean": 4581.0,
|
|
"valid_targets_min": 3729
|
|
},
|
|
{
|
|
"epoch": 6.264,
|
|
"grad_norm": 0.9762869527625367,
|
|
"learning_rate": 1.338035816454375e-06,
|
|
"loss": 0.0888,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07818831503391266,
|
|
"step": 3915,
|
|
"valid_targets_mean": 4512.4,
|
|
"valid_targets_min": 3087
|
|
},
|
|
{
|
|
"epoch": 6.272,
|
|
"grad_norm": 0.9698480885051443,
|
|
"learning_rate": 1.3094878402238887e-06,
|
|
"loss": 0.0955,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08287420868873596,
|
|
"step": 3920,
|
|
"valid_targets_mean": 4607.7,
|
|
"valid_targets_min": 3821
|
|
},
|
|
{
|
|
"epoch": 6.28,
|
|
"grad_norm": 0.9379508122214085,
|
|
"learning_rate": 1.2812373931632371e-06,
|
|
"loss": 0.0753,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07643862068653107,
|
|
"step": 3925,
|
|
"valid_targets_mean": 4286.2,
|
|
"valid_targets_min": 970
|
|
},
|
|
{
|
|
"epoch": 6.288,
|
|
"grad_norm": 0.9078008291530973,
|
|
"learning_rate": 1.2532849249835932e-06,
|
|
"loss": 0.0759,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06264795362949371,
|
|
"step": 3930,
|
|
"valid_targets_mean": 4882.1,
|
|
"valid_targets_min": 3888
|
|
},
|
|
{
|
|
"epoch": 6.296,
|
|
"grad_norm": 0.9381258715790302,
|
|
"learning_rate": 1.2256308806526774e-06,
|
|
"loss": 0.088,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06960931420326233,
|
|
"step": 3935,
|
|
"valid_targets_mean": 4037.6,
|
|
"valid_targets_min": 1452
|
|
},
|
|
{
|
|
"epoch": 6.304,
|
|
"grad_norm": 1.0628233447255107,
|
|
"learning_rate": 1.1982757003876855e-06,
|
|
"loss": 0.0793,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07392563670873642,
|
|
"step": 3940,
|
|
"valid_targets_mean": 4618.8,
|
|
"valid_targets_min": 2911
|
|
},
|
|
{
|
|
"epoch": 6.312,
|
|
"grad_norm": 1.118066310657044,
|
|
"learning_rate": 1.1712198196482793e-06,
|
|
"loss": 0.0822,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08308184891939163,
|
|
"step": 3945,
|
|
"valid_targets_mean": 4409.7,
|
|
"valid_targets_min": 3199
|
|
},
|
|
{
|
|
"epoch": 6.32,
|
|
"grad_norm": 0.9973845508032737,
|
|
"learning_rate": 1.1444636691296518e-06,
|
|
"loss": 0.0792,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07433116436004639,
|
|
"step": 3950,
|
|
"valid_targets_mean": 4323.4,
|
|
"valid_targets_min": 2807
|
|
},
|
|
{
|
|
"epoch": 6.328,
|
|
"grad_norm": 1.0629600729848958,
|
|
"learning_rate": 1.11800767475567e-06,
|
|
"loss": 0.0959,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09218830615282059,
|
|
"step": 3955,
|
|
"valid_targets_mean": 4416.7,
|
|
"valid_targets_min": 1356
|
|
},
|
|
{
|
|
"epoch": 6.336,
|
|
"grad_norm": 1.0642917209070468,
|
|
"learning_rate": 1.0918522576721014e-06,
|
|
"loss": 0.0892,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08946512639522552,
|
|
"step": 3960,
|
|
"valid_targets_mean": 4714.4,
|
|
"valid_targets_min": 3899
|
|
},
|
|
{
|
|
"epoch": 6.344,
|
|
"grad_norm": 1.0820081615593247,
|
|
"learning_rate": 1.0659978342399003e-06,
|
|
"loss": 0.0898,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1027638167142868,
|
|
"step": 3965,
|
|
"valid_targets_mean": 4428.8,
|
|
"valid_targets_min": 2182
|
|
},
|
|
{
|
|
"epoch": 6.352,
|
|
"grad_norm": 0.991755680878453,
|
|
"learning_rate": 1.0404448160285897e-06,
|
|
"loss": 0.0841,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08348552882671356,
|
|
"step": 3970,
|
|
"valid_targets_mean": 4510.2,
|
|
"valid_targets_min": 2212
|
|
},
|
|
{
|
|
"epoch": 6.36,
|
|
"grad_norm": 1.019210319907905,
|
|
"learning_rate": 1.0151936098097015e-06,
|
|
"loss": 0.0814,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07751256972551346,
|
|
"step": 3975,
|
|
"valid_targets_mean": 4551.8,
|
|
"valid_targets_min": 3278
|
|
},
|
|
{
|
|
"epoch": 6.368,
|
|
"grad_norm": 0.9598293564277333,
|
|
"learning_rate": 9.902446175503089e-07,
|
|
"loss": 0.0842,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08740901947021484,
|
|
"step": 3980,
|
|
"valid_targets_mean": 4624.4,
|
|
"valid_targets_min": 3006
|
|
},
|
|
{
|
|
"epoch": 6.376,
|
|
"grad_norm": 0.9069392164625397,
|
|
"learning_rate": 9.655982364066197e-07,
|
|
"loss": 0.0825,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07876341044902802,
|
|
"step": 3985,
|
|
"valid_targets_mean": 4249.2,
|
|
"valid_targets_min": 1585
|
|
},
|
|
{
|
|
"epoch": 6.384,
|
|
"grad_norm": 0.8848049157081566,
|
|
"learning_rate": 9.412548587176595e-07,
|
|
"loss": 0.0767,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06901217997074127,
|
|
"step": 3990,
|
|
"valid_targets_mean": 4678.8,
|
|
"valid_targets_min": 1623
|
|
},
|
|
{
|
|
"epoch": 6.392,
|
|
"grad_norm": 0.966895202526864,
|
|
"learning_rate": 9.172148719990237e-07,
|
|
"loss": 0.085,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07606752216815948,
|
|
"step": 3995,
|
|
"valid_targets_mean": 4289.1,
|
|
"valid_targets_min": 1502
|
|
},
|
|
{
|
|
"epoch": 6.4,
|
|
"grad_norm": 0.961257164391413,
|
|
"learning_rate": 8.934786589367106e-07,
|
|
"loss": 0.0833,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09865090996026993,
|
|
"step": 4000,
|
|
"valid_targets_mean": 4630.4,
|
|
"valid_targets_min": 2952
|
|
},
|
|
{
|
|
"epoch": 6.408,
|
|
"grad_norm": 0.9389390406936587,
|
|
"learning_rate": 8.700465973810246e-07,
|
|
"loss": 0.0841,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09862151741981506,
|
|
"step": 4005,
|
|
"valid_targets_mean": 4926.8,
|
|
"valid_targets_min": 3753
|
|
},
|
|
{
|
|
"epoch": 6.416,
|
|
"grad_norm": 1.549301812343912,
|
|
"learning_rate": 8.469190603405719e-07,
|
|
"loss": 0.077,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08590662479400635,
|
|
"step": 4010,
|
|
"valid_targets_mean": 4529.0,
|
|
"valid_targets_min": 2204
|
|
},
|
|
{
|
|
"epoch": 6.424,
|
|
"grad_norm": 0.9969758770528823,
|
|
"learning_rate": 8.240964159763121e-07,
|
|
"loss": 0.0886,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0942438393831253,
|
|
"step": 4015,
|
|
"valid_targets_mean": 4631.9,
|
|
"valid_targets_min": 3145
|
|
},
|
|
{
|
|
"epoch": 6.432,
|
|
"grad_norm": 0.9796640541530544,
|
|
"learning_rate": 8.015790275957003e-07,
|
|
"loss": 0.0827,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0823969841003418,
|
|
"step": 4020,
|
|
"valid_targets_mean": 4548.7,
|
|
"valid_targets_min": 3243
|
|
},
|
|
{
|
|
"epoch": 6.44,
|
|
"grad_norm": 1.012588243156625,
|
|
"learning_rate": 7.793672536469077e-07,
|
|
"loss": 0.0878,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08953895419836044,
|
|
"step": 4025,
|
|
"valid_targets_mean": 4258.1,
|
|
"valid_targets_min": 1585
|
|
},
|
|
{
|
|
"epoch": 6.448,
|
|
"grad_norm": 1.0394005736936405,
|
|
"learning_rate": 7.574614477131081e-07,
|
|
"loss": 0.0897,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1062498539686203,
|
|
"step": 4030,
|
|
"valid_targets_mean": 4623.8,
|
|
"valid_targets_min": 1681
|
|
},
|
|
{
|
|
"epoch": 6.456,
|
|
"grad_norm": 1.0187273540570003,
|
|
"learning_rate": 7.358619585068583e-07,
|
|
"loss": 0.0774,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08131261169910431,
|
|
"step": 4035,
|
|
"valid_targets_mean": 4504.7,
|
|
"valid_targets_min": 2801
|
|
},
|
|
{
|
|
"epoch": 6.464,
|
|
"grad_norm": 0.9216342500867519,
|
|
"learning_rate": 7.145691298645419e-07,
|
|
"loss": 0.091,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09456481039524078,
|
|
"step": 4040,
|
|
"valid_targets_mean": 4470.8,
|
|
"valid_targets_min": 2326
|
|
},
|
|
{
|
|
"epoch": 6.4719999999999995,
|
|
"grad_norm": 1.0710778073561171,
|
|
"learning_rate": 6.935833007408965e-07,
|
|
"loss": 0.0829,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07773569226264954,
|
|
"step": 4045,
|
|
"valid_targets_mean": 4556.8,
|
|
"valid_targets_min": 2945
|
|
},
|
|
{
|
|
"epoch": 6.48,
|
|
"grad_norm": 0.9927672728081478,
|
|
"learning_rate": 6.729048052036136e-07,
|
|
"loss": 0.0861,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08042934536933899,
|
|
"step": 4050,
|
|
"valid_targets_mean": 4438.1,
|
|
"valid_targets_min": 3141
|
|
},
|
|
{
|
|
"epoch": 6.4879999999999995,
|
|
"grad_norm": 0.9358082214133697,
|
|
"learning_rate": 6.52533972428031e-07,
|
|
"loss": 0.0825,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0787649154663086,
|
|
"step": 4055,
|
|
"valid_targets_mean": 4517.9,
|
|
"valid_targets_min": 3540
|
|
},
|
|
{
|
|
"epoch": 6.496,
|
|
"grad_norm": 1.0538737409757248,
|
|
"learning_rate": 6.324711266918826e-07,
|
|
"loss": 0.0894,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08961720019578934,
|
|
"step": 4060,
|
|
"valid_targets_mean": 4675.2,
|
|
"valid_targets_min": 3198
|
|
},
|
|
{
|
|
"epoch": 6.504,
|
|
"grad_norm": 1.1015278618105548,
|
|
"learning_rate": 6.127165873701457e-07,
|
|
"loss": 0.0894,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08255113661289215,
|
|
"step": 4065,
|
|
"valid_targets_mean": 4119.5,
|
|
"valid_targets_min": 1994
|
|
},
|
|
{
|
|
"epoch": 6.5120000000000005,
|
|
"grad_norm": 0.8635242987956312,
|
|
"learning_rate": 5.932706689299461e-07,
|
|
"loss": 0.0884,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07663701474666595,
|
|
"step": 4070,
|
|
"valid_targets_mean": 4817.1,
|
|
"valid_targets_min": 4046
|
|
},
|
|
{
|
|
"epoch": 6.52,
|
|
"grad_norm": 0.9847594843362265,
|
|
"learning_rate": 5.741336809255615e-07,
|
|
"loss": 0.0822,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08127310872077942,
|
|
"step": 4075,
|
|
"valid_targets_mean": 4428.3,
|
|
"valid_targets_min": 2792
|
|
},
|
|
{
|
|
"epoch": 6.5280000000000005,
|
|
"grad_norm": 0.9687578996590749,
|
|
"learning_rate": 5.553059279934902e-07,
|
|
"loss": 0.092,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08839556574821472,
|
|
"step": 4080,
|
|
"valid_targets_mean": 4432.8,
|
|
"valid_targets_min": 2221
|
|
},
|
|
{
|
|
"epoch": 6.536,
|
|
"grad_norm": 0.964467416705786,
|
|
"learning_rate": 5.36787709847597e-07,
|
|
"loss": 0.0767,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0892396941781044,
|
|
"step": 4085,
|
|
"valid_targets_mean": 4533.4,
|
|
"valid_targets_min": 2308
|
|
},
|
|
{
|
|
"epoch": 6.5440000000000005,
|
|
"grad_norm": 0.9650603214121655,
|
|
"learning_rate": 5.185793212743529e-07,
|
|
"loss": 0.0907,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09095023572444916,
|
|
"step": 4090,
|
|
"valid_targets_mean": 4822.0,
|
|
"valid_targets_min": 3879
|
|
},
|
|
{
|
|
"epoch": 6.552,
|
|
"grad_norm": 1.0738862583576596,
|
|
"learning_rate": 5.006810521281335e-07,
|
|
"loss": 0.0829,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10348834097385406,
|
|
"step": 4095,
|
|
"valid_targets_mean": 4305.9,
|
|
"valid_targets_min": 2204
|
|
},
|
|
{
|
|
"epoch": 6.5600000000000005,
|
|
"grad_norm": 0.9419949031473305,
|
|
"learning_rate": 4.830931873266065e-07,
|
|
"loss": 0.0881,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07656043767929077,
|
|
"step": 4100,
|
|
"valid_targets_mean": 4767.8,
|
|
"valid_targets_min": 3665
|
|
},
|
|
{
|
|
"epoch": 6.568,
|
|
"grad_norm": 0.9233911299976182,
|
|
"learning_rate": 4.658160068462025e-07,
|
|
"loss": 0.0787,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07702470570802689,
|
|
"step": 4105,
|
|
"valid_targets_mean": 4609.8,
|
|
"valid_targets_min": 3455
|
|
},
|
|
{
|
|
"epoch": 6.576,
|
|
"grad_norm": 0.940752752988166,
|
|
"learning_rate": 4.488497857176466e-07,
|
|
"loss": 0.0856,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0811225026845932,
|
|
"step": 4110,
|
|
"valid_targets_mean": 4342.5,
|
|
"valid_targets_min": 2461
|
|
},
|
|
{
|
|
"epoch": 6.584,
|
|
"grad_norm": 1.0222590928872706,
|
|
"learning_rate": 4.321947940215898e-07,
|
|
"loss": 0.0872,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08996831625699997,
|
|
"step": 4115,
|
|
"valid_targets_mean": 4536.8,
|
|
"valid_targets_min": 3011
|
|
},
|
|
{
|
|
"epoch": 6.592,
|
|
"grad_norm": 0.9419183793933782,
|
|
"learning_rate": 4.1585129688430425e-07,
|
|
"loss": 0.0881,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08329015225172043,
|
|
"step": 4120,
|
|
"valid_targets_mean": 4763.7,
|
|
"valid_targets_min": 3987
|
|
},
|
|
{
|
|
"epoch": 6.6,
|
|
"grad_norm": 0.9631283651084054,
|
|
"learning_rate": 3.998195544734706e-07,
|
|
"loss": 0.0801,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07111963629722595,
|
|
"step": 4125,
|
|
"valid_targets_mean": 4086.8,
|
|
"valid_targets_min": 1838
|
|
},
|
|
{
|
|
"epoch": 6.608,
|
|
"grad_norm": 1.0521774570474127,
|
|
"learning_rate": 3.840998219940284e-07,
|
|
"loss": 0.0796,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08146822452545166,
|
|
"step": 4130,
|
|
"valid_targets_mean": 4448.0,
|
|
"valid_targets_min": 2503
|
|
},
|
|
{
|
|
"epoch": 6.616,
|
|
"grad_norm": 0.9337025007610875,
|
|
"learning_rate": 3.6869234968411214e-07,
|
|
"loss": 0.0753,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06824478507041931,
|
|
"step": 4135,
|
|
"valid_targets_mean": 4912.3,
|
|
"valid_targets_min": 3775
|
|
},
|
|
{
|
|
"epoch": 6.624,
|
|
"grad_norm": 0.9595618136398398,
|
|
"learning_rate": 3.5359738281107504e-07,
|
|
"loss": 0.079,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0691223070025444,
|
|
"step": 4140,
|
|
"valid_targets_mean": 4639.1,
|
|
"valid_targets_min": 4024
|
|
},
|
|
{
|
|
"epoch": 6.632,
|
|
"grad_norm": 0.9988148531272858,
|
|
"learning_rate": 3.38815161667585e-07,
|
|
"loss": 0.089,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09183768928050995,
|
|
"step": 4145,
|
|
"valid_targets_mean": 4618.3,
|
|
"valid_targets_min": 3967
|
|
},
|
|
{
|
|
"epoch": 6.64,
|
|
"grad_norm": 1.0088896562319003,
|
|
"learning_rate": 3.24345921567788e-07,
|
|
"loss": 0.0745,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09073325991630554,
|
|
"step": 4150,
|
|
"valid_targets_mean": 4442.3,
|
|
"valid_targets_min": 3486
|
|
},
|
|
{
|
|
"epoch": 6.648,
|
|
"grad_norm": 1.1343085401760367,
|
|
"learning_rate": 3.101898928435754e-07,
|
|
"loss": 0.0896,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07714816927909851,
|
|
"step": 4155,
|
|
"valid_targets_mean": 4588.8,
|
|
"valid_targets_min": 2594
|
|
},
|
|
{
|
|
"epoch": 6.656,
|
|
"grad_norm": 0.9173207431705394,
|
|
"learning_rate": 2.9634730084091343e-07,
|
|
"loss": 0.0796,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07771845161914825,
|
|
"step": 4160,
|
|
"valid_targets_mean": 4411.8,
|
|
"valid_targets_min": 1663
|
|
},
|
|
{
|
|
"epoch": 6.664,
|
|
"grad_norm": 0.867456011724824,
|
|
"learning_rate": 2.8281836591624865e-07,
|
|
"loss": 0.0811,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07119958102703094,
|
|
"step": 4165,
|
|
"valid_targets_mean": 4480.9,
|
|
"valid_targets_min": 2633
|
|
},
|
|
{
|
|
"epoch": 6.672,
|
|
"grad_norm": 0.9688932407718841,
|
|
"learning_rate": 2.6960330343301033e-07,
|
|
"loss": 0.0876,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08467958867549896,
|
|
"step": 4170,
|
|
"valid_targets_mean": 4379.8,
|
|
"valid_targets_min": 2222
|
|
},
|
|
{
|
|
"epoch": 6.68,
|
|
"grad_norm": 0.8355311126219794,
|
|
"learning_rate": 2.5670232375817784e-07,
|
|
"loss": 0.0885,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05917223542928696,
|
|
"step": 4175,
|
|
"valid_targets_mean": 4283.1,
|
|
"valid_targets_min": 926
|
|
},
|
|
{
|
|
"epoch": 6.688,
|
|
"grad_norm": 1.0111069749149293,
|
|
"learning_rate": 2.441156322589322e-07,
|
|
"loss": 0.083,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09328457713127136,
|
|
"step": 4180,
|
|
"valid_targets_mean": 4477.6,
|
|
"valid_targets_min": 1201
|
|
},
|
|
{
|
|
"epoch": 6.696,
|
|
"grad_norm": 0.8959702879885959,
|
|
"learning_rate": 2.318434292993832e-07,
|
|
"loss": 0.0853,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08786676079034805,
|
|
"step": 4185,
|
|
"valid_targets_mean": 4335.6,
|
|
"valid_targets_min": 1828
|
|
},
|
|
{
|
|
"epoch": 6.704,
|
|
"grad_norm": 1.0295769209002925,
|
|
"learning_rate": 2.1988591023738514e-07,
|
|
"loss": 0.0827,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08741824328899384,
|
|
"step": 4190,
|
|
"valid_targets_mean": 4267.8,
|
|
"valid_targets_min": 1839
|
|
},
|
|
{
|
|
"epoch": 6.712,
|
|
"grad_norm": 1.1211881256879794,
|
|
"learning_rate": 2.0824326542142835e-07,
|
|
"loss": 0.0816,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07274734973907471,
|
|
"step": 4195,
|
|
"valid_targets_mean": 4282.2,
|
|
"valid_targets_min": 2594
|
|
},
|
|
{
|
|
"epoch": 6.72,
|
|
"grad_norm": 0.9734647006436701,
|
|
"learning_rate": 1.9691568018759931e-07,
|
|
"loss": 0.0783,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07601093500852585,
|
|
"step": 4200,
|
|
"valid_targets_mean": 4475.5,
|
|
"valid_targets_min": 3292
|
|
},
|
|
{
|
|
"epoch": 6.728,
|
|
"grad_norm": 0.9668324674665926,
|
|
"learning_rate": 1.8590333485664525e-07,
|
|
"loss": 0.0854,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08436590433120728,
|
|
"step": 4205,
|
|
"valid_targets_mean": 4644.7,
|
|
"valid_targets_min": 3964
|
|
},
|
|
{
|
|
"epoch": 6.736,
|
|
"grad_norm": 0.8998877767833496,
|
|
"learning_rate": 1.752064047310853e-07,
|
|
"loss": 0.0856,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0853387713432312,
|
|
"step": 4210,
|
|
"valid_targets_mean": 4551.5,
|
|
"valid_targets_min": 3827
|
|
},
|
|
{
|
|
"epoch": 6.744,
|
|
"grad_norm": 0.8630745005971612,
|
|
"learning_rate": 1.6482506009243949e-07,
|
|
"loss": 0.0775,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07094122469425201,
|
|
"step": 4215,
|
|
"valid_targets_mean": 4606.8,
|
|
"valid_targets_min": 3552
|
|
},
|
|
{
|
|
"epoch": 6.752,
|
|
"grad_norm": 1.0219573400106008,
|
|
"learning_rate": 1.5475946619850192e-07,
|
|
"loss": 0.0775,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09020616859197617,
|
|
"step": 4220,
|
|
"valid_targets_mean": 4183.8,
|
|
"valid_targets_min": 1803
|
|
},
|
|
{
|
|
"epoch": 6.76,
|
|
"grad_norm": 0.9685921361230694,
|
|
"learning_rate": 1.4500978328071845e-07,
|
|
"loss": 0.081,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09113042056560516,
|
|
"step": 4225,
|
|
"valid_targets_mean": 4713.9,
|
|
"valid_targets_min": 2858
|
|
},
|
|
{
|
|
"epoch": 6.768,
|
|
"grad_norm": 0.9858396339080018,
|
|
"learning_rate": 1.3557616654163775e-07,
|
|
"loss": 0.0829,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08655436336994171,
|
|
"step": 4230,
|
|
"valid_targets_mean": 4460.2,
|
|
"valid_targets_min": 3599
|
|
},
|
|
{
|
|
"epoch": 6.776,
|
|
"grad_norm": 0.7956434167199038,
|
|
"learning_rate": 1.264587661524308e-07,
|
|
"loss": 0.0862,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05951028689742088,
|
|
"step": 4235,
|
|
"valid_targets_mean": 4340.7,
|
|
"valid_targets_min": 1611
|
|
},
|
|
{
|
|
"epoch": 6.784,
|
|
"grad_norm": 1.011368405296906,
|
|
"learning_rate": 1.1765772725051084e-07,
|
|
"loss": 0.0891,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09062150865793228,
|
|
"step": 4240,
|
|
"valid_targets_mean": 4671.1,
|
|
"valid_targets_min": 1452
|
|
},
|
|
{
|
|
"epoch": 6.792,
|
|
"grad_norm": 0.9551280745958595,
|
|
"learning_rate": 1.0917318993721726e-07,
|
|
"loss": 0.0808,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07964757084846497,
|
|
"step": 4245,
|
|
"valid_targets_mean": 4136.3,
|
|
"valid_targets_min": 1221
|
|
},
|
|
{
|
|
"epoch": 6.8,
|
|
"grad_norm": 0.9033558461760581,
|
|
"learning_rate": 1.0100528927558861e-07,
|
|
"loss": 0.0885,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08310828357934952,
|
|
"step": 4250,
|
|
"valid_targets_mean": 4679.9,
|
|
"valid_targets_min": 3972
|
|
},
|
|
{
|
|
"epoch": 6.808,
|
|
"grad_norm": 1.0187651393109292,
|
|
"learning_rate": 9.31541552882087e-08,
|
|
"loss": 0.0767,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08091911673545837,
|
|
"step": 4255,
|
|
"valid_targets_mean": 4477.1,
|
|
"valid_targets_min": 3644
|
|
},
|
|
{
|
|
"epoch": 6.816,
|
|
"grad_norm": 0.9843554117589307,
|
|
"learning_rate": 8.561991295514161e-08,
|
|
"loss": 0.0838,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09663708508014679,
|
|
"step": 4260,
|
|
"valid_targets_mean": 4688.4,
|
|
"valid_targets_min": 3921
|
|
},
|
|
{
|
|
"epoch": 6.824,
|
|
"grad_norm": 0.9227416119142963,
|
|
"learning_rate": 7.840268221193548e-08,
|
|
"loss": 0.0729,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07944896072149277,
|
|
"step": 4265,
|
|
"valid_targets_mean": 4474.9,
|
|
"valid_targets_min": 1616
|
|
},
|
|
{
|
|
"epoch": 6.832,
|
|
"grad_norm": 1.01952166771798,
|
|
"learning_rate": 7.150257794772186e-08,
|
|
"loss": 0.0815,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08922655135393143,
|
|
"step": 4270,
|
|
"valid_targets_mean": 4556.6,
|
|
"valid_targets_min": 1997
|
|
},
|
|
{
|
|
"epoch": 6.84,
|
|
"grad_norm": 0.8968978451655257,
|
|
"learning_rate": 6.491971000337938e-08,
|
|
"loss": 0.084,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08347638696432114,
|
|
"step": 4275,
|
|
"valid_targets_mean": 4803.4,
|
|
"valid_targets_min": 3984
|
|
},
|
|
{
|
|
"epoch": 6.848,
|
|
"grad_norm": 1.0285763710073241,
|
|
"learning_rate": 5.8654183169788435e-08,
|
|
"loss": 0.08,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0849163606762886,
|
|
"step": 4280,
|
|
"valid_targets_mean": 4604.2,
|
|
"valid_targets_min": 3820
|
|
},
|
|
{
|
|
"epoch": 6.856,
|
|
"grad_norm": 0.8983928364725378,
|
|
"learning_rate": 5.270609718616593e-08,
|
|
"loss": 0.0863,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08616328239440918,
|
|
"step": 4285,
|
|
"valid_targets_mean": 4831.1,
|
|
"valid_targets_min": 3694
|
|
},
|
|
{
|
|
"epoch": 6.864,
|
|
"grad_norm": 0.9571598009684891,
|
|
"learning_rate": 4.70755467384687e-08,
|
|
"loss": 0.0893,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08254455029964447,
|
|
"step": 4290,
|
|
"valid_targets_mean": 4250.3,
|
|
"valid_targets_min": 1295
|
|
},
|
|
{
|
|
"epoch": 6.872,
|
|
"grad_norm": 0.9540142639753194,
|
|
"learning_rate": 4.176262145789478e-08,
|
|
"loss": 0.0831,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09365525841712952,
|
|
"step": 4295,
|
|
"valid_targets_mean": 4645.9,
|
|
"valid_targets_min": 1616
|
|
},
|
|
{
|
|
"epoch": 6.88,
|
|
"grad_norm": 1.1044258017436492,
|
|
"learning_rate": 3.676740591945782e-08,
|
|
"loss": 0.0864,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07931956648826599,
|
|
"step": 4300,
|
|
"valid_targets_mean": 4615.0,
|
|
"valid_targets_min": 3145
|
|
},
|
|
{
|
|
"epoch": 6.888,
|
|
"grad_norm": 0.9470844549451539,
|
|
"learning_rate": 3.208997964062821e-08,
|
|
"loss": 0.0773,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08684109896421432,
|
|
"step": 4305,
|
|
"valid_targets_mean": 4425.6,
|
|
"valid_targets_min": 2530
|
|
},
|
|
{
|
|
"epoch": 6.896,
|
|
"grad_norm": 0.9218232732418137,
|
|
"learning_rate": 2.773041708008295e-08,
|
|
"loss": 0.0866,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.082296222448349,
|
|
"step": 4310,
|
|
"valid_targets_mean": 4279.6,
|
|
"valid_targets_min": 1180
|
|
},
|
|
{
|
|
"epoch": 6.904,
|
|
"grad_norm": 0.9071249998095612,
|
|
"learning_rate": 2.3688787636511057e-08,
|
|
"loss": 0.0736,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08217165619134903,
|
|
"step": 4315,
|
|
"valid_targets_mean": 4544.4,
|
|
"valid_targets_min": 3597
|
|
},
|
|
{
|
|
"epoch": 6.912,
|
|
"grad_norm": 0.8891574178897317,
|
|
"learning_rate": 1.9965155647507782e-08,
|
|
"loss": 0.0825,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06934249401092529,
|
|
"step": 4320,
|
|
"valid_targets_mean": 4592.1,
|
|
"valid_targets_min": 3450
|
|
},
|
|
{
|
|
"epoch": 6.92,
|
|
"grad_norm": 0.9080492183145513,
|
|
"learning_rate": 1.655958038855765e-08,
|
|
"loss": 0.0815,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08091747760772705,
|
|
"step": 4325,
|
|
"valid_targets_mean": 4569.8,
|
|
"valid_targets_min": 3395
|
|
},
|
|
{
|
|
"epoch": 6.928,
|
|
"grad_norm": 0.8879644070977345,
|
|
"learning_rate": 1.3472116072084096e-08,
|
|
"loss": 0.0817,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07529698312282562,
|
|
"step": 4330,
|
|
"valid_targets_mean": 4701.6,
|
|
"valid_targets_min": 3017
|
|
},
|
|
{
|
|
"epoch": 6.936,
|
|
"grad_norm": 0.9129140226325475,
|
|
"learning_rate": 1.0702811846590167e-08,
|
|
"loss": 0.0789,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07977509498596191,
|
|
"step": 4335,
|
|
"valid_targets_mean": 4494.8,
|
|
"valid_targets_min": 2169
|
|
},
|
|
{
|
|
"epoch": 6.944,
|
|
"grad_norm": 1.104263961464604,
|
|
"learning_rate": 8.251711795876916e-09,
|
|
"loss": 0.0827,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09657025337219238,
|
|
"step": 4340,
|
|
"valid_targets_mean": 4092.5,
|
|
"valid_targets_min": 1800
|
|
},
|
|
{
|
|
"epoch": 6.952,
|
|
"grad_norm": 0.950385428088302,
|
|
"learning_rate": 6.1188549383373044e-09,
|
|
"loss": 0.0826,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0766601487994194,
|
|
"step": 4345,
|
|
"valid_targets_mean": 4405.0,
|
|
"valid_targets_min": 2952
|
|
},
|
|
{
|
|
"epoch": 6.96,
|
|
"grad_norm": 0.9897846459676481,
|
|
"learning_rate": 4.304275226338916e-09,
|
|
"loss": 0.087,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10060922801494598,
|
|
"step": 4350,
|
|
"valid_targets_mean": 4761.4,
|
|
"valid_targets_min": 4005
|
|
},
|
|
{
|
|
"epoch": 6.968,
|
|
"grad_norm": 1.0311363385103403,
|
|
"learning_rate": 2.8080015456799503e-09,
|
|
"loss": 0.0794,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08533407747745514,
|
|
"step": 4355,
|
|
"valid_targets_mean": 4302.2,
|
|
"valid_targets_min": 1915
|
|
},
|
|
{
|
|
"epoch": 6.976,
|
|
"grad_norm": 1.1144151745197948,
|
|
"learning_rate": 1.6300577151340257e-09,
|
|
"loss": 0.0926,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08315521478652954,
|
|
"step": 4360,
|
|
"valid_targets_mean": 4450.6,
|
|
"valid_targets_min": 926
|
|
},
|
|
{
|
|
"epoch": 6.984,
|
|
"grad_norm": 0.9457292158664942,
|
|
"learning_rate": 7.70462486070489e-10,
|
|
"loss": 0.0839,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0931163877248764,
|
|
"step": 4365,
|
|
"valid_targets_mean": 4652.3,
|
|
"valid_targets_min": 2608
|
|
},
|
|
{
|
|
"epoch": 6.992,
|
|
"grad_norm": 1.231726130549695,
|
|
"learning_rate": 2.2922954214799065e-10,
|
|
"loss": 0.0838,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0798584520816803,
|
|
"step": 4370,
|
|
"valid_targets_mean": 4554.8,
|
|
"valid_targets_min": 3754
|
|
},
|
|
{
|
|
"epoch": 7.0,
|
|
"grad_norm": 0.9959170907949623,
|
|
"learning_rate": 6.367499107984288e-12,
|
|
"loss": 0.0869,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07439238578081131,
|
|
"step": 4375,
|
|
"valid_targets_mean": 4424.4,
|
|
"valid_targets_min": 3274
|
|
},
|
|
{
|
|
"epoch": 7.0,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07439238578081131,
|
|
"step": 4375,
|
|
"total_flos": 2121000910061568.0,
|
|
"train_loss": 0.1883448345048087,
|
|
"train_runtime": 24429.3523,
|
|
"train_samples_per_second": 2.865,
|
|
"train_steps_per_second": 0.179,
|
|
"valid_targets_mean": 4424.4,
|
|
"valid_targets_min": 3274
|
|
}
|
|
],
|
|
"logging_steps": 5,
|
|
"max_steps": 4375,
|
|
"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": 2121000910061568.0,
|
|
"train_batch_size": 1,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|