9486 lines
263 KiB
JSON
9486 lines
263 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 7.0,
|
|
"eval_steps": 500,
|
|
"global_step": 4291,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.00816326530612245,
|
|
"grad_norm": 12.33506016365843,
|
|
"learning_rate": 3.7209302325581396e-07,
|
|
"loss": 0.75,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.31827449798583984,
|
|
"step": 5,
|
|
"valid_targets_mean": 3888.0,
|
|
"valid_targets_min": 2300
|
|
},
|
|
{
|
|
"epoch": 0.0163265306122449,
|
|
"grad_norm": 14.06557282235133,
|
|
"learning_rate": 8.372093023255814e-07,
|
|
"loss": 0.7413,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4188452959060669,
|
|
"step": 10,
|
|
"valid_targets_mean": 3943.4,
|
|
"valid_targets_min": 1645
|
|
},
|
|
{
|
|
"epoch": 0.024489795918367346,
|
|
"grad_norm": 12.64563346356677,
|
|
"learning_rate": 1.302325581395349e-06,
|
|
"loss": 0.7263,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4356209933757782,
|
|
"step": 15,
|
|
"valid_targets_mean": 5323.4,
|
|
"valid_targets_min": 2731
|
|
},
|
|
{
|
|
"epoch": 0.0326530612244898,
|
|
"grad_norm": 7.5861629954645045,
|
|
"learning_rate": 1.7674418604651164e-06,
|
|
"loss": 0.7189,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3566373586654663,
|
|
"step": 20,
|
|
"valid_targets_mean": 4771.5,
|
|
"valid_targets_min": 231
|
|
},
|
|
{
|
|
"epoch": 0.04081632653061224,
|
|
"grad_norm": 5.884180511617331,
|
|
"learning_rate": 2.232558139534884e-06,
|
|
"loss": 0.6782,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2910090386867523,
|
|
"step": 25,
|
|
"valid_targets_mean": 2981.0,
|
|
"valid_targets_min": 697
|
|
},
|
|
{
|
|
"epoch": 0.04897959183673469,
|
|
"grad_norm": 4.048322161590491,
|
|
"learning_rate": 2.6976744186046517e-06,
|
|
"loss": 0.5857,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2788548767566681,
|
|
"step": 30,
|
|
"valid_targets_mean": 4749.9,
|
|
"valid_targets_min": 2045
|
|
},
|
|
{
|
|
"epoch": 0.05714285714285714,
|
|
"grad_norm": 3.8721871492073894,
|
|
"learning_rate": 3.1627906976744187e-06,
|
|
"loss": 0.5584,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2676370143890381,
|
|
"step": 35,
|
|
"valid_targets_mean": 4417.5,
|
|
"valid_targets_min": 1155
|
|
},
|
|
{
|
|
"epoch": 0.0653061224489796,
|
|
"grad_norm": 2.1663266935097294,
|
|
"learning_rate": 3.6279069767441866e-06,
|
|
"loss": 0.5306,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28581827878952026,
|
|
"step": 40,
|
|
"valid_targets_mean": 4260.9,
|
|
"valid_targets_min": 2287
|
|
},
|
|
{
|
|
"epoch": 0.07346938775510205,
|
|
"grad_norm": 1.4335626902963403,
|
|
"learning_rate": 4.0930232558139536e-06,
|
|
"loss": 0.5073,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28230318427085876,
|
|
"step": 45,
|
|
"valid_targets_mean": 4514.6,
|
|
"valid_targets_min": 1947
|
|
},
|
|
{
|
|
"epoch": 0.08163265306122448,
|
|
"grad_norm": 0.8187625274651436,
|
|
"learning_rate": 4.558139534883721e-06,
|
|
"loss": 0.463,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1898300051689148,
|
|
"step": 50,
|
|
"valid_targets_mean": 6367.4,
|
|
"valid_targets_min": 1883
|
|
},
|
|
{
|
|
"epoch": 0.08979591836734693,
|
|
"grad_norm": 0.9581011591667019,
|
|
"learning_rate": 5.023255813953489e-06,
|
|
"loss": 0.4861,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22689494490623474,
|
|
"step": 55,
|
|
"valid_targets_mean": 3721.6,
|
|
"valid_targets_min": 2174
|
|
},
|
|
{
|
|
"epoch": 0.09795918367346938,
|
|
"grad_norm": 0.688459144076369,
|
|
"learning_rate": 5.488372093023256e-06,
|
|
"loss": 0.4585,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21300674974918365,
|
|
"step": 60,
|
|
"valid_targets_mean": 5311.5,
|
|
"valid_targets_min": 179
|
|
},
|
|
{
|
|
"epoch": 0.10612244897959183,
|
|
"grad_norm": 0.7211400009988276,
|
|
"learning_rate": 5.953488372093023e-06,
|
|
"loss": 0.4672,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2803425192832947,
|
|
"step": 65,
|
|
"valid_targets_mean": 5541.2,
|
|
"valid_targets_min": 225
|
|
},
|
|
{
|
|
"epoch": 0.11428571428571428,
|
|
"grad_norm": 0.6534824892302817,
|
|
"learning_rate": 6.418604651162791e-06,
|
|
"loss": 0.44,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2042653113603592,
|
|
"step": 70,
|
|
"valid_targets_mean": 3739.4,
|
|
"valid_targets_min": 1544
|
|
},
|
|
{
|
|
"epoch": 0.12244897959183673,
|
|
"grad_norm": 0.7238400984773256,
|
|
"learning_rate": 6.883720930232559e-06,
|
|
"loss": 0.4608,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2297825664281845,
|
|
"step": 75,
|
|
"valid_targets_mean": 3874.8,
|
|
"valid_targets_min": 2090
|
|
},
|
|
{
|
|
"epoch": 0.1306122448979592,
|
|
"grad_norm": 0.6261187440459337,
|
|
"learning_rate": 7.348837209302326e-06,
|
|
"loss": 0.4456,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20637601613998413,
|
|
"step": 80,
|
|
"valid_targets_mean": 4730.2,
|
|
"valid_targets_min": 2225
|
|
},
|
|
{
|
|
"epoch": 0.13877551020408163,
|
|
"grad_norm": 0.6083383898632471,
|
|
"learning_rate": 7.813953488372094e-06,
|
|
"loss": 0.4337,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21255721151828766,
|
|
"step": 85,
|
|
"valid_targets_mean": 4527.2,
|
|
"valid_targets_min": 2565
|
|
},
|
|
{
|
|
"epoch": 0.1469387755102041,
|
|
"grad_norm": 0.49647727411141274,
|
|
"learning_rate": 8.279069767441861e-06,
|
|
"loss": 0.3833,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18475192785263062,
|
|
"step": 90,
|
|
"valid_targets_mean": 5922.8,
|
|
"valid_targets_min": 2468
|
|
},
|
|
{
|
|
"epoch": 0.15510204081632653,
|
|
"grad_norm": 0.6578636753926649,
|
|
"learning_rate": 8.74418604651163e-06,
|
|
"loss": 0.4174,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19449181854724884,
|
|
"step": 95,
|
|
"valid_targets_mean": 4025.9,
|
|
"valid_targets_min": 2181
|
|
},
|
|
{
|
|
"epoch": 0.16326530612244897,
|
|
"grad_norm": 0.5554069864757591,
|
|
"learning_rate": 9.209302325581397e-06,
|
|
"loss": 0.4086,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2075560986995697,
|
|
"step": 100,
|
|
"valid_targets_mean": 5168.0,
|
|
"valid_targets_min": 2191
|
|
},
|
|
{
|
|
"epoch": 0.17142857142857143,
|
|
"grad_norm": 0.5442721349991868,
|
|
"learning_rate": 9.674418604651164e-06,
|
|
"loss": 0.3998,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17283964157104492,
|
|
"step": 105,
|
|
"valid_targets_mean": 4438.6,
|
|
"valid_targets_min": 1285
|
|
},
|
|
{
|
|
"epoch": 0.17959183673469387,
|
|
"grad_norm": 0.5267931849085247,
|
|
"learning_rate": 1.0139534883720932e-05,
|
|
"loss": 0.3843,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20762407779693604,
|
|
"step": 110,
|
|
"valid_targets_mean": 6056.4,
|
|
"valid_targets_min": 2910
|
|
},
|
|
{
|
|
"epoch": 0.18775510204081633,
|
|
"grad_norm": 0.5536849451475337,
|
|
"learning_rate": 1.0604651162790698e-05,
|
|
"loss": 0.3954,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20768246054649353,
|
|
"step": 115,
|
|
"valid_targets_mean": 4792.0,
|
|
"valid_targets_min": 864
|
|
},
|
|
{
|
|
"epoch": 0.19591836734693877,
|
|
"grad_norm": 0.5236424454710094,
|
|
"learning_rate": 1.1069767441860466e-05,
|
|
"loss": 0.3875,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21655716001987457,
|
|
"step": 120,
|
|
"valid_targets_mean": 6482.5,
|
|
"valid_targets_min": 1500
|
|
},
|
|
{
|
|
"epoch": 0.20408163265306123,
|
|
"grad_norm": 0.5025306927483865,
|
|
"learning_rate": 1.1534883720930235e-05,
|
|
"loss": 0.3829,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17163898050785065,
|
|
"step": 125,
|
|
"valid_targets_mean": 4536.0,
|
|
"valid_targets_min": 2532
|
|
},
|
|
{
|
|
"epoch": 0.21224489795918366,
|
|
"grad_norm": 0.4284847923366298,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 0.3573,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13200239837169647,
|
|
"step": 130,
|
|
"valid_targets_mean": 4378.4,
|
|
"valid_targets_min": 1834
|
|
},
|
|
{
|
|
"epoch": 0.22040816326530613,
|
|
"grad_norm": 0.5724886335853256,
|
|
"learning_rate": 1.2465116279069769e-05,
|
|
"loss": 0.3791,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17680321633815765,
|
|
"step": 135,
|
|
"valid_targets_mean": 4725.5,
|
|
"valid_targets_min": 2184
|
|
},
|
|
{
|
|
"epoch": 0.22857142857142856,
|
|
"grad_norm": 0.5262974819415349,
|
|
"learning_rate": 1.2930232558139534e-05,
|
|
"loss": 0.3689,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1830161213874817,
|
|
"step": 140,
|
|
"valid_targets_mean": 5207.1,
|
|
"valid_targets_min": 1525
|
|
},
|
|
{
|
|
"epoch": 0.23673469387755103,
|
|
"grad_norm": 0.5943237872366759,
|
|
"learning_rate": 1.3395348837209303e-05,
|
|
"loss": 0.3841,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16007442772388458,
|
|
"step": 145,
|
|
"valid_targets_mean": 4077.4,
|
|
"valid_targets_min": 3143
|
|
},
|
|
{
|
|
"epoch": 0.24489795918367346,
|
|
"grad_norm": 0.5011629694901417,
|
|
"learning_rate": 1.3860465116279072e-05,
|
|
"loss": 0.3655,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21169030666351318,
|
|
"step": 150,
|
|
"valid_targets_mean": 7955.6,
|
|
"valid_targets_min": 2451
|
|
},
|
|
{
|
|
"epoch": 0.2530612244897959,
|
|
"grad_norm": 0.5412588141865855,
|
|
"learning_rate": 1.4325581395348837e-05,
|
|
"loss": 0.3682,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1689492166042328,
|
|
"step": 155,
|
|
"valid_targets_mean": 5123.8,
|
|
"valid_targets_min": 2164
|
|
},
|
|
{
|
|
"epoch": 0.2612244897959184,
|
|
"grad_norm": 0.6643295450246909,
|
|
"learning_rate": 1.4790697674418606e-05,
|
|
"loss": 0.3562,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1859186887741089,
|
|
"step": 160,
|
|
"valid_targets_mean": 6049.8,
|
|
"valid_targets_min": 3021
|
|
},
|
|
{
|
|
"epoch": 0.2693877551020408,
|
|
"grad_norm": 0.597344608075607,
|
|
"learning_rate": 1.5255813953488374e-05,
|
|
"loss": 0.3692,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15211470425128937,
|
|
"step": 165,
|
|
"valid_targets_mean": 3266.6,
|
|
"valid_targets_min": 1135
|
|
},
|
|
{
|
|
"epoch": 0.27755102040816326,
|
|
"grad_norm": 0.6202606312284262,
|
|
"learning_rate": 1.572093023255814e-05,
|
|
"loss": 0.3689,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16817247867584229,
|
|
"step": 170,
|
|
"valid_targets_mean": 3775.8,
|
|
"valid_targets_min": 2136
|
|
},
|
|
{
|
|
"epoch": 0.2857142857142857,
|
|
"grad_norm": 0.5839399121332536,
|
|
"learning_rate": 1.618604651162791e-05,
|
|
"loss": 0.3431,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15334473550319672,
|
|
"step": 175,
|
|
"valid_targets_mean": 4572.5,
|
|
"valid_targets_min": 2188
|
|
},
|
|
{
|
|
"epoch": 0.2938775510204082,
|
|
"grad_norm": 0.46689156867921305,
|
|
"learning_rate": 1.6651162790697674e-05,
|
|
"loss": 0.3154,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1432073712348938,
|
|
"step": 180,
|
|
"valid_targets_mean": 5915.1,
|
|
"valid_targets_min": 2132
|
|
},
|
|
{
|
|
"epoch": 0.3020408163265306,
|
|
"grad_norm": 0.5309548508364261,
|
|
"learning_rate": 1.7116279069767444e-05,
|
|
"loss": 0.3333,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20056375861167908,
|
|
"step": 185,
|
|
"valid_targets_mean": 6440.8,
|
|
"valid_targets_min": 2163
|
|
},
|
|
{
|
|
"epoch": 0.31020408163265306,
|
|
"grad_norm": 0.5604620864213024,
|
|
"learning_rate": 1.758139534883721e-05,
|
|
"loss": 0.3352,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1434939056634903,
|
|
"step": 190,
|
|
"valid_targets_mean": 4429.0,
|
|
"valid_targets_min": 2288
|
|
},
|
|
{
|
|
"epoch": 0.3183673469387755,
|
|
"grad_norm": 0.5450998899560288,
|
|
"learning_rate": 1.8046511627906978e-05,
|
|
"loss": 0.345,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17527905106544495,
|
|
"step": 195,
|
|
"valid_targets_mean": 5860.5,
|
|
"valid_targets_min": 1828
|
|
},
|
|
{
|
|
"epoch": 0.32653061224489793,
|
|
"grad_norm": 0.5852906718193894,
|
|
"learning_rate": 1.8511627906976745e-05,
|
|
"loss": 0.3586,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15143322944641113,
|
|
"step": 200,
|
|
"valid_targets_mean": 4496.5,
|
|
"valid_targets_min": 1416
|
|
},
|
|
{
|
|
"epoch": 0.3346938775510204,
|
|
"grad_norm": 0.6271764625641036,
|
|
"learning_rate": 1.8976744186046516e-05,
|
|
"loss": 0.3368,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14210546016693115,
|
|
"step": 205,
|
|
"valid_targets_mean": 3699.0,
|
|
"valid_targets_min": 2282
|
|
},
|
|
{
|
|
"epoch": 0.34285714285714286,
|
|
"grad_norm": 0.5616628504866027,
|
|
"learning_rate": 1.944186046511628e-05,
|
|
"loss": 0.3404,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2095171958208084,
|
|
"step": 210,
|
|
"valid_targets_mean": 5037.4,
|
|
"valid_targets_min": 1231
|
|
},
|
|
{
|
|
"epoch": 0.3510204081632653,
|
|
"grad_norm": 0.5615575210169799,
|
|
"learning_rate": 1.990697674418605e-05,
|
|
"loss": 0.3455,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1549168974161148,
|
|
"step": 215,
|
|
"valid_targets_mean": 4484.1,
|
|
"valid_targets_min": 665
|
|
},
|
|
{
|
|
"epoch": 0.35918367346938773,
|
|
"grad_norm": 0.507550652268599,
|
|
"learning_rate": 2.0372093023255813e-05,
|
|
"loss": 0.3256,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1841312050819397,
|
|
"step": 220,
|
|
"valid_targets_mean": 6250.6,
|
|
"valid_targets_min": 3309
|
|
},
|
|
{
|
|
"epoch": 0.3673469387755102,
|
|
"grad_norm": 0.5094360914962326,
|
|
"learning_rate": 2.0837209302325584e-05,
|
|
"loss": 0.2986,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1224166676402092,
|
|
"step": 225,
|
|
"valid_targets_mean": 3814.6,
|
|
"valid_targets_min": 2198
|
|
},
|
|
{
|
|
"epoch": 0.37551020408163266,
|
|
"grad_norm": 0.6246439139516935,
|
|
"learning_rate": 2.130232558139535e-05,
|
|
"loss": 0.3275,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1374599039554596,
|
|
"step": 230,
|
|
"valid_targets_mean": 3553.8,
|
|
"valid_targets_min": 217
|
|
},
|
|
{
|
|
"epoch": 0.3836734693877551,
|
|
"grad_norm": 0.5951069221562814,
|
|
"learning_rate": 2.176744186046512e-05,
|
|
"loss": 0.3489,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18446165323257446,
|
|
"step": 235,
|
|
"valid_targets_mean": 4747.2,
|
|
"valid_targets_min": 1414
|
|
},
|
|
{
|
|
"epoch": 0.39183673469387753,
|
|
"grad_norm": 0.6523853794671483,
|
|
"learning_rate": 2.2232558139534888e-05,
|
|
"loss": 0.333,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15122082829475403,
|
|
"step": 240,
|
|
"valid_targets_mean": 3160.5,
|
|
"valid_targets_min": 1638
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"grad_norm": 0.4725934988884912,
|
|
"learning_rate": 2.269767441860465e-05,
|
|
"loss": 0.3198,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13313765823841095,
|
|
"step": 245,
|
|
"valid_targets_mean": 6339.0,
|
|
"valid_targets_min": 1913
|
|
},
|
|
{
|
|
"epoch": 0.40816326530612246,
|
|
"grad_norm": 0.49423205201340153,
|
|
"learning_rate": 2.316279069767442e-05,
|
|
"loss": 0.3368,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16928038001060486,
|
|
"step": 250,
|
|
"valid_targets_mean": 7214.2,
|
|
"valid_targets_min": 3139
|
|
},
|
|
{
|
|
"epoch": 0.4163265306122449,
|
|
"grad_norm": 0.4974690927027136,
|
|
"learning_rate": 2.362790697674419e-05,
|
|
"loss": 0.3345,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16831715404987335,
|
|
"step": 255,
|
|
"valid_targets_mean": 5773.4,
|
|
"valid_targets_min": 2589
|
|
},
|
|
{
|
|
"epoch": 0.42448979591836733,
|
|
"grad_norm": 0.5602390366587728,
|
|
"learning_rate": 2.4093023255813956e-05,
|
|
"loss": 0.3444,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17655014991760254,
|
|
"step": 260,
|
|
"valid_targets_mean": 5242.8,
|
|
"valid_targets_min": 2248
|
|
},
|
|
{
|
|
"epoch": 0.4326530612244898,
|
|
"grad_norm": 0.6143492193154322,
|
|
"learning_rate": 2.4558139534883726e-05,
|
|
"loss": 0.3311,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16245931386947632,
|
|
"step": 265,
|
|
"valid_targets_mean": 4020.8,
|
|
"valid_targets_min": 1845
|
|
},
|
|
{
|
|
"epoch": 0.44081632653061226,
|
|
"grad_norm": 0.5185982721162242,
|
|
"learning_rate": 2.502325581395349e-05,
|
|
"loss": 0.3395,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.169978067278862,
|
|
"step": 270,
|
|
"valid_targets_mean": 5750.5,
|
|
"valid_targets_min": 1634
|
|
},
|
|
{
|
|
"epoch": 0.4489795918367347,
|
|
"grad_norm": 0.5510287595838287,
|
|
"learning_rate": 2.5488372093023257e-05,
|
|
"loss": 0.3273,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15716497600078583,
|
|
"step": 275,
|
|
"valid_targets_mean": 6445.8,
|
|
"valid_targets_min": 1772
|
|
},
|
|
{
|
|
"epoch": 0.45714285714285713,
|
|
"grad_norm": 0.5410082300355423,
|
|
"learning_rate": 2.5953488372093024e-05,
|
|
"loss": 0.3111,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19414874911308289,
|
|
"step": 280,
|
|
"valid_targets_mean": 5556.1,
|
|
"valid_targets_min": 1816
|
|
},
|
|
{
|
|
"epoch": 0.46530612244897956,
|
|
"grad_norm": 0.5707835911746898,
|
|
"learning_rate": 2.6418604651162794e-05,
|
|
"loss": 0.2959,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16951584815979004,
|
|
"step": 285,
|
|
"valid_targets_mean": 6885.9,
|
|
"valid_targets_min": 1210
|
|
},
|
|
{
|
|
"epoch": 0.47346938775510206,
|
|
"grad_norm": 0.6040168533013804,
|
|
"learning_rate": 2.688372093023256e-05,
|
|
"loss": 0.3089,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1755496859550476,
|
|
"step": 290,
|
|
"valid_targets_mean": 4725.0,
|
|
"valid_targets_min": 1232
|
|
},
|
|
{
|
|
"epoch": 0.4816326530612245,
|
|
"grad_norm": 0.5954085643873348,
|
|
"learning_rate": 2.7348837209302325e-05,
|
|
"loss": 0.3217,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13682439923286438,
|
|
"step": 295,
|
|
"valid_targets_mean": 4376.2,
|
|
"valid_targets_min": 2137
|
|
},
|
|
{
|
|
"epoch": 0.4897959183673469,
|
|
"grad_norm": 0.6431457999458166,
|
|
"learning_rate": 2.7813953488372095e-05,
|
|
"loss": 0.3085,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1722913682460785,
|
|
"step": 300,
|
|
"valid_targets_mean": 4122.9,
|
|
"valid_targets_min": 2335
|
|
},
|
|
{
|
|
"epoch": 0.49795918367346936,
|
|
"grad_norm": 0.5475282840206034,
|
|
"learning_rate": 2.8279069767441862e-05,
|
|
"loss": 0.3217,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14570221304893494,
|
|
"step": 305,
|
|
"valid_targets_mean": 3578.1,
|
|
"valid_targets_min": 2075
|
|
},
|
|
{
|
|
"epoch": 0.5061224489795918,
|
|
"grad_norm": 0.5977501757018006,
|
|
"learning_rate": 2.874418604651163e-05,
|
|
"loss": 0.3227,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13097640872001648,
|
|
"step": 310,
|
|
"valid_targets_mean": 3568.1,
|
|
"valid_targets_min": 1950
|
|
},
|
|
{
|
|
"epoch": 0.5142857142857142,
|
|
"grad_norm": 0.6119511908814466,
|
|
"learning_rate": 2.92093023255814e-05,
|
|
"loss": 0.3193,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1389477699995041,
|
|
"step": 315,
|
|
"valid_targets_mean": 3503.9,
|
|
"valid_targets_min": 1251
|
|
},
|
|
{
|
|
"epoch": 0.5224489795918368,
|
|
"grad_norm": 0.5252404379494299,
|
|
"learning_rate": 2.9674418604651167e-05,
|
|
"loss": 0.3192,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1406920701265335,
|
|
"step": 320,
|
|
"valid_targets_mean": 4817.9,
|
|
"valid_targets_min": 2302
|
|
},
|
|
{
|
|
"epoch": 0.5306122448979592,
|
|
"grad_norm": 0.5672213393499427,
|
|
"learning_rate": 3.013953488372093e-05,
|
|
"loss": 0.3121,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17177443206310272,
|
|
"step": 325,
|
|
"valid_targets_mean": 5428.4,
|
|
"valid_targets_min": 1917
|
|
},
|
|
{
|
|
"epoch": 0.5387755102040817,
|
|
"grad_norm": 0.5873908296751622,
|
|
"learning_rate": 3.06046511627907e-05,
|
|
"loss": 0.3149,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1694718897342682,
|
|
"step": 330,
|
|
"valid_targets_mean": 4601.1,
|
|
"valid_targets_min": 2799
|
|
},
|
|
{
|
|
"epoch": 0.5469387755102041,
|
|
"grad_norm": 0.5200193608156581,
|
|
"learning_rate": 3.1069767441860465e-05,
|
|
"loss": 0.3143,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1358920931816101,
|
|
"step": 335,
|
|
"valid_targets_mean": 5289.6,
|
|
"valid_targets_min": 3964
|
|
},
|
|
{
|
|
"epoch": 0.5551020408163265,
|
|
"grad_norm": 0.6481598582680852,
|
|
"learning_rate": 3.1534883720930235e-05,
|
|
"loss": 0.3014,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17161871492862701,
|
|
"step": 340,
|
|
"valid_targets_mean": 4119.6,
|
|
"valid_targets_min": 1883
|
|
},
|
|
{
|
|
"epoch": 0.563265306122449,
|
|
"grad_norm": 0.5109022013798562,
|
|
"learning_rate": 3.2000000000000005e-05,
|
|
"loss": 0.313,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13416041433811188,
|
|
"step": 345,
|
|
"valid_targets_mean": 4991.4,
|
|
"valid_targets_min": 657
|
|
},
|
|
{
|
|
"epoch": 0.5714285714285714,
|
|
"grad_norm": 0.5339499139390554,
|
|
"learning_rate": 3.246511627906977e-05,
|
|
"loss": 0.2967,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14657235145568848,
|
|
"step": 350,
|
|
"valid_targets_mean": 5172.1,
|
|
"valid_targets_min": 2580
|
|
},
|
|
{
|
|
"epoch": 0.5795918367346938,
|
|
"grad_norm": 0.6023496113208678,
|
|
"learning_rate": 3.293023255813954e-05,
|
|
"loss": 0.3015,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1426255851984024,
|
|
"step": 355,
|
|
"valid_targets_mean": 5163.6,
|
|
"valid_targets_min": 2359
|
|
},
|
|
{
|
|
"epoch": 0.5877551020408164,
|
|
"grad_norm": 0.6011243129435078,
|
|
"learning_rate": 3.33953488372093e-05,
|
|
"loss": 0.3067,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19121070206165314,
|
|
"step": 360,
|
|
"valid_targets_mean": 4767.0,
|
|
"valid_targets_min": 2148
|
|
},
|
|
{
|
|
"epoch": 0.5959183673469388,
|
|
"grad_norm": 0.6749077948841615,
|
|
"learning_rate": 3.386046511627907e-05,
|
|
"loss": 0.3107,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18281207978725433,
|
|
"step": 365,
|
|
"valid_targets_mean": 3520.6,
|
|
"valid_targets_min": 1283
|
|
},
|
|
{
|
|
"epoch": 0.6040816326530613,
|
|
"grad_norm": 0.5445436044445798,
|
|
"learning_rate": 3.4325581395348844e-05,
|
|
"loss": 0.2927,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12598849833011627,
|
|
"step": 370,
|
|
"valid_targets_mean": 5079.1,
|
|
"valid_targets_min": 1151
|
|
},
|
|
{
|
|
"epoch": 0.6122448979591837,
|
|
"grad_norm": 0.698927419975243,
|
|
"learning_rate": 3.479069767441861e-05,
|
|
"loss": 0.306,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1887175440788269,
|
|
"step": 375,
|
|
"valid_targets_mean": 5366.5,
|
|
"valid_targets_min": 1548
|
|
},
|
|
{
|
|
"epoch": 0.6204081632653061,
|
|
"grad_norm": 0.5846609611760529,
|
|
"learning_rate": 3.525581395348837e-05,
|
|
"loss": 0.3136,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15284685790538788,
|
|
"step": 380,
|
|
"valid_targets_mean": 3979.0,
|
|
"valid_targets_min": 1392
|
|
},
|
|
{
|
|
"epoch": 0.6285714285714286,
|
|
"grad_norm": 0.5376221163617384,
|
|
"learning_rate": 3.572093023255814e-05,
|
|
"loss": 0.2908,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1378687173128128,
|
|
"step": 385,
|
|
"valid_targets_mean": 4058.5,
|
|
"valid_targets_min": 1270
|
|
},
|
|
{
|
|
"epoch": 0.636734693877551,
|
|
"grad_norm": 0.5530069475016968,
|
|
"learning_rate": 3.618604651162791e-05,
|
|
"loss": 0.3002,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13155528903007507,
|
|
"step": 390,
|
|
"valid_targets_mean": 4221.2,
|
|
"valid_targets_min": 1822
|
|
},
|
|
{
|
|
"epoch": 0.6448979591836734,
|
|
"grad_norm": 0.4978881241002247,
|
|
"learning_rate": 3.6651162790697675e-05,
|
|
"loss": 0.3036,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11640087515115738,
|
|
"step": 395,
|
|
"valid_targets_mean": 3738.6,
|
|
"valid_targets_min": 2168
|
|
},
|
|
{
|
|
"epoch": 0.6530612244897959,
|
|
"grad_norm": 0.5430912992210294,
|
|
"learning_rate": 3.7116279069767446e-05,
|
|
"loss": 0.2896,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.155136376619339,
|
|
"step": 400,
|
|
"valid_targets_mean": 5601.9,
|
|
"valid_targets_min": 3271
|
|
},
|
|
{
|
|
"epoch": 0.6612244897959184,
|
|
"grad_norm": 0.49528034996711595,
|
|
"learning_rate": 3.758139534883721e-05,
|
|
"loss": 0.3105,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13687986135482788,
|
|
"step": 405,
|
|
"valid_targets_mean": 5079.8,
|
|
"valid_targets_min": 2260
|
|
},
|
|
{
|
|
"epoch": 0.6693877551020408,
|
|
"grad_norm": 0.5921551637945828,
|
|
"learning_rate": 3.804651162790698e-05,
|
|
"loss": 0.3186,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11712108552455902,
|
|
"step": 410,
|
|
"valid_targets_mean": 2763.8,
|
|
"valid_targets_min": 1469
|
|
},
|
|
{
|
|
"epoch": 0.6775510204081633,
|
|
"grad_norm": 0.43782810606243927,
|
|
"learning_rate": 3.851162790697675e-05,
|
|
"loss": 0.29,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09977801144123077,
|
|
"step": 415,
|
|
"valid_targets_mean": 5919.1,
|
|
"valid_targets_min": 2181
|
|
},
|
|
{
|
|
"epoch": 0.6857142857142857,
|
|
"grad_norm": 0.4890700657801529,
|
|
"learning_rate": 3.8976744186046514e-05,
|
|
"loss": 0.2986,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11042569577693939,
|
|
"step": 420,
|
|
"valid_targets_mean": 3841.8,
|
|
"valid_targets_min": 1487
|
|
},
|
|
{
|
|
"epoch": 0.6938775510204082,
|
|
"grad_norm": 0.5529585097612127,
|
|
"learning_rate": 3.9441860465116284e-05,
|
|
"loss": 0.2987,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14143112301826477,
|
|
"step": 425,
|
|
"valid_targets_mean": 4780.0,
|
|
"valid_targets_min": 1625
|
|
},
|
|
{
|
|
"epoch": 0.7020408163265306,
|
|
"grad_norm": 0.5508262572064092,
|
|
"learning_rate": 3.990697674418605e-05,
|
|
"loss": 0.2894,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1426255702972412,
|
|
"step": 430,
|
|
"valid_targets_mean": 5180.5,
|
|
"valid_targets_min": 2789
|
|
},
|
|
{
|
|
"epoch": 0.710204081632653,
|
|
"grad_norm": 0.5066137206557315,
|
|
"learning_rate": 3.99998940698124e-05,
|
|
"loss": 0.3196,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16803446412086487,
|
|
"step": 435,
|
|
"valid_targets_mean": 5961.6,
|
|
"valid_targets_min": 3011
|
|
},
|
|
{
|
|
"epoch": 0.7183673469387755,
|
|
"grad_norm": 0.5025096770678751,
|
|
"learning_rate": 3.999946373034843e-05,
|
|
"loss": 0.2733,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15143230557441711,
|
|
"step": 440,
|
|
"valid_targets_mean": 6720.8,
|
|
"valid_targets_min": 3227
|
|
},
|
|
{
|
|
"epoch": 0.726530612244898,
|
|
"grad_norm": 0.6002164385003982,
|
|
"learning_rate": 3.999870236808868e-05,
|
|
"loss": 0.2884,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1483706831932068,
|
|
"step": 445,
|
|
"valid_targets_mean": 4109.9,
|
|
"valid_targets_min": 215
|
|
},
|
|
{
|
|
"epoch": 0.7346938775510204,
|
|
"grad_norm": 0.7700094830439378,
|
|
"learning_rate": 3.999760999563491e-05,
|
|
"loss": 0.2935,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14717252552509308,
|
|
"step": 450,
|
|
"valid_targets_mean": 4119.0,
|
|
"valid_targets_min": 2148
|
|
},
|
|
{
|
|
"epoch": 0.7428571428571429,
|
|
"grad_norm": 0.46482998773287904,
|
|
"learning_rate": 3.999618663106761e-05,
|
|
"loss": 0.2933,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1267111450433731,
|
|
"step": 455,
|
|
"valid_targets_mean": 7414.5,
|
|
"valid_targets_min": 1461
|
|
},
|
|
{
|
|
"epoch": 0.7510204081632653,
|
|
"grad_norm": 0.5950339538598457,
|
|
"learning_rate": 3.999443229794572e-05,
|
|
"loss": 0.2951,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14380259811878204,
|
|
"step": 460,
|
|
"valid_targets_mean": 3798.4,
|
|
"valid_targets_min": 1922
|
|
},
|
|
{
|
|
"epoch": 0.7591836734693878,
|
|
"grad_norm": 0.5308130390938092,
|
|
"learning_rate": 3.9992347025306236e-05,
|
|
"loss": 0.2931,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1623890995979309,
|
|
"step": 465,
|
|
"valid_targets_mean": 5759.2,
|
|
"valid_targets_min": 2835
|
|
},
|
|
{
|
|
"epoch": 0.7673469387755102,
|
|
"grad_norm": 0.48879962344124256,
|
|
"learning_rate": 3.9989930847663706e-05,
|
|
"loss": 0.2778,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13271786272525787,
|
|
"step": 470,
|
|
"valid_targets_mean": 5688.9,
|
|
"valid_targets_min": 1937
|
|
},
|
|
{
|
|
"epoch": 0.7755102040816326,
|
|
"grad_norm": 0.5934438944710013,
|
|
"learning_rate": 3.998718380500971e-05,
|
|
"loss": 0.292,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1767587959766388,
|
|
"step": 475,
|
|
"valid_targets_mean": 4823.4,
|
|
"valid_targets_min": 2594
|
|
},
|
|
{
|
|
"epoch": 0.7836734693877551,
|
|
"grad_norm": 0.5617745043664201,
|
|
"learning_rate": 3.998410594281215e-05,
|
|
"loss": 0.2923,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14815029501914978,
|
|
"step": 480,
|
|
"valid_targets_mean": 4843.0,
|
|
"valid_targets_min": 2702
|
|
},
|
|
{
|
|
"epoch": 0.7918367346938775,
|
|
"grad_norm": 0.5280138408881649,
|
|
"learning_rate": 3.998069731201452e-05,
|
|
"loss": 0.2956,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15829509496688843,
|
|
"step": 485,
|
|
"valid_targets_mean": 5016.4,
|
|
"valid_targets_min": 1744
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"grad_norm": 0.5374828402758218,
|
|
"learning_rate": 3.997695796903506e-05,
|
|
"loss": 0.2764,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14506173133850098,
|
|
"step": 490,
|
|
"valid_targets_mean": 5647.2,
|
|
"valid_targets_min": 556
|
|
},
|
|
{
|
|
"epoch": 0.8081632653061225,
|
|
"grad_norm": 0.5363773773173921,
|
|
"learning_rate": 3.997288797576581e-05,
|
|
"loss": 0.2857,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13211554288864136,
|
|
"step": 495,
|
|
"valid_targets_mean": 5134.4,
|
|
"valid_targets_min": 2075
|
|
},
|
|
{
|
|
"epoch": 0.8163265306122449,
|
|
"grad_norm": 0.5371091766380987,
|
|
"learning_rate": 3.996848739957162e-05,
|
|
"loss": 0.2943,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14844098687171936,
|
|
"step": 500,
|
|
"valid_targets_mean": 4823.0,
|
|
"valid_targets_min": 2984
|
|
},
|
|
{
|
|
"epoch": 0.8244897959183674,
|
|
"grad_norm": 0.5553565489257202,
|
|
"learning_rate": 3.996375631328898e-05,
|
|
"loss": 0.3024,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.174277663230896,
|
|
"step": 505,
|
|
"valid_targets_mean": 5254.9,
|
|
"valid_targets_min": 2011
|
|
},
|
|
{
|
|
"epoch": 0.8326530612244898,
|
|
"grad_norm": 0.6386966753151678,
|
|
"learning_rate": 3.995869479522486e-05,
|
|
"loss": 0.3079,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14175033569335938,
|
|
"step": 510,
|
|
"valid_targets_mean": 3835.4,
|
|
"valid_targets_min": 2278
|
|
},
|
|
{
|
|
"epoch": 0.8408163265306122,
|
|
"grad_norm": 0.5488250127766011,
|
|
"learning_rate": 3.99533029291554e-05,
|
|
"loss": 0.2807,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14564122259616852,
|
|
"step": 515,
|
|
"valid_targets_mean": 3774.8,
|
|
"valid_targets_min": 1732
|
|
},
|
|
{
|
|
"epoch": 0.8489795918367347,
|
|
"grad_norm": 0.47254565574720186,
|
|
"learning_rate": 3.994758080432453e-05,
|
|
"loss": 0.2915,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1435089409351349,
|
|
"step": 520,
|
|
"valid_targets_mean": 4876.8,
|
|
"valid_targets_min": 1882
|
|
},
|
|
{
|
|
"epoch": 0.8571428571428571,
|
|
"grad_norm": 0.5007153163086048,
|
|
"learning_rate": 3.994152851544248e-05,
|
|
"loss": 0.287,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1679384410381317,
|
|
"step": 525,
|
|
"valid_targets_mean": 5747.9,
|
|
"valid_targets_min": 1080
|
|
},
|
|
{
|
|
"epoch": 0.8653061224489796,
|
|
"grad_norm": 1.2081947619323428,
|
|
"learning_rate": 3.9935146162684206e-05,
|
|
"loss": 0.2786,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12986889481544495,
|
|
"step": 530,
|
|
"valid_targets_mean": 5772.0,
|
|
"valid_targets_min": 1397
|
|
},
|
|
{
|
|
"epoch": 0.8734693877551021,
|
|
"grad_norm": 0.4993470220190206,
|
|
"learning_rate": 3.992843385168775e-05,
|
|
"loss": 0.2746,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1556703746318817,
|
|
"step": 535,
|
|
"valid_targets_mean": 6776.5,
|
|
"valid_targets_min": 1956
|
|
},
|
|
{
|
|
"epoch": 0.8816326530612245,
|
|
"grad_norm": 0.5776232745040066,
|
|
"learning_rate": 3.99213916935525e-05,
|
|
"loss": 0.2898,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13914015889167786,
|
|
"step": 540,
|
|
"valid_targets_mean": 4373.4,
|
|
"valid_targets_min": 703
|
|
},
|
|
{
|
|
"epoch": 0.889795918367347,
|
|
"grad_norm": 0.4812680988740483,
|
|
"learning_rate": 3.991401980483731e-05,
|
|
"loss": 0.2792,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1518934965133667,
|
|
"step": 545,
|
|
"valid_targets_mean": 6895.9,
|
|
"valid_targets_min": 2243
|
|
},
|
|
{
|
|
"epoch": 0.8979591836734694,
|
|
"grad_norm": 0.5738308130244382,
|
|
"learning_rate": 3.990631830755863e-05,
|
|
"loss": 0.2863,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1788262128829956,
|
|
"step": 550,
|
|
"valid_targets_mean": 4816.4,
|
|
"valid_targets_min": 2138
|
|
},
|
|
{
|
|
"epoch": 0.9061224489795918,
|
|
"grad_norm": 0.5224458825908416,
|
|
"learning_rate": 3.9898287329188426e-05,
|
|
"loss": 0.292,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14131759107112885,
|
|
"step": 555,
|
|
"valid_targets_mean": 4659.0,
|
|
"valid_targets_min": 1993
|
|
},
|
|
{
|
|
"epoch": 0.9142857142857143,
|
|
"grad_norm": 0.5698616740709329,
|
|
"learning_rate": 3.988992700265211e-05,
|
|
"loss": 0.2751,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14461690187454224,
|
|
"step": 560,
|
|
"valid_targets_mean": 3929.5,
|
|
"valid_targets_min": 1063
|
|
},
|
|
{
|
|
"epoch": 0.9224489795918367,
|
|
"grad_norm": 0.543077032885466,
|
|
"learning_rate": 3.988123746632632e-05,
|
|
"loss": 0.2664,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13961417973041534,
|
|
"step": 565,
|
|
"valid_targets_mean": 4097.1,
|
|
"valid_targets_min": 1526
|
|
},
|
|
{
|
|
"epoch": 0.9306122448979591,
|
|
"grad_norm": 0.5536871068210806,
|
|
"learning_rate": 3.9872218864036656e-05,
|
|
"loss": 0.2805,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12596586346626282,
|
|
"step": 570,
|
|
"valid_targets_mean": 4359.4,
|
|
"valid_targets_min": 1569
|
|
},
|
|
{
|
|
"epoch": 0.9387755102040817,
|
|
"grad_norm": 0.5691487205064515,
|
|
"learning_rate": 3.986287134505525e-05,
|
|
"loss": 0.2866,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11891940236091614,
|
|
"step": 575,
|
|
"valid_targets_mean": 4227.9,
|
|
"valid_targets_min": 2328
|
|
},
|
|
{
|
|
"epoch": 0.9469387755102041,
|
|
"grad_norm": 0.5175770297779654,
|
|
"learning_rate": 3.985319506409836e-05,
|
|
"loss": 0.2864,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14197929203510284,
|
|
"step": 580,
|
|
"valid_targets_mean": 4714.2,
|
|
"valid_targets_min": 2851
|
|
},
|
|
{
|
|
"epoch": 0.9551020408163265,
|
|
"grad_norm": 0.4841626589697453,
|
|
"learning_rate": 3.9843190181323744e-05,
|
|
"loss": 0.2805,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1410512924194336,
|
|
"step": 585,
|
|
"valid_targets_mean": 6033.8,
|
|
"valid_targets_min": 2286
|
|
},
|
|
{
|
|
"epoch": 0.963265306122449,
|
|
"grad_norm": 0.5070780388430243,
|
|
"learning_rate": 3.9832856862328074e-05,
|
|
"loss": 0.2883,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14377368986606598,
|
|
"step": 590,
|
|
"valid_targets_mean": 5524.2,
|
|
"valid_targets_min": 2156
|
|
},
|
|
{
|
|
"epoch": 0.9714285714285714,
|
|
"grad_norm": 0.5420150728069583,
|
|
"learning_rate": 3.982219527814413e-05,
|
|
"loss": 0.2788,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16339737176895142,
|
|
"step": 595,
|
|
"valid_targets_mean": 5497.6,
|
|
"valid_targets_min": 2958
|
|
},
|
|
{
|
|
"epoch": 0.9795918367346939,
|
|
"grad_norm": 0.4456547760032798,
|
|
"learning_rate": 3.981120560523802e-05,
|
|
"loss": 0.2731,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13854515552520752,
|
|
"step": 600,
|
|
"valid_targets_mean": 6230.0,
|
|
"valid_targets_min": 2322
|
|
},
|
|
{
|
|
"epoch": 0.9877551020408163,
|
|
"grad_norm": 0.5810811547000726,
|
|
"learning_rate": 3.979988802550622e-05,
|
|
"loss": 0.2832,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1590469479560852,
|
|
"step": 605,
|
|
"valid_targets_mean": 4077.9,
|
|
"valid_targets_min": 2857
|
|
},
|
|
{
|
|
"epoch": 0.9959183673469387,
|
|
"grad_norm": 0.4891857085078456,
|
|
"learning_rate": 3.978824272627262e-05,
|
|
"loss": 0.2689,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11531509459018707,
|
|
"step": 610,
|
|
"valid_targets_mean": 4370.0,
|
|
"valid_targets_min": 1733
|
|
},
|
|
{
|
|
"epoch": 1.003265306122449,
|
|
"grad_norm": 0.5532155013235863,
|
|
"learning_rate": 3.977626990028535e-05,
|
|
"loss": 0.2645,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11366681754589081,
|
|
"step": 615,
|
|
"valid_targets_mean": 3601.6,
|
|
"valid_targets_min": 2097
|
|
},
|
|
{
|
|
"epoch": 1.0114285714285713,
|
|
"grad_norm": 0.45978091826871326,
|
|
"learning_rate": 3.976396974571364e-05,
|
|
"loss": 0.2485,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11303171515464783,
|
|
"step": 620,
|
|
"valid_targets_mean": 6528.6,
|
|
"valid_targets_min": 2089
|
|
},
|
|
{
|
|
"epoch": 1.019591836734694,
|
|
"grad_norm": 0.4523914061946981,
|
|
"learning_rate": 3.9751342466144526e-05,
|
|
"loss": 0.2686,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1494063138961792,
|
|
"step": 625,
|
|
"valid_targets_mean": 7086.5,
|
|
"valid_targets_min": 2442
|
|
},
|
|
{
|
|
"epoch": 1.0277551020408162,
|
|
"grad_norm": 0.544225989094849,
|
|
"learning_rate": 3.9738388270579484e-05,
|
|
"loss": 0.2776,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14501337707042694,
|
|
"step": 630,
|
|
"valid_targets_mean": 4636.4,
|
|
"valid_targets_min": 2459
|
|
},
|
|
{
|
|
"epoch": 1.0359183673469388,
|
|
"grad_norm": 0.4867306661490911,
|
|
"learning_rate": 3.972510737343097e-05,
|
|
"loss": 0.2474,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10430733859539032,
|
|
"step": 635,
|
|
"valid_targets_mean": 4400.8,
|
|
"valid_targets_min": 2464
|
|
},
|
|
{
|
|
"epoch": 1.0440816326530613,
|
|
"grad_norm": 0.4444263422457097,
|
|
"learning_rate": 3.971149999451886e-05,
|
|
"loss": 0.2585,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11613980680704117,
|
|
"step": 640,
|
|
"valid_targets_mean": 5829.5,
|
|
"valid_targets_min": 1063
|
|
},
|
|
{
|
|
"epoch": 1.0522448979591836,
|
|
"grad_norm": 0.5230816472689035,
|
|
"learning_rate": 3.9697566359066805e-05,
|
|
"loss": 0.2661,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.146030992269516,
|
|
"step": 645,
|
|
"valid_targets_mean": 5094.1,
|
|
"valid_targets_min": 2429
|
|
},
|
|
{
|
|
"epoch": 1.0604081632653062,
|
|
"grad_norm": 0.47487568623337845,
|
|
"learning_rate": 3.968330669769857e-05,
|
|
"loss": 0.2544,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1315501481294632,
|
|
"step": 650,
|
|
"valid_targets_mean": 5665.8,
|
|
"valid_targets_min": 2710
|
|
},
|
|
{
|
|
"epoch": 1.0685714285714285,
|
|
"grad_norm": 0.49799195804012064,
|
|
"learning_rate": 3.966872124643411e-05,
|
|
"loss": 0.2703,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1552223563194275,
|
|
"step": 655,
|
|
"valid_targets_mean": 5938.4,
|
|
"valid_targets_min": 1687
|
|
},
|
|
{
|
|
"epoch": 1.076734693877551,
|
|
"grad_norm": 0.5297556483893657,
|
|
"learning_rate": 3.9653810246685746e-05,
|
|
"loss": 0.2701,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13517478108406067,
|
|
"step": 660,
|
|
"valid_targets_mean": 4919.1,
|
|
"valid_targets_min": 1791
|
|
},
|
|
{
|
|
"epoch": 1.0848979591836734,
|
|
"grad_norm": 0.5601908106981581,
|
|
"learning_rate": 3.963857394525413e-05,
|
|
"loss": 0.2726,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1491164267063141,
|
|
"step": 665,
|
|
"valid_targets_mean": 5137.8,
|
|
"valid_targets_min": 2264
|
|
},
|
|
{
|
|
"epoch": 1.093061224489796,
|
|
"grad_norm": 0.4525683784590435,
|
|
"learning_rate": 3.96230125943242e-05,
|
|
"loss": 0.2559,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.113084577023983,
|
|
"step": 670,
|
|
"valid_targets_mean": 5150.8,
|
|
"valid_targets_min": 2168
|
|
},
|
|
{
|
|
"epoch": 1.1012244897959185,
|
|
"grad_norm": 0.519392037612358,
|
|
"learning_rate": 3.960712645146094e-05,
|
|
"loss": 0.2717,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13259488344192505,
|
|
"step": 675,
|
|
"valid_targets_mean": 4962.0,
|
|
"valid_targets_min": 2038
|
|
},
|
|
{
|
|
"epoch": 1.1093877551020408,
|
|
"grad_norm": 0.5539538809708434,
|
|
"learning_rate": 3.959091577960517e-05,
|
|
"loss": 0.267,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14472344517707825,
|
|
"step": 680,
|
|
"valid_targets_mean": 4567.8,
|
|
"valid_targets_min": 2960
|
|
},
|
|
{
|
|
"epoch": 1.1175510204081633,
|
|
"grad_norm": 0.4887317566930953,
|
|
"learning_rate": 3.957438084706919e-05,
|
|
"loss": 0.2567,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10747165977954865,
|
|
"step": 685,
|
|
"valid_targets_mean": 4679.0,
|
|
"valid_targets_min": 2378
|
|
},
|
|
{
|
|
"epoch": 1.1257142857142857,
|
|
"grad_norm": 0.555310737006963,
|
|
"learning_rate": 3.955752192753231e-05,
|
|
"loss": 0.2669,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12046556919813156,
|
|
"step": 690,
|
|
"valid_targets_mean": 4157.1,
|
|
"valid_targets_min": 2181
|
|
},
|
|
{
|
|
"epoch": 1.1338775510204082,
|
|
"grad_norm": 0.5121166772677435,
|
|
"learning_rate": 3.954033930003634e-05,
|
|
"loss": 0.2782,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1369376927614212,
|
|
"step": 695,
|
|
"valid_targets_mean": 5442.1,
|
|
"valid_targets_min": 3151
|
|
},
|
|
{
|
|
"epoch": 1.1420408163265305,
|
|
"grad_norm": 0.531503784823624,
|
|
"learning_rate": 3.9522833248980995e-05,
|
|
"loss": 0.2636,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11154456436634064,
|
|
"step": 700,
|
|
"valid_targets_mean": 4064.0,
|
|
"valid_targets_min": 2285
|
|
},
|
|
{
|
|
"epoch": 1.150204081632653,
|
|
"grad_norm": 0.5489557045500898,
|
|
"learning_rate": 3.950500406411913e-05,
|
|
"loss": 0.2559,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12069879472255707,
|
|
"step": 705,
|
|
"valid_targets_mean": 3922.4,
|
|
"valid_targets_min": 160
|
|
},
|
|
{
|
|
"epoch": 1.1583673469387756,
|
|
"grad_norm": 0.5440676931905495,
|
|
"learning_rate": 3.948685204055199e-05,
|
|
"loss": 0.2745,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15134495496749878,
|
|
"step": 710,
|
|
"valid_targets_mean": 4761.1,
|
|
"valid_targets_min": 1830
|
|
},
|
|
{
|
|
"epoch": 1.166530612244898,
|
|
"grad_norm": 0.4377127575255182,
|
|
"learning_rate": 3.946837747872431e-05,
|
|
"loss": 0.2773,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12781107425689697,
|
|
"step": 715,
|
|
"valid_targets_mean": 6668.2,
|
|
"valid_targets_min": 2050
|
|
},
|
|
{
|
|
"epoch": 1.1746938775510205,
|
|
"grad_norm": 0.46747115033774317,
|
|
"learning_rate": 3.9449580684419346e-05,
|
|
"loss": 0.2403,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1349138617515564,
|
|
"step": 720,
|
|
"valid_targets_mean": 7172.1,
|
|
"valid_targets_min": 2796
|
|
},
|
|
{
|
|
"epoch": 1.1828571428571428,
|
|
"grad_norm": 0.46477551009936313,
|
|
"learning_rate": 3.943046196875381e-05,
|
|
"loss": 0.2483,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09906481206417084,
|
|
"step": 725,
|
|
"valid_targets_mean": 3456.6,
|
|
"valid_targets_min": 971
|
|
},
|
|
{
|
|
"epoch": 1.1910204081632654,
|
|
"grad_norm": 0.5174344184504699,
|
|
"learning_rate": 3.941102164817272e-05,
|
|
"loss": 0.2741,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1233099177479744,
|
|
"step": 730,
|
|
"valid_targets_mean": 4068.8,
|
|
"valid_targets_min": 2132
|
|
},
|
|
{
|
|
"epoch": 1.1991836734693877,
|
|
"grad_norm": 0.4559852424556182,
|
|
"learning_rate": 3.939126004444417e-05,
|
|
"loss": 0.2581,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12599629163742065,
|
|
"step": 735,
|
|
"valid_targets_mean": 5618.4,
|
|
"valid_targets_min": 1192
|
|
},
|
|
{
|
|
"epoch": 1.2073469387755102,
|
|
"grad_norm": 0.6236070648619523,
|
|
"learning_rate": 3.9371177484654e-05,
|
|
"loss": 0.2621,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1476593315601349,
|
|
"step": 740,
|
|
"valid_targets_mean": 4005.5,
|
|
"valid_targets_min": 1569
|
|
},
|
|
{
|
|
"epoch": 1.2155102040816326,
|
|
"grad_norm": 0.4113790501215956,
|
|
"learning_rate": 3.9350774301200374e-05,
|
|
"loss": 0.256,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10646657645702362,
|
|
"step": 745,
|
|
"valid_targets_mean": 5235.4,
|
|
"valid_targets_min": 2214
|
|
},
|
|
{
|
|
"epoch": 1.2236734693877551,
|
|
"grad_norm": 0.4400441587321837,
|
|
"learning_rate": 3.933005083178828e-05,
|
|
"loss": 0.2486,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09953781217336655,
|
|
"step": 750,
|
|
"valid_targets_mean": 4607.5,
|
|
"valid_targets_min": 1155
|
|
},
|
|
{
|
|
"epoch": 1.2318367346938777,
|
|
"grad_norm": 0.49581897488480603,
|
|
"learning_rate": 3.930900741942397e-05,
|
|
"loss": 0.2696,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12676215171813965,
|
|
"step": 755,
|
|
"valid_targets_mean": 4551.0,
|
|
"valid_targets_min": 845
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"grad_norm": 0.44448319184762836,
|
|
"learning_rate": 3.928764441240922e-05,
|
|
"loss": 0.27,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13121074438095093,
|
|
"step": 760,
|
|
"valid_targets_mean": 6703.2,
|
|
"valid_targets_min": 3792
|
|
},
|
|
{
|
|
"epoch": 1.2481632653061225,
|
|
"grad_norm": 0.6007678431285308,
|
|
"learning_rate": 3.9265962164335656e-05,
|
|
"loss": 0.2669,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12405592948198318,
|
|
"step": 765,
|
|
"valid_targets_mean": 3433.2,
|
|
"valid_targets_min": 574
|
|
},
|
|
{
|
|
"epoch": 1.2563265306122449,
|
|
"grad_norm": 0.5103975132542746,
|
|
"learning_rate": 3.92439610340788e-05,
|
|
"loss": 0.2558,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12031939625740051,
|
|
"step": 770,
|
|
"valid_targets_mean": 4843.1,
|
|
"valid_targets_min": 2665
|
|
},
|
|
{
|
|
"epoch": 1.2644897959183674,
|
|
"grad_norm": 0.5233070874477446,
|
|
"learning_rate": 3.92216413857922e-05,
|
|
"loss": 0.2397,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1319732964038849,
|
|
"step": 775,
|
|
"valid_targets_mean": 4303.9,
|
|
"valid_targets_min": 1636
|
|
},
|
|
{
|
|
"epoch": 1.2726530612244897,
|
|
"grad_norm": 0.45711695489790305,
|
|
"learning_rate": 3.9199003588901375e-05,
|
|
"loss": 0.2744,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12586191296577454,
|
|
"step": 780,
|
|
"valid_targets_mean": 5050.6,
|
|
"valid_targets_min": 1917
|
|
},
|
|
{
|
|
"epoch": 1.2808163265306123,
|
|
"grad_norm": 0.4912001376959002,
|
|
"learning_rate": 3.917604801809774e-05,
|
|
"loss": 0.2691,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13443773984909058,
|
|
"step": 785,
|
|
"valid_targets_mean": 4459.2,
|
|
"valid_targets_min": 2587
|
|
},
|
|
{
|
|
"epoch": 1.2889795918367346,
|
|
"grad_norm": 0.5195769905666631,
|
|
"learning_rate": 3.9152775053332305e-05,
|
|
"loss": 0.2611,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1408383846282959,
|
|
"step": 790,
|
|
"valid_targets_mean": 4617.0,
|
|
"valid_targets_min": 1967
|
|
},
|
|
{
|
|
"epoch": 1.2971428571428572,
|
|
"grad_norm": 0.5616503318329411,
|
|
"learning_rate": 3.912918507980952e-05,
|
|
"loss": 0.2589,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14663545787334442,
|
|
"step": 795,
|
|
"valid_targets_mean": 4577.2,
|
|
"valid_targets_min": 917
|
|
},
|
|
{
|
|
"epoch": 1.3053061224489797,
|
|
"grad_norm": 0.5302531747449998,
|
|
"learning_rate": 3.91052784879808e-05,
|
|
"loss": 0.2609,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12137632071971893,
|
|
"step": 800,
|
|
"valid_targets_mean": 3798.2,
|
|
"valid_targets_min": 1148
|
|
},
|
|
{
|
|
"epoch": 1.313469387755102,
|
|
"grad_norm": 0.5111215932189485,
|
|
"learning_rate": 3.9081055673538093e-05,
|
|
"loss": 0.2711,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12059567123651505,
|
|
"step": 805,
|
|
"valid_targets_mean": 4635.2,
|
|
"valid_targets_min": 385
|
|
},
|
|
{
|
|
"epoch": 1.3216326530612246,
|
|
"grad_norm": 0.48693696302052697,
|
|
"learning_rate": 3.9056517037407334e-05,
|
|
"loss": 0.2571,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1377648413181305,
|
|
"step": 810,
|
|
"valid_targets_mean": 5086.8,
|
|
"valid_targets_min": 3186
|
|
},
|
|
{
|
|
"epoch": 1.329795918367347,
|
|
"grad_norm": 0.5020090791110657,
|
|
"learning_rate": 3.903166298574181e-05,
|
|
"loss": 0.2583,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12303146719932556,
|
|
"step": 815,
|
|
"valid_targets_mean": 4312.8,
|
|
"valid_targets_min": 1772
|
|
},
|
|
{
|
|
"epoch": 1.3379591836734694,
|
|
"grad_norm": 0.5428165968206224,
|
|
"learning_rate": 3.900649392991545e-05,
|
|
"loss": 0.2646,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13852190971374512,
|
|
"step": 820,
|
|
"valid_targets_mean": 4192.0,
|
|
"valid_targets_min": 736
|
|
},
|
|
{
|
|
"epoch": 1.3461224489795918,
|
|
"grad_norm": 0.507038409712324,
|
|
"learning_rate": 3.898101028651597e-05,
|
|
"loss": 0.2741,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11246784031391144,
|
|
"step": 825,
|
|
"valid_targets_mean": 4226.5,
|
|
"valid_targets_min": 1760
|
|
},
|
|
{
|
|
"epoch": 1.3542857142857143,
|
|
"grad_norm": 0.4846864352304685,
|
|
"learning_rate": 3.895521247733803e-05,
|
|
"loss": 0.2617,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15112227201461792,
|
|
"step": 830,
|
|
"valid_targets_mean": 4990.0,
|
|
"valid_targets_min": 2134
|
|
},
|
|
{
|
|
"epoch": 1.3624489795918366,
|
|
"grad_norm": 0.5498149576955207,
|
|
"learning_rate": 3.892910092937622e-05,
|
|
"loss": 0.249,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1215394139289856,
|
|
"step": 835,
|
|
"valid_targets_mean": 4482.6,
|
|
"valid_targets_min": 3359
|
|
},
|
|
{
|
|
"epoch": 1.3706122448979592,
|
|
"grad_norm": 0.506582687397877,
|
|
"learning_rate": 3.890267607481801e-05,
|
|
"loss": 0.2603,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14072394371032715,
|
|
"step": 840,
|
|
"valid_targets_mean": 4253.4,
|
|
"valid_targets_min": 2900
|
|
},
|
|
{
|
|
"epoch": 1.3787755102040817,
|
|
"grad_norm": 0.5310004085942591,
|
|
"learning_rate": 3.887593835103659e-05,
|
|
"loss": 0.2741,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11263677477836609,
|
|
"step": 845,
|
|
"valid_targets_mean": 3621.0,
|
|
"valid_targets_min": 2181
|
|
},
|
|
{
|
|
"epoch": 1.386938775510204,
|
|
"grad_norm": 0.4339415251534723,
|
|
"learning_rate": 3.8848888200583637e-05,
|
|
"loss": 0.2703,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10042256116867065,
|
|
"step": 850,
|
|
"valid_targets_mean": 4672.8,
|
|
"valid_targets_min": 2327
|
|
},
|
|
{
|
|
"epoch": 1.3951020408163266,
|
|
"grad_norm": 0.5441893667906397,
|
|
"learning_rate": 3.882152607118197e-05,
|
|
"loss": 0.2695,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12515777349472046,
|
|
"step": 855,
|
|
"valid_targets_mean": 3982.5,
|
|
"valid_targets_min": 2314
|
|
},
|
|
{
|
|
"epoch": 1.403265306122449,
|
|
"grad_norm": 0.503757452633956,
|
|
"learning_rate": 3.879385241571817e-05,
|
|
"loss": 0.2583,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12633231282234192,
|
|
"step": 860,
|
|
"valid_targets_mean": 5279.4,
|
|
"valid_targets_min": 1544
|
|
},
|
|
{
|
|
"epoch": 1.4114285714285715,
|
|
"grad_norm": 0.4894760287198677,
|
|
"learning_rate": 3.876586769223505e-05,
|
|
"loss": 0.249,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12801289558410645,
|
|
"step": 865,
|
|
"valid_targets_mean": 4574.9,
|
|
"valid_targets_min": 2731
|
|
},
|
|
{
|
|
"epoch": 1.4195918367346938,
|
|
"grad_norm": 0.5146349691454388,
|
|
"learning_rate": 3.87375723639241e-05,
|
|
"loss": 0.2586,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10573422908782959,
|
|
"step": 870,
|
|
"valid_targets_mean": 5117.6,
|
|
"valid_targets_min": 726
|
|
},
|
|
{
|
|
"epoch": 1.4277551020408163,
|
|
"grad_norm": 0.41242003835171465,
|
|
"learning_rate": 3.8708966899117815e-05,
|
|
"loss": 0.2525,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15078118443489075,
|
|
"step": 875,
|
|
"valid_targets_mean": 8537.8,
|
|
"valid_targets_min": 4366
|
|
},
|
|
{
|
|
"epoch": 1.4359183673469387,
|
|
"grad_norm": 0.45629679864975425,
|
|
"learning_rate": 3.868005177128193e-05,
|
|
"loss": 0.2422,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10194846987724304,
|
|
"step": 880,
|
|
"valid_targets_mean": 3460.2,
|
|
"valid_targets_min": 469
|
|
},
|
|
{
|
|
"epoch": 1.4440816326530612,
|
|
"grad_norm": 0.5171964688688151,
|
|
"learning_rate": 3.8650827459007615e-05,
|
|
"loss": 0.274,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14588257670402527,
|
|
"step": 885,
|
|
"valid_targets_mean": 5561.6,
|
|
"valid_targets_min": 3511
|
|
},
|
|
{
|
|
"epoch": 1.4522448979591838,
|
|
"grad_norm": 0.4084420445366524,
|
|
"learning_rate": 3.8621294446003504e-05,
|
|
"loss": 0.2664,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15651914477348328,
|
|
"step": 890,
|
|
"valid_targets_mean": 7917.4,
|
|
"valid_targets_min": 3301
|
|
},
|
|
{
|
|
"epoch": 1.460408163265306,
|
|
"grad_norm": 0.5083640639561413,
|
|
"learning_rate": 3.859145322108774e-05,
|
|
"loss": 0.2375,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1252627670764923,
|
|
"step": 895,
|
|
"valid_targets_mean": 3932.4,
|
|
"valid_targets_min": 1273
|
|
},
|
|
{
|
|
"epoch": 1.4685714285714286,
|
|
"grad_norm": 0.5443678273198022,
|
|
"learning_rate": 3.856130427817984e-05,
|
|
"loss": 0.2744,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13401755690574646,
|
|
"step": 900,
|
|
"valid_targets_mean": 4187.8,
|
|
"valid_targets_min": 1703
|
|
},
|
|
{
|
|
"epoch": 1.476734693877551,
|
|
"grad_norm": 0.46848277024034957,
|
|
"learning_rate": 3.853084811629257e-05,
|
|
"loss": 0.2561,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13730978965759277,
|
|
"step": 905,
|
|
"valid_targets_mean": 5213.6,
|
|
"valid_targets_min": 1923
|
|
},
|
|
{
|
|
"epoch": 1.4848979591836735,
|
|
"grad_norm": 0.5269763713171643,
|
|
"learning_rate": 3.8500085239523624e-05,
|
|
"loss": 0.2484,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14742831885814667,
|
|
"step": 910,
|
|
"valid_targets_mean": 5609.5,
|
|
"valid_targets_min": 1810
|
|
},
|
|
{
|
|
"epoch": 1.4930612244897958,
|
|
"grad_norm": 0.4918260517625046,
|
|
"learning_rate": 3.846901615704734e-05,
|
|
"loss": 0.2398,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12141484022140503,
|
|
"step": 915,
|
|
"valid_targets_mean": 4149.9,
|
|
"valid_targets_min": 755
|
|
},
|
|
{
|
|
"epoch": 1.5012244897959184,
|
|
"grad_norm": 0.48992607672168265,
|
|
"learning_rate": 3.843764138310622e-05,
|
|
"loss": 0.2615,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16299812495708466,
|
|
"step": 920,
|
|
"valid_targets_mean": 5494.4,
|
|
"valid_targets_min": 1927
|
|
},
|
|
{
|
|
"epoch": 1.5093877551020407,
|
|
"grad_norm": 0.47445400914699815,
|
|
"learning_rate": 3.840596143700246e-05,
|
|
"loss": 0.2483,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13037662208080292,
|
|
"step": 925,
|
|
"valid_targets_mean": 5328.8,
|
|
"valid_targets_min": 2703
|
|
},
|
|
{
|
|
"epoch": 1.5175510204081633,
|
|
"grad_norm": 0.5085959632245128,
|
|
"learning_rate": 3.8373976843089355e-05,
|
|
"loss": 0.2527,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14229968190193176,
|
|
"step": 930,
|
|
"valid_targets_mean": 4230.6,
|
|
"valid_targets_min": 503
|
|
},
|
|
{
|
|
"epoch": 1.5257142857142858,
|
|
"grad_norm": 0.4982028747496671,
|
|
"learning_rate": 3.8341688130762574e-05,
|
|
"loss": 0.2682,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14528988301753998,
|
|
"step": 935,
|
|
"valid_targets_mean": 5524.4,
|
|
"valid_targets_min": 1399
|
|
},
|
|
{
|
|
"epoch": 1.5338775510204081,
|
|
"grad_norm": 0.5614943316039491,
|
|
"learning_rate": 3.830909583445143e-05,
|
|
"loss": 0.257,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14664828777313232,
|
|
"step": 940,
|
|
"valid_targets_mean": 3903.5,
|
|
"valid_targets_min": 720
|
|
},
|
|
{
|
|
"epoch": 1.5420408163265305,
|
|
"grad_norm": 0.4217324955618739,
|
|
"learning_rate": 3.827620049361004e-05,
|
|
"loss": 0.2681,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13761885464191437,
|
|
"step": 945,
|
|
"valid_targets_mean": 4984.2,
|
|
"valid_targets_min": 2669
|
|
},
|
|
{
|
|
"epoch": 1.550204081632653,
|
|
"grad_norm": 0.5031297388057706,
|
|
"learning_rate": 3.82430026527084e-05,
|
|
"loss": 0.2613,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11942078918218613,
|
|
"step": 950,
|
|
"valid_targets_mean": 4265.5,
|
|
"valid_targets_min": 2414
|
|
},
|
|
{
|
|
"epoch": 1.5583673469387755,
|
|
"grad_norm": 0.594624788084472,
|
|
"learning_rate": 3.8209502861223344e-05,
|
|
"loss": 0.2737,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14711017906665802,
|
|
"step": 955,
|
|
"valid_targets_mean": 4126.1,
|
|
"valid_targets_min": 1653
|
|
},
|
|
{
|
|
"epoch": 1.566530612244898,
|
|
"grad_norm": 0.5147712518084654,
|
|
"learning_rate": 3.817570167362947e-05,
|
|
"loss": 0.2552,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11636977642774582,
|
|
"step": 960,
|
|
"valid_targets_mean": 4158.8,
|
|
"valid_targets_min": 533
|
|
},
|
|
{
|
|
"epoch": 1.5746938775510204,
|
|
"grad_norm": 0.49811050757421693,
|
|
"learning_rate": 3.814159964938996e-05,
|
|
"loss": 0.2476,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09341903030872345,
|
|
"step": 965,
|
|
"valid_targets_mean": 4023.4,
|
|
"valid_targets_min": 1942
|
|
},
|
|
{
|
|
"epoch": 1.5828571428571427,
|
|
"grad_norm": 0.4865312743365694,
|
|
"learning_rate": 3.810719735294731e-05,
|
|
"loss": 0.2548,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11299189925193787,
|
|
"step": 970,
|
|
"valid_targets_mean": 4597.5,
|
|
"valid_targets_min": 1141
|
|
},
|
|
{
|
|
"epoch": 1.5910204081632653,
|
|
"grad_norm": 0.5076603149396371,
|
|
"learning_rate": 3.8072495353714025e-05,
|
|
"loss": 0.2542,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16264669597148895,
|
|
"step": 975,
|
|
"valid_targets_mean": 5229.2,
|
|
"valid_targets_min": 1843
|
|
},
|
|
{
|
|
"epoch": 1.5991836734693878,
|
|
"grad_norm": 0.5194180616689766,
|
|
"learning_rate": 3.8037494226063126e-05,
|
|
"loss": 0.2603,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15082842111587524,
|
|
"step": 980,
|
|
"valid_targets_mean": 4919.4,
|
|
"valid_targets_min": 2627
|
|
},
|
|
{
|
|
"epoch": 1.6073469387755102,
|
|
"grad_norm": 0.596887240652063,
|
|
"learning_rate": 3.800219454931871e-05,
|
|
"loss": 0.271,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1672152429819107,
|
|
"step": 985,
|
|
"valid_targets_mean": 4447.2,
|
|
"valid_targets_min": 2166
|
|
},
|
|
{
|
|
"epoch": 1.6155102040816327,
|
|
"grad_norm": 0.5350107457946114,
|
|
"learning_rate": 3.7966596907746344e-05,
|
|
"loss": 0.2595,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1205388754606247,
|
|
"step": 990,
|
|
"valid_targets_mean": 3446.4,
|
|
"valid_targets_min": 1205
|
|
},
|
|
{
|
|
"epoch": 1.623673469387755,
|
|
"grad_norm": 0.5081950112157592,
|
|
"learning_rate": 3.793070189054335e-05,
|
|
"loss": 0.2548,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12550537288188934,
|
|
"step": 995,
|
|
"valid_targets_mean": 4020.2,
|
|
"valid_targets_min": 2696
|
|
},
|
|
{
|
|
"epoch": 1.6318367346938776,
|
|
"grad_norm": 0.5318057562656638,
|
|
"learning_rate": 3.789451009182911e-05,
|
|
"loss": 0.2688,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13900066912174225,
|
|
"step": 1000,
|
|
"valid_targets_mean": 4368.8,
|
|
"valid_targets_min": 2759
|
|
},
|
|
{
|
|
"epoch": 1.6400000000000001,
|
|
"grad_norm": 0.510742602562625,
|
|
"learning_rate": 3.78580221106352e-05,
|
|
"loss": 0.241,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1156846210360527,
|
|
"step": 1005,
|
|
"valid_targets_mean": 3944.9,
|
|
"valid_targets_min": 1110
|
|
},
|
|
{
|
|
"epoch": 1.6481632653061224,
|
|
"grad_norm": 0.4750127711790795,
|
|
"learning_rate": 3.78212385508955e-05,
|
|
"loss": 0.2569,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16291871666908264,
|
|
"step": 1010,
|
|
"valid_targets_mean": 6779.4,
|
|
"valid_targets_min": 3330
|
|
},
|
|
{
|
|
"epoch": 1.6563265306122448,
|
|
"grad_norm": 0.478006168965345,
|
|
"learning_rate": 3.778416002143617e-05,
|
|
"loss": 0.2678,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1426718533039093,
|
|
"step": 1015,
|
|
"valid_targets_mean": 6106.1,
|
|
"valid_targets_min": 2608
|
|
},
|
|
{
|
|
"epoch": 1.6644897959183673,
|
|
"grad_norm": 0.47169805607015175,
|
|
"learning_rate": 3.774678713596559e-05,
|
|
"loss": 0.2459,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11533847451210022,
|
|
"step": 1020,
|
|
"valid_targets_mean": 5635.5,
|
|
"valid_targets_min": 2118
|
|
},
|
|
{
|
|
"epoch": 1.6726530612244899,
|
|
"grad_norm": 0.5401353076658538,
|
|
"learning_rate": 3.7709120513064196e-05,
|
|
"loss": 0.2655,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.148171067237854,
|
|
"step": 1025,
|
|
"valid_targets_mean": 5144.6,
|
|
"valid_targets_min": 1841
|
|
},
|
|
{
|
|
"epoch": 1.6808163265306122,
|
|
"grad_norm": 0.5251628231233804,
|
|
"learning_rate": 3.767116077617426e-05,
|
|
"loss": 0.2589,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12557116150856018,
|
|
"step": 1030,
|
|
"valid_targets_mean": 4335.6,
|
|
"valid_targets_min": 2394
|
|
},
|
|
{
|
|
"epoch": 1.6889795918367347,
|
|
"grad_norm": 0.4650011640177102,
|
|
"learning_rate": 3.763290855358952e-05,
|
|
"loss": 0.2527,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15282702445983887,
|
|
"step": 1035,
|
|
"valid_targets_mean": 6148.4,
|
|
"valid_targets_min": 552
|
|
},
|
|
{
|
|
"epoch": 1.697142857142857,
|
|
"grad_norm": 0.47198000057835904,
|
|
"learning_rate": 3.759436447844485e-05,
|
|
"loss": 0.2496,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09859016537666321,
|
|
"step": 1040,
|
|
"valid_targets_mean": 4540.5,
|
|
"valid_targets_min": 2642
|
|
},
|
|
{
|
|
"epoch": 1.7053061224489796,
|
|
"grad_norm": 0.550851013931262,
|
|
"learning_rate": 3.7555529188705716e-05,
|
|
"loss": 0.2622,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1216202899813652,
|
|
"step": 1045,
|
|
"valid_targets_mean": 3473.2,
|
|
"valid_targets_min": 2111
|
|
},
|
|
{
|
|
"epoch": 1.7134693877551022,
|
|
"grad_norm": 0.37040303534509933,
|
|
"learning_rate": 3.751640332715768e-05,
|
|
"loss": 0.2394,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10425127297639847,
|
|
"step": 1050,
|
|
"valid_targets_mean": 6519.0,
|
|
"valid_targets_min": 3227
|
|
},
|
|
{
|
|
"epoch": 1.7216326530612245,
|
|
"grad_norm": 0.53785718338026,
|
|
"learning_rate": 3.74769875413957e-05,
|
|
"loss": 0.2453,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13041691482067108,
|
|
"step": 1055,
|
|
"valid_targets_mean": 4362.1,
|
|
"valid_targets_min": 2065
|
|
},
|
|
{
|
|
"epoch": 1.7297959183673468,
|
|
"grad_norm": 0.440296077925039,
|
|
"learning_rate": 3.7437282483813455e-05,
|
|
"loss": 0.2516,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1237720474600792,
|
|
"step": 1060,
|
|
"valid_targets_mean": 5525.8,
|
|
"valid_targets_min": 2969
|
|
},
|
|
{
|
|
"epoch": 1.7379591836734694,
|
|
"grad_norm": 0.4894636219664195,
|
|
"learning_rate": 3.739728881159252e-05,
|
|
"loss": 0.2542,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1411234736442566,
|
|
"step": 1065,
|
|
"valid_targets_mean": 4698.5,
|
|
"valid_targets_min": 2461
|
|
},
|
|
{
|
|
"epoch": 1.746122448979592,
|
|
"grad_norm": 0.44036148587883767,
|
|
"learning_rate": 3.735700718669151e-05,
|
|
"loss": 0.2481,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13445939123630524,
|
|
"step": 1070,
|
|
"valid_targets_mean": 6161.1,
|
|
"valid_targets_min": 2236
|
|
},
|
|
{
|
|
"epoch": 1.7542857142857144,
|
|
"grad_norm": 0.5008003842103482,
|
|
"learning_rate": 3.73164382758351e-05,
|
|
"loss": 0.2649,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14120852947235107,
|
|
"step": 1075,
|
|
"valid_targets_mean": 5492.5,
|
|
"valid_targets_min": 3082
|
|
},
|
|
{
|
|
"epoch": 1.7624489795918368,
|
|
"grad_norm": 0.6470980556874778,
|
|
"learning_rate": 3.727558275050301e-05,
|
|
"loss": 0.2657,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15491387248039246,
|
|
"step": 1080,
|
|
"valid_targets_mean": 6985.4,
|
|
"valid_targets_min": 1928
|
|
},
|
|
{
|
|
"epoch": 1.770612244897959,
|
|
"grad_norm": 0.4683575217814649,
|
|
"learning_rate": 3.723444128691888e-05,
|
|
"loss": 0.2575,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13208508491516113,
|
|
"step": 1085,
|
|
"valid_targets_mean": 4950.8,
|
|
"valid_targets_min": 2287
|
|
},
|
|
{
|
|
"epoch": 1.7787755102040816,
|
|
"grad_norm": 0.36384695238620846,
|
|
"learning_rate": 3.71930145660391e-05,
|
|
"loss": 0.2479,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06327690929174423,
|
|
"step": 1090,
|
|
"valid_targets_mean": 3406.9,
|
|
"valid_targets_min": 2404
|
|
},
|
|
{
|
|
"epoch": 1.7869387755102042,
|
|
"grad_norm": 0.46447438386840667,
|
|
"learning_rate": 3.715130327354148e-05,
|
|
"loss": 0.2459,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12327778339385986,
|
|
"step": 1095,
|
|
"valid_targets_mean": 4916.1,
|
|
"valid_targets_min": 1133
|
|
},
|
|
{
|
|
"epoch": 1.7951020408163265,
|
|
"grad_norm": 0.4993865990758053,
|
|
"learning_rate": 3.7109308099813986e-05,
|
|
"loss": 0.2477,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13568077981472015,
|
|
"step": 1100,
|
|
"valid_targets_mean": 5312.5,
|
|
"valid_targets_min": 2093
|
|
},
|
|
{
|
|
"epoch": 1.8032653061224488,
|
|
"grad_norm": 0.45257355005085736,
|
|
"learning_rate": 3.706702973994323e-05,
|
|
"loss": 0.2567,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1562289148569107,
|
|
"step": 1105,
|
|
"valid_targets_mean": 7640.9,
|
|
"valid_targets_min": 5005
|
|
},
|
|
{
|
|
"epoch": 1.8114285714285714,
|
|
"grad_norm": 0.4722207457039624,
|
|
"learning_rate": 3.702446889370303e-05,
|
|
"loss": 0.2476,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11828452348709106,
|
|
"step": 1110,
|
|
"valid_targets_mean": 4442.2,
|
|
"valid_targets_min": 1883
|
|
},
|
|
{
|
|
"epoch": 1.819591836734694,
|
|
"grad_norm": 0.4484385738226761,
|
|
"learning_rate": 3.698162626554279e-05,
|
|
"loss": 0.2478,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13094788789749146,
|
|
"step": 1115,
|
|
"valid_targets_mean": 6618.0,
|
|
"valid_targets_min": 2405
|
|
},
|
|
{
|
|
"epoch": 1.8277551020408165,
|
|
"grad_norm": 0.4955177645979594,
|
|
"learning_rate": 3.693850256457585e-05,
|
|
"loss": 0.2385,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16945160925388336,
|
|
"step": 1120,
|
|
"valid_targets_mean": 6220.6,
|
|
"valid_targets_min": 3866
|
|
},
|
|
{
|
|
"epoch": 1.8359183673469388,
|
|
"grad_norm": 0.5870278446195637,
|
|
"learning_rate": 3.689509850456774e-05,
|
|
"loss": 0.2372,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11503128707408905,
|
|
"step": 1125,
|
|
"valid_targets_mean": 4211.6,
|
|
"valid_targets_min": 2709
|
|
},
|
|
{
|
|
"epoch": 1.8440816326530611,
|
|
"grad_norm": 0.5875782256134718,
|
|
"learning_rate": 3.685141480392441e-05,
|
|
"loss": 0.2538,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10766669362783432,
|
|
"step": 1130,
|
|
"valid_targets_mean": 3444.6,
|
|
"valid_targets_min": 1349
|
|
},
|
|
{
|
|
"epoch": 1.8522448979591837,
|
|
"grad_norm": 0.43013135938747526,
|
|
"learning_rate": 3.680745218568026e-05,
|
|
"loss": 0.2475,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1484908163547516,
|
|
"step": 1135,
|
|
"valid_targets_mean": 7711.4,
|
|
"valid_targets_min": 1330
|
|
},
|
|
{
|
|
"epoch": 1.8604081632653062,
|
|
"grad_norm": 0.4031243286403073,
|
|
"learning_rate": 3.676321137748626e-05,
|
|
"loss": 0.25,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11534357070922852,
|
|
"step": 1140,
|
|
"valid_targets_mean": 6376.8,
|
|
"valid_targets_min": 1852
|
|
},
|
|
{
|
|
"epoch": 1.8685714285714285,
|
|
"grad_norm": 0.4706254490578847,
|
|
"learning_rate": 3.671869311159783e-05,
|
|
"loss": 0.2548,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1212092861533165,
|
|
"step": 1145,
|
|
"valid_targets_mean": 5040.8,
|
|
"valid_targets_min": 1542
|
|
},
|
|
{
|
|
"epoch": 1.8767346938775509,
|
|
"grad_norm": 0.3795817061920972,
|
|
"learning_rate": 3.6673898124862776e-05,
|
|
"loss": 0.2519,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13475094735622406,
|
|
"step": 1150,
|
|
"valid_targets_mean": 8024.8,
|
|
"valid_targets_min": 2130
|
|
},
|
|
{
|
|
"epoch": 1.8848979591836734,
|
|
"grad_norm": 0.5252621439889926,
|
|
"learning_rate": 3.662882715870907e-05,
|
|
"loss": 0.2687,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12451595067977905,
|
|
"step": 1155,
|
|
"valid_targets_mean": 4560.5,
|
|
"valid_targets_min": 3147
|
|
},
|
|
{
|
|
"epoch": 1.893061224489796,
|
|
"grad_norm": 0.5148704957978237,
|
|
"learning_rate": 3.658348095913257e-05,
|
|
"loss": 0.2469,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1368659883737564,
|
|
"step": 1160,
|
|
"valid_targets_mean": 4340.2,
|
|
"valid_targets_min": 1934
|
|
},
|
|
{
|
|
"epoch": 1.9012244897959185,
|
|
"grad_norm": 0.6121916844321755,
|
|
"learning_rate": 3.6537860276684686e-05,
|
|
"loss": 0.2402,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12297011911869049,
|
|
"step": 1165,
|
|
"valid_targets_mean": 5257.2,
|
|
"valid_targets_min": 2234
|
|
},
|
|
{
|
|
"epoch": 1.9093877551020408,
|
|
"grad_norm": 0.5724642448538444,
|
|
"learning_rate": 3.6491965866459966e-05,
|
|
"loss": 0.2605,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12487930059432983,
|
|
"step": 1170,
|
|
"valid_targets_mean": 3325.8,
|
|
"valid_targets_min": 1434
|
|
},
|
|
{
|
|
"epoch": 1.9175510204081632,
|
|
"grad_norm": 0.46933023905049503,
|
|
"learning_rate": 3.6445798488083576e-05,
|
|
"loss": 0.2224,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11442330479621887,
|
|
"step": 1175,
|
|
"valid_targets_mean": 4419.6,
|
|
"valid_targets_min": 1587
|
|
},
|
|
{
|
|
"epoch": 1.9257142857142857,
|
|
"grad_norm": 0.5036546733874333,
|
|
"learning_rate": 3.639935890569873e-05,
|
|
"loss": 0.2504,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12574106454849243,
|
|
"step": 1180,
|
|
"valid_targets_mean": 4680.1,
|
|
"valid_targets_min": 1697
|
|
},
|
|
{
|
|
"epoch": 1.9338775510204083,
|
|
"grad_norm": 0.43369263982057915,
|
|
"learning_rate": 3.635264788795407e-05,
|
|
"loss": 0.2548,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11851448565721512,
|
|
"step": 1185,
|
|
"valid_targets_mean": 5754.1,
|
|
"valid_targets_min": 2418
|
|
},
|
|
{
|
|
"epoch": 1.9420408163265306,
|
|
"grad_norm": 0.5419617720778334,
|
|
"learning_rate": 3.6305666207990886e-05,
|
|
"loss": 0.2537,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11305820941925049,
|
|
"step": 1190,
|
|
"valid_targets_mean": 3559.0,
|
|
"valid_targets_min": 197
|
|
},
|
|
{
|
|
"epoch": 1.950204081632653,
|
|
"grad_norm": 0.5314154428545551,
|
|
"learning_rate": 3.625841464343042e-05,
|
|
"loss": 0.2597,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13911515474319458,
|
|
"step": 1195,
|
|
"valid_targets_mean": 3929.6,
|
|
"valid_targets_min": 1821
|
|
},
|
|
{
|
|
"epoch": 1.9583673469387755,
|
|
"grad_norm": 0.4896698812099476,
|
|
"learning_rate": 3.621089397636086e-05,
|
|
"loss": 0.2587,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14098498225212097,
|
|
"step": 1200,
|
|
"valid_targets_mean": 4840.2,
|
|
"valid_targets_min": 2400
|
|
},
|
|
{
|
|
"epoch": 1.966530612244898,
|
|
"grad_norm": 0.5856569580560823,
|
|
"learning_rate": 3.616310499332451e-05,
|
|
"loss": 0.2514,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11941181123256683,
|
|
"step": 1205,
|
|
"valid_targets_mean": 3408.5,
|
|
"valid_targets_min": 1840
|
|
},
|
|
{
|
|
"epoch": 1.9746938775510205,
|
|
"grad_norm": 0.4982878645027437,
|
|
"learning_rate": 3.6115048485304736e-05,
|
|
"loss": 0.2492,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15822862088680267,
|
|
"step": 1210,
|
|
"valid_targets_mean": 4985.1,
|
|
"valid_targets_min": 1426
|
|
},
|
|
{
|
|
"epoch": 1.9828571428571429,
|
|
"grad_norm": 0.5618752039712213,
|
|
"learning_rate": 3.606672524771281e-05,
|
|
"loss": 0.2538,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1203514039516449,
|
|
"step": 1215,
|
|
"valid_targets_mean": 3678.4,
|
|
"valid_targets_min": 1868
|
|
},
|
|
{
|
|
"epoch": 1.9910204081632652,
|
|
"grad_norm": 0.5291588003096855,
|
|
"learning_rate": 3.601813608037488e-05,
|
|
"loss": 0.2525,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12971103191375732,
|
|
"step": 1220,
|
|
"valid_targets_mean": 4032.5,
|
|
"valid_targets_min": 2465
|
|
},
|
|
{
|
|
"epoch": 1.9991836734693877,
|
|
"grad_norm": 0.45963462836895563,
|
|
"learning_rate": 3.596928178751858e-05,
|
|
"loss": 0.2563,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16318891942501068,
|
|
"step": 1225,
|
|
"valid_targets_mean": 7103.4,
|
|
"valid_targets_min": 2453
|
|
},
|
|
{
|
|
"epoch": 2.006530612244898,
|
|
"grad_norm": 0.43056684768932413,
|
|
"learning_rate": 3.592016317775984e-05,
|
|
"loss": 0.2295,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1186676174402237,
|
|
"step": 1230,
|
|
"valid_targets_mean": 8096.0,
|
|
"valid_targets_min": 3368
|
|
},
|
|
{
|
|
"epoch": 2.0146938775510206,
|
|
"grad_norm": 0.48079760996523624,
|
|
"learning_rate": 3.587078106408942e-05,
|
|
"loss": 0.2196,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10643745213747025,
|
|
"step": 1235,
|
|
"valid_targets_mean": 4640.6,
|
|
"valid_targets_min": 2642
|
|
},
|
|
{
|
|
"epoch": 2.0228571428571427,
|
|
"grad_norm": 0.5739864792377497,
|
|
"learning_rate": 3.582113626385951e-05,
|
|
"loss": 0.2343,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12379193305969238,
|
|
"step": 1240,
|
|
"valid_targets_mean": 3463.2,
|
|
"valid_targets_min": 1642
|
|
},
|
|
{
|
|
"epoch": 2.0310204081632652,
|
|
"grad_norm": 0.4995233604930039,
|
|
"learning_rate": 3.577122959877017e-05,
|
|
"loss": 0.2393,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13157770037651062,
|
|
"step": 1245,
|
|
"valid_targets_mean": 4963.4,
|
|
"valid_targets_min": 2360
|
|
},
|
|
{
|
|
"epoch": 2.039183673469388,
|
|
"grad_norm": 0.5292051470490394,
|
|
"learning_rate": 3.5721061894855756e-05,
|
|
"loss": 0.2212,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1389136165380478,
|
|
"step": 1250,
|
|
"valid_targets_mean": 5193.1,
|
|
"valid_targets_min": 2899
|
|
},
|
|
{
|
|
"epoch": 2.0473469387755103,
|
|
"grad_norm": 0.4888907093559062,
|
|
"learning_rate": 3.5670633982471185e-05,
|
|
"loss": 0.236,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15732908248901367,
|
|
"step": 1255,
|
|
"valid_targets_mean": 6197.2,
|
|
"valid_targets_min": 922
|
|
},
|
|
{
|
|
"epoch": 2.0555102040816324,
|
|
"grad_norm": 0.46552539231145457,
|
|
"learning_rate": 3.561994669627828e-05,
|
|
"loss": 0.2375,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10380949079990387,
|
|
"step": 1260,
|
|
"valid_targets_mean": 4741.5,
|
|
"valid_targets_min": 1876
|
|
},
|
|
{
|
|
"epoch": 2.063673469387755,
|
|
"grad_norm": 0.502350861185688,
|
|
"learning_rate": 3.55690008752319e-05,
|
|
"loss": 0.2315,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1098252683877945,
|
|
"step": 1265,
|
|
"valid_targets_mean": 4458.0,
|
|
"valid_targets_min": 426
|
|
},
|
|
{
|
|
"epoch": 2.0718367346938775,
|
|
"grad_norm": 0.5119241705569415,
|
|
"learning_rate": 3.551779736256604e-05,
|
|
"loss": 0.2266,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10564523935317993,
|
|
"step": 1270,
|
|
"valid_targets_mean": 4026.6,
|
|
"valid_targets_min": 1932
|
|
},
|
|
{
|
|
"epoch": 2.08,
|
|
"grad_norm": 0.5445752890676937,
|
|
"learning_rate": 3.546633700577994e-05,
|
|
"loss": 0.2335,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13011673092842102,
|
|
"step": 1275,
|
|
"valid_targets_mean": 4863.9,
|
|
"valid_targets_min": 2948
|
|
},
|
|
{
|
|
"epoch": 2.0881632653061226,
|
|
"grad_norm": 0.503922154527504,
|
|
"learning_rate": 3.541462065662399e-05,
|
|
"loss": 0.2299,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11520066112279892,
|
|
"step": 1280,
|
|
"valid_targets_mean": 5264.2,
|
|
"valid_targets_min": 699
|
|
},
|
|
{
|
|
"epoch": 2.0963265306122447,
|
|
"grad_norm": 0.5078588427716441,
|
|
"learning_rate": 3.536264917108567e-05,
|
|
"loss": 0.225,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13688494265079498,
|
|
"step": 1285,
|
|
"valid_targets_mean": 5433.9,
|
|
"valid_targets_min": 428
|
|
},
|
|
{
|
|
"epoch": 2.1044897959183673,
|
|
"grad_norm": 0.5346239759699418,
|
|
"learning_rate": 3.531042340937537e-05,
|
|
"loss": 0.2234,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09234683960676193,
|
|
"step": 1290,
|
|
"valid_targets_mean": 3514.4,
|
|
"valid_targets_min": 1542
|
|
},
|
|
{
|
|
"epoch": 2.11265306122449,
|
|
"grad_norm": 0.48695969669957334,
|
|
"learning_rate": 3.525794423591214e-05,
|
|
"loss": 0.2397,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10513603687286377,
|
|
"step": 1295,
|
|
"valid_targets_mean": 4443.8,
|
|
"valid_targets_min": 2414
|
|
},
|
|
{
|
|
"epoch": 2.1208163265306124,
|
|
"grad_norm": 0.3494135233315135,
|
|
"learning_rate": 3.520521251930941e-05,
|
|
"loss": 0.2111,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08017104119062424,
|
|
"step": 1300,
|
|
"valid_targets_mean": 4831.8,
|
|
"valid_targets_min": 1849
|
|
},
|
|
{
|
|
"epoch": 2.1289795918367345,
|
|
"grad_norm": 0.51369999487573,
|
|
"learning_rate": 3.51522291323606e-05,
|
|
"loss": 0.2318,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12431921064853668,
|
|
"step": 1305,
|
|
"valid_targets_mean": 4205.5,
|
|
"valid_targets_min": 1934
|
|
},
|
|
{
|
|
"epoch": 2.137142857142857,
|
|
"grad_norm": 0.48865254132392544,
|
|
"learning_rate": 3.509899495202467e-05,
|
|
"loss": 0.2425,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11693074554204941,
|
|
"step": 1310,
|
|
"valid_targets_mean": 4546.9,
|
|
"valid_targets_min": 1284
|
|
},
|
|
{
|
|
"epoch": 2.1453061224489796,
|
|
"grad_norm": 0.44405836098264856,
|
|
"learning_rate": 3.50455108594116e-05,
|
|
"loss": 0.2292,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09972664713859558,
|
|
"step": 1315,
|
|
"valid_targets_mean": 5387.9,
|
|
"valid_targets_min": 2268
|
|
},
|
|
{
|
|
"epoch": 2.153469387755102,
|
|
"grad_norm": 0.5265273408182132,
|
|
"learning_rate": 3.499177773976781e-05,
|
|
"loss": 0.2274,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13111922144889832,
|
|
"step": 1320,
|
|
"valid_targets_mean": 4415.8,
|
|
"valid_targets_min": 2872
|
|
},
|
|
{
|
|
"epoch": 2.1616326530612247,
|
|
"grad_norm": 0.5205910134073096,
|
|
"learning_rate": 3.493779648246154e-05,
|
|
"loss": 0.2266,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1394486427307129,
|
|
"step": 1325,
|
|
"valid_targets_mean": 5611.8,
|
|
"valid_targets_min": 1843
|
|
},
|
|
{
|
|
"epoch": 2.1697959183673468,
|
|
"grad_norm": 0.5319230026891569,
|
|
"learning_rate": 3.488356798096808e-05,
|
|
"loss": 0.2326,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09613202512264252,
|
|
"step": 1330,
|
|
"valid_targets_mean": 3297.8,
|
|
"valid_targets_min": 2093
|
|
},
|
|
{
|
|
"epoch": 2.1779591836734693,
|
|
"grad_norm": 0.4979347957927288,
|
|
"learning_rate": 3.482909313285499e-05,
|
|
"loss": 0.2218,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10724276304244995,
|
|
"step": 1335,
|
|
"valid_targets_mean": 3485.8,
|
|
"valid_targets_min": 1248
|
|
},
|
|
{
|
|
"epoch": 2.186122448979592,
|
|
"grad_norm": 0.4937495740699081,
|
|
"learning_rate": 3.47743728397673e-05,
|
|
"loss": 0.2284,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12111664563417435,
|
|
"step": 1340,
|
|
"valid_targets_mean": 4159.6,
|
|
"valid_targets_min": 1846
|
|
},
|
|
{
|
|
"epoch": 2.1942857142857144,
|
|
"grad_norm": 0.4324928163179517,
|
|
"learning_rate": 3.4719408007412494e-05,
|
|
"loss": 0.2197,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08307406306266785,
|
|
"step": 1345,
|
|
"valid_targets_mean": 3331.8,
|
|
"valid_targets_min": 1356
|
|
},
|
|
{
|
|
"epoch": 2.202448979591837,
|
|
"grad_norm": 0.5400878779239833,
|
|
"learning_rate": 3.4664199545545586e-05,
|
|
"loss": 0.2252,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11346576362848282,
|
|
"step": 1350,
|
|
"valid_targets_mean": 4346.8,
|
|
"valid_targets_min": 1594
|
|
},
|
|
{
|
|
"epoch": 2.210612244897959,
|
|
"grad_norm": 0.5751437287217012,
|
|
"learning_rate": 3.4608748367954064e-05,
|
|
"loss": 0.2474,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12245942652225494,
|
|
"step": 1355,
|
|
"valid_targets_mean": 3260.6,
|
|
"valid_targets_min": 2321
|
|
},
|
|
{
|
|
"epoch": 2.2187755102040816,
|
|
"grad_norm": 0.47449273447519114,
|
|
"learning_rate": 3.455305539244272e-05,
|
|
"loss": 0.2397,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11719304323196411,
|
|
"step": 1360,
|
|
"valid_targets_mean": 5256.8,
|
|
"valid_targets_min": 1937
|
|
},
|
|
{
|
|
"epoch": 2.226938775510204,
|
|
"grad_norm": 0.5472839028591866,
|
|
"learning_rate": 3.44971215408185e-05,
|
|
"loss": 0.2319,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14066477119922638,
|
|
"step": 1365,
|
|
"valid_targets_mean": 4968.9,
|
|
"valid_targets_min": 2880
|
|
},
|
|
{
|
|
"epoch": 2.2351020408163267,
|
|
"grad_norm": 0.46090788127961696,
|
|
"learning_rate": 3.4440947738875215e-05,
|
|
"loss": 0.2322,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13428354263305664,
|
|
"step": 1370,
|
|
"valid_targets_mean": 5852.8,
|
|
"valid_targets_min": 2568
|
|
},
|
|
{
|
|
"epoch": 2.243265306122449,
|
|
"grad_norm": 0.5012503306811049,
|
|
"learning_rate": 3.438453491637825e-05,
|
|
"loss": 0.212,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12726007401943207,
|
|
"step": 1375,
|
|
"valid_targets_mean": 4681.1,
|
|
"valid_targets_min": 2095
|
|
},
|
|
{
|
|
"epoch": 2.2514285714285713,
|
|
"grad_norm": 0.5273115750782535,
|
|
"learning_rate": 3.432788400704915e-05,
|
|
"loss": 0.2289,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13001614809036255,
|
|
"step": 1380,
|
|
"valid_targets_mean": 5218.6,
|
|
"valid_targets_min": 2828
|
|
},
|
|
{
|
|
"epoch": 2.259591836734694,
|
|
"grad_norm": 0.4808418482598062,
|
|
"learning_rate": 3.4270995948550175e-05,
|
|
"loss": 0.2238,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11392636597156525,
|
|
"step": 1385,
|
|
"valid_targets_mean": 4804.0,
|
|
"valid_targets_min": 1839
|
|
},
|
|
{
|
|
"epoch": 2.2677551020408164,
|
|
"grad_norm": 0.49225698613286484,
|
|
"learning_rate": 3.421387168246879e-05,
|
|
"loss": 0.235,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13683432340621948,
|
|
"step": 1390,
|
|
"valid_targets_mean": 5497.4,
|
|
"valid_targets_min": 2835
|
|
},
|
|
{
|
|
"epoch": 2.2759183673469385,
|
|
"grad_norm": 0.4707613798201057,
|
|
"learning_rate": 3.415651215430205e-05,
|
|
"loss": 0.2203,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11409764736890793,
|
|
"step": 1395,
|
|
"valid_targets_mean": 4437.4,
|
|
"valid_targets_min": 2262
|
|
},
|
|
{
|
|
"epoch": 2.284081632653061,
|
|
"grad_norm": 0.5457878781003122,
|
|
"learning_rate": 3.409891831344099e-05,
|
|
"loss": 0.2295,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15483315289020538,
|
|
"step": 1400,
|
|
"valid_targets_mean": 4922.2,
|
|
"valid_targets_min": 2454
|
|
},
|
|
{
|
|
"epoch": 2.2922448979591836,
|
|
"grad_norm": 0.526618791139005,
|
|
"learning_rate": 3.404109111315487e-05,
|
|
"loss": 0.2443,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1386374831199646,
|
|
"step": 1405,
|
|
"valid_targets_mean": 5216.4,
|
|
"valid_targets_min": 2441
|
|
},
|
|
{
|
|
"epoch": 2.300408163265306,
|
|
"grad_norm": 0.5133511190962673,
|
|
"learning_rate": 3.398303151057543e-05,
|
|
"loss": 0.2353,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1506226360797882,
|
|
"step": 1410,
|
|
"valid_targets_mean": 5939.0,
|
|
"valid_targets_min": 2086
|
|
},
|
|
{
|
|
"epoch": 2.3085714285714287,
|
|
"grad_norm": 0.5246066974685838,
|
|
"learning_rate": 3.392474046668105e-05,
|
|
"loss": 0.2354,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1148165687918663,
|
|
"step": 1415,
|
|
"valid_targets_mean": 3559.0,
|
|
"valid_targets_min": 319
|
|
},
|
|
{
|
|
"epoch": 2.3167346938775513,
|
|
"grad_norm": 0.45674426779962213,
|
|
"learning_rate": 3.38662189462808e-05,
|
|
"loss": 0.2396,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10377395153045654,
|
|
"step": 1420,
|
|
"valid_targets_mean": 4903.6,
|
|
"valid_targets_min": 1524
|
|
},
|
|
{
|
|
"epoch": 2.3248979591836734,
|
|
"grad_norm": 0.5549953930504047,
|
|
"learning_rate": 3.3807467917998534e-05,
|
|
"loss": 0.2294,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10187067836523056,
|
|
"step": 1425,
|
|
"valid_targets_mean": 5022.4,
|
|
"valid_targets_min": 179
|
|
},
|
|
{
|
|
"epoch": 2.333061224489796,
|
|
"grad_norm": 0.5667679888700494,
|
|
"learning_rate": 3.374848835425679e-05,
|
|
"loss": 0.2148,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10153408348560333,
|
|
"step": 1430,
|
|
"valid_targets_mean": 4056.2,
|
|
"valid_targets_min": 514
|
|
},
|
|
{
|
|
"epoch": 2.3412244897959185,
|
|
"grad_norm": 0.38849681119607615,
|
|
"learning_rate": 3.3689281231260755e-05,
|
|
"loss": 0.2216,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09394226968288422,
|
|
"step": 1435,
|
|
"valid_targets_mean": 7920.5,
|
|
"valid_targets_min": 2036
|
|
},
|
|
{
|
|
"epoch": 2.349387755102041,
|
|
"grad_norm": 0.4849228280895079,
|
|
"learning_rate": 3.362984752898208e-05,
|
|
"loss": 0.228,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10231403261423111,
|
|
"step": 1440,
|
|
"valid_targets_mean": 5006.1,
|
|
"valid_targets_min": 1883
|
|
},
|
|
{
|
|
"epoch": 2.357551020408163,
|
|
"grad_norm": 0.507994637708824,
|
|
"learning_rate": 3.3570188231142647e-05,
|
|
"loss": 0.2438,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11404062807559967,
|
|
"step": 1445,
|
|
"valid_targets_mean": 4506.0,
|
|
"valid_targets_min": 2781
|
|
},
|
|
{
|
|
"epoch": 2.3657142857142857,
|
|
"grad_norm": 0.47807993839798824,
|
|
"learning_rate": 3.351030432519832e-05,
|
|
"loss": 0.222,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11850126087665558,
|
|
"step": 1450,
|
|
"valid_targets_mean": 5020.4,
|
|
"valid_targets_min": 3208
|
|
},
|
|
{
|
|
"epoch": 2.373877551020408,
|
|
"grad_norm": 0.48000969417395906,
|
|
"learning_rate": 3.345019680232256e-05,
|
|
"loss": 0.2175,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1052422970533371,
|
|
"step": 1455,
|
|
"valid_targets_mean": 4933.5,
|
|
"valid_targets_min": 2331
|
|
},
|
|
{
|
|
"epoch": 2.3820408163265308,
|
|
"grad_norm": 0.7487096129240244,
|
|
"learning_rate": 3.3389866657390074e-05,
|
|
"loss": 0.239,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11607220768928528,
|
|
"step": 1460,
|
|
"valid_targets_mean": 4892.9,
|
|
"valid_targets_min": 1966
|
|
},
|
|
{
|
|
"epoch": 2.3902040816326533,
|
|
"grad_norm": 0.48365221971953487,
|
|
"learning_rate": 3.332931488896029e-05,
|
|
"loss": 0.2384,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11191040277481079,
|
|
"step": 1465,
|
|
"valid_targets_mean": 4724.5,
|
|
"valid_targets_min": 274
|
|
},
|
|
{
|
|
"epoch": 2.3983673469387754,
|
|
"grad_norm": 0.4928050884387631,
|
|
"learning_rate": 3.3268542499260874e-05,
|
|
"loss": 0.2354,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11471347510814667,
|
|
"step": 1470,
|
|
"valid_targets_mean": 4724.8,
|
|
"valid_targets_min": 729
|
|
},
|
|
{
|
|
"epoch": 2.406530612244898,
|
|
"grad_norm": 0.5035592274557454,
|
|
"learning_rate": 3.32075504941711e-05,
|
|
"loss": 0.218,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1193801611661911,
|
|
"step": 1475,
|
|
"valid_targets_mean": 4907.6,
|
|
"valid_targets_min": 2960
|
|
},
|
|
{
|
|
"epoch": 2.4146938775510205,
|
|
"grad_norm": 0.5363290738713465,
|
|
"learning_rate": 3.3146339883205243e-05,
|
|
"loss": 0.2342,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09244148433208466,
|
|
"step": 1480,
|
|
"valid_targets_mean": 4252.2,
|
|
"valid_targets_min": 1957
|
|
},
|
|
{
|
|
"epoch": 2.422857142857143,
|
|
"grad_norm": 0.5780978434020073,
|
|
"learning_rate": 3.308491167949585e-05,
|
|
"loss": 0.2275,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09925157576799393,
|
|
"step": 1485,
|
|
"valid_targets_mean": 4348.6,
|
|
"valid_targets_min": 1251
|
|
},
|
|
{
|
|
"epoch": 2.431020408163265,
|
|
"grad_norm": 0.42097104143162,
|
|
"learning_rate": 3.302326689977697e-05,
|
|
"loss": 0.2258,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10086309164762497,
|
|
"step": 1490,
|
|
"valid_targets_mean": 4883.4,
|
|
"valid_targets_min": 1399
|
|
},
|
|
{
|
|
"epoch": 2.4391836734693877,
|
|
"grad_norm": 0.5932340797258087,
|
|
"learning_rate": 3.2961406564367335e-05,
|
|
"loss": 0.2247,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11674109101295471,
|
|
"step": 1495,
|
|
"valid_targets_mean": 3973.2,
|
|
"valid_targets_min": 1920
|
|
},
|
|
{
|
|
"epoch": 2.4473469387755102,
|
|
"grad_norm": 0.4972235138038265,
|
|
"learning_rate": 3.2899331697153445e-05,
|
|
"loss": 0.2415,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1610492318868637,
|
|
"step": 1500,
|
|
"valid_targets_mean": 7308.4,
|
|
"valid_targets_min": 3836
|
|
},
|
|
{
|
|
"epoch": 2.455510204081633,
|
|
"grad_norm": 0.5172073645695415,
|
|
"learning_rate": 3.283704332557266e-05,
|
|
"loss": 0.2351,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10869985818862915,
|
|
"step": 1505,
|
|
"valid_targets_mean": 4040.6,
|
|
"valid_targets_min": 755
|
|
},
|
|
{
|
|
"epoch": 2.4636734693877553,
|
|
"grad_norm": 0.5477449704840818,
|
|
"learning_rate": 3.2774542480596165e-05,
|
|
"loss": 0.2223,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10822654515504837,
|
|
"step": 1510,
|
|
"valid_targets_mean": 3207.1,
|
|
"valid_targets_min": 580
|
|
},
|
|
{
|
|
"epoch": 2.4718367346938774,
|
|
"grad_norm": 0.5657022689186633,
|
|
"learning_rate": 3.2711830196711926e-05,
|
|
"loss": 0.2221,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11191504448652267,
|
|
"step": 1515,
|
|
"valid_targets_mean": 4077.8,
|
|
"valid_targets_min": 2333
|
|
},
|
|
{
|
|
"epoch": 2.48,
|
|
"grad_norm": 0.4381042129056105,
|
|
"learning_rate": 3.2648907511907544e-05,
|
|
"loss": 0.2231,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10943110287189484,
|
|
"step": 1520,
|
|
"valid_targets_mean": 6497.8,
|
|
"valid_targets_min": 2968
|
|
},
|
|
{
|
|
"epoch": 2.4881632653061225,
|
|
"grad_norm": 0.4131104301772404,
|
|
"learning_rate": 3.258577546765311e-05,
|
|
"loss": 0.2274,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10738101601600647,
|
|
"step": 1525,
|
|
"valid_targets_mean": 7473.2,
|
|
"valid_targets_min": 2792
|
|
},
|
|
{
|
|
"epoch": 2.496326530612245,
|
|
"grad_norm": 0.4381525165508198,
|
|
"learning_rate": 3.2522435108883925e-05,
|
|
"loss": 0.2246,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09731510281562805,
|
|
"step": 1530,
|
|
"valid_targets_mean": 4936.4,
|
|
"valid_targets_min": 1337
|
|
},
|
|
{
|
|
"epoch": 2.504489795918367,
|
|
"grad_norm": 0.4643511154489369,
|
|
"learning_rate": 3.245888748398322e-05,
|
|
"loss": 0.2305,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12030883133411407,
|
|
"step": 1535,
|
|
"valid_targets_mean": 4852.2,
|
|
"valid_targets_min": 1923
|
|
},
|
|
{
|
|
"epoch": 2.5126530612244897,
|
|
"grad_norm": 0.5157394733487625,
|
|
"learning_rate": 3.239513364476484e-05,
|
|
"loss": 0.2247,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10848742723464966,
|
|
"step": 1540,
|
|
"valid_targets_mean": 3704.1,
|
|
"valid_targets_min": 2390
|
|
},
|
|
{
|
|
"epoch": 2.5208163265306123,
|
|
"grad_norm": 0.431398191953788,
|
|
"learning_rate": 3.233117464645577e-05,
|
|
"loss": 0.2272,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10247580707073212,
|
|
"step": 1545,
|
|
"valid_targets_mean": 4868.9,
|
|
"valid_targets_min": 2738
|
|
},
|
|
{
|
|
"epoch": 2.528979591836735,
|
|
"grad_norm": 0.49163079122173614,
|
|
"learning_rate": 3.2267011547678735e-05,
|
|
"loss": 0.2399,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10695581138134003,
|
|
"step": 1550,
|
|
"valid_targets_mean": 4697.6,
|
|
"valid_targets_min": 1898
|
|
},
|
|
{
|
|
"epoch": 2.5371428571428574,
|
|
"grad_norm": 0.5377249490301479,
|
|
"learning_rate": 3.220264541043461e-05,
|
|
"loss": 0.2248,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14529483020305634,
|
|
"step": 1555,
|
|
"valid_targets_mean": 4557.8,
|
|
"valid_targets_min": 2132
|
|
},
|
|
{
|
|
"epoch": 2.5453061224489795,
|
|
"grad_norm": 0.4657234274209734,
|
|
"learning_rate": 3.21380773000849e-05,
|
|
"loss": 0.2327,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07366371154785156,
|
|
"step": 1560,
|
|
"valid_targets_mean": 3585.4,
|
|
"valid_targets_min": 172
|
|
},
|
|
{
|
|
"epoch": 2.553469387755102,
|
|
"grad_norm": 0.5015836388954693,
|
|
"learning_rate": 3.2073308285334085e-05,
|
|
"loss": 0.2059,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08729308098554611,
|
|
"step": 1565,
|
|
"valid_targets_mean": 3744.1,
|
|
"valid_targets_min": 503
|
|
},
|
|
{
|
|
"epoch": 2.5616326530612246,
|
|
"grad_norm": 0.4722130178319186,
|
|
"learning_rate": 3.200833943821192e-05,
|
|
"loss": 0.2271,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11033571511507034,
|
|
"step": 1570,
|
|
"valid_targets_mean": 5072.1,
|
|
"valid_targets_min": 2320
|
|
},
|
|
{
|
|
"epoch": 2.5697959183673467,
|
|
"grad_norm": 0.5032636329867763,
|
|
"learning_rate": 3.194317183405573e-05,
|
|
"loss": 0.2249,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10128222405910492,
|
|
"step": 1575,
|
|
"valid_targets_mean": 4459.9,
|
|
"valid_targets_min": 1882
|
|
},
|
|
{
|
|
"epoch": 2.577959183673469,
|
|
"grad_norm": 0.44825229490940477,
|
|
"learning_rate": 3.187780655149253e-05,
|
|
"loss": 0.2208,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10649105161428452,
|
|
"step": 1580,
|
|
"valid_targets_mean": 5630.6,
|
|
"valid_targets_min": 2670
|
|
},
|
|
{
|
|
"epoch": 2.5861224489795918,
|
|
"grad_norm": 0.4744318297919026,
|
|
"learning_rate": 3.1812244672421293e-05,
|
|
"loss": 0.2358,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12488362193107605,
|
|
"step": 1585,
|
|
"valid_targets_mean": 5510.8,
|
|
"valid_targets_min": 1727
|
|
},
|
|
{
|
|
"epoch": 2.5942857142857143,
|
|
"grad_norm": 0.46273596039967474,
|
|
"learning_rate": 3.174648728199491e-05,
|
|
"loss": 0.2257,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09472452104091644,
|
|
"step": 1590,
|
|
"valid_targets_mean": 4258.9,
|
|
"valid_targets_min": 1055
|
|
},
|
|
{
|
|
"epoch": 2.602448979591837,
|
|
"grad_norm": 0.6295062662437589,
|
|
"learning_rate": 3.168053546860234e-05,
|
|
"loss": 0.2274,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13179591298103333,
|
|
"step": 1595,
|
|
"valid_targets_mean": 4865.5,
|
|
"valid_targets_min": 991
|
|
},
|
|
{
|
|
"epoch": 2.6106122448979594,
|
|
"grad_norm": 0.47096472069028195,
|
|
"learning_rate": 3.1614390323850545e-05,
|
|
"loss": 0.2171,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08240728825330734,
|
|
"step": 1600,
|
|
"valid_targets_mean": 3505.2,
|
|
"valid_targets_min": 2442
|
|
},
|
|
{
|
|
"epoch": 2.6187755102040815,
|
|
"grad_norm": 0.46338642405080044,
|
|
"learning_rate": 3.15480529425464e-05,
|
|
"loss": 0.227,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17011001706123352,
|
|
"step": 1605,
|
|
"valid_targets_mean": 6040.1,
|
|
"valid_targets_min": 2068
|
|
},
|
|
{
|
|
"epoch": 2.626938775510204,
|
|
"grad_norm": 0.4699772712282333,
|
|
"learning_rate": 3.148152442267863e-05,
|
|
"loss": 0.2199,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07595792412757874,
|
|
"step": 1610,
|
|
"valid_targets_mean": 2887.2,
|
|
"valid_targets_min": 580
|
|
},
|
|
{
|
|
"epoch": 2.6351020408163266,
|
|
"grad_norm": 0.5128860468493535,
|
|
"learning_rate": 3.141480586539958e-05,
|
|
"loss": 0.23,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11134334653615952,
|
|
"step": 1615,
|
|
"valid_targets_mean": 3978.0,
|
|
"valid_targets_min": 1984
|
|
},
|
|
{
|
|
"epoch": 2.643265306122449,
|
|
"grad_norm": 0.48807682019439635,
|
|
"learning_rate": 3.1347898375007046e-05,
|
|
"loss": 0.2286,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10528915375471115,
|
|
"step": 1620,
|
|
"valid_targets_mean": 4671.4,
|
|
"valid_targets_min": 2219
|
|
},
|
|
{
|
|
"epoch": 2.6514285714285712,
|
|
"grad_norm": 0.5321071877691989,
|
|
"learning_rate": 3.128080305892593e-05,
|
|
"loss": 0.2354,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11503913253545761,
|
|
"step": 1625,
|
|
"valid_targets_mean": 4098.6,
|
|
"valid_targets_min": 2062
|
|
},
|
|
{
|
|
"epoch": 2.659591836734694,
|
|
"grad_norm": 0.47764688293351737,
|
|
"learning_rate": 3.121352102768998e-05,
|
|
"loss": 0.2344,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13357749581336975,
|
|
"step": 1630,
|
|
"valid_targets_mean": 5889.1,
|
|
"valid_targets_min": 1806
|
|
},
|
|
{
|
|
"epoch": 2.6677551020408163,
|
|
"grad_norm": 0.43425010363034666,
|
|
"learning_rate": 3.114605339492335e-05,
|
|
"loss": 0.2179,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08906761556863785,
|
|
"step": 1635,
|
|
"valid_targets_mean": 4261.2,
|
|
"valid_targets_min": 1482
|
|
},
|
|
{
|
|
"epoch": 2.675918367346939,
|
|
"grad_norm": 0.4790146606325519,
|
|
"learning_rate": 3.107840127732221e-05,
|
|
"loss": 0.2297,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09749139845371246,
|
|
"step": 1640,
|
|
"valid_targets_mean": 3916.4,
|
|
"valid_targets_min": 1664
|
|
},
|
|
{
|
|
"epoch": 2.6840816326530614,
|
|
"grad_norm": 0.47802026551929605,
|
|
"learning_rate": 3.1010565794636236e-05,
|
|
"loss": 0.2196,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0704977810382843,
|
|
"step": 1645,
|
|
"valid_targets_mean": 2540.2,
|
|
"valid_targets_min": 408
|
|
},
|
|
{
|
|
"epoch": 2.6922448979591835,
|
|
"grad_norm": 0.4139517698882833,
|
|
"learning_rate": 3.0942548069650085e-05,
|
|
"loss": 0.2262,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08792874217033386,
|
|
"step": 1650,
|
|
"valid_targets_mean": 4686.0,
|
|
"valid_targets_min": 385
|
|
},
|
|
{
|
|
"epoch": 2.700408163265306,
|
|
"grad_norm": 0.49595022031258346,
|
|
"learning_rate": 3.0874349228164836e-05,
|
|
"loss": 0.2254,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11520551890134811,
|
|
"step": 1655,
|
|
"valid_targets_mean": 4993.1,
|
|
"valid_targets_min": 2719
|
|
},
|
|
{
|
|
"epoch": 2.7085714285714286,
|
|
"grad_norm": 0.4840889074246445,
|
|
"learning_rate": 3.08059703989793e-05,
|
|
"loss": 0.2378,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09348023682832718,
|
|
"step": 1660,
|
|
"valid_targets_mean": 3658.9,
|
|
"valid_targets_min": 1500
|
|
},
|
|
{
|
|
"epoch": 2.716734693877551,
|
|
"grad_norm": 0.46075186086680664,
|
|
"learning_rate": 3.073741271387138e-05,
|
|
"loss": 0.2153,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09797358512878418,
|
|
"step": 1665,
|
|
"valid_targets_mean": 4534.5,
|
|
"valid_targets_min": 182
|
|
},
|
|
{
|
|
"epoch": 2.7248979591836733,
|
|
"grad_norm": 0.44333047710393725,
|
|
"learning_rate": 3.0668677307579344e-05,
|
|
"loss": 0.2284,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11998357623815536,
|
|
"step": 1670,
|
|
"valid_targets_mean": 5285.8,
|
|
"valid_targets_min": 1452
|
|
},
|
|
{
|
|
"epoch": 2.733061224489796,
|
|
"grad_norm": 0.48610888044017014,
|
|
"learning_rate": 3.059976531778301e-05,
|
|
"loss": 0.235,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12351099401712418,
|
|
"step": 1675,
|
|
"valid_targets_mean": 5781.0,
|
|
"valid_targets_min": 2358
|
|
},
|
|
{
|
|
"epoch": 2.7412244897959184,
|
|
"grad_norm": 0.5038017541644728,
|
|
"learning_rate": 3.0530677885084946e-05,
|
|
"loss": 0.2242,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11074423789978027,
|
|
"step": 1680,
|
|
"valid_targets_mean": 3891.0,
|
|
"valid_targets_min": 231
|
|
},
|
|
{
|
|
"epoch": 2.749387755102041,
|
|
"grad_norm": 0.39086467925287205,
|
|
"learning_rate": 3.0461416152991555e-05,
|
|
"loss": 0.214,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10134381055831909,
|
|
"step": 1685,
|
|
"valid_targets_mean": 5829.8,
|
|
"valid_targets_min": 2469
|
|
},
|
|
{
|
|
"epoch": 2.7575510204081635,
|
|
"grad_norm": 0.5035516790420694,
|
|
"learning_rate": 3.0391981267894195e-05,
|
|
"loss": 0.2332,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13245612382888794,
|
|
"step": 1690,
|
|
"valid_targets_mean": 5677.1,
|
|
"valid_targets_min": 2756
|
|
},
|
|
{
|
|
"epoch": 2.7657142857142856,
|
|
"grad_norm": 0.534397662500551,
|
|
"learning_rate": 3.0322374379050166e-05,
|
|
"loss": 0.234,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10749916732311249,
|
|
"step": 1695,
|
|
"valid_targets_mean": 3503.0,
|
|
"valid_targets_min": 612
|
|
},
|
|
{
|
|
"epoch": 2.773877551020408,
|
|
"grad_norm": 0.5164717311339068,
|
|
"learning_rate": 3.0252596638563717e-05,
|
|
"loss": 0.2336,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09786826372146606,
|
|
"step": 1700,
|
|
"valid_targets_mean": 4356.8,
|
|
"valid_targets_min": 2568
|
|
},
|
|
{
|
|
"epoch": 2.7820408163265307,
|
|
"grad_norm": 0.4634558853870556,
|
|
"learning_rate": 3.0182649201366944e-05,
|
|
"loss": 0.2263,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12021247297525406,
|
|
"step": 1705,
|
|
"valid_targets_mean": 5664.5,
|
|
"valid_targets_min": 2675
|
|
},
|
|
{
|
|
"epoch": 2.790204081632653,
|
|
"grad_norm": 0.4967608998125744,
|
|
"learning_rate": 3.0112533225200694e-05,
|
|
"loss": 0.2095,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07486874610185623,
|
|
"step": 1710,
|
|
"valid_targets_mean": 3141.6,
|
|
"valid_targets_min": 1362
|
|
},
|
|
{
|
|
"epoch": 2.7983673469387753,
|
|
"grad_norm": 0.4872167497296232,
|
|
"learning_rate": 3.0042249870595424e-05,
|
|
"loss": 0.2282,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10342056304216385,
|
|
"step": 1715,
|
|
"valid_targets_mean": 4081.1,
|
|
"valid_targets_min": 497
|
|
},
|
|
{
|
|
"epoch": 2.806530612244898,
|
|
"grad_norm": 0.5608970204261299,
|
|
"learning_rate": 2.9971800300851947e-05,
|
|
"loss": 0.2358,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09813547134399414,
|
|
"step": 1720,
|
|
"valid_targets_mean": 3014.8,
|
|
"valid_targets_min": 842
|
|
},
|
|
{
|
|
"epoch": 2.8146938775510204,
|
|
"grad_norm": 0.4983776943603883,
|
|
"learning_rate": 2.9901185682022208e-05,
|
|
"loss": 0.2224,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1242903396487236,
|
|
"step": 1725,
|
|
"valid_targets_mean": 4135.0,
|
|
"valid_targets_min": 2853
|
|
},
|
|
{
|
|
"epoch": 2.822857142857143,
|
|
"grad_norm": 0.4616169148945666,
|
|
"learning_rate": 2.9830407182889966e-05,
|
|
"loss": 0.2267,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13832835853099823,
|
|
"step": 1730,
|
|
"valid_targets_mean": 5644.0,
|
|
"valid_targets_min": 2770
|
|
},
|
|
{
|
|
"epoch": 2.8310204081632655,
|
|
"grad_norm": 0.43901300519214903,
|
|
"learning_rate": 2.9759465974951473e-05,
|
|
"loss": 0.2304,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1145050972700119,
|
|
"step": 1735,
|
|
"valid_targets_mean": 5897.6,
|
|
"valid_targets_min": 2242
|
|
},
|
|
{
|
|
"epoch": 2.8391836734693876,
|
|
"grad_norm": 0.37515014808594804,
|
|
"learning_rate": 2.9688363232396056e-05,
|
|
"loss": 0.2322,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10348987579345703,
|
|
"step": 1740,
|
|
"valid_targets_mean": 6573.2,
|
|
"valid_targets_min": 3306
|
|
},
|
|
{
|
|
"epoch": 2.84734693877551,
|
|
"grad_norm": 0.44557975300400043,
|
|
"learning_rate": 2.9617100132086705e-05,
|
|
"loss": 0.2453,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1028202623128891,
|
|
"step": 1745,
|
|
"valid_targets_mean": 4791.2,
|
|
"valid_targets_min": 2091
|
|
},
|
|
{
|
|
"epoch": 2.8555102040816327,
|
|
"grad_norm": 0.4618656993079806,
|
|
"learning_rate": 2.9545677853540584e-05,
|
|
"loss": 0.219,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12122482061386108,
|
|
"step": 1750,
|
|
"valid_targets_mean": 5584.4,
|
|
"valid_targets_min": 2143
|
|
},
|
|
{
|
|
"epoch": 2.8636734693877552,
|
|
"grad_norm": 0.5201277039799321,
|
|
"learning_rate": 2.9474097578909507e-05,
|
|
"loss": 0.2268,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12581226229667664,
|
|
"step": 1755,
|
|
"valid_targets_mean": 4061.0,
|
|
"valid_targets_min": 871
|
|
},
|
|
{
|
|
"epoch": 2.8718367346938773,
|
|
"grad_norm": 0.3861661290634797,
|
|
"learning_rate": 2.9402360492960377e-05,
|
|
"loss": 0.2207,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09558389335870743,
|
|
"step": 1760,
|
|
"valid_targets_mean": 7256.2,
|
|
"valid_targets_min": 1539
|
|
},
|
|
{
|
|
"epoch": 2.88,
|
|
"grad_norm": 0.46182446582565007,
|
|
"learning_rate": 2.933046778305556e-05,
|
|
"loss": 0.2261,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11080855131149292,
|
|
"step": 1765,
|
|
"valid_targets_mean": 4730.5,
|
|
"valid_targets_min": 1340
|
|
},
|
|
{
|
|
"epoch": 2.8881632653061224,
|
|
"grad_norm": 0.5463925298521676,
|
|
"learning_rate": 2.9258420639133253e-05,
|
|
"loss": 0.2329,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14682868123054504,
|
|
"step": 1770,
|
|
"valid_targets_mean": 4417.8,
|
|
"valid_targets_min": 1779
|
|
},
|
|
{
|
|
"epoch": 2.896326530612245,
|
|
"grad_norm": 0.4948282687111177,
|
|
"learning_rate": 2.918622025368779e-05,
|
|
"loss": 0.2305,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10451407730579376,
|
|
"step": 1775,
|
|
"valid_targets_mean": 4030.2,
|
|
"valid_targets_min": 1111
|
|
},
|
|
{
|
|
"epoch": 2.9044897959183675,
|
|
"grad_norm": 0.4208329318988087,
|
|
"learning_rate": 2.9113867821749883e-05,
|
|
"loss": 0.2261,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10942228138446808,
|
|
"step": 1780,
|
|
"valid_targets_mean": 6234.6,
|
|
"valid_targets_min": 2505
|
|
},
|
|
{
|
|
"epoch": 2.9126530612244896,
|
|
"grad_norm": 0.46262665189609575,
|
|
"learning_rate": 2.9041364540866855e-05,
|
|
"loss": 0.2202,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1118980124592781,
|
|
"step": 1785,
|
|
"valid_targets_mean": 5584.4,
|
|
"valid_targets_min": 2249
|
|
},
|
|
{
|
|
"epoch": 2.920816326530612,
|
|
"grad_norm": 0.4653192043672795,
|
|
"learning_rate": 2.8968711611082825e-05,
|
|
"loss": 0.2316,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11189016699790955,
|
|
"step": 1790,
|
|
"valid_targets_mean": 4856.0,
|
|
"valid_targets_min": 1638
|
|
},
|
|
{
|
|
"epoch": 2.9289795918367347,
|
|
"grad_norm": 0.5381032783967711,
|
|
"learning_rate": 2.8895910234918828e-05,
|
|
"loss": 0.231,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12157237529754639,
|
|
"step": 1795,
|
|
"valid_targets_mean": 4281.2,
|
|
"valid_targets_min": 1729
|
|
},
|
|
{
|
|
"epoch": 2.9371428571428573,
|
|
"grad_norm": 0.45749805411420635,
|
|
"learning_rate": 2.882296161735293e-05,
|
|
"loss": 0.2268,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17637786269187927,
|
|
"step": 1800,
|
|
"valid_targets_mean": 7033.2,
|
|
"valid_targets_min": 3103
|
|
},
|
|
{
|
|
"epoch": 2.9453061224489794,
|
|
"grad_norm": 0.5185768036987373,
|
|
"learning_rate": 2.874986696580028e-05,
|
|
"loss": 0.2238,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1181664690375328,
|
|
"step": 1805,
|
|
"valid_targets_mean": 4420.5,
|
|
"valid_targets_min": 1700
|
|
},
|
|
{
|
|
"epoch": 2.953469387755102,
|
|
"grad_norm": 0.530425221551735,
|
|
"learning_rate": 2.86766274900931e-05,
|
|
"loss": 0.2331,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0907604768872261,
|
|
"step": 1810,
|
|
"valid_targets_mean": 3195.0,
|
|
"valid_targets_min": 868
|
|
},
|
|
{
|
|
"epoch": 2.9616326530612245,
|
|
"grad_norm": 0.463810084198026,
|
|
"learning_rate": 2.8603244402460703e-05,
|
|
"loss": 0.228,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09564442187547684,
|
|
"step": 1815,
|
|
"valid_targets_mean": 4673.5,
|
|
"valid_targets_min": 1035
|
|
},
|
|
{
|
|
"epoch": 2.969795918367347,
|
|
"grad_norm": 0.4874888396808933,
|
|
"learning_rate": 2.8529718917509393e-05,
|
|
"loss": 0.2218,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14039471745491028,
|
|
"step": 1820,
|
|
"valid_targets_mean": 5813.9,
|
|
"valid_targets_min": 2879
|
|
},
|
|
{
|
|
"epoch": 2.9779591836734696,
|
|
"grad_norm": 0.5728968555651159,
|
|
"learning_rate": 2.8456052252202385e-05,
|
|
"loss": 0.2212,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1152806431055069,
|
|
"step": 1825,
|
|
"valid_targets_mean": 3103.2,
|
|
"valid_targets_min": 394
|
|
},
|
|
{
|
|
"epoch": 2.9861224489795917,
|
|
"grad_norm": 0.4835541798438136,
|
|
"learning_rate": 2.838224562583965e-05,
|
|
"loss": 0.2297,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09890221059322357,
|
|
"step": 1830,
|
|
"valid_targets_mean": 3995.4,
|
|
"valid_targets_min": 1916
|
|
},
|
|
{
|
|
"epoch": 2.994285714285714,
|
|
"grad_norm": 0.5015938870952967,
|
|
"learning_rate": 2.8308300260037734e-05,
|
|
"loss": 0.2321,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11656036227941513,
|
|
"step": 1835,
|
|
"valid_targets_mean": 4335.5,
|
|
"valid_targets_min": 2135
|
|
},
|
|
{
|
|
"epoch": 3.0016326530612245,
|
|
"grad_norm": 0.5164883314650555,
|
|
"learning_rate": 2.8234217378709533e-05,
|
|
"loss": 0.2193,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1306457817554474,
|
|
"step": 1840,
|
|
"valid_targets_mean": 5073.4,
|
|
"valid_targets_min": 3300
|
|
},
|
|
{
|
|
"epoch": 3.009795918367347,
|
|
"grad_norm": 0.4295219668494872,
|
|
"learning_rate": 2.8159998208044054e-05,
|
|
"loss": 0.21,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09993826597929001,
|
|
"step": 1845,
|
|
"valid_targets_mean": 5941.2,
|
|
"valid_targets_min": 2309
|
|
},
|
|
{
|
|
"epoch": 3.0179591836734696,
|
|
"grad_norm": 0.47919659318523444,
|
|
"learning_rate": 2.80856439764861e-05,
|
|
"loss": 0.2094,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09795888513326645,
|
|
"step": 1850,
|
|
"valid_targets_mean": 4683.8,
|
|
"valid_targets_min": 1422
|
|
},
|
|
{
|
|
"epoch": 3.0261224489795917,
|
|
"grad_norm": 0.40561110927902666,
|
|
"learning_rate": 2.8011155914715966e-05,
|
|
"loss": 0.1975,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08308486640453339,
|
|
"step": 1855,
|
|
"valid_targets_mean": 4387.9,
|
|
"valid_targets_min": 1836
|
|
},
|
|
{
|
|
"epoch": 3.0342857142857143,
|
|
"grad_norm": 0.4522047892147258,
|
|
"learning_rate": 2.7936535255629012e-05,
|
|
"loss": 0.2058,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07871349155902863,
|
|
"step": 1860,
|
|
"valid_targets_mean": 4495.1,
|
|
"valid_targets_min": 494
|
|
},
|
|
{
|
|
"epoch": 3.042448979591837,
|
|
"grad_norm": 0.5334853924211195,
|
|
"learning_rate": 2.786178323431534e-05,
|
|
"loss": 0.2012,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08362503349781036,
|
|
"step": 1865,
|
|
"valid_targets_mean": 3050.2,
|
|
"valid_targets_min": 228
|
|
},
|
|
{
|
|
"epoch": 3.0506122448979593,
|
|
"grad_norm": 0.5357442255015453,
|
|
"learning_rate": 2.7786901088039252e-05,
|
|
"loss": 0.2073,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1317649483680725,
|
|
"step": 1870,
|
|
"valid_targets_mean": 5041.1,
|
|
"valid_targets_min": 2540
|
|
},
|
|
{
|
|
"epoch": 3.0587755102040814,
|
|
"grad_norm": 0.3913265665588566,
|
|
"learning_rate": 2.771189005621887e-05,
|
|
"loss": 0.2011,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09924866259098053,
|
|
"step": 1875,
|
|
"valid_targets_mean": 7752.0,
|
|
"valid_targets_min": 2987
|
|
},
|
|
{
|
|
"epoch": 3.066938775510204,
|
|
"grad_norm": 0.5177456280517249,
|
|
"learning_rate": 2.7636751380405548e-05,
|
|
"loss": 0.2192,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08077879995107651,
|
|
"step": 1880,
|
|
"valid_targets_mean": 3772.6,
|
|
"valid_targets_min": 428
|
|
},
|
|
{
|
|
"epoch": 3.0751020408163265,
|
|
"grad_norm": 0.554211366794004,
|
|
"learning_rate": 2.7561486304263368e-05,
|
|
"loss": 0.2113,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0937132015824318,
|
|
"step": 1885,
|
|
"valid_targets_mean": 3425.9,
|
|
"valid_targets_min": 1557
|
|
},
|
|
{
|
|
"epoch": 3.083265306122449,
|
|
"grad_norm": 0.5273556149528048,
|
|
"learning_rate": 2.7486096073548525e-05,
|
|
"loss": 0.2043,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10002836585044861,
|
|
"step": 1890,
|
|
"valid_targets_mean": 3580.9,
|
|
"valid_targets_min": 347
|
|
},
|
|
{
|
|
"epoch": 3.0914285714285716,
|
|
"grad_norm": 0.4568666684236704,
|
|
"learning_rate": 2.7410581936088724e-05,
|
|
"loss": 0.2124,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12599283456802368,
|
|
"step": 1895,
|
|
"valid_targets_mean": 5725.8,
|
|
"valid_targets_min": 1676
|
|
},
|
|
{
|
|
"epoch": 3.0995918367346937,
|
|
"grad_norm": 0.5178725047067185,
|
|
"learning_rate": 2.733494514176253e-05,
|
|
"loss": 0.2096,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14653289318084717,
|
|
"step": 1900,
|
|
"valid_targets_mean": 5193.2,
|
|
"valid_targets_min": 2352
|
|
},
|
|
{
|
|
"epoch": 3.1077551020408163,
|
|
"grad_norm": 0.5266790142927706,
|
|
"learning_rate": 2.7259186942478656e-05,
|
|
"loss": 0.2004,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1208992749452591,
|
|
"step": 1905,
|
|
"valid_targets_mean": 5259.6,
|
|
"valid_targets_min": 3002
|
|
},
|
|
{
|
|
"epoch": 3.115918367346939,
|
|
"grad_norm": 0.45599128475629863,
|
|
"learning_rate": 2.718330859215528e-05,
|
|
"loss": 0.2042,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09286023676395416,
|
|
"step": 1910,
|
|
"valid_targets_mean": 4556.6,
|
|
"valid_targets_min": 1514
|
|
},
|
|
{
|
|
"epoch": 3.1240816326530614,
|
|
"grad_norm": 0.4647475863679506,
|
|
"learning_rate": 2.710731134669926e-05,
|
|
"loss": 0.2063,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1082848533987999,
|
|
"step": 1915,
|
|
"valid_targets_mean": 5435.9,
|
|
"valid_targets_min": 2714
|
|
},
|
|
{
|
|
"epoch": 3.1322448979591835,
|
|
"grad_norm": 0.5485809884441133,
|
|
"learning_rate": 2.7031196463985355e-05,
|
|
"loss": 0.2146,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1032031774520874,
|
|
"step": 1920,
|
|
"valid_targets_mean": 4039.2,
|
|
"valid_targets_min": 2406
|
|
},
|
|
{
|
|
"epoch": 3.140408163265306,
|
|
"grad_norm": 0.4890342953844648,
|
|
"learning_rate": 2.69549652038354e-05,
|
|
"loss": 0.2118,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09243686497211456,
|
|
"step": 1925,
|
|
"valid_targets_mean": 4101.8,
|
|
"valid_targets_min": 1504
|
|
},
|
|
{
|
|
"epoch": 3.1485714285714286,
|
|
"grad_norm": 0.550436781586566,
|
|
"learning_rate": 2.687861882799748e-05,
|
|
"loss": 0.201,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09199976921081543,
|
|
"step": 1930,
|
|
"valid_targets_mean": 4970.2,
|
|
"valid_targets_min": 1889
|
|
},
|
|
{
|
|
"epoch": 3.156734693877551,
|
|
"grad_norm": 0.5099546687084738,
|
|
"learning_rate": 2.6802158600125022e-05,
|
|
"loss": 0.2129,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07879088819026947,
|
|
"step": 1935,
|
|
"valid_targets_mean": 3468.5,
|
|
"valid_targets_min": 2111
|
|
},
|
|
{
|
|
"epoch": 3.1648979591836737,
|
|
"grad_norm": 0.555886818946411,
|
|
"learning_rate": 2.6725585785755873e-05,
|
|
"loss": 0.2002,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.109479621052742,
|
|
"step": 1940,
|
|
"valid_targets_mean": 5166.8,
|
|
"valid_targets_min": 1744
|
|
},
|
|
{
|
|
"epoch": 3.1730612244897958,
|
|
"grad_norm": 0.4420617318565779,
|
|
"learning_rate": 2.6648901652291366e-05,
|
|
"loss": 0.2059,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10851933062076569,
|
|
"step": 1945,
|
|
"valid_targets_mean": 6564.5,
|
|
"valid_targets_min": 755
|
|
},
|
|
{
|
|
"epoch": 3.1812244897959183,
|
|
"grad_norm": 0.4094155215148865,
|
|
"learning_rate": 2.657210746897535e-05,
|
|
"loss": 0.1962,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08861730992794037,
|
|
"step": 1950,
|
|
"valid_targets_mean": 4720.9,
|
|
"valid_targets_min": 2427
|
|
},
|
|
{
|
|
"epoch": 3.189387755102041,
|
|
"grad_norm": 0.47205587038214847,
|
|
"learning_rate": 2.649520450687317e-05,
|
|
"loss": 0.2031,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0953550860285759,
|
|
"step": 1955,
|
|
"valid_targets_mean": 4718.2,
|
|
"valid_targets_min": 1569
|
|
},
|
|
{
|
|
"epoch": 3.1975510204081634,
|
|
"grad_norm": 0.5460699888433964,
|
|
"learning_rate": 2.6418194038850634e-05,
|
|
"loss": 0.2058,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12797680497169495,
|
|
"step": 1960,
|
|
"valid_targets_mean": 5194.0,
|
|
"valid_targets_min": 2231
|
|
},
|
|
{
|
|
"epoch": 3.2057142857142855,
|
|
"grad_norm": 0.5232868644903416,
|
|
"learning_rate": 2.6341077339552938e-05,
|
|
"loss": 0.1986,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0944906622171402,
|
|
"step": 1965,
|
|
"valid_targets_mean": 4287.2,
|
|
"valid_targets_min": 2086
|
|
},
|
|
{
|
|
"epoch": 3.213877551020408,
|
|
"grad_norm": 0.4393077695920715,
|
|
"learning_rate": 2.6263855685383585e-05,
|
|
"loss": 0.1889,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0835411548614502,
|
|
"step": 1970,
|
|
"valid_targets_mean": 4544.9,
|
|
"valid_targets_min": 2044
|
|
},
|
|
{
|
|
"epoch": 3.2220408163265306,
|
|
"grad_norm": 0.5802825664153698,
|
|
"learning_rate": 2.6186530354483224e-05,
|
|
"loss": 0.2088,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07232633233070374,
|
|
"step": 1975,
|
|
"valid_targets_mean": 3895.1,
|
|
"valid_targets_min": 1513
|
|
},
|
|
{
|
|
"epoch": 3.230204081632653,
|
|
"grad_norm": 0.7167557743459537,
|
|
"learning_rate": 2.6109102626708547e-05,
|
|
"loss": 0.2026,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10113707929849625,
|
|
"step": 1980,
|
|
"valid_targets_mean": 3797.6,
|
|
"valid_targets_min": 875
|
|
},
|
|
{
|
|
"epoch": 3.2383673469387757,
|
|
"grad_norm": 0.4045153165661452,
|
|
"learning_rate": 2.6031573783611053e-05,
|
|
"loss": 0.203,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.131624236702919,
|
|
"step": 1985,
|
|
"valid_targets_mean": 7943.2,
|
|
"valid_targets_min": 2568
|
|
},
|
|
{
|
|
"epoch": 3.246530612244898,
|
|
"grad_norm": 0.5833282836855167,
|
|
"learning_rate": 2.5953945108415883e-05,
|
|
"loss": 0.2118,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10405785590410233,
|
|
"step": 1990,
|
|
"valid_targets_mean": 3333.1,
|
|
"valid_targets_min": 1429
|
|
},
|
|
{
|
|
"epoch": 3.2546938775510204,
|
|
"grad_norm": 0.508945720636517,
|
|
"learning_rate": 2.5876217886000526e-05,
|
|
"loss": 0.2243,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11219657957553864,
|
|
"step": 1995,
|
|
"valid_targets_mean": 4835.9,
|
|
"valid_targets_min": 629
|
|
},
|
|
{
|
|
"epoch": 3.262857142857143,
|
|
"grad_norm": 0.5076274988600784,
|
|
"learning_rate": 2.5798393402873614e-05,
|
|
"loss": 0.2172,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0813908651471138,
|
|
"step": 2000,
|
|
"valid_targets_mean": 3677.1,
|
|
"valid_targets_min": 1703
|
|
},
|
|
{
|
|
"epoch": 3.2710204081632654,
|
|
"grad_norm": 0.5434822373214382,
|
|
"learning_rate": 2.5720472947153582e-05,
|
|
"loss": 0.201,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12669488787651062,
|
|
"step": 2005,
|
|
"valid_targets_mean": 4550.0,
|
|
"valid_targets_min": 1585
|
|
},
|
|
{
|
|
"epoch": 3.2791836734693875,
|
|
"grad_norm": 0.451823896304527,
|
|
"learning_rate": 2.5642457808547365e-05,
|
|
"loss": 0.2076,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09495006501674652,
|
|
"step": 2010,
|
|
"valid_targets_mean": 5470.9,
|
|
"valid_targets_min": 2327
|
|
},
|
|
{
|
|
"epoch": 3.28734693877551,
|
|
"grad_norm": 0.6430875308815419,
|
|
"learning_rate": 2.5564349278329056e-05,
|
|
"loss": 0.2099,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08256812393665314,
|
|
"step": 2015,
|
|
"valid_targets_mean": 2062.4,
|
|
"valid_targets_min": 998
|
|
},
|
|
{
|
|
"epoch": 3.2955102040816326,
|
|
"grad_norm": 0.43098561769347765,
|
|
"learning_rate": 2.5486148649318514e-05,
|
|
"loss": 0.2014,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08821088820695877,
|
|
"step": 2020,
|
|
"valid_targets_mean": 5786.9,
|
|
"valid_targets_min": 2578
|
|
},
|
|
{
|
|
"epoch": 3.303673469387755,
|
|
"grad_norm": 0.6356272149997206,
|
|
"learning_rate": 2.540785721586e-05,
|
|
"loss": 0.1946,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09246645122766495,
|
|
"step": 2025,
|
|
"valid_targets_mean": 3517.8,
|
|
"valid_targets_min": 2197
|
|
},
|
|
{
|
|
"epoch": 3.3118367346938777,
|
|
"grad_norm": 0.4454137575172535,
|
|
"learning_rate": 2.53294762738007e-05,
|
|
"loss": 0.2151,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10923437029123306,
|
|
"step": 2030,
|
|
"valid_targets_mean": 6211.1,
|
|
"valid_targets_min": 1175
|
|
},
|
|
{
|
|
"epoch": 3.32,
|
|
"grad_norm": 0.5257535142881744,
|
|
"learning_rate": 2.5251007120469347e-05,
|
|
"loss": 0.2101,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13460084795951843,
|
|
"step": 2035,
|
|
"valid_targets_mean": 5227.9,
|
|
"valid_targets_min": 2360
|
|
},
|
|
{
|
|
"epoch": 3.3281632653061224,
|
|
"grad_norm": 0.5088064453423816,
|
|
"learning_rate": 2.5172451054654685e-05,
|
|
"loss": 0.2121,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10537538677453995,
|
|
"step": 2040,
|
|
"valid_targets_mean": 4792.5,
|
|
"valid_targets_min": 1346
|
|
},
|
|
{
|
|
"epoch": 3.336326530612245,
|
|
"grad_norm": 0.49456466651480946,
|
|
"learning_rate": 2.5093809376584006e-05,
|
|
"loss": 0.2177,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08632659167051315,
|
|
"step": 2045,
|
|
"valid_targets_mean": 4183.0,
|
|
"valid_targets_min": 2064
|
|
},
|
|
{
|
|
"epoch": 3.3444897959183675,
|
|
"grad_norm": 0.5409869448697774,
|
|
"learning_rate": 2.501508338790162e-05,
|
|
"loss": 0.2069,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12297812104225159,
|
|
"step": 2050,
|
|
"valid_targets_mean": 5444.6,
|
|
"valid_targets_min": 2788
|
|
},
|
|
{
|
|
"epoch": 3.3526530612244896,
|
|
"grad_norm": 0.5093822050798533,
|
|
"learning_rate": 2.49362743916473e-05,
|
|
"loss": 0.2135,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09434787929058075,
|
|
"step": 2055,
|
|
"valid_targets_mean": 3569.2,
|
|
"valid_targets_min": 2053
|
|
},
|
|
{
|
|
"epoch": 3.360816326530612,
|
|
"grad_norm": 0.47603529261568117,
|
|
"learning_rate": 2.4857383692234742e-05,
|
|
"loss": 0.2021,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12311601638793945,
|
|
"step": 2060,
|
|
"valid_targets_mean": 6423.6,
|
|
"valid_targets_min": 2096
|
|
},
|
|
{
|
|
"epoch": 3.3689795918367347,
|
|
"grad_norm": 0.534211960665438,
|
|
"learning_rate": 2.4778412595429955e-05,
|
|
"loss": 0.2085,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10466711223125458,
|
|
"step": 2065,
|
|
"valid_targets_mean": 4221.9,
|
|
"valid_targets_min": 1956
|
|
},
|
|
{
|
|
"epoch": 3.3771428571428572,
|
|
"grad_norm": 0.5472416262428637,
|
|
"learning_rate": 2.4699362408329646e-05,
|
|
"loss": 0.2,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10589661449193954,
|
|
"step": 2070,
|
|
"valid_targets_mean": 3920.5,
|
|
"valid_targets_min": 620
|
|
},
|
|
{
|
|
"epoch": 3.3853061224489798,
|
|
"grad_norm": 0.547677700443028,
|
|
"learning_rate": 2.46202344393396e-05,
|
|
"loss": 0.2168,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1077793538570404,
|
|
"step": 2075,
|
|
"valid_targets_mean": 3886.6,
|
|
"valid_targets_min": 684
|
|
},
|
|
{
|
|
"epoch": 3.393469387755102,
|
|
"grad_norm": 0.5831345321921154,
|
|
"learning_rate": 2.4541029998153e-05,
|
|
"loss": 0.1992,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1060214415192604,
|
|
"step": 2080,
|
|
"valid_targets_mean": 3710.5,
|
|
"valid_targets_min": 1638
|
|
},
|
|
{
|
|
"epoch": 3.4016326530612244,
|
|
"grad_norm": 0.493306213537092,
|
|
"learning_rate": 2.446175039572878e-05,
|
|
"loss": 0.2018,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11661600321531296,
|
|
"step": 2085,
|
|
"valid_targets_mean": 5181.2,
|
|
"valid_targets_min": 1399
|
|
},
|
|
{
|
|
"epoch": 3.409795918367347,
|
|
"grad_norm": 0.5871415626775445,
|
|
"learning_rate": 2.4382396944269912e-05,
|
|
"loss": 0.2076,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09259402751922607,
|
|
"step": 2090,
|
|
"valid_targets_mean": 3459.8,
|
|
"valid_targets_min": 1845
|
|
},
|
|
{
|
|
"epoch": 3.4179591836734695,
|
|
"grad_norm": 0.5745344129842763,
|
|
"learning_rate": 2.430297095720168e-05,
|
|
"loss": 0.2141,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09636922925710678,
|
|
"step": 2095,
|
|
"valid_targets_mean": 3819.6,
|
|
"valid_targets_min": 186
|
|
},
|
|
{
|
|
"epoch": 3.4261224489795916,
|
|
"grad_norm": 0.4721493101920305,
|
|
"learning_rate": 2.4223473749149937e-05,
|
|
"loss": 0.1987,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09481912851333618,
|
|
"step": 2100,
|
|
"valid_targets_mean": 4979.5,
|
|
"valid_targets_min": 1675
|
|
},
|
|
{
|
|
"epoch": 3.434285714285714,
|
|
"grad_norm": 0.49501466744059974,
|
|
"learning_rate": 2.4143906635919385e-05,
|
|
"loss": 0.2079,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08848510682582855,
|
|
"step": 2105,
|
|
"valid_targets_mean": 4279.2,
|
|
"valid_targets_min": 839
|
|
},
|
|
{
|
|
"epoch": 3.4424489795918367,
|
|
"grad_norm": 0.3912763685993033,
|
|
"learning_rate": 2.4064270934471735e-05,
|
|
"loss": 0.2105,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0881570428609848,
|
|
"step": 2110,
|
|
"valid_targets_mean": 5399.5,
|
|
"valid_targets_min": 2613
|
|
},
|
|
{
|
|
"epoch": 3.4506122448979593,
|
|
"grad_norm": 0.43492407833873475,
|
|
"learning_rate": 2.3984567962903968e-05,
|
|
"loss": 0.2021,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1029646098613739,
|
|
"step": 2115,
|
|
"valid_targets_mean": 4683.2,
|
|
"valid_targets_min": 3756
|
|
},
|
|
{
|
|
"epoch": 3.458775510204082,
|
|
"grad_norm": 0.5260597065670592,
|
|
"learning_rate": 2.3904799040426476e-05,
|
|
"loss": 0.2074,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09860092401504517,
|
|
"step": 2120,
|
|
"valid_targets_mean": 3959.4,
|
|
"valid_targets_min": 2219
|
|
},
|
|
{
|
|
"epoch": 3.466938775510204,
|
|
"grad_norm": 0.47356375082893604,
|
|
"learning_rate": 2.3824965487341247e-05,
|
|
"loss": 0.2102,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11778430640697479,
|
|
"step": 2125,
|
|
"valid_targets_mean": 5864.9,
|
|
"valid_targets_min": 3196
|
|
},
|
|
{
|
|
"epoch": 3.4751020408163265,
|
|
"grad_norm": 0.5462984356735526,
|
|
"learning_rate": 2.3745068625020008e-05,
|
|
"loss": 0.2144,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.119639553129673,
|
|
"step": 2130,
|
|
"valid_targets_mean": 4661.4,
|
|
"valid_targets_min": 2031
|
|
},
|
|
{
|
|
"epoch": 3.483265306122449,
|
|
"grad_norm": 0.5206942495860799,
|
|
"learning_rate": 2.3665109775882352e-05,
|
|
"loss": 0.2071,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08757778257131577,
|
|
"step": 2135,
|
|
"valid_targets_mean": 3686.8,
|
|
"valid_targets_min": 1095
|
|
},
|
|
{
|
|
"epoch": 3.4914285714285715,
|
|
"grad_norm": 0.5018763236389352,
|
|
"learning_rate": 2.3585090263373868e-05,
|
|
"loss": 0.2106,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11070487648248672,
|
|
"step": 2140,
|
|
"valid_targets_mean": 4611.9,
|
|
"valid_targets_min": 1927
|
|
},
|
|
{
|
|
"epoch": 3.4995918367346937,
|
|
"grad_norm": 0.47575628659027597,
|
|
"learning_rate": 2.350501141194419e-05,
|
|
"loss": 0.2066,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10880281031131744,
|
|
"step": 2145,
|
|
"valid_targets_mean": 5161.1,
|
|
"valid_targets_min": 1502
|
|
},
|
|
{
|
|
"epoch": 3.507755102040816,
|
|
"grad_norm": 0.4928847665812081,
|
|
"learning_rate": 2.3424874547025125e-05,
|
|
"loss": 0.2127,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11148449778556824,
|
|
"step": 2150,
|
|
"valid_targets_mean": 5960.8,
|
|
"valid_targets_min": 1937
|
|
},
|
|
{
|
|
"epoch": 3.5159183673469387,
|
|
"grad_norm": 0.4433641303426999,
|
|
"learning_rate": 2.33446809950087e-05,
|
|
"loss": 0.2134,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1022956520318985,
|
|
"step": 2155,
|
|
"valid_targets_mean": 5677.5,
|
|
"valid_targets_min": 2168
|
|
},
|
|
{
|
|
"epoch": 3.5240816326530613,
|
|
"grad_norm": 0.46201695625079175,
|
|
"learning_rate": 2.326443208322519e-05,
|
|
"loss": 0.2091,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1019848957657814,
|
|
"step": 2160,
|
|
"valid_targets_mean": 5709.8,
|
|
"valid_targets_min": 2873
|
|
},
|
|
{
|
|
"epoch": 3.532244897959184,
|
|
"grad_norm": 0.4932375115901645,
|
|
"learning_rate": 2.3184129139921162e-05,
|
|
"loss": 0.2054,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0880424901843071,
|
|
"step": 2165,
|
|
"valid_targets_mean": 4456.4,
|
|
"valid_targets_min": 754
|
|
},
|
|
{
|
|
"epoch": 3.5404081632653064,
|
|
"grad_norm": 0.4415593903313715,
|
|
"learning_rate": 2.3103773494237504e-05,
|
|
"loss": 0.1912,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08950482308864594,
|
|
"step": 2170,
|
|
"valid_targets_mean": 4576.4,
|
|
"valid_targets_min": 2308
|
|
},
|
|
{
|
|
"epoch": 3.5485714285714285,
|
|
"grad_norm": 0.6079075078157,
|
|
"learning_rate": 2.3023366476187402e-05,
|
|
"loss": 0.2069,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1052299439907074,
|
|
"step": 2175,
|
|
"valid_targets_mean": 2974.9,
|
|
"valid_targets_min": 1248
|
|
},
|
|
{
|
|
"epoch": 3.556734693877551,
|
|
"grad_norm": 0.3989429150863732,
|
|
"learning_rate": 2.2942909416634326e-05,
|
|
"loss": 0.1996,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10140019655227661,
|
|
"step": 2180,
|
|
"valid_targets_mean": 5633.8,
|
|
"valid_targets_min": 3165
|
|
},
|
|
{
|
|
"epoch": 3.5648979591836736,
|
|
"grad_norm": 0.4904211159678837,
|
|
"learning_rate": 2.2862403647270033e-05,
|
|
"loss": 0.2169,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10210586339235306,
|
|
"step": 2185,
|
|
"valid_targets_mean": 4799.4,
|
|
"valid_targets_min": 589
|
|
},
|
|
{
|
|
"epoch": 3.5730612244897957,
|
|
"grad_norm": 0.5484079285629692,
|
|
"learning_rate": 2.2781850500592487e-05,
|
|
"loss": 0.2071,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11891067028045654,
|
|
"step": 2190,
|
|
"valid_targets_mean": 5383.9,
|
|
"valid_targets_min": 3293
|
|
},
|
|
{
|
|
"epoch": 3.5812244897959182,
|
|
"grad_norm": 0.6008177442360957,
|
|
"learning_rate": 2.2701251309883837e-05,
|
|
"loss": 0.2055,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10325556993484497,
|
|
"step": 2195,
|
|
"valid_targets_mean": 3358.9,
|
|
"valid_targets_min": 1192
|
|
},
|
|
{
|
|
"epoch": 3.5893877551020408,
|
|
"grad_norm": 0.5227749077095252,
|
|
"learning_rate": 2.2620607409188326e-05,
|
|
"loss": 0.2104,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09844657778739929,
|
|
"step": 2200,
|
|
"valid_targets_mean": 4752.5,
|
|
"valid_targets_min": 2204
|
|
},
|
|
{
|
|
"epoch": 3.5975510204081633,
|
|
"grad_norm": 0.5258378975483752,
|
|
"learning_rate": 2.2539920133290208e-05,
|
|
"loss": 0.2077,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11812394112348557,
|
|
"step": 2205,
|
|
"valid_targets_mean": 5001.0,
|
|
"valid_targets_min": 1199
|
|
},
|
|
{
|
|
"epoch": 3.605714285714286,
|
|
"grad_norm": 0.5339466282169013,
|
|
"learning_rate": 2.245919081769168e-05,
|
|
"loss": 0.2064,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10490740835666656,
|
|
"step": 2210,
|
|
"valid_targets_mean": 3964.6,
|
|
"valid_targets_min": 1520
|
|
},
|
|
{
|
|
"epoch": 3.6138775510204084,
|
|
"grad_norm": 0.4045810022832643,
|
|
"learning_rate": 2.2378420798590764e-05,
|
|
"loss": 0.2085,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08282206952571869,
|
|
"step": 2215,
|
|
"valid_targets_mean": 6330.6,
|
|
"valid_targets_min": 1416
|
|
},
|
|
{
|
|
"epoch": 3.6220408163265305,
|
|
"grad_norm": 0.4675388517138899,
|
|
"learning_rate": 2.2297611412859178e-05,
|
|
"loss": 0.213,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12077487260103226,
|
|
"step": 2220,
|
|
"valid_targets_mean": 6203.0,
|
|
"valid_targets_min": 2988
|
|
},
|
|
{
|
|
"epoch": 3.630204081632653,
|
|
"grad_norm": 0.4974295688370866,
|
|
"learning_rate": 2.2216763998020222e-05,
|
|
"loss": 0.2065,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08385932445526123,
|
|
"step": 2225,
|
|
"valid_targets_mean": 3604.4,
|
|
"valid_targets_min": 1055
|
|
},
|
|
{
|
|
"epoch": 3.6383673469387756,
|
|
"grad_norm": 0.40766596357353385,
|
|
"learning_rate": 2.2135879892226643e-05,
|
|
"loss": 0.1884,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0922253355383873,
|
|
"step": 2230,
|
|
"valid_targets_mean": 6555.6,
|
|
"valid_targets_min": 2428
|
|
},
|
|
{
|
|
"epoch": 3.6465306122448977,
|
|
"grad_norm": 0.5442667722758696,
|
|
"learning_rate": 2.205496043423849e-05,
|
|
"loss": 0.2129,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10492303967475891,
|
|
"step": 2235,
|
|
"valid_targets_mean": 4267.6,
|
|
"valid_targets_min": 508
|
|
},
|
|
{
|
|
"epoch": 3.6546938775510203,
|
|
"grad_norm": 0.47975005964527107,
|
|
"learning_rate": 2.197400696340091e-05,
|
|
"loss": 0.2045,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13115760684013367,
|
|
"step": 2240,
|
|
"valid_targets_mean": 6332.2,
|
|
"valid_targets_min": 4055
|
|
},
|
|
{
|
|
"epoch": 3.662857142857143,
|
|
"grad_norm": 0.5520568799725798,
|
|
"learning_rate": 2.189302081962207e-05,
|
|
"loss": 0.2147,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09570181369781494,
|
|
"step": 2245,
|
|
"valid_targets_mean": 3627.2,
|
|
"valid_targets_min": 2068
|
|
},
|
|
{
|
|
"epoch": 3.6710204081632654,
|
|
"grad_norm": 0.5183011327920499,
|
|
"learning_rate": 2.1812003343350886e-05,
|
|
"loss": 0.2095,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12069438397884369,
|
|
"step": 2250,
|
|
"valid_targets_mean": 5529.8,
|
|
"valid_targets_min": 1055
|
|
},
|
|
{
|
|
"epoch": 3.679183673469388,
|
|
"grad_norm": 0.40494166134398335,
|
|
"learning_rate": 2.1730955875554894e-05,
|
|
"loss": 0.1989,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11352106928825378,
|
|
"step": 2255,
|
|
"valid_targets_mean": 7916.5,
|
|
"valid_targets_min": 2533
|
|
},
|
|
{
|
|
"epoch": 3.6873469387755105,
|
|
"grad_norm": 0.5377042271103643,
|
|
"learning_rate": 2.164987975769803e-05,
|
|
"loss": 0.2041,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08446522057056427,
|
|
"step": 2260,
|
|
"valid_targets_mean": 3122.2,
|
|
"valid_targets_min": 1664
|
|
},
|
|
{
|
|
"epoch": 3.6955102040816326,
|
|
"grad_norm": 0.4714079493569945,
|
|
"learning_rate": 2.1568776331718428e-05,
|
|
"loss": 0.2108,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12201312184333801,
|
|
"step": 2265,
|
|
"valid_targets_mean": 5722.1,
|
|
"valid_targets_min": 2148
|
|
},
|
|
{
|
|
"epoch": 3.703673469387755,
|
|
"grad_norm": 0.6138764873320968,
|
|
"learning_rate": 2.148764694000624e-05,
|
|
"loss": 0.2105,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12149910628795624,
|
|
"step": 2270,
|
|
"valid_targets_mean": 3841.4,
|
|
"valid_targets_min": 581
|
|
},
|
|
{
|
|
"epoch": 3.7118367346938776,
|
|
"grad_norm": 0.6152682833954068,
|
|
"learning_rate": 2.1406492925381358e-05,
|
|
"loss": 0.1967,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1131722629070282,
|
|
"step": 2275,
|
|
"valid_targets_mean": 3731.0,
|
|
"valid_targets_min": 231
|
|
},
|
|
{
|
|
"epoch": 3.7199999999999998,
|
|
"grad_norm": 0.4810706951283212,
|
|
"learning_rate": 2.1325315631071257e-05,
|
|
"loss": 0.1972,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12919890880584717,
|
|
"step": 2280,
|
|
"valid_targets_mean": 5619.1,
|
|
"valid_targets_min": 2068
|
|
},
|
|
{
|
|
"epoch": 3.7281632653061223,
|
|
"grad_norm": 0.4109059218621779,
|
|
"learning_rate": 2.124411640068869e-05,
|
|
"loss": 0.2078,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0750608742237091,
|
|
"step": 2285,
|
|
"valid_targets_mean": 4065.9,
|
|
"valid_targets_min": 186
|
|
},
|
|
{
|
|
"epoch": 3.736326530612245,
|
|
"grad_norm": 0.5452804250141515,
|
|
"learning_rate": 2.1162896578209517e-05,
|
|
"loss": 0.219,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13831168413162231,
|
|
"step": 2290,
|
|
"valid_targets_mean": 4769.5,
|
|
"valid_targets_min": 2192
|
|
},
|
|
{
|
|
"epoch": 3.7444897959183674,
|
|
"grad_norm": 0.46277655460159545,
|
|
"learning_rate": 2.1081657507950423e-05,
|
|
"loss": 0.2048,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10686130821704865,
|
|
"step": 2295,
|
|
"valid_targets_mean": 5485.1,
|
|
"valid_targets_min": 1555
|
|
},
|
|
{
|
|
"epoch": 3.75265306122449,
|
|
"grad_norm": 0.4422005430051166,
|
|
"learning_rate": 2.100040053454666e-05,
|
|
"loss": 0.2145,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12035848200321198,
|
|
"step": 2300,
|
|
"valid_targets_mean": 6698.6,
|
|
"valid_targets_min": 2586
|
|
},
|
|
{
|
|
"epoch": 3.7608163265306125,
|
|
"grad_norm": 0.44658354686920304,
|
|
"learning_rate": 2.091912700292983e-05,
|
|
"loss": 0.2033,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10489645600318909,
|
|
"step": 2305,
|
|
"valid_targets_mean": 5490.9,
|
|
"valid_targets_min": 2301
|
|
},
|
|
{
|
|
"epoch": 3.7689795918367346,
|
|
"grad_norm": 0.5729406847308828,
|
|
"learning_rate": 2.083783825830557e-05,
|
|
"loss": 0.2076,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09271632134914398,
|
|
"step": 2310,
|
|
"valid_targets_mean": 3407.8,
|
|
"valid_targets_min": 1110
|
|
},
|
|
{
|
|
"epoch": 3.777142857142857,
|
|
"grad_norm": 0.5408692265981219,
|
|
"learning_rate": 2.075653564613134e-05,
|
|
"loss": 0.2055,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08484731614589691,
|
|
"step": 2315,
|
|
"valid_targets_mean": 3749.0,
|
|
"valid_targets_min": 2862
|
|
},
|
|
{
|
|
"epoch": 3.7853061224489797,
|
|
"grad_norm": 0.5281560420325818,
|
|
"learning_rate": 2.0675220512094124e-05,
|
|
"loss": 0.2042,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12320023775100708,
|
|
"step": 2320,
|
|
"valid_targets_mean": 5550.1,
|
|
"valid_targets_min": 2278
|
|
},
|
|
{
|
|
"epoch": 3.793469387755102,
|
|
"grad_norm": 0.5655352291536425,
|
|
"learning_rate": 2.059389420208816e-05,
|
|
"loss": 0.2156,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12888170778751373,
|
|
"step": 2325,
|
|
"valid_targets_mean": 3960.6,
|
|
"valid_targets_min": 2243
|
|
},
|
|
{
|
|
"epoch": 3.8016326530612243,
|
|
"grad_norm": 0.5643861980349051,
|
|
"learning_rate": 2.0512558062192655e-05,
|
|
"loss": 0.2117,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14314799010753632,
|
|
"step": 2330,
|
|
"valid_targets_mean": 4727.9,
|
|
"valid_targets_min": 2596
|
|
},
|
|
{
|
|
"epoch": 3.809795918367347,
|
|
"grad_norm": 0.6586685049239898,
|
|
"learning_rate": 2.043121343864954e-05,
|
|
"loss": 0.209,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10116365551948547,
|
|
"step": 2335,
|
|
"valid_targets_mean": 5758.2,
|
|
"valid_targets_min": 2438
|
|
},
|
|
{
|
|
"epoch": 3.8179591836734694,
|
|
"grad_norm": 0.7771486185341817,
|
|
"learning_rate": 2.0349861677841148e-05,
|
|
"loss": 0.2123,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1095040887594223,
|
|
"step": 2340,
|
|
"valid_targets_mean": 4051.2,
|
|
"valid_targets_min": 1823
|
|
},
|
|
{
|
|
"epoch": 3.826122448979592,
|
|
"grad_norm": 0.3983669329609547,
|
|
"learning_rate": 2.0268504126267952e-05,
|
|
"loss": 0.2062,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08110864460468292,
|
|
"step": 2345,
|
|
"valid_targets_mean": 5468.2,
|
|
"valid_targets_min": 1297
|
|
},
|
|
{
|
|
"epoch": 3.8342857142857145,
|
|
"grad_norm": 0.5066751984407911,
|
|
"learning_rate": 2.0187142130526266e-05,
|
|
"loss": 0.206,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11485150456428528,
|
|
"step": 2350,
|
|
"valid_targets_mean": 4955.1,
|
|
"valid_targets_min": 865
|
|
},
|
|
{
|
|
"epoch": 3.8424489795918366,
|
|
"grad_norm": 0.5141538006278502,
|
|
"learning_rate": 2.0105777037285957e-05,
|
|
"loss": 0.2108,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09995323419570923,
|
|
"step": 2355,
|
|
"valid_targets_mean": 4502.9,
|
|
"valid_targets_min": 1772
|
|
},
|
|
{
|
|
"epoch": 3.850612244897959,
|
|
"grad_norm": 0.6241468197660187,
|
|
"learning_rate": 2.0024410193268177e-05,
|
|
"loss": 0.2055,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09824708104133606,
|
|
"step": 2360,
|
|
"valid_targets_mean": 3143.4,
|
|
"valid_targets_min": 1446
|
|
},
|
|
{
|
|
"epoch": 3.8587755102040817,
|
|
"grad_norm": 0.4833346466208415,
|
|
"learning_rate": 1.994304294522304e-05,
|
|
"loss": 0.201,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08165107667446136,
|
|
"step": 2365,
|
|
"valid_targets_mean": 4159.0,
|
|
"valid_targets_min": 532
|
|
},
|
|
{
|
|
"epoch": 3.866938775510204,
|
|
"grad_norm": 0.49067284008168804,
|
|
"learning_rate": 1.9861676639907358e-05,
|
|
"loss": 0.2133,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09489576518535614,
|
|
"step": 2370,
|
|
"valid_targets_mean": 4256.8,
|
|
"valid_targets_min": 2423
|
|
},
|
|
{
|
|
"epoch": 3.8751020408163264,
|
|
"grad_norm": 0.4899645525965157,
|
|
"learning_rate": 1.978031262406233e-05,
|
|
"loss": 0.2151,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09761510044336319,
|
|
"step": 2375,
|
|
"valid_targets_mean": 4272.1,
|
|
"valid_targets_min": 2099
|
|
},
|
|
{
|
|
"epoch": 3.883265306122449,
|
|
"grad_norm": 0.4823414520905191,
|
|
"learning_rate": 1.9698952244391273e-05,
|
|
"loss": 0.2147,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09190724790096283,
|
|
"step": 2380,
|
|
"valid_targets_mean": 4303.9,
|
|
"valid_targets_min": 1791
|
|
},
|
|
{
|
|
"epoch": 3.8914285714285715,
|
|
"grad_norm": 0.4036421152941417,
|
|
"learning_rate": 1.9617596847537314e-05,
|
|
"loss": 0.2015,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09016790986061096,
|
|
"step": 2385,
|
|
"valid_targets_mean": 6770.0,
|
|
"valid_targets_min": 2283
|
|
},
|
|
{
|
|
"epoch": 3.899591836734694,
|
|
"grad_norm": 0.5304022072294978,
|
|
"learning_rate": 1.953624778006109e-05,
|
|
"loss": 0.216,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11460654437541962,
|
|
"step": 2390,
|
|
"valid_targets_mean": 4059.8,
|
|
"valid_targets_min": 2678
|
|
},
|
|
{
|
|
"epoch": 3.9077551020408166,
|
|
"grad_norm": 0.5475140898120625,
|
|
"learning_rate": 1.9454906388418507e-05,
|
|
"loss": 0.2018,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09916389733552933,
|
|
"step": 2395,
|
|
"valid_targets_mean": 4083.2,
|
|
"valid_targets_min": 1999
|
|
},
|
|
{
|
|
"epoch": 3.9159183673469387,
|
|
"grad_norm": 0.5272651986791439,
|
|
"learning_rate": 1.93735740189384e-05,
|
|
"loss": 0.1993,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08801892399787903,
|
|
"step": 2400,
|
|
"valid_targets_mean": 3487.5,
|
|
"valid_targets_min": 2172
|
|
},
|
|
{
|
|
"epoch": 3.924081632653061,
|
|
"grad_norm": 0.4996257867275813,
|
|
"learning_rate": 1.929225201780029e-05,
|
|
"loss": 0.1962,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12266403436660767,
|
|
"step": 2405,
|
|
"valid_targets_mean": 5626.6,
|
|
"valid_targets_min": 2293
|
|
},
|
|
{
|
|
"epoch": 3.9322448979591837,
|
|
"grad_norm": 0.4803588830227563,
|
|
"learning_rate": 1.9210941731012058e-05,
|
|
"loss": 0.2034,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09042441099882126,
|
|
"step": 2410,
|
|
"valid_targets_mean": 4183.6,
|
|
"valid_targets_min": 657
|
|
},
|
|
{
|
|
"epoch": 3.940408163265306,
|
|
"grad_norm": 0.49225801562943217,
|
|
"learning_rate": 1.9129644504387738e-05,
|
|
"loss": 0.2118,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09421433508396149,
|
|
"step": 2415,
|
|
"valid_targets_mean": 4357.8,
|
|
"valid_targets_min": 1251
|
|
},
|
|
{
|
|
"epoch": 3.9485714285714284,
|
|
"grad_norm": 0.4419635106773158,
|
|
"learning_rate": 1.9048361683525155e-05,
|
|
"loss": 0.1873,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08827171474695206,
|
|
"step": 2420,
|
|
"valid_targets_mean": 5476.1,
|
|
"valid_targets_min": 2469
|
|
},
|
|
{
|
|
"epoch": 3.956734693877551,
|
|
"grad_norm": 0.5346905119601936,
|
|
"learning_rate": 1.8967094613783726e-05,
|
|
"loss": 0.1998,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08471651375293732,
|
|
"step": 2425,
|
|
"valid_targets_mean": 3831.2,
|
|
"valid_targets_min": 2011
|
|
},
|
|
{
|
|
"epoch": 3.9648979591836735,
|
|
"grad_norm": 0.4829207115923738,
|
|
"learning_rate": 1.888584464026213e-05,
|
|
"loss": 0.1969,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11744900792837143,
|
|
"step": 2430,
|
|
"valid_targets_mean": 5821.8,
|
|
"valid_targets_min": 726
|
|
},
|
|
{
|
|
"epoch": 3.973061224489796,
|
|
"grad_norm": 0.3872676695587164,
|
|
"learning_rate": 1.8804613107776118e-05,
|
|
"loss": 0.201,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08249861001968384,
|
|
"step": 2435,
|
|
"valid_targets_mean": 6168.4,
|
|
"valid_targets_min": 2748
|
|
},
|
|
{
|
|
"epoch": 3.9812244897959186,
|
|
"grad_norm": 0.44806085077099195,
|
|
"learning_rate": 1.8723401360836178e-05,
|
|
"loss": 0.2127,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11369358003139496,
|
|
"step": 2440,
|
|
"valid_targets_mean": 6784.9,
|
|
"valid_targets_min": 752
|
|
},
|
|
{
|
|
"epoch": 3.9893877551020407,
|
|
"grad_norm": 0.4710951192858675,
|
|
"learning_rate": 1.8642210743625325e-05,
|
|
"loss": 0.212,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12091173231601715,
|
|
"step": 2445,
|
|
"valid_targets_mean": 6086.6,
|
|
"valid_targets_min": 2989
|
|
},
|
|
{
|
|
"epoch": 3.9975510204081632,
|
|
"grad_norm": 0.5450661144221071,
|
|
"learning_rate": 1.8561042599976854e-05,
|
|
"loss": 0.2127,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1176106184720993,
|
|
"step": 2450,
|
|
"valid_targets_mean": 4691.8,
|
|
"valid_targets_min": 2349
|
|
},
|
|
{
|
|
"epoch": 4.0048979591836735,
|
|
"grad_norm": 0.49684305530383777,
|
|
"learning_rate": 1.8479898273352084e-05,
|
|
"loss": 0.1961,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09347330778837204,
|
|
"step": 2455,
|
|
"valid_targets_mean": 4379.8,
|
|
"valid_targets_min": 2255
|
|
},
|
|
{
|
|
"epoch": 4.013061224489796,
|
|
"grad_norm": 0.5102719717526412,
|
|
"learning_rate": 1.8398779106818115e-05,
|
|
"loss": 0.1964,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10384904593229294,
|
|
"step": 2460,
|
|
"valid_targets_mean": 5453.8,
|
|
"valid_targets_min": 2491
|
|
},
|
|
{
|
|
"epoch": 4.021224489795919,
|
|
"grad_norm": 0.4928662125813874,
|
|
"learning_rate": 1.83176864430256e-05,
|
|
"loss": 0.1966,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08712229132652283,
|
|
"step": 2465,
|
|
"valid_targets_mean": 4610.5,
|
|
"valid_targets_min": 1697
|
|
},
|
|
{
|
|
"epoch": 4.029387755102041,
|
|
"grad_norm": 0.5021600618126286,
|
|
"learning_rate": 1.823662162418657e-05,
|
|
"loss": 0.1912,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09403041005134583,
|
|
"step": 2470,
|
|
"valid_targets_mean": 5114.4,
|
|
"valid_targets_min": 1426
|
|
},
|
|
{
|
|
"epoch": 4.037551020408163,
|
|
"grad_norm": 0.5398163844157312,
|
|
"learning_rate": 1.815558599205213e-05,
|
|
"loss": 0.1975,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09044396877288818,
|
|
"step": 2475,
|
|
"valid_targets_mean": 3718.1,
|
|
"valid_targets_min": 1760
|
|
},
|
|
{
|
|
"epoch": 4.045714285714285,
|
|
"grad_norm": 0.9492538214781632,
|
|
"learning_rate": 1.8074580887890317e-05,
|
|
"loss": 0.2029,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10314054042100906,
|
|
"step": 2480,
|
|
"valid_targets_mean": 3802.1,
|
|
"valid_targets_min": 1962
|
|
},
|
|
{
|
|
"epoch": 4.053877551020408,
|
|
"grad_norm": 0.4125298411521795,
|
|
"learning_rate": 1.79936076524639e-05,
|
|
"loss": 0.1926,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09343983232975006,
|
|
"step": 2485,
|
|
"valid_targets_mean": 5443.4,
|
|
"valid_targets_min": 2814
|
|
},
|
|
{
|
|
"epoch": 4.0620408163265305,
|
|
"grad_norm": 0.5505051260285522,
|
|
"learning_rate": 1.791266762600815e-05,
|
|
"loss": 0.1995,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10079911351203918,
|
|
"step": 2490,
|
|
"valid_targets_mean": 4203.0,
|
|
"valid_targets_min": 1870
|
|
},
|
|
{
|
|
"epoch": 4.070204081632653,
|
|
"grad_norm": 0.4793360527143795,
|
|
"learning_rate": 1.7831762148208672e-05,
|
|
"loss": 0.1856,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0887790098786354,
|
|
"step": 2495,
|
|
"valid_targets_mean": 4890.6,
|
|
"valid_targets_min": 2429
|
|
},
|
|
{
|
|
"epoch": 4.078367346938776,
|
|
"grad_norm": 0.5006011398859094,
|
|
"learning_rate": 1.775089255817925e-05,
|
|
"loss": 0.1975,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08640684187412262,
|
|
"step": 2500,
|
|
"valid_targets_mean": 4471.8,
|
|
"valid_targets_min": 729
|
|
},
|
|
{
|
|
"epoch": 4.086530612244898,
|
|
"grad_norm": 0.4823608555582478,
|
|
"learning_rate": 1.7670060194439665e-05,
|
|
"loss": 0.2024,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09267100691795349,
|
|
"step": 2505,
|
|
"valid_targets_mean": 6793.4,
|
|
"valid_targets_min": 3312
|
|
},
|
|
{
|
|
"epoch": 4.094693877551021,
|
|
"grad_norm": 0.4950168084238867,
|
|
"learning_rate": 1.758926639489354e-05,
|
|
"loss": 0.1854,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09220251441001892,
|
|
"step": 2510,
|
|
"valid_targets_mean": 4972.6,
|
|
"valid_targets_min": 3191
|
|
},
|
|
{
|
|
"epoch": 4.102857142857143,
|
|
"grad_norm": 0.5397701907428266,
|
|
"learning_rate": 1.7508512496806202e-05,
|
|
"loss": 0.1932,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11720062792301178,
|
|
"step": 2515,
|
|
"valid_targets_mean": 5273.1,
|
|
"valid_targets_min": 2051
|
|
},
|
|
{
|
|
"epoch": 4.111020408163265,
|
|
"grad_norm": 0.5151168385770297,
|
|
"learning_rate": 1.7427799836782554e-05,
|
|
"loss": 0.1941,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07783135771751404,
|
|
"step": 2520,
|
|
"valid_targets_mean": 3984.2,
|
|
"valid_targets_min": 2415
|
|
},
|
|
{
|
|
"epoch": 4.119183673469387,
|
|
"grad_norm": 0.5475399954363521,
|
|
"learning_rate": 1.7347129750744923e-05,
|
|
"loss": 0.2046,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1131369099020958,
|
|
"step": 2525,
|
|
"valid_targets_mean": 5291.9,
|
|
"valid_targets_min": 1966
|
|
},
|
|
{
|
|
"epoch": 4.12734693877551,
|
|
"grad_norm": 0.4256526467582212,
|
|
"learning_rate": 1.7266503573910995e-05,
|
|
"loss": 0.1975,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09197886288166046,
|
|
"step": 2530,
|
|
"valid_targets_mean": 4821.0,
|
|
"valid_targets_min": 2732
|
|
},
|
|
{
|
|
"epoch": 4.1355102040816325,
|
|
"grad_norm": 0.572725873451191,
|
|
"learning_rate": 1.7185922640771657e-05,
|
|
"loss": 0.1773,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09851355850696564,
|
|
"step": 2535,
|
|
"valid_targets_mean": 3625.0,
|
|
"valid_targets_min": 2596
|
|
},
|
|
{
|
|
"epoch": 4.143673469387755,
|
|
"grad_norm": 0.48901114744309754,
|
|
"learning_rate": 1.710538828506898e-05,
|
|
"loss": 0.1867,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09258966147899628,
|
|
"step": 2540,
|
|
"valid_targets_mean": 5980.6,
|
|
"valid_targets_min": 2077
|
|
},
|
|
{
|
|
"epoch": 4.151836734693878,
|
|
"grad_norm": 0.4713249554585069,
|
|
"learning_rate": 1.7024901839774074e-05,
|
|
"loss": 0.1936,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11058518290519714,
|
|
"step": 2545,
|
|
"valid_targets_mean": 5908.0,
|
|
"valid_targets_min": 2449
|
|
},
|
|
{
|
|
"epoch": 4.16,
|
|
"grad_norm": 0.46189356108689933,
|
|
"learning_rate": 1.6944464637065062e-05,
|
|
"loss": 0.1796,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09432432055473328,
|
|
"step": 2550,
|
|
"valid_targets_mean": 6157.5,
|
|
"valid_targets_min": 2201
|
|
},
|
|
{
|
|
"epoch": 4.168163265306123,
|
|
"grad_norm": 0.49702673341480896,
|
|
"learning_rate": 1.686407800830504e-05,
|
|
"loss": 0.1828,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07615435123443604,
|
|
"step": 2555,
|
|
"valid_targets_mean": 4524.0,
|
|
"valid_targets_min": 796
|
|
},
|
|
{
|
|
"epoch": 4.176326530612245,
|
|
"grad_norm": 0.5819775268795271,
|
|
"learning_rate": 1.678374328402001e-05,
|
|
"loss": 0.194,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11597669124603271,
|
|
"step": 2560,
|
|
"valid_targets_mean": 4323.1,
|
|
"valid_targets_min": 1681
|
|
},
|
|
{
|
|
"epoch": 4.184489795918367,
|
|
"grad_norm": 0.5634760624657875,
|
|
"learning_rate": 1.6703461793876876e-05,
|
|
"loss": 0.1864,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07644718140363693,
|
|
"step": 2565,
|
|
"valid_targets_mean": 3535.8,
|
|
"valid_targets_min": 1396
|
|
},
|
|
{
|
|
"epoch": 4.192653061224489,
|
|
"grad_norm": 0.5663166930782293,
|
|
"learning_rate": 1.6623234866661425e-05,
|
|
"loss": 0.1895,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09392023831605911,
|
|
"step": 2570,
|
|
"valid_targets_mean": 4895.1,
|
|
"valid_targets_min": 2242
|
|
},
|
|
{
|
|
"epoch": 4.200816326530612,
|
|
"grad_norm": 0.4898814924321093,
|
|
"learning_rate": 1.6543063830256365e-05,
|
|
"loss": 0.191,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.074742391705513,
|
|
"step": 2575,
|
|
"valid_targets_mean": 3640.9,
|
|
"valid_targets_min": 302
|
|
},
|
|
{
|
|
"epoch": 4.2089795918367345,
|
|
"grad_norm": 0.5329704151981115,
|
|
"learning_rate": 1.6462950011619294e-05,
|
|
"loss": 0.1888,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10386515408754349,
|
|
"step": 2580,
|
|
"valid_targets_mean": 4789.4,
|
|
"valid_targets_min": 1685
|
|
},
|
|
{
|
|
"epoch": 4.217142857142857,
|
|
"grad_norm": 0.5654217835025043,
|
|
"learning_rate": 1.638289473676079e-05,
|
|
"loss": 0.1819,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10331140458583832,
|
|
"step": 2585,
|
|
"valid_targets_mean": 4166.1,
|
|
"valid_targets_min": 1248
|
|
},
|
|
{
|
|
"epoch": 4.22530612244898,
|
|
"grad_norm": 0.4477703963173839,
|
|
"learning_rate": 1.6302899330722408e-05,
|
|
"loss": 0.1759,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07035226374864578,
|
|
"step": 2590,
|
|
"valid_targets_mean": 4336.6,
|
|
"valid_targets_min": 1659
|
|
},
|
|
{
|
|
"epoch": 4.233469387755102,
|
|
"grad_norm": 0.5146643374207694,
|
|
"learning_rate": 1.622296511755482e-05,
|
|
"loss": 0.182,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10337590426206589,
|
|
"step": 2595,
|
|
"valid_targets_mean": 5123.9,
|
|
"valid_targets_min": 1762
|
|
},
|
|
{
|
|
"epoch": 4.241632653061225,
|
|
"grad_norm": 0.5283593915985714,
|
|
"learning_rate": 1.6143093420295823e-05,
|
|
"loss": 0.1902,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11016341298818588,
|
|
"step": 2600,
|
|
"valid_targets_mean": 5239.8,
|
|
"valid_targets_min": 1520
|
|
},
|
|
{
|
|
"epoch": 4.249795918367347,
|
|
"grad_norm": 0.44759186145233176,
|
|
"learning_rate": 1.6063285560948492e-05,
|
|
"loss": 0.1892,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09216510504484177,
|
|
"step": 2605,
|
|
"valid_targets_mean": 5240.5,
|
|
"valid_targets_min": 1707
|
|
},
|
|
{
|
|
"epoch": 4.257959183673469,
|
|
"grad_norm": 0.6073616846345756,
|
|
"learning_rate": 1.5983542860459292e-05,
|
|
"loss": 0.1899,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.065554678440094,
|
|
"step": 2610,
|
|
"valid_targets_mean": 2690.9,
|
|
"valid_targets_min": 556
|
|
},
|
|
{
|
|
"epoch": 4.2661224489795915,
|
|
"grad_norm": 0.494273576410177,
|
|
"learning_rate": 1.590386663869619e-05,
|
|
"loss": 0.1843,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11738970875740051,
|
|
"step": 2615,
|
|
"valid_targets_mean": 5866.9,
|
|
"valid_targets_min": 3054
|
|
},
|
|
{
|
|
"epoch": 4.274285714285714,
|
|
"grad_norm": 0.47365274825672116,
|
|
"learning_rate": 1.5824258214426833e-05,
|
|
"loss": 0.1874,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08815792948007584,
|
|
"step": 2620,
|
|
"valid_targets_mean": 4863.2,
|
|
"valid_targets_min": 1469
|
|
},
|
|
{
|
|
"epoch": 4.282448979591837,
|
|
"grad_norm": 0.3951324311720292,
|
|
"learning_rate": 1.57447189052967e-05,
|
|
"loss": 0.1972,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07140126824378967,
|
|
"step": 2625,
|
|
"valid_targets_mean": 5357.2,
|
|
"valid_targets_min": 2212
|
|
},
|
|
{
|
|
"epoch": 4.290612244897959,
|
|
"grad_norm": 0.3899464760912804,
|
|
"learning_rate": 1.5665250027807326e-05,
|
|
"loss": 0.1807,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07976418733596802,
|
|
"step": 2630,
|
|
"valid_targets_mean": 6317.6,
|
|
"valid_targets_min": 870
|
|
},
|
|
{
|
|
"epoch": 4.298775510204082,
|
|
"grad_norm": 0.4477544191318257,
|
|
"learning_rate": 1.558585289729448e-05,
|
|
"loss": 0.1886,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08820460736751556,
|
|
"step": 2635,
|
|
"valid_targets_mean": 5996.9,
|
|
"valid_targets_min": 1867
|
|
},
|
|
{
|
|
"epoch": 4.306938775510204,
|
|
"grad_norm": 0.49235787781606305,
|
|
"learning_rate": 1.55065288279064e-05,
|
|
"loss": 0.1931,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07115676999092102,
|
|
"step": 2640,
|
|
"valid_targets_mean": 4227.9,
|
|
"valid_targets_min": 2739
|
|
},
|
|
{
|
|
"epoch": 4.315102040816327,
|
|
"grad_norm": 0.5161672595037532,
|
|
"learning_rate": 1.5427279132582055e-05,
|
|
"loss": 0.195,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10170981287956238,
|
|
"step": 2645,
|
|
"valid_targets_mean": 5293.6,
|
|
"valid_targets_min": 2011
|
|
},
|
|
{
|
|
"epoch": 4.323265306122449,
|
|
"grad_norm": 0.5046151825642934,
|
|
"learning_rate": 1.5348105123029404e-05,
|
|
"loss": 0.1861,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08673232793807983,
|
|
"step": 2650,
|
|
"valid_targets_mean": 4813.0,
|
|
"valid_targets_min": 1235
|
|
},
|
|
{
|
|
"epoch": 4.331428571428571,
|
|
"grad_norm": 0.4703623860473787,
|
|
"learning_rate": 1.5269008109703676e-05,
|
|
"loss": 0.1949,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07027247548103333,
|
|
"step": 2655,
|
|
"valid_targets_mean": 4028.8,
|
|
"valid_targets_min": 1481
|
|
},
|
|
{
|
|
"epoch": 4.3395918367346935,
|
|
"grad_norm": 0.5046661473253734,
|
|
"learning_rate": 1.5189989401785698e-05,
|
|
"loss": 0.1928,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1191679909825325,
|
|
"step": 2660,
|
|
"valid_targets_mean": 5938.4,
|
|
"valid_targets_min": 3219
|
|
},
|
|
{
|
|
"epoch": 4.347755102040816,
|
|
"grad_norm": 0.4248241877322586,
|
|
"learning_rate": 1.5111050307160237e-05,
|
|
"loss": 0.1893,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07479171454906464,
|
|
"step": 2665,
|
|
"valid_targets_mean": 6721.6,
|
|
"valid_targets_min": 424
|
|
},
|
|
{
|
|
"epoch": 4.355918367346939,
|
|
"grad_norm": 0.3557581340270684,
|
|
"learning_rate": 1.5032192132394305e-05,
|
|
"loss": 0.1784,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07210604846477509,
|
|
"step": 2670,
|
|
"valid_targets_mean": 6250.0,
|
|
"valid_targets_min": 2950
|
|
},
|
|
{
|
|
"epoch": 4.364081632653061,
|
|
"grad_norm": 0.5382046936001726,
|
|
"learning_rate": 1.4953416182715566e-05,
|
|
"loss": 0.1923,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07258506119251251,
|
|
"step": 2675,
|
|
"valid_targets_mean": 3225.1,
|
|
"valid_targets_min": 2163
|
|
},
|
|
{
|
|
"epoch": 4.372244897959184,
|
|
"grad_norm": 0.4899944797037151,
|
|
"learning_rate": 1.4874723761990751e-05,
|
|
"loss": 0.1946,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07114744931459427,
|
|
"step": 2680,
|
|
"valid_targets_mean": 4258.1,
|
|
"valid_targets_min": 704
|
|
},
|
|
{
|
|
"epoch": 4.380408163265306,
|
|
"grad_norm": 0.49284944300678446,
|
|
"learning_rate": 1.4796116172704032e-05,
|
|
"loss": 0.1991,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10904914885759354,
|
|
"step": 2685,
|
|
"valid_targets_mean": 6167.9,
|
|
"valid_targets_min": 838
|
|
},
|
|
{
|
|
"epoch": 4.388571428571429,
|
|
"grad_norm": 0.5432323972808517,
|
|
"learning_rate": 1.4717594715935498e-05,
|
|
"loss": 0.197,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09540468454360962,
|
|
"step": 2690,
|
|
"valid_targets_mean": 3727.9,
|
|
"valid_targets_min": 3180
|
|
},
|
|
{
|
|
"epoch": 4.396734693877551,
|
|
"grad_norm": 0.4854316131867214,
|
|
"learning_rate": 1.4639160691339593e-05,
|
|
"loss": 0.1954,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09044314920902252,
|
|
"step": 2695,
|
|
"valid_targets_mean": 4713.0,
|
|
"valid_targets_min": 2170
|
|
},
|
|
{
|
|
"epoch": 4.404897959183674,
|
|
"grad_norm": 0.41399018384818026,
|
|
"learning_rate": 1.456081539712365e-05,
|
|
"loss": 0.1885,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07750784605741501,
|
|
"step": 2700,
|
|
"valid_targets_mean": 6339.5,
|
|
"valid_targets_min": 988
|
|
},
|
|
{
|
|
"epoch": 4.4130612244897955,
|
|
"grad_norm": 0.45239236721280934,
|
|
"learning_rate": 1.4482560130026342e-05,
|
|
"loss": 0.187,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09557928144931793,
|
|
"step": 2705,
|
|
"valid_targets_mean": 5853.2,
|
|
"valid_targets_min": 2915
|
|
},
|
|
{
|
|
"epoch": 4.421224489795918,
|
|
"grad_norm": 0.5584019817797293,
|
|
"learning_rate": 1.4404396185296268e-05,
|
|
"loss": 0.1916,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08847010135650635,
|
|
"step": 2710,
|
|
"valid_targets_mean": 4407.9,
|
|
"valid_targets_min": 1298
|
|
},
|
|
{
|
|
"epoch": 4.429387755102041,
|
|
"grad_norm": 0.48880926965662114,
|
|
"learning_rate": 1.4326324856670494e-05,
|
|
"loss": 0.1904,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07466118037700653,
|
|
"step": 2715,
|
|
"valid_targets_mean": 4065.1,
|
|
"valid_targets_min": 2074
|
|
},
|
|
{
|
|
"epoch": 4.437551020408163,
|
|
"grad_norm": 0.4277135480825074,
|
|
"learning_rate": 1.4248347436353147e-05,
|
|
"loss": 0.1908,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0795067697763443,
|
|
"step": 2720,
|
|
"valid_targets_mean": 5436.0,
|
|
"valid_targets_min": 1588
|
|
},
|
|
{
|
|
"epoch": 4.445714285714286,
|
|
"grad_norm": 0.5413528832361828,
|
|
"learning_rate": 1.4170465214994011e-05,
|
|
"loss": 0.1885,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10669746994972229,
|
|
"step": 2725,
|
|
"valid_targets_mean": 4772.5,
|
|
"valid_targets_min": 1882
|
|
},
|
|
{
|
|
"epoch": 4.453877551020408,
|
|
"grad_norm": 0.4731571122632174,
|
|
"learning_rate": 1.409267948166718e-05,
|
|
"loss": 0.192,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08785238116979599,
|
|
"step": 2730,
|
|
"valid_targets_mean": 5070.4,
|
|
"valid_targets_min": 2187
|
|
},
|
|
{
|
|
"epoch": 4.462040816326531,
|
|
"grad_norm": 0.5039704029344154,
|
|
"learning_rate": 1.4014991523849726e-05,
|
|
"loss": 0.1984,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10380153357982635,
|
|
"step": 2735,
|
|
"valid_targets_mean": 4711.8,
|
|
"valid_targets_min": 1346
|
|
},
|
|
{
|
|
"epoch": 4.470204081632653,
|
|
"grad_norm": 0.542316983338691,
|
|
"learning_rate": 1.3937402627400384e-05,
|
|
"loss": 0.1971,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13545742630958557,
|
|
"step": 2740,
|
|
"valid_targets_mean": 6321.9,
|
|
"valid_targets_min": 1489
|
|
},
|
|
{
|
|
"epoch": 4.478367346938776,
|
|
"grad_norm": 0.47877743495837094,
|
|
"learning_rate": 1.3859914076538245e-05,
|
|
"loss": 0.1944,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09974990785121918,
|
|
"step": 2745,
|
|
"valid_targets_mean": 5603.5,
|
|
"valid_targets_min": 2886
|
|
},
|
|
{
|
|
"epoch": 4.486530612244898,
|
|
"grad_norm": 0.5437938225021358,
|
|
"learning_rate": 1.3782527153821554e-05,
|
|
"loss": 0.1797,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1074422299861908,
|
|
"step": 2750,
|
|
"valid_targets_mean": 4773.6,
|
|
"valid_targets_min": 2787
|
|
},
|
|
{
|
|
"epoch": 4.49469387755102,
|
|
"grad_norm": 0.5296213977075995,
|
|
"learning_rate": 1.370524314012642e-05,
|
|
"loss": 0.1829,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1249004527926445,
|
|
"step": 2755,
|
|
"valid_targets_mean": 5794.8,
|
|
"valid_targets_min": 2462
|
|
},
|
|
{
|
|
"epoch": 4.502857142857143,
|
|
"grad_norm": 0.4448019181831348,
|
|
"learning_rate": 1.3628063314625655e-05,
|
|
"loss": 0.1785,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07824553549289703,
|
|
"step": 2760,
|
|
"valid_targets_mean": 5313.8,
|
|
"valid_targets_min": 1962
|
|
},
|
|
{
|
|
"epoch": 4.511020408163265,
|
|
"grad_norm": 0.5684669670820979,
|
|
"learning_rate": 1.355098895476759e-05,
|
|
"loss": 0.1913,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10887186229228973,
|
|
"step": 2765,
|
|
"valid_targets_mean": 4395.8,
|
|
"valid_targets_min": 2515
|
|
},
|
|
{
|
|
"epoch": 4.519183673469388,
|
|
"grad_norm": 0.37621169472643107,
|
|
"learning_rate": 1.347402133625495e-05,
|
|
"loss": 0.1762,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07432250678539276,
|
|
"step": 2770,
|
|
"valid_targets_mean": 5932.8,
|
|
"valid_targets_min": 1810
|
|
},
|
|
{
|
|
"epoch": 4.52734693877551,
|
|
"grad_norm": 0.5608359554833122,
|
|
"learning_rate": 1.3397161733023698e-05,
|
|
"loss": 0.1892,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0860264003276825,
|
|
"step": 2775,
|
|
"valid_targets_mean": 3904.0,
|
|
"valid_targets_min": 1149
|
|
},
|
|
{
|
|
"epoch": 4.535510204081633,
|
|
"grad_norm": 0.4795668691745758,
|
|
"learning_rate": 1.3320411417221974e-05,
|
|
"loss": 0.1967,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10846267640590668,
|
|
"step": 2780,
|
|
"valid_targets_mean": 5842.1,
|
|
"valid_targets_min": 2204
|
|
},
|
|
{
|
|
"epoch": 4.543673469387755,
|
|
"grad_norm": 0.375899081661828,
|
|
"learning_rate": 1.324377165918906e-05,
|
|
"loss": 0.1875,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.055310845375061035,
|
|
"step": 2785,
|
|
"valid_targets_mean": 5695.2,
|
|
"valid_targets_min": 1003
|
|
},
|
|
{
|
|
"epoch": 4.551836734693877,
|
|
"grad_norm": 0.48026420303225553,
|
|
"learning_rate": 1.3167243727434307e-05,
|
|
"loss": 0.1974,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10313354432582855,
|
|
"step": 2790,
|
|
"valid_targets_mean": 5187.1,
|
|
"valid_targets_min": 3179
|
|
},
|
|
{
|
|
"epoch": 4.5600000000000005,
|
|
"grad_norm": 0.42389844482180566,
|
|
"learning_rate": 1.3090828888616183e-05,
|
|
"loss": 0.1794,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0987774059176445,
|
|
"step": 2795,
|
|
"valid_targets_mean": 6063.2,
|
|
"valid_targets_min": 1414
|
|
},
|
|
{
|
|
"epoch": 4.568163265306122,
|
|
"grad_norm": 0.492898294974917,
|
|
"learning_rate": 1.301452840752127e-05,
|
|
"loss": 0.1902,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09452690184116364,
|
|
"step": 2800,
|
|
"valid_targets_mean": 5125.5,
|
|
"valid_targets_min": 1920
|
|
},
|
|
{
|
|
"epoch": 4.576326530612245,
|
|
"grad_norm": 0.5922819701493808,
|
|
"learning_rate": 1.293834354704337e-05,
|
|
"loss": 0.1879,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1072399765253067,
|
|
"step": 2805,
|
|
"valid_targets_mean": 4184.5,
|
|
"valid_targets_min": 2043
|
|
},
|
|
{
|
|
"epoch": 4.584489795918367,
|
|
"grad_norm": 0.5009712991870972,
|
|
"learning_rate": 1.2862275568162566e-05,
|
|
"loss": 0.189,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12208345532417297,
|
|
"step": 2810,
|
|
"valid_targets_mean": 5785.0,
|
|
"valid_targets_min": 1694
|
|
},
|
|
{
|
|
"epoch": 4.59265306122449,
|
|
"grad_norm": 0.5019101438266571,
|
|
"learning_rate": 1.2786325729924362e-05,
|
|
"loss": 0.1857,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11706788837909698,
|
|
"step": 2815,
|
|
"valid_targets_mean": 6469.1,
|
|
"valid_targets_min": 1605
|
|
},
|
|
{
|
|
"epoch": 4.600816326530612,
|
|
"grad_norm": 0.5425198767387878,
|
|
"learning_rate": 1.2710495289418869e-05,
|
|
"loss": 0.1898,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08403375744819641,
|
|
"step": 2820,
|
|
"valid_targets_mean": 3734.6,
|
|
"valid_targets_min": 1120
|
|
},
|
|
{
|
|
"epoch": 4.608979591836735,
|
|
"grad_norm": 0.47797549079237955,
|
|
"learning_rate": 1.2634785501759957e-05,
|
|
"loss": 0.1881,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08239451050758362,
|
|
"step": 2825,
|
|
"valid_targets_mean": 4446.0,
|
|
"valid_targets_min": 2288
|
|
},
|
|
{
|
|
"epoch": 4.617142857142857,
|
|
"grad_norm": 0.5237661865631653,
|
|
"learning_rate": 1.2559197620064501e-05,
|
|
"loss": 0.1804,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11617680639028549,
|
|
"step": 2830,
|
|
"valid_targets_mean": 5307.2,
|
|
"valid_targets_min": 2390
|
|
},
|
|
{
|
|
"epoch": 4.625306122448979,
|
|
"grad_norm": 0.49168671805845504,
|
|
"learning_rate": 1.2483732895431637e-05,
|
|
"loss": 0.2024,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1150754764676094,
|
|
"step": 2835,
|
|
"valid_targets_mean": 6200.5,
|
|
"valid_targets_min": 2398
|
|
},
|
|
{
|
|
"epoch": 4.6334693877551025,
|
|
"grad_norm": 0.48627176014261897,
|
|
"learning_rate": 1.2408392576922075e-05,
|
|
"loss": 0.1959,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08021821081638336,
|
|
"step": 2840,
|
|
"valid_targets_mean": 5957.8,
|
|
"valid_targets_min": 2327
|
|
},
|
|
{
|
|
"epoch": 4.641632653061224,
|
|
"grad_norm": 0.44639892122473207,
|
|
"learning_rate": 1.2333177911537396e-05,
|
|
"loss": 0.1953,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08871643245220184,
|
|
"step": 2845,
|
|
"valid_targets_mean": 5963.0,
|
|
"valid_targets_min": 2091
|
|
},
|
|
{
|
|
"epoch": 4.649795918367347,
|
|
"grad_norm": 0.5226387363165819,
|
|
"learning_rate": 1.2258090144199409e-05,
|
|
"loss": 0.1869,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08227618038654327,
|
|
"step": 2850,
|
|
"valid_targets_mean": 3925.1,
|
|
"valid_targets_min": 2181
|
|
},
|
|
{
|
|
"epoch": 4.657959183673469,
|
|
"grad_norm": 0.4332462492586316,
|
|
"learning_rate": 1.2183130517729586e-05,
|
|
"loss": 0.1831,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.081905797123909,
|
|
"step": 2855,
|
|
"valid_targets_mean": 5561.9,
|
|
"valid_targets_min": 2039
|
|
},
|
|
{
|
|
"epoch": 4.666122448979592,
|
|
"grad_norm": 0.5683336516651115,
|
|
"learning_rate": 1.2108300272828446e-05,
|
|
"loss": 0.1939,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10651646554470062,
|
|
"step": 2860,
|
|
"valid_targets_mean": 5148.9,
|
|
"valid_targets_min": 2376
|
|
},
|
|
{
|
|
"epoch": 4.674285714285714,
|
|
"grad_norm": 0.44839763343296324,
|
|
"learning_rate": 1.2033600648055038e-05,
|
|
"loss": 0.1966,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08386656641960144,
|
|
"step": 2865,
|
|
"valid_targets_mean": 5665.2,
|
|
"valid_targets_min": 228
|
|
},
|
|
{
|
|
"epoch": 4.682448979591837,
|
|
"grad_norm": 0.5024860369005062,
|
|
"learning_rate": 1.1959032879806442e-05,
|
|
"loss": 0.1918,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09913516044616699,
|
|
"step": 2870,
|
|
"valid_targets_mean": 4861.0,
|
|
"valid_targets_min": 2413
|
|
},
|
|
{
|
|
"epoch": 4.6906122448979595,
|
|
"grad_norm": 0.49564273386489915,
|
|
"learning_rate": 1.1884598202297327e-05,
|
|
"loss": 0.1908,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10049331933259964,
|
|
"step": 2875,
|
|
"valid_targets_mean": 5382.5,
|
|
"valid_targets_min": 1733
|
|
},
|
|
{
|
|
"epoch": 4.698775510204082,
|
|
"grad_norm": 0.46680473946107015,
|
|
"learning_rate": 1.1810297847539464e-05,
|
|
"loss": 0.2062,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08834764361381531,
|
|
"step": 2880,
|
|
"valid_targets_mean": 4403.9,
|
|
"valid_targets_min": 1545
|
|
},
|
|
{
|
|
"epoch": 4.706938775510205,
|
|
"grad_norm": 0.5099284522866282,
|
|
"learning_rate": 1.1736133045321387e-05,
|
|
"loss": 0.1886,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09604378044605255,
|
|
"step": 2885,
|
|
"valid_targets_mean": 5398.1,
|
|
"valid_targets_min": 2002
|
|
},
|
|
{
|
|
"epoch": 4.715102040816326,
|
|
"grad_norm": 0.44693345599274104,
|
|
"learning_rate": 1.1662105023188027e-05,
|
|
"loss": 0.1797,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04823494702577591,
|
|
"step": 2890,
|
|
"valid_targets_mean": 3021.4,
|
|
"valid_targets_min": 350
|
|
},
|
|
{
|
|
"epoch": 4.723265306122449,
|
|
"grad_norm": 0.5114382893021212,
|
|
"learning_rate": 1.1588215006420374e-05,
|
|
"loss": 0.1928,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08465202152729034,
|
|
"step": 2895,
|
|
"valid_targets_mean": 3635.6,
|
|
"valid_targets_min": 574
|
|
},
|
|
{
|
|
"epoch": 4.731428571428571,
|
|
"grad_norm": 0.5061445498959004,
|
|
"learning_rate": 1.1514464218015218e-05,
|
|
"loss": 0.1862,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08960931748151779,
|
|
"step": 2900,
|
|
"valid_targets_mean": 5353.8,
|
|
"valid_targets_min": 2572
|
|
},
|
|
{
|
|
"epoch": 4.739591836734694,
|
|
"grad_norm": 1.103649871943999,
|
|
"learning_rate": 1.1440853878664902e-05,
|
|
"loss": 0.1928,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11720076203346252,
|
|
"step": 2905,
|
|
"valid_targets_mean": 5939.9,
|
|
"valid_targets_min": 1889
|
|
},
|
|
{
|
|
"epoch": 4.747755102040816,
|
|
"grad_norm": 0.4954748711036458,
|
|
"learning_rate": 1.1367385206737106e-05,
|
|
"loss": 0.1846,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07366099953651428,
|
|
"step": 2910,
|
|
"valid_targets_mean": 3961.4,
|
|
"valid_targets_min": 2342
|
|
},
|
|
{
|
|
"epoch": 4.755918367346939,
|
|
"grad_norm": 0.49215147868377257,
|
|
"learning_rate": 1.129405941825471e-05,
|
|
"loss": 0.1934,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08957966417074203,
|
|
"step": 2915,
|
|
"valid_targets_mean": 4154.5,
|
|
"valid_targets_min": 1411
|
|
},
|
|
{
|
|
"epoch": 4.7640816326530615,
|
|
"grad_norm": 0.5405766702143414,
|
|
"learning_rate": 1.1220877726875603e-05,
|
|
"loss": 0.1896,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10561898350715637,
|
|
"step": 2920,
|
|
"valid_targets_mean": 4969.0,
|
|
"valid_targets_min": 2225
|
|
},
|
|
{
|
|
"epoch": 4.772244897959184,
|
|
"grad_norm": 0.460865615051362,
|
|
"learning_rate": 1.1147841343872699e-05,
|
|
"loss": 0.1815,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07561326026916504,
|
|
"step": 2925,
|
|
"valid_targets_mean": 4488.9,
|
|
"valid_targets_min": 1389
|
|
},
|
|
{
|
|
"epoch": 4.780408163265307,
|
|
"grad_norm": 0.5238153626279427,
|
|
"learning_rate": 1.1074951478113774e-05,
|
|
"loss": 0.2038,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09728378057479858,
|
|
"step": 2930,
|
|
"valid_targets_mean": 5040.6,
|
|
"valid_targets_min": 2158
|
|
},
|
|
{
|
|
"epoch": 4.788571428571428,
|
|
"grad_norm": 0.44238272143257545,
|
|
"learning_rate": 1.1002209336041542e-05,
|
|
"loss": 0.184,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08578670024871826,
|
|
"step": 2935,
|
|
"valid_targets_mean": 4627.5,
|
|
"valid_targets_min": 1124
|
|
},
|
|
{
|
|
"epoch": 4.796734693877551,
|
|
"grad_norm": 0.43479390970924037,
|
|
"learning_rate": 1.092961612165365e-05,
|
|
"loss": 0.1815,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06755372136831284,
|
|
"step": 2940,
|
|
"valid_targets_mean": 6472.2,
|
|
"valid_targets_min": 1480
|
|
},
|
|
{
|
|
"epoch": 4.804897959183673,
|
|
"grad_norm": 0.5853935194690773,
|
|
"learning_rate": 1.085717303648275e-05,
|
|
"loss": 0.1966,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11776626110076904,
|
|
"step": 2945,
|
|
"valid_targets_mean": 4299.6,
|
|
"valid_targets_min": 148
|
|
},
|
|
{
|
|
"epoch": 4.813061224489796,
|
|
"grad_norm": 0.5131471997499106,
|
|
"learning_rate": 1.0784881279576635e-05,
|
|
"loss": 0.1917,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.096668541431427,
|
|
"step": 2950,
|
|
"valid_targets_mean": 4616.1,
|
|
"valid_targets_min": 2963
|
|
},
|
|
{
|
|
"epoch": 4.8212244897959184,
|
|
"grad_norm": 0.5679561757392165,
|
|
"learning_rate": 1.0712742047478346e-05,
|
|
"loss": 0.1937,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07576416432857513,
|
|
"step": 2955,
|
|
"valid_targets_mean": 3169.6,
|
|
"valid_targets_min": 1363
|
|
},
|
|
{
|
|
"epoch": 4.829387755102041,
|
|
"grad_norm": 0.5331028375004665,
|
|
"learning_rate": 1.0640756534206412e-05,
|
|
"loss": 0.1837,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09721846878528595,
|
|
"step": 2960,
|
|
"valid_targets_mean": 3602.8,
|
|
"valid_targets_min": 987
|
|
},
|
|
{
|
|
"epoch": 4.8375510204081635,
|
|
"grad_norm": 0.5645277289402424,
|
|
"learning_rate": 1.0568925931235095e-05,
|
|
"loss": 0.1898,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11298727989196777,
|
|
"step": 2965,
|
|
"valid_targets_mean": 4520.5,
|
|
"valid_targets_min": 2114
|
|
},
|
|
{
|
|
"epoch": 4.845714285714286,
|
|
"grad_norm": 0.4145507098470101,
|
|
"learning_rate": 1.0497251427474602e-05,
|
|
"loss": 0.1896,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09271155297756195,
|
|
"step": 2970,
|
|
"valid_targets_mean": 6519.9,
|
|
"valid_targets_min": 2358
|
|
},
|
|
{
|
|
"epoch": 4.853877551020409,
|
|
"grad_norm": 0.5313111617754027,
|
|
"learning_rate": 1.0425734209251486e-05,
|
|
"loss": 0.2019,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09830297529697418,
|
|
"step": 2975,
|
|
"valid_targets_mean": 4779.5,
|
|
"valid_targets_min": 522
|
|
},
|
|
{
|
|
"epoch": 4.86204081632653,
|
|
"grad_norm": 0.5361791729354184,
|
|
"learning_rate": 1.035437546028896e-05,
|
|
"loss": 0.1796,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10205239057540894,
|
|
"step": 2980,
|
|
"valid_targets_mean": 4452.4,
|
|
"valid_targets_min": 2395
|
|
},
|
|
{
|
|
"epoch": 4.870204081632653,
|
|
"grad_norm": 0.5333208117311148,
|
|
"learning_rate": 1.028317636168732e-05,
|
|
"loss": 0.1872,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10787051916122437,
|
|
"step": 2985,
|
|
"valid_targets_mean": 5309.5,
|
|
"valid_targets_min": 1707
|
|
},
|
|
{
|
|
"epoch": 4.878367346938775,
|
|
"grad_norm": 0.4620967725507521,
|
|
"learning_rate": 1.0212138091904409e-05,
|
|
"loss": 0.1945,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10534008592367172,
|
|
"step": 2990,
|
|
"valid_targets_mean": 5099.0,
|
|
"valid_targets_min": 1974
|
|
},
|
|
{
|
|
"epoch": 4.886530612244898,
|
|
"grad_norm": 0.6802090469740967,
|
|
"learning_rate": 1.0141261826736067e-05,
|
|
"loss": 0.1889,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07617118954658508,
|
|
"step": 2995,
|
|
"valid_targets_mean": 2980.5,
|
|
"valid_targets_min": 557
|
|
},
|
|
{
|
|
"epoch": 4.8946938775510205,
|
|
"grad_norm": 0.5014624004024499,
|
|
"learning_rate": 1.0070548739296746e-05,
|
|
"loss": 0.1766,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06977512687444687,
|
|
"step": 3000,
|
|
"valid_targets_mean": 3862.0,
|
|
"valid_targets_min": 1983
|
|
},
|
|
{
|
|
"epoch": 4.902857142857143,
|
|
"grad_norm": 0.5733710902901722,
|
|
"learning_rate": 1.0000000000000006e-05,
|
|
"loss": 0.1967,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0930173248052597,
|
|
"step": 3005,
|
|
"valid_targets_mean": 4976.5,
|
|
"valid_targets_min": 2294
|
|
},
|
|
{
|
|
"epoch": 4.911020408163266,
|
|
"grad_norm": 0.5198355148927837,
|
|
"learning_rate": 9.929616776539199e-06,
|
|
"loss": 0.1966,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09638778120279312,
|
|
"step": 3010,
|
|
"valid_targets_mean": 5231.0,
|
|
"valid_targets_min": 1368
|
|
},
|
|
{
|
|
"epoch": 4.919183673469388,
|
|
"grad_norm": 0.49759517932274444,
|
|
"learning_rate": 9.859400233868159e-06,
|
|
"loss": 0.1829,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11821071803569794,
|
|
"step": 3015,
|
|
"valid_targets_mean": 5549.5,
|
|
"valid_targets_min": 3325
|
|
},
|
|
{
|
|
"epoch": 4.927346938775511,
|
|
"grad_norm": 0.5400270338184223,
|
|
"learning_rate": 9.789351534181837e-06,
|
|
"loss": 0.187,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09970055520534515,
|
|
"step": 3020,
|
|
"valid_targets_mean": 4339.2,
|
|
"valid_targets_min": 2592
|
|
},
|
|
{
|
|
"epoch": 4.935510204081632,
|
|
"grad_norm": 0.518057347117569,
|
|
"learning_rate": 9.719471836897156e-06,
|
|
"loss": 0.1794,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10430602729320526,
|
|
"step": 3025,
|
|
"valid_targets_mean": 4877.8,
|
|
"valid_targets_min": 779
|
|
},
|
|
{
|
|
"epoch": 4.943673469387755,
|
|
"grad_norm": 0.5336142059133698,
|
|
"learning_rate": 9.649762298633747e-06,
|
|
"loss": 0.2045,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1243668645620346,
|
|
"step": 3030,
|
|
"valid_targets_mean": 5056.6,
|
|
"valid_targets_min": 2234
|
|
},
|
|
{
|
|
"epoch": 4.951836734693877,
|
|
"grad_norm": 0.5272384020096476,
|
|
"learning_rate": 9.58022407319488e-06,
|
|
"loss": 0.192,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09622962027788162,
|
|
"step": 3035,
|
|
"valid_targets_mean": 4204.1,
|
|
"valid_targets_min": 172
|
|
},
|
|
{
|
|
"epoch": 4.96,
|
|
"grad_norm": 0.4502669212181341,
|
|
"learning_rate": 9.510858311548306e-06,
|
|
"loss": 0.1919,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09366258978843689,
|
|
"step": 3040,
|
|
"valid_targets_mean": 6401.0,
|
|
"valid_targets_min": 1612
|
|
},
|
|
{
|
|
"epoch": 4.9681632653061225,
|
|
"grad_norm": 0.5272654404132141,
|
|
"learning_rate": 9.441666161807206e-06,
|
|
"loss": 0.1943,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10728336870670319,
|
|
"step": 3045,
|
|
"valid_targets_mean": 4781.5,
|
|
"valid_targets_min": 786
|
|
},
|
|
{
|
|
"epoch": 4.976326530612245,
|
|
"grad_norm": 0.40424101442932664,
|
|
"learning_rate": 9.372648769211259e-06,
|
|
"loss": 0.1918,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07456552982330322,
|
|
"step": 3050,
|
|
"valid_targets_mean": 4632.0,
|
|
"valid_targets_min": 2145
|
|
},
|
|
{
|
|
"epoch": 4.984489795918368,
|
|
"grad_norm": 0.5600282228675467,
|
|
"learning_rate": 9.30380727610759e-06,
|
|
"loss": 0.1999,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10491231083869934,
|
|
"step": 3055,
|
|
"valid_targets_mean": 5399.2,
|
|
"valid_targets_min": 2148
|
|
},
|
|
{
|
|
"epoch": 4.99265306122449,
|
|
"grad_norm": 0.3787674156466776,
|
|
"learning_rate": 9.235142821931928e-06,
|
|
"loss": 0.1861,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08586746454238892,
|
|
"step": 3060,
|
|
"valid_targets_mean": 8370.4,
|
|
"valid_targets_min": 4348
|
|
},
|
|
{
|
|
"epoch": 5.0,
|
|
"grad_norm": 0.7067710949929604,
|
|
"learning_rate": 9.166656543189733e-06,
|
|
"loss": 0.184,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18785783648490906,
|
|
"step": 3065,
|
|
"valid_targets_mean": 5040.1,
|
|
"valid_targets_min": 2867
|
|
},
|
|
{
|
|
"epoch": 5.0081632653061225,
|
|
"grad_norm": 0.4788393552290572,
|
|
"learning_rate": 9.098349573437368e-06,
|
|
"loss": 0.1672,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09013266861438751,
|
|
"step": 3070,
|
|
"valid_targets_mean": 4971.5,
|
|
"valid_targets_min": 2198
|
|
},
|
|
{
|
|
"epoch": 5.016326530612245,
|
|
"grad_norm": 0.5452058782222013,
|
|
"learning_rate": 9.03022304326336e-06,
|
|
"loss": 0.1836,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12489248812198639,
|
|
"step": 3075,
|
|
"valid_targets_mean": 6247.5,
|
|
"valid_targets_min": 3417
|
|
},
|
|
{
|
|
"epoch": 5.024489795918368,
|
|
"grad_norm": 0.4958317510599947,
|
|
"learning_rate": 8.962278080269649e-06,
|
|
"loss": 0.1868,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09371688961982727,
|
|
"step": 3080,
|
|
"valid_targets_mean": 5708.5,
|
|
"valid_targets_min": 2568
|
|
},
|
|
{
|
|
"epoch": 5.03265306122449,
|
|
"grad_norm": 0.4874809774022175,
|
|
"learning_rate": 8.89451580905299e-06,
|
|
"loss": 0.1822,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09922803193330765,
|
|
"step": 3085,
|
|
"valid_targets_mean": 5210.1,
|
|
"valid_targets_min": 2680
|
|
},
|
|
{
|
|
"epoch": 5.040816326530612,
|
|
"grad_norm": 0.43812239820329635,
|
|
"learning_rate": 8.82693735118628e-06,
|
|
"loss": 0.1642,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0819573774933815,
|
|
"step": 3090,
|
|
"valid_targets_mean": 4633.8,
|
|
"valid_targets_min": 1840
|
|
},
|
|
{
|
|
"epoch": 5.048979591836734,
|
|
"grad_norm": 0.5062954532674303,
|
|
"learning_rate": 8.759543825200001e-06,
|
|
"loss": 0.1879,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12129976600408554,
|
|
"step": 3095,
|
|
"valid_targets_mean": 5984.4,
|
|
"valid_targets_min": 2152
|
|
},
|
|
{
|
|
"epoch": 5.057142857142857,
|
|
"grad_norm": 0.4048917822614195,
|
|
"learning_rate": 8.692336346563735e-06,
|
|
"loss": 0.1739,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05262184143066406,
|
|
"step": 3100,
|
|
"valid_targets_mean": 4425.0,
|
|
"valid_targets_min": 1267
|
|
},
|
|
{
|
|
"epoch": 5.0653061224489795,
|
|
"grad_norm": 0.5806920910174197,
|
|
"learning_rate": 8.625316027667685e-06,
|
|
"loss": 0.1794,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09791389107704163,
|
|
"step": 3105,
|
|
"valid_targets_mean": 4411.6,
|
|
"valid_targets_min": 1893
|
|
},
|
|
{
|
|
"epoch": 5.073469387755102,
|
|
"grad_norm": 0.5005518459398053,
|
|
"learning_rate": 8.558483977804265e-06,
|
|
"loss": 0.1792,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07806237041950226,
|
|
"step": 3110,
|
|
"valid_targets_mean": 4310.2,
|
|
"valid_targets_min": 1229
|
|
},
|
|
{
|
|
"epoch": 5.081632653061225,
|
|
"grad_norm": 0.4362967329293203,
|
|
"learning_rate": 8.491841303149728e-06,
|
|
"loss": 0.1831,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0794868916273117,
|
|
"step": 3115,
|
|
"valid_targets_mean": 6788.8,
|
|
"valid_targets_min": 1302
|
|
},
|
|
{
|
|
"epoch": 5.089795918367347,
|
|
"grad_norm": 0.5707602601433491,
|
|
"learning_rate": 8.425389106745881e-06,
|
|
"loss": 0.1739,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09465576708316803,
|
|
"step": 3120,
|
|
"valid_targets_mean": 4276.5,
|
|
"valid_targets_min": 1630
|
|
},
|
|
{
|
|
"epoch": 5.09795918367347,
|
|
"grad_norm": 0.5475463253467584,
|
|
"learning_rate": 8.35912848848181e-06,
|
|
"loss": 0.1748,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07283581793308258,
|
|
"step": 3125,
|
|
"valid_targets_mean": 3742.4,
|
|
"valid_targets_min": 2123
|
|
},
|
|
{
|
|
"epoch": 5.106122448979592,
|
|
"grad_norm": 0.5650276786893733,
|
|
"learning_rate": 8.293060545075667e-06,
|
|
"loss": 0.1844,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07957779616117477,
|
|
"step": 3130,
|
|
"valid_targets_mean": 3538.2,
|
|
"valid_targets_min": 209
|
|
},
|
|
{
|
|
"epoch": 5.114285714285714,
|
|
"grad_norm": 0.4742456395275569,
|
|
"learning_rate": 8.227186370056546e-06,
|
|
"loss": 0.1803,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08860338479280472,
|
|
"step": 3135,
|
|
"valid_targets_mean": 6125.4,
|
|
"valid_targets_min": 1943
|
|
},
|
|
{
|
|
"epoch": 5.122448979591836,
|
|
"grad_norm": 0.4533762535853237,
|
|
"learning_rate": 8.16150705374636e-06,
|
|
"loss": 0.1826,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06944446265697479,
|
|
"step": 3140,
|
|
"valid_targets_mean": 5134.6,
|
|
"valid_targets_min": 1513
|
|
},
|
|
{
|
|
"epoch": 5.130612244897959,
|
|
"grad_norm": 0.5879056841548346,
|
|
"learning_rate": 8.096023683241805e-06,
|
|
"loss": 0.1794,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08228239417076111,
|
|
"step": 3145,
|
|
"valid_targets_mean": 3871.9,
|
|
"valid_targets_min": 2275
|
|
},
|
|
{
|
|
"epoch": 5.1387755102040815,
|
|
"grad_norm": 0.45175621395683185,
|
|
"learning_rate": 8.030737342396362e-06,
|
|
"loss": 0.1816,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08341191709041595,
|
|
"step": 3150,
|
|
"valid_targets_mean": 6288.5,
|
|
"valid_targets_min": 1892
|
|
},
|
|
{
|
|
"epoch": 5.146938775510204,
|
|
"grad_norm": 0.5069533232855673,
|
|
"learning_rate": 7.965649111802358e-06,
|
|
"loss": 0.178,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08151441812515259,
|
|
"step": 3155,
|
|
"valid_targets_mean": 4190.8,
|
|
"valid_targets_min": 2441
|
|
},
|
|
{
|
|
"epoch": 5.155102040816327,
|
|
"grad_norm": 0.38454817482765574,
|
|
"learning_rate": 7.900760068773083e-06,
|
|
"loss": 0.1818,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08950209617614746,
|
|
"step": 3160,
|
|
"valid_targets_mean": 6927.8,
|
|
"valid_targets_min": 3518
|
|
},
|
|
{
|
|
"epoch": 5.163265306122449,
|
|
"grad_norm": 0.5503677993367798,
|
|
"learning_rate": 7.836071287324965e-06,
|
|
"loss": 0.1878,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08563916385173798,
|
|
"step": 3165,
|
|
"valid_targets_mean": 3916.5,
|
|
"valid_targets_min": 2768
|
|
},
|
|
{
|
|
"epoch": 5.171428571428572,
|
|
"grad_norm": 0.4633224599111477,
|
|
"learning_rate": 7.771583838159756e-06,
|
|
"loss": 0.171,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0768350288271904,
|
|
"step": 3170,
|
|
"valid_targets_mean": 5713.6,
|
|
"valid_targets_min": 2432
|
|
},
|
|
{
|
|
"epoch": 5.179591836734694,
|
|
"grad_norm": 0.547107579940367,
|
|
"learning_rate": 7.70729878864689e-06,
|
|
"loss": 0.1774,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08860382437705994,
|
|
"step": 3175,
|
|
"valid_targets_mean": 5152.6,
|
|
"valid_targets_min": 2714
|
|
},
|
|
{
|
|
"epoch": 5.187755102040816,
|
|
"grad_norm": 0.5499484727345023,
|
|
"learning_rate": 7.643217202805715e-06,
|
|
"loss": 0.175,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10342984646558762,
|
|
"step": 3180,
|
|
"valid_targets_mean": 5412.6,
|
|
"valid_targets_min": 2343
|
|
},
|
|
{
|
|
"epoch": 5.1959183673469385,
|
|
"grad_norm": 0.4611447181076451,
|
|
"learning_rate": 7.579340141287965e-06,
|
|
"loss": 0.1806,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0803723856806755,
|
|
"step": 3185,
|
|
"valid_targets_mean": 5043.5,
|
|
"valid_targets_min": 2553
|
|
},
|
|
{
|
|
"epoch": 5.204081632653061,
|
|
"grad_norm": 0.4987990372400629,
|
|
"learning_rate": 7.515668661360162e-06,
|
|
"loss": 0.1885,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08560259640216827,
|
|
"step": 3190,
|
|
"valid_targets_mean": 4151.0,
|
|
"valid_targets_min": 818
|
|
},
|
|
{
|
|
"epoch": 5.2122448979591836,
|
|
"grad_norm": 0.6385599519928179,
|
|
"learning_rate": 7.452203816886132e-06,
|
|
"loss": 0.182,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09238561242818832,
|
|
"step": 3195,
|
|
"valid_targets_mean": 2983.1,
|
|
"valid_targets_min": 1985
|
|
},
|
|
{
|
|
"epoch": 5.220408163265306,
|
|
"grad_norm": 0.49972296239749786,
|
|
"learning_rate": 7.388946658309557e-06,
|
|
"loss": 0.1725,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07128877937793732,
|
|
"step": 3200,
|
|
"valid_targets_mean": 4395.1,
|
|
"valid_targets_min": 838
|
|
},
|
|
{
|
|
"epoch": 5.228571428571429,
|
|
"grad_norm": 0.4251048049138713,
|
|
"learning_rate": 7.3258982326365656e-06,
|
|
"loss": 0.1797,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08632293343544006,
|
|
"step": 3205,
|
|
"valid_targets_mean": 7603.4,
|
|
"valid_targets_min": 3155
|
|
},
|
|
{
|
|
"epoch": 5.236734693877551,
|
|
"grad_norm": 0.5029326144715,
|
|
"learning_rate": 7.263059583418473e-06,
|
|
"loss": 0.1807,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07719817012548447,
|
|
"step": 3210,
|
|
"valid_targets_mean": 4976.4,
|
|
"valid_targets_min": 3287
|
|
},
|
|
{
|
|
"epoch": 5.244897959183674,
|
|
"grad_norm": 0.61537978028631,
|
|
"learning_rate": 7.200431750734414e-06,
|
|
"loss": 0.1887,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07119373232126236,
|
|
"step": 3215,
|
|
"valid_targets_mean": 3600.5,
|
|
"valid_targets_min": 845
|
|
},
|
|
{
|
|
"epoch": 5.253061224489796,
|
|
"grad_norm": 0.6170700381181095,
|
|
"learning_rate": 7.138015771174196e-06,
|
|
"loss": 0.1709,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08307771384716034,
|
|
"step": 3220,
|
|
"valid_targets_mean": 4144.8,
|
|
"valid_targets_min": 2515
|
|
},
|
|
{
|
|
"epoch": 5.261224489795918,
|
|
"grad_norm": 0.56626652835952,
|
|
"learning_rate": 7.075812677821145e-06,
|
|
"loss": 0.1807,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10066160559654236,
|
|
"step": 3225,
|
|
"valid_targets_mean": 3968.5,
|
|
"valid_targets_min": 927
|
|
},
|
|
{
|
|
"epoch": 5.2693877551020405,
|
|
"grad_norm": 0.48288948191263076,
|
|
"learning_rate": 7.013823500234938e-06,
|
|
"loss": 0.1757,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08535900712013245,
|
|
"step": 3230,
|
|
"valid_targets_mean": 5515.0,
|
|
"valid_targets_min": 2714
|
|
},
|
|
{
|
|
"epoch": 5.277551020408163,
|
|
"grad_norm": 0.4865175316764564,
|
|
"learning_rate": 6.952049264434644e-06,
|
|
"loss": 0.1711,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0826706737279892,
|
|
"step": 3235,
|
|
"valid_targets_mean": 5179.6,
|
|
"valid_targets_min": 1055
|
|
},
|
|
{
|
|
"epoch": 5.285714285714286,
|
|
"grad_norm": 0.5086830041890779,
|
|
"learning_rate": 6.890490992881671e-06,
|
|
"loss": 0.1656,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08522078394889832,
|
|
"step": 3240,
|
|
"valid_targets_mean": 7090.1,
|
|
"valid_targets_min": 3606
|
|
},
|
|
{
|
|
"epoch": 5.293877551020408,
|
|
"grad_norm": 0.5137504324453931,
|
|
"learning_rate": 6.829149704462918e-06,
|
|
"loss": 0.1784,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08339082449674606,
|
|
"step": 3245,
|
|
"valid_targets_mean": 5650.1,
|
|
"valid_targets_min": 2900
|
|
},
|
|
{
|
|
"epoch": 5.302040816326531,
|
|
"grad_norm": 0.5367587892690145,
|
|
"learning_rate": 6.768026414473845e-06,
|
|
"loss": 0.1806,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09846776723861694,
|
|
"step": 3250,
|
|
"valid_targets_mean": 5025.4,
|
|
"valid_targets_min": 2477
|
|
},
|
|
{
|
|
"epoch": 5.310204081632653,
|
|
"grad_norm": 0.4109130891833122,
|
|
"learning_rate": 6.707122134601676e-06,
|
|
"loss": 0.1732,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09323020279407501,
|
|
"step": 3255,
|
|
"valid_targets_mean": 6455.5,
|
|
"valid_targets_min": 2791
|
|
},
|
|
{
|
|
"epoch": 5.318367346938776,
|
|
"grad_norm": 0.4363547754110573,
|
|
"learning_rate": 6.6464378729087196e-06,
|
|
"loss": 0.1698,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0628901794552803,
|
|
"step": 3260,
|
|
"valid_targets_mean": 3648.1,
|
|
"valid_targets_min": 1569
|
|
},
|
|
{
|
|
"epoch": 5.326530612244898,
|
|
"grad_norm": 0.6083738609525996,
|
|
"learning_rate": 6.585974633815584e-06,
|
|
"loss": 0.1833,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12459546327590942,
|
|
"step": 3265,
|
|
"valid_targets_mean": 5369.5,
|
|
"valid_targets_min": 2892
|
|
},
|
|
{
|
|
"epoch": 5.33469387755102,
|
|
"grad_norm": 0.5271929880336159,
|
|
"learning_rate": 6.525733418084635e-06,
|
|
"loss": 0.1849,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09798870235681534,
|
|
"step": 3270,
|
|
"valid_targets_mean": 5105.4,
|
|
"valid_targets_min": 1418
|
|
},
|
|
{
|
|
"epoch": 5.3428571428571425,
|
|
"grad_norm": 0.5526213337564067,
|
|
"learning_rate": 6.465715222803386e-06,
|
|
"loss": 0.178,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08597704768180847,
|
|
"step": 3275,
|
|
"valid_targets_mean": 4135.8,
|
|
"valid_targets_min": 2845
|
|
},
|
|
{
|
|
"epoch": 5.351020408163265,
|
|
"grad_norm": 0.5433902287654617,
|
|
"learning_rate": 6.4059210413680175e-06,
|
|
"loss": 0.1827,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09743403643369675,
|
|
"step": 3280,
|
|
"valid_targets_mean": 4350.8,
|
|
"valid_targets_min": 1555
|
|
},
|
|
{
|
|
"epoch": 5.359183673469388,
|
|
"grad_norm": 0.4767498259019695,
|
|
"learning_rate": 6.346351863466926e-06,
|
|
"loss": 0.1723,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07725220918655396,
|
|
"step": 3285,
|
|
"valid_targets_mean": 4635.6,
|
|
"valid_targets_min": 3068
|
|
},
|
|
{
|
|
"epoch": 5.36734693877551,
|
|
"grad_norm": 0.47409096876292317,
|
|
"learning_rate": 6.2870086750643215e-06,
|
|
"loss": 0.1831,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07922723889350891,
|
|
"step": 3290,
|
|
"valid_targets_mean": 4892.5,
|
|
"valid_targets_min": 1427
|
|
},
|
|
{
|
|
"epoch": 5.375510204081633,
|
|
"grad_norm": 0.5084499811899263,
|
|
"learning_rate": 6.2278924583839665e-06,
|
|
"loss": 0.1777,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06283068656921387,
|
|
"step": 3295,
|
|
"valid_targets_mean": 3258.6,
|
|
"valid_targets_min": 1718
|
|
},
|
|
{
|
|
"epoch": 5.383673469387755,
|
|
"grad_norm": 0.6078381929182485,
|
|
"learning_rate": 6.16900419189286e-06,
|
|
"loss": 0.1859,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10925310105085373,
|
|
"step": 3300,
|
|
"valid_targets_mean": 4435.9,
|
|
"valid_targets_min": 1413
|
|
},
|
|
{
|
|
"epoch": 5.391836734693878,
|
|
"grad_norm": 0.6378926459368445,
|
|
"learning_rate": 6.110344850285059e-06,
|
|
"loss": 0.1834,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10677415132522583,
|
|
"step": 3305,
|
|
"valid_targets_mean": 3772.9,
|
|
"valid_targets_min": 1762
|
|
},
|
|
{
|
|
"epoch": 5.4,
|
|
"grad_norm": 0.4589984510633392,
|
|
"learning_rate": 6.0519154044655674e-06,
|
|
"loss": 0.1813,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09519022703170776,
|
|
"step": 3310,
|
|
"valid_targets_mean": 6158.0,
|
|
"valid_targets_min": 1700
|
|
},
|
|
{
|
|
"epoch": 5.408163265306122,
|
|
"grad_norm": 0.6567603574653257,
|
|
"learning_rate": 5.9937168215342345e-06,
|
|
"loss": 0.182,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10160709172487259,
|
|
"step": 3315,
|
|
"valid_targets_mean": 4658.2,
|
|
"valid_targets_min": 655
|
|
},
|
|
{
|
|
"epoch": 5.416326530612245,
|
|
"grad_norm": 0.5338399798016586,
|
|
"learning_rate": 5.9357500647697786e-06,
|
|
"loss": 0.1711,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11896622180938721,
|
|
"step": 3320,
|
|
"valid_targets_mean": 5882.9,
|
|
"valid_targets_min": 3456
|
|
},
|
|
{
|
|
"epoch": 5.424489795918367,
|
|
"grad_norm": 0.5616038365610015,
|
|
"learning_rate": 5.878016093613816e-06,
|
|
"loss": 0.1936,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07236294448375702,
|
|
"step": 3325,
|
|
"valid_targets_mean": 3982.4,
|
|
"valid_targets_min": 1704
|
|
},
|
|
{
|
|
"epoch": 5.43265306122449,
|
|
"grad_norm": 0.515377184603261,
|
|
"learning_rate": 5.820515863654999e-06,
|
|
"loss": 0.1745,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07034678012132645,
|
|
"step": 3330,
|
|
"valid_targets_mean": 3606.8,
|
|
"valid_targets_min": 1744
|
|
},
|
|
{
|
|
"epoch": 5.440816326530612,
|
|
"grad_norm": 0.5377914529022676,
|
|
"learning_rate": 5.7632503266131925e-06,
|
|
"loss": 0.1734,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09335826337337494,
|
|
"step": 3335,
|
|
"valid_targets_mean": 4594.0,
|
|
"valid_targets_min": 2669
|
|
},
|
|
{
|
|
"epoch": 5.448979591836735,
|
|
"grad_norm": 0.5255517607613142,
|
|
"learning_rate": 5.70622043032371e-06,
|
|
"loss": 0.1784,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0681927353143692,
|
|
"step": 3340,
|
|
"valid_targets_mean": 5194.6,
|
|
"valid_targets_min": 1358
|
|
},
|
|
{
|
|
"epoch": 5.457142857142857,
|
|
"grad_norm": 0.5117051396865993,
|
|
"learning_rate": 5.649427118721655e-06,
|
|
"loss": 0.1809,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08759818226099014,
|
|
"step": 3345,
|
|
"valid_targets_mean": 4859.9,
|
|
"valid_targets_min": 2107
|
|
},
|
|
{
|
|
"epoch": 5.46530612244898,
|
|
"grad_norm": 0.5608179829600851,
|
|
"learning_rate": 5.592871331826271e-06,
|
|
"loss": 0.1803,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.092364102602005,
|
|
"step": 3350,
|
|
"valid_targets_mean": 4509.1,
|
|
"valid_targets_min": 182
|
|
},
|
|
{
|
|
"epoch": 5.473469387755102,
|
|
"grad_norm": 0.523015667386693,
|
|
"learning_rate": 5.536554005725392e-06,
|
|
"loss": 0.1687,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07398107647895813,
|
|
"step": 3355,
|
|
"valid_targets_mean": 3645.1,
|
|
"valid_targets_min": 802
|
|
},
|
|
{
|
|
"epoch": 5.481632653061224,
|
|
"grad_norm": 0.5440603818142956,
|
|
"learning_rate": 5.480476072559955e-06,
|
|
"loss": 0.1772,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10536229610443115,
|
|
"step": 3360,
|
|
"valid_targets_mean": 5585.2,
|
|
"valid_targets_min": 3009
|
|
},
|
|
{
|
|
"epoch": 5.489795918367347,
|
|
"grad_norm": 0.5106202139769874,
|
|
"learning_rate": 5.42463846050854e-06,
|
|
"loss": 0.1792,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08815483748912811,
|
|
"step": 3365,
|
|
"valid_targets_mean": 4938.4,
|
|
"valid_targets_min": 436
|
|
},
|
|
{
|
|
"epoch": 5.497959183673469,
|
|
"grad_norm": 0.5792561830469595,
|
|
"learning_rate": 5.369042093772068e-06,
|
|
"loss": 0.1864,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08561515808105469,
|
|
"step": 3370,
|
|
"valid_targets_mean": 3781.2,
|
|
"valid_targets_min": 2797
|
|
},
|
|
{
|
|
"epoch": 5.506122448979592,
|
|
"grad_norm": 0.5716983079904695,
|
|
"learning_rate": 5.313687892558455e-06,
|
|
"loss": 0.1797,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07345669716596603,
|
|
"step": 3375,
|
|
"valid_targets_mean": 3362.6,
|
|
"valid_targets_min": 385
|
|
},
|
|
{
|
|
"epoch": 5.514285714285714,
|
|
"grad_norm": 0.5214998517548255,
|
|
"learning_rate": 5.258576773067366e-06,
|
|
"loss": 0.1731,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0814686045050621,
|
|
"step": 3380,
|
|
"valid_targets_mean": 4806.2,
|
|
"valid_targets_min": 2774
|
|
},
|
|
{
|
|
"epoch": 5.522448979591837,
|
|
"grad_norm": 0.5036864461547301,
|
|
"learning_rate": 5.203709647475128e-06,
|
|
"loss": 0.177,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09993688762187958,
|
|
"step": 3385,
|
|
"valid_targets_mean": 5561.8,
|
|
"valid_targets_min": 988
|
|
},
|
|
{
|
|
"epoch": 5.530612244897959,
|
|
"grad_norm": 0.45864417905162363,
|
|
"learning_rate": 5.149087423919541e-06,
|
|
"loss": 0.1638,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07489781081676483,
|
|
"step": 3390,
|
|
"valid_targets_mean": 5958.8,
|
|
"valid_targets_min": 1967
|
|
},
|
|
{
|
|
"epoch": 5.538775510204082,
|
|
"grad_norm": 0.4694299423214727,
|
|
"learning_rate": 5.094711006484907e-06,
|
|
"loss": 0.175,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08611179888248444,
|
|
"step": 3395,
|
|
"valid_targets_mean": 5369.8,
|
|
"valid_targets_min": 1875
|
|
},
|
|
{
|
|
"epoch": 5.546938775510204,
|
|
"grad_norm": 0.5643467092019754,
|
|
"learning_rate": 5.040581295187048e-06,
|
|
"loss": 0.1736,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09875790774822235,
|
|
"step": 3400,
|
|
"valid_targets_mean": 4500.5,
|
|
"valid_targets_min": 2867
|
|
},
|
|
{
|
|
"epoch": 5.555102040816326,
|
|
"grad_norm": 0.4390252283312783,
|
|
"learning_rate": 4.986699185958403e-06,
|
|
"loss": 0.1745,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07030566036701202,
|
|
"step": 3405,
|
|
"valid_targets_mean": 4411.1,
|
|
"valid_targets_min": 2515
|
|
},
|
|
{
|
|
"epoch": 5.563265306122449,
|
|
"grad_norm": 0.4638144447712849,
|
|
"learning_rate": 4.933065570633215e-06,
|
|
"loss": 0.1768,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07658712565898895,
|
|
"step": 3410,
|
|
"valid_targets_mean": 4903.1,
|
|
"valid_targets_min": 2572
|
|
},
|
|
{
|
|
"epoch": 5.571428571428571,
|
|
"grad_norm": 0.5537043507081186,
|
|
"learning_rate": 4.879681336932733e-06,
|
|
"loss": 0.1804,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11388731002807617,
|
|
"step": 3415,
|
|
"valid_targets_mean": 4796.4,
|
|
"valid_targets_min": 1771
|
|
},
|
|
{
|
|
"epoch": 5.579591836734694,
|
|
"grad_norm": 0.4731889808421747,
|
|
"learning_rate": 4.8265473684505844e-06,
|
|
"loss": 0.1799,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09846645593643188,
|
|
"step": 3420,
|
|
"valid_targets_mean": 6823.2,
|
|
"valid_targets_min": 3185
|
|
},
|
|
{
|
|
"epoch": 5.587755102040816,
|
|
"grad_norm": 0.4240237470679026,
|
|
"learning_rate": 4.773664544638079e-06,
|
|
"loss": 0.1751,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06629286706447601,
|
|
"step": 3425,
|
|
"valid_targets_mean": 5260.6,
|
|
"valid_targets_min": 755
|
|
},
|
|
{
|
|
"epoch": 5.595918367346939,
|
|
"grad_norm": 0.48722456054449315,
|
|
"learning_rate": 4.721033740789696e-06,
|
|
"loss": 0.1845,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12045013904571533,
|
|
"step": 3430,
|
|
"valid_targets_mean": 6874.5,
|
|
"valid_targets_min": 3698
|
|
},
|
|
{
|
|
"epoch": 5.604081632653061,
|
|
"grad_norm": 0.561411868538588,
|
|
"learning_rate": 4.6686558280285875e-06,
|
|
"loss": 0.1838,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09412881731987,
|
|
"step": 3435,
|
|
"valid_targets_mean": 4171.6,
|
|
"valid_targets_min": 1389
|
|
},
|
|
{
|
|
"epoch": 5.612244897959184,
|
|
"grad_norm": 0.5100313882410629,
|
|
"learning_rate": 4.6165316732921525e-06,
|
|
"loss": 0.1797,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09020973742008209,
|
|
"step": 3440,
|
|
"valid_targets_mean": 4960.4,
|
|
"valid_targets_min": 3477
|
|
},
|
|
{
|
|
"epoch": 5.6204081632653065,
|
|
"grad_norm": 0.48037905657238017,
|
|
"learning_rate": 4.5646621393177e-06,
|
|
"loss": 0.1865,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10528279840946198,
|
|
"step": 3445,
|
|
"valid_targets_mean": 5070.9,
|
|
"valid_targets_min": 2828
|
|
},
|
|
{
|
|
"epoch": 5.628571428571428,
|
|
"grad_norm": 0.6129409856060618,
|
|
"learning_rate": 4.513048084628153e-06,
|
|
"loss": 0.1765,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06684743613004684,
|
|
"step": 3450,
|
|
"valid_targets_mean": 3636.2,
|
|
"valid_targets_min": 2132
|
|
},
|
|
{
|
|
"epoch": 5.636734693877551,
|
|
"grad_norm": 0.5188406585305368,
|
|
"learning_rate": 4.461690363517856e-06,
|
|
"loss": 0.1793,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0819583535194397,
|
|
"step": 3455,
|
|
"valid_targets_mean": 4631.9,
|
|
"valid_targets_min": 1520
|
|
},
|
|
{
|
|
"epoch": 5.644897959183673,
|
|
"grad_norm": 0.5261488357902977,
|
|
"learning_rate": 4.410589826038427e-06,
|
|
"loss": 0.1835,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09630433470010757,
|
|
"step": 3460,
|
|
"valid_targets_mean": 4737.4,
|
|
"valid_targets_min": 2158
|
|
},
|
|
{
|
|
"epoch": 5.653061224489796,
|
|
"grad_norm": 0.5564718916899687,
|
|
"learning_rate": 4.35974731798467e-06,
|
|
"loss": 0.17,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09089779853820801,
|
|
"step": 3465,
|
|
"valid_targets_mean": 4074.0,
|
|
"valid_targets_min": 2613
|
|
},
|
|
{
|
|
"epoch": 5.661224489795918,
|
|
"grad_norm": 0.49359667139456637,
|
|
"learning_rate": 4.309163680880619e-06,
|
|
"loss": 0.1857,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09821240603923798,
|
|
"step": 3470,
|
|
"valid_targets_mean": 5946.6,
|
|
"valid_targets_min": 1546
|
|
},
|
|
{
|
|
"epoch": 5.669387755102041,
|
|
"grad_norm": 0.5695633164379076,
|
|
"learning_rate": 4.258839751965566e-06,
|
|
"loss": 0.1803,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08557415753602982,
|
|
"step": 3475,
|
|
"valid_targets_mean": 4009.6,
|
|
"valid_targets_min": 2417
|
|
},
|
|
{
|
|
"epoch": 5.677551020408163,
|
|
"grad_norm": 0.4790980762705507,
|
|
"learning_rate": 4.208776364180231e-06,
|
|
"loss": 0.1775,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09636128693819046,
|
|
"step": 3480,
|
|
"valid_targets_mean": 6290.5,
|
|
"valid_targets_min": 1357
|
|
},
|
|
{
|
|
"epoch": 5.685714285714286,
|
|
"grad_norm": 0.49444289079244524,
|
|
"learning_rate": 4.15897434615296e-06,
|
|
"loss": 0.1768,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07259830832481384,
|
|
"step": 3485,
|
|
"valid_targets_mean": 4316.1,
|
|
"valid_targets_min": 264
|
|
},
|
|
{
|
|
"epoch": 5.6938775510204085,
|
|
"grad_norm": 0.5166295899116976,
|
|
"learning_rate": 4.10943452218602e-06,
|
|
"loss": 0.1922,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1207120418548584,
|
|
"step": 3490,
|
|
"valid_targets_mean": 5905.9,
|
|
"valid_targets_min": 3679
|
|
},
|
|
{
|
|
"epoch": 5.70204081632653,
|
|
"grad_norm": 0.5250771519188173,
|
|
"learning_rate": 4.06015771224195e-06,
|
|
"loss": 0.1863,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09034144878387451,
|
|
"step": 3495,
|
|
"valid_targets_mean": 4832.6,
|
|
"valid_targets_min": 2040
|
|
},
|
|
{
|
|
"epoch": 5.710204081632653,
|
|
"grad_norm": 0.5483295619114166,
|
|
"learning_rate": 4.011144731929981e-06,
|
|
"loss": 0.1909,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11903704702854156,
|
|
"step": 3500,
|
|
"valid_targets_mean": 5562.9,
|
|
"valid_targets_min": 2074
|
|
},
|
|
{
|
|
"epoch": 5.718367346938775,
|
|
"grad_norm": 0.5227938259641866,
|
|
"learning_rate": 3.962396392492554e-06,
|
|
"loss": 0.1835,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09354373812675476,
|
|
"step": 3505,
|
|
"valid_targets_mean": 4609.5,
|
|
"valid_targets_min": 827
|
|
},
|
|
{
|
|
"epoch": 5.726530612244898,
|
|
"grad_norm": 0.46671920374295456,
|
|
"learning_rate": 3.9139135007919e-06,
|
|
"loss": 0.1815,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10033905506134033,
|
|
"step": 3510,
|
|
"valid_targets_mean": 6064.2,
|
|
"valid_targets_min": 1319
|
|
},
|
|
{
|
|
"epoch": 5.73469387755102,
|
|
"grad_norm": 0.4910469291437556,
|
|
"learning_rate": 3.865696859296635e-06,
|
|
"loss": 0.1815,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10732349753379822,
|
|
"step": 3515,
|
|
"valid_targets_mean": 5148.2,
|
|
"valid_targets_min": 2218
|
|
},
|
|
{
|
|
"epoch": 5.742857142857143,
|
|
"grad_norm": 0.43353585019350355,
|
|
"learning_rate": 3.817747266068541e-06,
|
|
"loss": 0.1756,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08565760403871536,
|
|
"step": 3520,
|
|
"valid_targets_mean": 4565.4,
|
|
"valid_targets_min": 1967
|
|
},
|
|
{
|
|
"epoch": 5.751020408163265,
|
|
"grad_norm": 0.5310832117874204,
|
|
"learning_rate": 3.770065514749317e-06,
|
|
"loss": 0.1731,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1077844426035881,
|
|
"step": 3525,
|
|
"valid_targets_mean": 5440.5,
|
|
"valid_targets_min": 2018
|
|
},
|
|
{
|
|
"epoch": 5.759183673469388,
|
|
"grad_norm": 0.5169834534688904,
|
|
"learning_rate": 3.7226523945474567e-06,
|
|
"loss": 0.1839,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10681848227977753,
|
|
"step": 3530,
|
|
"valid_targets_mean": 5627.4,
|
|
"valid_targets_min": 2586
|
|
},
|
|
{
|
|
"epoch": 5.7673469387755105,
|
|
"grad_norm": 0.5484383627618905,
|
|
"learning_rate": 3.6755086902251847e-06,
|
|
"loss": 0.1838,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0877349004149437,
|
|
"step": 3535,
|
|
"valid_targets_mean": 5774.1,
|
|
"valid_targets_min": 2666
|
|
},
|
|
{
|
|
"epoch": 5.775510204081632,
|
|
"grad_norm": 0.591238582407821,
|
|
"learning_rate": 3.6286351820854514e-06,
|
|
"loss": 0.1856,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09708157181739807,
|
|
"step": 3540,
|
|
"valid_targets_mean": 3561.5,
|
|
"valid_targets_min": 1409
|
|
},
|
|
{
|
|
"epoch": 5.783673469387755,
|
|
"grad_norm": 0.4832269943744314,
|
|
"learning_rate": 3.582032645959066e-06,
|
|
"loss": 0.1729,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08354311436414719,
|
|
"step": 3545,
|
|
"valid_targets_mean": 5434.9,
|
|
"valid_targets_min": 3216
|
|
},
|
|
{
|
|
"epoch": 5.791836734693877,
|
|
"grad_norm": 0.533477273302161,
|
|
"learning_rate": 3.535701853191784e-06,
|
|
"loss": 0.1854,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10452941060066223,
|
|
"step": 3550,
|
|
"valid_targets_mean": 4881.2,
|
|
"valid_targets_min": 2425
|
|
},
|
|
{
|
|
"epoch": 5.8,
|
|
"grad_norm": 0.4388726457273361,
|
|
"learning_rate": 3.4896435706316e-06,
|
|
"loss": 0.1783,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0693935751914978,
|
|
"step": 3555,
|
|
"valid_targets_mean": 5472.5,
|
|
"valid_targets_min": 677
|
|
},
|
|
{
|
|
"epoch": 5.808163265306122,
|
|
"grad_norm": 0.6045179245376615,
|
|
"learning_rate": 3.443858560616029e-06,
|
|
"loss": 0.1716,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07769778370857239,
|
|
"step": 3560,
|
|
"valid_targets_mean": 2920.1,
|
|
"valid_targets_min": 1135
|
|
},
|
|
{
|
|
"epoch": 5.816326530612245,
|
|
"grad_norm": 0.4748517425016773,
|
|
"learning_rate": 3.3983475809594892e-06,
|
|
"loss": 0.1762,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07783222198486328,
|
|
"step": 3565,
|
|
"valid_targets_mean": 5231.5,
|
|
"valid_targets_min": 2791
|
|
},
|
|
{
|
|
"epoch": 5.8244897959183675,
|
|
"grad_norm": 0.5094001344648212,
|
|
"learning_rate": 3.353111384940775e-06,
|
|
"loss": 0.1751,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12287256121635437,
|
|
"step": 3570,
|
|
"valid_targets_mean": 6160.6,
|
|
"valid_targets_min": 3278
|
|
},
|
|
{
|
|
"epoch": 5.83265306122449,
|
|
"grad_norm": 0.5476080239908376,
|
|
"learning_rate": 3.308150721290546e-06,
|
|
"loss": 0.1829,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06965255737304688,
|
|
"step": 3575,
|
|
"valid_targets_mean": 3738.8,
|
|
"valid_targets_min": 2244
|
|
},
|
|
{
|
|
"epoch": 5.840816326530613,
|
|
"grad_norm": 0.5134167564842722,
|
|
"learning_rate": 3.263466334179002e-06,
|
|
"loss": 0.1783,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09871196001768112,
|
|
"step": 3580,
|
|
"valid_targets_mean": 4696.0,
|
|
"valid_targets_min": 845
|
|
},
|
|
{
|
|
"epoch": 5.848979591836734,
|
|
"grad_norm": 0.5552614954950037,
|
|
"learning_rate": 3.2190589632035117e-06,
|
|
"loss": 0.181,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08760549128055573,
|
|
"step": 3585,
|
|
"valid_targets_mean": 4645.8,
|
|
"valid_targets_min": 1498
|
|
},
|
|
{
|
|
"epoch": 5.857142857142857,
|
|
"grad_norm": 0.6017051275561736,
|
|
"learning_rate": 3.174929343376374e-06,
|
|
"loss": 0.1877,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09327945113182068,
|
|
"step": 3590,
|
|
"valid_targets_mean": 3505.9,
|
|
"valid_targets_min": 2137
|
|
},
|
|
{
|
|
"epoch": 5.865306122448979,
|
|
"grad_norm": 0.5419286866691388,
|
|
"learning_rate": 3.1310782051126986e-06,
|
|
"loss": 0.1859,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10793288052082062,
|
|
"step": 3595,
|
|
"valid_targets_mean": 5848.5,
|
|
"valid_targets_min": 2733
|
|
},
|
|
{
|
|
"epoch": 5.873469387755102,
|
|
"grad_norm": 0.8665054567909115,
|
|
"learning_rate": 3.0875062742182526e-06,
|
|
"loss": 0.1735,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08918233215808868,
|
|
"step": 3600,
|
|
"valid_targets_mean": 5910.8,
|
|
"valid_targets_min": 1124
|
|
},
|
|
{
|
|
"epoch": 5.881632653061224,
|
|
"grad_norm": 0.4675391830481981,
|
|
"learning_rate": 3.0442142718774992e-06,
|
|
"loss": 0.1804,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09749846905469894,
|
|
"step": 3605,
|
|
"valid_targets_mean": 6391.9,
|
|
"valid_targets_min": 4286
|
|
},
|
|
{
|
|
"epoch": 5.889795918367347,
|
|
"grad_norm": 0.5536901285759013,
|
|
"learning_rate": 3.001202914641628e-06,
|
|
"loss": 0.1786,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08519423753023148,
|
|
"step": 3610,
|
|
"valid_targets_mean": 4084.0,
|
|
"valid_targets_min": 2777
|
|
},
|
|
{
|
|
"epoch": 5.8979591836734695,
|
|
"grad_norm": 0.5507724458934895,
|
|
"learning_rate": 2.958472914416721e-06,
|
|
"loss": 0.1814,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09706813097000122,
|
|
"step": 3615,
|
|
"valid_targets_mean": 3863.4,
|
|
"valid_targets_min": 1727
|
|
},
|
|
{
|
|
"epoch": 5.906122448979592,
|
|
"grad_norm": 0.49819909262377643,
|
|
"learning_rate": 2.916024978451948e-06,
|
|
"loss": 0.1914,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0789080485701561,
|
|
"step": 3620,
|
|
"valid_targets_mean": 5023.1,
|
|
"valid_targets_min": 1140
|
|
},
|
|
{
|
|
"epoch": 5.914285714285715,
|
|
"grad_norm": 0.5883950716908385,
|
|
"learning_rate": 2.8738598093278593e-06,
|
|
"loss": 0.1772,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09466518461704254,
|
|
"step": 3625,
|
|
"valid_targets_mean": 3618.1,
|
|
"valid_targets_min": 170
|
|
},
|
|
{
|
|
"epoch": 5.922448979591836,
|
|
"grad_norm": 0.5746947171019235,
|
|
"learning_rate": 2.8319781049447923e-06,
|
|
"loss": 0.1755,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09892357885837555,
|
|
"step": 3630,
|
|
"valid_targets_mean": 3903.4,
|
|
"valid_targets_min": 1718
|
|
},
|
|
{
|
|
"epoch": 5.930612244897959,
|
|
"grad_norm": 0.5179995327668454,
|
|
"learning_rate": 2.7903805585112674e-06,
|
|
"loss": 0.1751,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0768948346376419,
|
|
"step": 3635,
|
|
"valid_targets_mean": 4118.9,
|
|
"valid_targets_min": 2274
|
|
},
|
|
{
|
|
"epoch": 5.938775510204081,
|
|
"grad_norm": 0.4513464329838285,
|
|
"learning_rate": 2.7490678585325526e-06,
|
|
"loss": 0.1699,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06318273395299911,
|
|
"step": 3640,
|
|
"valid_targets_mean": 3500.8,
|
|
"valid_targets_min": 1673
|
|
},
|
|
{
|
|
"epoch": 5.946938775510204,
|
|
"grad_norm": 0.5186761444260091,
|
|
"learning_rate": 2.708040688799258e-06,
|
|
"loss": 0.1806,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09379004687070847,
|
|
"step": 3645,
|
|
"valid_targets_mean": 4657.5,
|
|
"valid_targets_min": 4182
|
|
},
|
|
{
|
|
"epoch": 5.955102040816326,
|
|
"grad_norm": 0.606778731280153,
|
|
"learning_rate": 2.6672997283760137e-06,
|
|
"loss": 0.1825,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11004135757684708,
|
|
"step": 3650,
|
|
"valid_targets_mean": 4382.9,
|
|
"valid_targets_min": 2676
|
|
},
|
|
{
|
|
"epoch": 5.963265306122449,
|
|
"grad_norm": 0.5229929454307417,
|
|
"learning_rate": 2.6268456515902283e-06,
|
|
"loss": 0.1801,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09963811188936234,
|
|
"step": 3655,
|
|
"valid_targets_mean": 5354.8,
|
|
"valid_targets_min": 2892
|
|
},
|
|
{
|
|
"epoch": 5.9714285714285715,
|
|
"grad_norm": 0.5001733152158864,
|
|
"learning_rate": 2.5866791280209368e-06,
|
|
"loss": 0.1735,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07187163084745407,
|
|
"step": 3660,
|
|
"valid_targets_mean": 3210.0,
|
|
"valid_targets_min": 684
|
|
},
|
|
{
|
|
"epoch": 5.979591836734694,
|
|
"grad_norm": 0.6159746098198164,
|
|
"learning_rate": 2.546800822487714e-06,
|
|
"loss": 0.1768,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.052930839359760284,
|
|
"step": 3665,
|
|
"valid_targets_mean": 3052.1,
|
|
"valid_targets_min": 1593
|
|
},
|
|
{
|
|
"epoch": 5.987755102040817,
|
|
"grad_norm": 0.4193135572158273,
|
|
"learning_rate": 2.507211395039668e-06,
|
|
"loss": 0.1783,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0880434513092041,
|
|
"step": 3670,
|
|
"valid_targets_mean": 8106.5,
|
|
"valid_targets_min": 177
|
|
},
|
|
{
|
|
"epoch": 5.995918367346938,
|
|
"grad_norm": 0.5144999909128704,
|
|
"learning_rate": 2.4679115009445132e-06,
|
|
"loss": 0.1636,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08727358281612396,
|
|
"step": 3675,
|
|
"valid_targets_mean": 4467.9,
|
|
"valid_targets_min": 3024
|
|
},
|
|
{
|
|
"epoch": 6.003265306122449,
|
|
"grad_norm": 0.47252041481974333,
|
|
"learning_rate": 2.428901790677731e-06,
|
|
"loss": 0.1667,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10447318851947784,
|
|
"step": 3680,
|
|
"valid_targets_mean": 6420.9,
|
|
"valid_targets_min": 2059
|
|
},
|
|
{
|
|
"epoch": 6.011428571428572,
|
|
"grad_norm": 0.4986949893337232,
|
|
"learning_rate": 2.3901829099118044e-06,
|
|
"loss": 0.1717,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08693604916334152,
|
|
"step": 3685,
|
|
"valid_targets_mean": 4444.5,
|
|
"valid_targets_min": 2091
|
|
},
|
|
{
|
|
"epoch": 6.019591836734694,
|
|
"grad_norm": 0.5681658717936597,
|
|
"learning_rate": 2.351755499505526e-06,
|
|
"loss": 0.1726,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1184157133102417,
|
|
"step": 3690,
|
|
"valid_targets_mean": 4972.2,
|
|
"valid_targets_min": 2219
|
|
},
|
|
{
|
|
"epoch": 6.027755102040817,
|
|
"grad_norm": 0.4921539159180312,
|
|
"learning_rate": 2.313620195493389e-06,
|
|
"loss": 0.1731,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08808198571205139,
|
|
"step": 3695,
|
|
"valid_targets_mean": 4892.5,
|
|
"valid_targets_min": 2057
|
|
},
|
|
{
|
|
"epoch": 6.035918367346939,
|
|
"grad_norm": 0.49040858813536775,
|
|
"learning_rate": 2.2757776290750687e-06,
|
|
"loss": 0.1757,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1032937616109848,
|
|
"step": 3700,
|
|
"valid_targets_mean": 5997.8,
|
|
"valid_targets_min": 2168
|
|
},
|
|
{
|
|
"epoch": 6.044081632653061,
|
|
"grad_norm": 0.5624209287229496,
|
|
"learning_rate": 2.238228426604967e-06,
|
|
"loss": 0.1753,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0846734493970871,
|
|
"step": 3705,
|
|
"valid_targets_mean": 3722.6,
|
|
"valid_targets_min": 2394
|
|
},
|
|
{
|
|
"epoch": 6.052244897959183,
|
|
"grad_norm": 0.5530054487383088,
|
|
"learning_rate": 2.2009732095818425e-06,
|
|
"loss": 0.176,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10292991250753403,
|
|
"step": 3710,
|
|
"valid_targets_mean": 4475.9,
|
|
"valid_targets_min": 1713
|
|
},
|
|
{
|
|
"epoch": 6.060408163265306,
|
|
"grad_norm": 0.49926699110953343,
|
|
"learning_rate": 2.1640125946385317e-06,
|
|
"loss": 0.1603,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08036244660615921,
|
|
"step": 3715,
|
|
"valid_targets_mean": 4533.5,
|
|
"valid_targets_min": 2572
|
|
},
|
|
{
|
|
"epoch": 6.0685714285714285,
|
|
"grad_norm": 0.5923091072775398,
|
|
"learning_rate": 2.127347193531757e-06,
|
|
"loss": 0.1647,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.053380295634269714,
|
|
"step": 3720,
|
|
"valid_targets_mean": 2151.5,
|
|
"valid_targets_min": 1027
|
|
},
|
|
{
|
|
"epoch": 6.076734693877551,
|
|
"grad_norm": 0.4861946492399618,
|
|
"learning_rate": 2.090977613131955e-06,
|
|
"loss": 0.1756,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08133894950151443,
|
|
"step": 3725,
|
|
"valid_targets_mean": 5612.1,
|
|
"valid_targets_min": 230
|
|
},
|
|
{
|
|
"epoch": 6.084897959183674,
|
|
"grad_norm": 0.47935348621159685,
|
|
"learning_rate": 2.0549044554132824e-06,
|
|
"loss": 0.169,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09912076592445374,
|
|
"step": 3730,
|
|
"valid_targets_mean": 6868.9,
|
|
"valid_targets_min": 2386
|
|
},
|
|
{
|
|
"epoch": 6.093061224489796,
|
|
"grad_norm": 0.47252848078690995,
|
|
"learning_rate": 2.019128317443626e-06,
|
|
"loss": 0.1807,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06253606826066971,
|
|
"step": 3735,
|
|
"valid_targets_mean": 4328.8,
|
|
"valid_targets_min": 2804
|
|
},
|
|
{
|
|
"epoch": 6.101224489795919,
|
|
"grad_norm": 0.4222132937119648,
|
|
"learning_rate": 1.98364979137472e-06,
|
|
"loss": 0.1757,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07729291915893555,
|
|
"step": 3740,
|
|
"valid_targets_mean": 4720.0,
|
|
"valid_targets_min": 2142
|
|
},
|
|
{
|
|
"epoch": 6.109387755102041,
|
|
"grad_norm": 0.40428888724347384,
|
|
"learning_rate": 1.948469464432359e-06,
|
|
"loss": 0.176,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05827842652797699,
|
|
"step": 3745,
|
|
"valid_targets_mean": 5187.4,
|
|
"valid_targets_min": 1394
|
|
},
|
|
{
|
|
"epoch": 6.117551020408163,
|
|
"grad_norm": 0.5559197845200394,
|
|
"learning_rate": 1.9135879189066562e-06,
|
|
"loss": 0.1708,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10092753916978836,
|
|
"step": 3750,
|
|
"valid_targets_mean": 5105.6,
|
|
"valid_targets_min": 2940
|
|
},
|
|
{
|
|
"epoch": 6.1257142857142854,
|
|
"grad_norm": 0.6036578135073778,
|
|
"learning_rate": 1.8790057321424427e-06,
|
|
"loss": 0.1768,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07620693743228912,
|
|
"step": 3755,
|
|
"valid_targets_mean": 3435.4,
|
|
"valid_targets_min": 751
|
|
},
|
|
{
|
|
"epoch": 6.133877551020408,
|
|
"grad_norm": 0.581844618201763,
|
|
"learning_rate": 1.8447234765296617e-06,
|
|
"loss": 0.1755,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08816036581993103,
|
|
"step": 3760,
|
|
"valid_targets_mean": 4790.1,
|
|
"valid_targets_min": 581
|
|
},
|
|
{
|
|
"epoch": 6.1420408163265305,
|
|
"grad_norm": 0.48917153992312773,
|
|
"learning_rate": 1.810741719493936e-06,
|
|
"loss": 0.1696,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07217949628829956,
|
|
"step": 3765,
|
|
"valid_targets_mean": 4503.4,
|
|
"valid_targets_min": 2449
|
|
},
|
|
{
|
|
"epoch": 6.150204081632653,
|
|
"grad_norm": 0.44242782826133753,
|
|
"learning_rate": 1.7770610234871588e-06,
|
|
"loss": 0.1654,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07949961721897125,
|
|
"step": 3770,
|
|
"valid_targets_mean": 6213.2,
|
|
"valid_targets_min": 2654
|
|
},
|
|
{
|
|
"epoch": 6.158367346938776,
|
|
"grad_norm": 0.5507329562626103,
|
|
"learning_rate": 1.743681945978184e-06,
|
|
"loss": 0.1786,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11011023819446564,
|
|
"step": 3775,
|
|
"valid_targets_mean": 5885.9,
|
|
"valid_targets_min": 2652
|
|
},
|
|
{
|
|
"epoch": 6.166530612244898,
|
|
"grad_norm": 0.4550015614312,
|
|
"learning_rate": 1.710605039443609e-06,
|
|
"loss": 0.1759,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07752910256385803,
|
|
"step": 3780,
|
|
"valid_targets_mean": 6354.8,
|
|
"valid_targets_min": 710
|
|
},
|
|
{
|
|
"epoch": 6.174693877551021,
|
|
"grad_norm": 0.5207522547176406,
|
|
"learning_rate": 1.6778308513586084e-06,
|
|
"loss": 0.1647,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08761303126811981,
|
|
"step": 3785,
|
|
"valid_targets_mean": 4992.5,
|
|
"valid_targets_min": 2047
|
|
},
|
|
{
|
|
"epoch": 6.182857142857143,
|
|
"grad_norm": 0.5273537447410652,
|
|
"learning_rate": 1.6453599241879014e-06,
|
|
"loss": 0.1634,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07733944058418274,
|
|
"step": 3790,
|
|
"valid_targets_mean": 4691.1,
|
|
"valid_targets_min": 961
|
|
},
|
|
{
|
|
"epoch": 6.191020408163265,
|
|
"grad_norm": 0.4161651854234828,
|
|
"learning_rate": 1.6131927953767567e-06,
|
|
"loss": 0.1674,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0671846866607666,
|
|
"step": 3795,
|
|
"valid_targets_mean": 6539.9,
|
|
"valid_targets_min": 1133
|
|
},
|
|
{
|
|
"epoch": 6.1991836734693875,
|
|
"grad_norm": 0.5463330466733015,
|
|
"learning_rate": 1.581329997342087e-06,
|
|
"loss": 0.1682,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09839221835136414,
|
|
"step": 3800,
|
|
"valid_targets_mean": 4389.6,
|
|
"valid_targets_min": 1840
|
|
},
|
|
{
|
|
"epoch": 6.20734693877551,
|
|
"grad_norm": 0.583054445182557,
|
|
"learning_rate": 1.5497720574636632e-06,
|
|
"loss": 0.1778,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09077779203653336,
|
|
"step": 3805,
|
|
"valid_targets_mean": 4176.5,
|
|
"valid_targets_min": 2000
|
|
},
|
|
{
|
|
"epoch": 6.215510204081633,
|
|
"grad_norm": 0.47287015564436347,
|
|
"learning_rate": 1.518519498075366e-06,
|
|
"loss": 0.1696,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08704468607902527,
|
|
"step": 3810,
|
|
"valid_targets_mean": 5991.4,
|
|
"valid_targets_min": 3677
|
|
},
|
|
{
|
|
"epoch": 6.223673469387755,
|
|
"grad_norm": 0.48414217836633555,
|
|
"learning_rate": 1.4875728364565411e-06,
|
|
"loss": 0.1652,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08780664950609207,
|
|
"step": 3815,
|
|
"valid_targets_mean": 5148.4,
|
|
"valid_targets_min": 2605
|
|
},
|
|
{
|
|
"epoch": 6.231836734693878,
|
|
"grad_norm": 0.48754025209367896,
|
|
"learning_rate": 1.4569325848234473e-06,
|
|
"loss": 0.1755,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07883431017398834,
|
|
"step": 3820,
|
|
"valid_targets_mean": 5642.8,
|
|
"valid_targets_min": 2511
|
|
},
|
|
{
|
|
"epoch": 6.24,
|
|
"grad_norm": 0.5239266949028567,
|
|
"learning_rate": 1.4265992503207704e-06,
|
|
"loss": 0.1671,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09598593413829803,
|
|
"step": 3825,
|
|
"valid_targets_mean": 5202.4,
|
|
"valid_targets_min": 2352
|
|
},
|
|
{
|
|
"epoch": 6.248163265306123,
|
|
"grad_norm": 0.48189216098707605,
|
|
"learning_rate": 1.396573335013236e-06,
|
|
"loss": 0.1831,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07325664907693863,
|
|
"step": 3830,
|
|
"valid_targets_mean": 4080.9,
|
|
"valid_targets_min": 1525
|
|
},
|
|
{
|
|
"epoch": 6.256326530612245,
|
|
"grad_norm": 0.3815761140634215,
|
|
"learning_rate": 1.3668553358772863e-06,
|
|
"loss": 0.163,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07809938490390778,
|
|
"step": 3835,
|
|
"valid_targets_mean": 6897.1,
|
|
"valid_targets_min": 4401
|
|
},
|
|
{
|
|
"epoch": 6.264489795918367,
|
|
"grad_norm": 0.5540990052785123,
|
|
"learning_rate": 1.337445744792867e-06,
|
|
"loss": 0.1776,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09904327988624573,
|
|
"step": 3840,
|
|
"valid_targets_mean": 4890.2,
|
|
"valid_targets_min": 1419
|
|
},
|
|
{
|
|
"epoch": 6.2726530612244895,
|
|
"grad_norm": 0.566758209601489,
|
|
"learning_rate": 1.3083450485352867e-06,
|
|
"loss": 0.168,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12004449963569641,
|
|
"step": 3845,
|
|
"valid_targets_mean": 4876.9,
|
|
"valid_targets_min": 2036
|
|
},
|
|
{
|
|
"epoch": 6.280816326530612,
|
|
"grad_norm": 0.5280459490954896,
|
|
"learning_rate": 1.279553728767151e-06,
|
|
"loss": 0.1706,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09008440375328064,
|
|
"step": 3850,
|
|
"valid_targets_mean": 4055.4,
|
|
"valid_targets_min": 2205
|
|
},
|
|
{
|
|
"epoch": 6.288979591836735,
|
|
"grad_norm": 0.4787362534748566,
|
|
"learning_rate": 1.2510722620303972e-06,
|
|
"loss": 0.1733,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11373571306467056,
|
|
"step": 3855,
|
|
"valid_targets_mean": 7750.4,
|
|
"valid_targets_min": 3349
|
|
},
|
|
{
|
|
"epoch": 6.297142857142857,
|
|
"grad_norm": 0.5015600982806857,
|
|
"learning_rate": 1.2229011197384021e-06,
|
|
"loss": 0.1695,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07090511918067932,
|
|
"step": 3860,
|
|
"valid_targets_mean": 5515.6,
|
|
"valid_targets_min": 3275
|
|
},
|
|
{
|
|
"epoch": 6.30530612244898,
|
|
"grad_norm": 0.5866542509141897,
|
|
"learning_rate": 1.1950407681681831e-06,
|
|
"loss": 0.1707,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07953464984893799,
|
|
"step": 3865,
|
|
"valid_targets_mean": 3868.6,
|
|
"valid_targets_min": 1761
|
|
},
|
|
{
|
|
"epoch": 6.313469387755102,
|
|
"grad_norm": 0.517511096201679,
|
|
"learning_rate": 1.1674916684526782e-06,
|
|
"loss": 0.1785,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08121711760759354,
|
|
"step": 3870,
|
|
"valid_targets_mean": 4257.8,
|
|
"valid_targets_min": 2048
|
|
},
|
|
{
|
|
"epoch": 6.321632653061225,
|
|
"grad_norm": 0.4774974122619391,
|
|
"learning_rate": 1.1402542765731072e-06,
|
|
"loss": 0.1683,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0860278308391571,
|
|
"step": 3875,
|
|
"valid_targets_mean": 5665.1,
|
|
"valid_targets_min": 1144
|
|
},
|
|
{
|
|
"epoch": 6.329795918367347,
|
|
"grad_norm": 0.4311198664284439,
|
|
"learning_rate": 1.1133290433514522e-06,
|
|
"loss": 0.1691,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07120402157306671,
|
|
"step": 3880,
|
|
"valid_targets_mean": 4677.0,
|
|
"valid_targets_min": 2435
|
|
},
|
|
{
|
|
"epoch": 6.337959183673469,
|
|
"grad_norm": 0.5839912382789675,
|
|
"learning_rate": 1.086716414442952e-06,
|
|
"loss": 0.1736,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.075423963367939,
|
|
"step": 3885,
|
|
"valid_targets_mean": 4535.6,
|
|
"valid_targets_min": 1917
|
|
},
|
|
{
|
|
"epoch": 6.3461224489795915,
|
|
"grad_norm": 0.5308773482359681,
|
|
"learning_rate": 1.0604168303287588e-06,
|
|
"loss": 0.1711,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07817616313695908,
|
|
"step": 3890,
|
|
"valid_targets_mean": 3928.2,
|
|
"valid_targets_min": 3006
|
|
},
|
|
{
|
|
"epoch": 6.354285714285714,
|
|
"grad_norm": 0.4985657759227005,
|
|
"learning_rate": 1.0344307263086417e-06,
|
|
"loss": 0.1744,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08368688821792603,
|
|
"step": 3895,
|
|
"valid_targets_mean": 5520.8,
|
|
"valid_targets_min": 2725
|
|
},
|
|
{
|
|
"epoch": 6.362448979591837,
|
|
"grad_norm": 0.47797322901288064,
|
|
"learning_rate": 1.008758532493772e-06,
|
|
"loss": 0.1734,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06277412176132202,
|
|
"step": 3900,
|
|
"valid_targets_mean": 4010.4,
|
|
"valid_targets_min": 844
|
|
},
|
|
{
|
|
"epoch": 6.370612244897959,
|
|
"grad_norm": 0.4862421427522955,
|
|
"learning_rate": 9.834006737996126e-07,
|
|
"loss": 0.1625,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0716070681810379,
|
|
"step": 3905,
|
|
"valid_targets_mean": 4795.8,
|
|
"valid_targets_min": 2504
|
|
},
|
|
{
|
|
"epoch": 6.378775510204082,
|
|
"grad_norm": 0.5535941599303137,
|
|
"learning_rate": 9.583575699388748e-07,
|
|
"loss": 0.1777,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09365393966436386,
|
|
"step": 3910,
|
|
"valid_targets_mean": 4736.4,
|
|
"valid_targets_min": 1733
|
|
},
|
|
{
|
|
"epoch": 6.386938775510204,
|
|
"grad_norm": 0.42700673020514873,
|
|
"learning_rate": 9.336296354145969e-07,
|
|
"loss": 0.17,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06854044646024704,
|
|
"step": 3915,
|
|
"valid_targets_mean": 4657.2,
|
|
"valid_targets_min": 2264
|
|
},
|
|
{
|
|
"epoch": 6.395102040816327,
|
|
"grad_norm": 0.6328165837439812,
|
|
"learning_rate": 9.092172795132459e-07,
|
|
"loss": 0.1727,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0929318219423294,
|
|
"step": 3920,
|
|
"valid_targets_mean": 3779.5,
|
|
"valid_targets_min": 1418
|
|
},
|
|
{
|
|
"epoch": 6.403265306122449,
|
|
"grad_norm": 0.5416202029620326,
|
|
"learning_rate": 8.851209062979715e-07,
|
|
"loss": 0.1805,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10915129631757736,
|
|
"step": 3925,
|
|
"valid_targets_mean": 5157.9,
|
|
"valid_targets_min": 2472
|
|
},
|
|
{
|
|
"epoch": 6.411428571428571,
|
|
"grad_norm": 0.5240492479008688,
|
|
"learning_rate": 8.613409146019158e-07,
|
|
"loss": 0.1786,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09709012508392334,
|
|
"step": 3930,
|
|
"valid_targets_mean": 4688.8,
|
|
"valid_targets_min": 2118
|
|
},
|
|
{
|
|
"epoch": 6.419591836734694,
|
|
"grad_norm": 0.42698292167145513,
|
|
"learning_rate": 8.378776980215919e-07,
|
|
"loss": 0.1779,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07900860905647278,
|
|
"step": 3935,
|
|
"valid_targets_mean": 5823.5,
|
|
"valid_targets_min": 3881
|
|
},
|
|
{
|
|
"epoch": 6.427755102040816,
|
|
"grad_norm": 0.5744560900858467,
|
|
"learning_rate": 8.147316449103959e-07,
|
|
"loss": 0.1741,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09831557422876358,
|
|
"step": 3940,
|
|
"valid_targets_mean": 4894.8,
|
|
"valid_targets_min": 469
|
|
},
|
|
{
|
|
"epoch": 6.435918367346939,
|
|
"grad_norm": 0.5299921373805195,
|
|
"learning_rate": 7.919031383721565e-07,
|
|
"loss": 0.1793,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08028759062290192,
|
|
"step": 3945,
|
|
"valid_targets_mean": 4523.4,
|
|
"valid_targets_min": 2392
|
|
},
|
|
{
|
|
"epoch": 6.444081632653061,
|
|
"grad_norm": 0.44582836009607935,
|
|
"learning_rate": 7.69392556254811e-07,
|
|
"loss": 0.166,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09207530319690704,
|
|
"step": 3950,
|
|
"valid_targets_mean": 6267.1,
|
|
"valid_targets_min": 3488
|
|
},
|
|
{
|
|
"epoch": 6.452244897959184,
|
|
"grad_norm": 0.4376049853971939,
|
|
"learning_rate": 7.472002711441418e-07,
|
|
"loss": 0.1782,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06368829309940338,
|
|
"step": 3955,
|
|
"valid_targets_mean": 4596.1,
|
|
"valid_targets_min": 2451
|
|
},
|
|
{
|
|
"epoch": 6.460408163265306,
|
|
"grad_norm": 0.5848626971055868,
|
|
"learning_rate": 7.253266503576051e-07,
|
|
"loss": 0.1787,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08905494213104248,
|
|
"step": 3960,
|
|
"valid_targets_mean": 3593.2,
|
|
"valid_targets_min": 2068
|
|
},
|
|
{
|
|
"epoch": 6.468571428571429,
|
|
"grad_norm": 0.555314669268285,
|
|
"learning_rate": 7.037720559382699e-07,
|
|
"loss": 0.1743,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08827593922615051,
|
|
"step": 3965,
|
|
"valid_targets_mean": 3865.5,
|
|
"valid_targets_min": 1420
|
|
},
|
|
{
|
|
"epoch": 6.476734693877551,
|
|
"grad_norm": 0.5514660209987079,
|
|
"learning_rate": 6.825368446488068e-07,
|
|
"loss": 0.1759,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10137411952018738,
|
|
"step": 3970,
|
|
"valid_targets_mean": 5435.1,
|
|
"valid_targets_min": 2282
|
|
},
|
|
{
|
|
"epoch": 6.484897959183673,
|
|
"grad_norm": 0.5054757359761405,
|
|
"learning_rate": 6.616213679655881e-07,
|
|
"loss": 0.1727,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0968596339225769,
|
|
"step": 3975,
|
|
"valid_targets_mean": 5499.6,
|
|
"valid_targets_min": 2797
|
|
},
|
|
{
|
|
"epoch": 6.493061224489796,
|
|
"grad_norm": 0.5159849008385041,
|
|
"learning_rate": 6.410259720728751e-07,
|
|
"loss": 0.17,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08823905885219574,
|
|
"step": 3980,
|
|
"valid_targets_mean": 4579.8,
|
|
"valid_targets_min": 2799
|
|
},
|
|
{
|
|
"epoch": 6.501224489795918,
|
|
"grad_norm": 0.44717355523296626,
|
|
"learning_rate": 6.207509978570848e-07,
|
|
"loss": 0.1648,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07926648110151291,
|
|
"step": 3985,
|
|
"valid_targets_mean": 5532.9,
|
|
"valid_targets_min": 2673
|
|
},
|
|
{
|
|
"epoch": 6.509387755102041,
|
|
"grad_norm": 0.5259133348625326,
|
|
"learning_rate": 6.007967809011517e-07,
|
|
"loss": 0.1694,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06820449978113174,
|
|
"step": 3990,
|
|
"valid_targets_mean": 3630.4,
|
|
"valid_targets_min": 1967
|
|
},
|
|
{
|
|
"epoch": 6.517551020408163,
|
|
"grad_norm": 0.4803407867104553,
|
|
"learning_rate": 5.811636514789598e-07,
|
|
"loss": 0.181,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09790891408920288,
|
|
"step": 3995,
|
|
"valid_targets_mean": 5995.4,
|
|
"valid_targets_min": 230
|
|
},
|
|
{
|
|
"epoch": 6.525714285714286,
|
|
"grad_norm": 0.5829612841676888,
|
|
"learning_rate": 5.618519345498952e-07,
|
|
"loss": 0.1637,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07763680815696716,
|
|
"step": 4000,
|
|
"valid_targets_mean": 3797.0,
|
|
"valid_targets_min": 1159
|
|
},
|
|
{
|
|
"epoch": 6.533877551020408,
|
|
"grad_norm": 0.5543971635143561,
|
|
"learning_rate": 5.428619497534548e-07,
|
|
"loss": 0.1779,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0884358286857605,
|
|
"step": 4005,
|
|
"valid_targets_mean": 4588.5,
|
|
"valid_targets_min": 1806
|
|
},
|
|
{
|
|
"epoch": 6.542040816326531,
|
|
"grad_norm": 0.504734374948056,
|
|
"learning_rate": 5.2419401140396e-07,
|
|
"loss": 0.17,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07078321278095245,
|
|
"step": 4010,
|
|
"valid_targets_mean": 4664.2,
|
|
"valid_targets_min": 1845
|
|
},
|
|
{
|
|
"epoch": 6.550204081632653,
|
|
"grad_norm": 0.4242043874771602,
|
|
"learning_rate": 5.058484284853494e-07,
|
|
"loss": 0.1625,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07394272089004517,
|
|
"step": 4015,
|
|
"valid_targets_mean": 5321.9,
|
|
"valid_targets_min": 2081
|
|
},
|
|
{
|
|
"epoch": 6.558367346938775,
|
|
"grad_norm": 0.6707218641944701,
|
|
"learning_rate": 4.878255046460755e-07,
|
|
"loss": 0.179,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07270123064517975,
|
|
"step": 4020,
|
|
"valid_targets_mean": 2379.8,
|
|
"valid_targets_min": 1694
|
|
},
|
|
{
|
|
"epoch": 6.566530612244898,
|
|
"grad_norm": 0.54487029299083,
|
|
"learning_rate": 4.701255381940684e-07,
|
|
"loss": 0.1809,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11994723975658417,
|
|
"step": 4025,
|
|
"valid_targets_mean": 6316.6,
|
|
"valid_targets_min": 4205
|
|
},
|
|
{
|
|
"epoch": 6.57469387755102,
|
|
"grad_norm": 0.5040060467257438,
|
|
"learning_rate": 4.5274882209180195e-07,
|
|
"loss": 0.1795,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08845013380050659,
|
|
"step": 4030,
|
|
"valid_targets_mean": 5180.4,
|
|
"valid_targets_min": 2284
|
|
},
|
|
{
|
|
"epoch": 6.582857142857143,
|
|
"grad_norm": 0.5126754142819413,
|
|
"learning_rate": 4.3569564395144595e-07,
|
|
"loss": 0.1757,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07828289270401001,
|
|
"step": 4035,
|
|
"valid_targets_mean": 4312.9,
|
|
"valid_targets_min": 2007
|
|
},
|
|
{
|
|
"epoch": 6.591020408163265,
|
|
"grad_norm": 0.5011495581492582,
|
|
"learning_rate": 4.189662860301069e-07,
|
|
"loss": 0.1832,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09557807445526123,
|
|
"step": 4040,
|
|
"valid_targets_mean": 6264.6,
|
|
"valid_targets_min": 3218
|
|
},
|
|
{
|
|
"epoch": 6.599183673469388,
|
|
"grad_norm": 0.4387049581215748,
|
|
"learning_rate": 4.025610252251433e-07,
|
|
"loss": 0.1747,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06180455535650253,
|
|
"step": 4045,
|
|
"valid_targets_mean": 5429.1,
|
|
"valid_targets_min": 802
|
|
},
|
|
{
|
|
"epoch": 6.60734693877551,
|
|
"grad_norm": 0.5220540147993991,
|
|
"learning_rate": 3.8648013306960664e-07,
|
|
"loss": 0.1721,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08135996013879776,
|
|
"step": 4050,
|
|
"valid_targets_mean": 4716.4,
|
|
"valid_targets_min": 2988
|
|
},
|
|
{
|
|
"epoch": 6.615510204081633,
|
|
"grad_norm": 0.4869619477437,
|
|
"learning_rate": 3.7072387572772317e-07,
|
|
"loss": 0.1798,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08093626797199249,
|
|
"step": 4055,
|
|
"valid_targets_mean": 5446.0,
|
|
"valid_targets_min": 201
|
|
},
|
|
{
|
|
"epoch": 6.6236734693877555,
|
|
"grad_norm": 0.5111229087974583,
|
|
"learning_rate": 3.552925139905083e-07,
|
|
"loss": 0.1815,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11379435658454895,
|
|
"step": 4060,
|
|
"valid_targets_mean": 6852.5,
|
|
"valid_targets_min": 3094
|
|
},
|
|
{
|
|
"epoch": 6.631836734693877,
|
|
"grad_norm": 0.6008745315447422,
|
|
"learning_rate": 3.4018630327143474e-07,
|
|
"loss": 0.1734,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10556414723396301,
|
|
"step": 4065,
|
|
"valid_targets_mean": 4508.4,
|
|
"valid_targets_min": 1910
|
|
},
|
|
{
|
|
"epoch": 6.64,
|
|
"grad_norm": 0.5057466157957375,
|
|
"learning_rate": 3.2540549360221105e-07,
|
|
"loss": 0.176,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09385024011135101,
|
|
"step": 4070,
|
|
"valid_targets_mean": 5681.0,
|
|
"valid_targets_min": 2830
|
|
},
|
|
{
|
|
"epoch": 6.648163265306122,
|
|
"grad_norm": 0.4568487337498165,
|
|
"learning_rate": 3.109503296286431e-07,
|
|
"loss": 0.1665,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0856882780790329,
|
|
"step": 4075,
|
|
"valid_targets_mean": 5635.8,
|
|
"valid_targets_min": 3360
|
|
},
|
|
{
|
|
"epoch": 6.656326530612245,
|
|
"grad_norm": 0.5328631009846894,
|
|
"learning_rate": 2.968210506065883e-07,
|
|
"loss": 0.1687,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0812063068151474,
|
|
"step": 4080,
|
|
"valid_targets_mean": 5037.9,
|
|
"valid_targets_min": 1807
|
|
},
|
|
{
|
|
"epoch": 6.664489795918367,
|
|
"grad_norm": 0.5606746811313319,
|
|
"learning_rate": 2.8301789039798523e-07,
|
|
"loss": 0.1722,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09690988063812256,
|
|
"step": 4085,
|
|
"valid_targets_mean": 4049.2,
|
|
"valid_targets_min": 1713
|
|
},
|
|
{
|
|
"epoch": 6.67265306122449,
|
|
"grad_norm": 0.477663793530669,
|
|
"learning_rate": 2.695410774669993e-07,
|
|
"loss": 0.1724,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07050501555204391,
|
|
"step": 4090,
|
|
"valid_targets_mean": 5059.9,
|
|
"valid_targets_min": 2563
|
|
},
|
|
{
|
|
"epoch": 6.680816326530612,
|
|
"grad_norm": 0.4907864488143317,
|
|
"learning_rate": 2.5639083487622116e-07,
|
|
"loss": 0.1777,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09305992722511292,
|
|
"step": 4095,
|
|
"valid_targets_mean": 4990.8,
|
|
"valid_targets_min": 1928
|
|
},
|
|
{
|
|
"epoch": 6.688979591836735,
|
|
"grad_norm": 0.5661942419711758,
|
|
"learning_rate": 2.435673802829941e-07,
|
|
"loss": 0.1822,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10230745375156403,
|
|
"step": 4100,
|
|
"valid_targets_mean": 4283.2,
|
|
"valid_targets_min": 1520
|
|
},
|
|
{
|
|
"epoch": 6.6971428571428575,
|
|
"grad_norm": 0.4826553611509298,
|
|
"learning_rate": 2.3107092593579905e-07,
|
|
"loss": 0.1735,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08301982283592224,
|
|
"step": 4105,
|
|
"valid_targets_mean": 4510.2,
|
|
"valid_targets_min": 1516
|
|
},
|
|
{
|
|
"epoch": 6.705306122448979,
|
|
"grad_norm": 0.49400269132368113,
|
|
"learning_rate": 2.1890167867074208e-07,
|
|
"loss": 0.1664,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07874350249767303,
|
|
"step": 4110,
|
|
"valid_targets_mean": 4519.6,
|
|
"valid_targets_min": 2366
|
|
},
|
|
{
|
|
"epoch": 6.713469387755102,
|
|
"grad_norm": 0.5404819175512501,
|
|
"learning_rate": 2.0705983990813915e-07,
|
|
"loss": 0.187,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09719543904066086,
|
|
"step": 4115,
|
|
"valid_targets_mean": 5836.2,
|
|
"valid_targets_min": 3722
|
|
},
|
|
{
|
|
"epoch": 6.721632653061224,
|
|
"grad_norm": 0.46187951904073626,
|
|
"learning_rate": 1.9554560564917001e-07,
|
|
"loss": 0.1713,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0860443189740181,
|
|
"step": 4120,
|
|
"valid_targets_mean": 6422.9,
|
|
"valid_targets_min": 1910
|
|
},
|
|
{
|
|
"epoch": 6.729795918367347,
|
|
"grad_norm": 0.5749101600772076,
|
|
"learning_rate": 1.843591664726496e-07,
|
|
"loss": 0.1804,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08156899362802505,
|
|
"step": 4125,
|
|
"valid_targets_mean": 3767.6,
|
|
"valid_targets_min": 1886
|
|
},
|
|
{
|
|
"epoch": 6.737959183673469,
|
|
"grad_norm": 0.5184829896214048,
|
|
"learning_rate": 1.7350070753186176e-07,
|
|
"loss": 0.175,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0800006240606308,
|
|
"step": 4130,
|
|
"valid_targets_mean": 4578.6,
|
|
"valid_targets_min": 2523
|
|
},
|
|
{
|
|
"epoch": 6.746122448979592,
|
|
"grad_norm": 0.5493945003783083,
|
|
"learning_rate": 1.6297040855149493e-07,
|
|
"loss": 0.1785,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11112290620803833,
|
|
"step": 4135,
|
|
"valid_targets_mean": 4718.0,
|
|
"valid_targets_min": 1843
|
|
},
|
|
{
|
|
"epoch": 6.7542857142857144,
|
|
"grad_norm": 0.5579444536692725,
|
|
"learning_rate": 1.5276844382467125e-07,
|
|
"loss": 0.1704,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08533066511154175,
|
|
"step": 4140,
|
|
"valid_targets_mean": 3846.4,
|
|
"valid_targets_min": 1548
|
|
},
|
|
{
|
|
"epoch": 6.762448979591837,
|
|
"grad_norm": 0.5697066680123272,
|
|
"learning_rate": 1.4289498221006893e-07,
|
|
"loss": 0.1775,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07441797852516174,
|
|
"step": 4145,
|
|
"valid_targets_mean": 3496.2,
|
|
"valid_targets_min": 1210
|
|
},
|
|
{
|
|
"epoch": 6.7706122448979595,
|
|
"grad_norm": 0.5354391536115609,
|
|
"learning_rate": 1.333501871291065e-07,
|
|
"loss": 0.1598,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08926458656787872,
|
|
"step": 4150,
|
|
"valid_targets_mean": 5302.8,
|
|
"valid_targets_min": 2814
|
|
},
|
|
{
|
|
"epoch": 6.778775510204081,
|
|
"grad_norm": 0.4695044365380142,
|
|
"learning_rate": 1.2413421656326307e-07,
|
|
"loss": 0.1788,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10014846175909042,
|
|
"step": 4155,
|
|
"valid_targets_mean": 6973.4,
|
|
"valid_targets_min": 1842
|
|
},
|
|
{
|
|
"epoch": 6.786938775510204,
|
|
"grad_norm": 0.5838126357143397,
|
|
"learning_rate": 1.1524722305144231e-07,
|
|
"loss": 0.1791,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07630135864019394,
|
|
"step": 4160,
|
|
"valid_targets_mean": 4727.8,
|
|
"valid_targets_min": 1813
|
|
},
|
|
{
|
|
"epoch": 6.795102040816326,
|
|
"grad_norm": 0.5070279879159841,
|
|
"learning_rate": 1.0668935368746136e-07,
|
|
"loss": 0.1801,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12365451455116272,
|
|
"step": 4165,
|
|
"valid_targets_mean": 6618.4,
|
|
"valid_targets_min": 4383
|
|
},
|
|
{
|
|
"epoch": 6.803265306122449,
|
|
"grad_norm": 0.5953179410730541,
|
|
"learning_rate": 9.846075011761048e-08,
|
|
"loss": 0.1712,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09822401404380798,
|
|
"step": 4170,
|
|
"valid_targets_mean": 3929.1,
|
|
"valid_targets_min": 1845
|
|
},
|
|
{
|
|
"epoch": 6.811428571428571,
|
|
"grad_norm": 0.697967990542345,
|
|
"learning_rate": 9.056154853830823e-08,
|
|
"loss": 0.1786,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08507151901721954,
|
|
"step": 4175,
|
|
"valid_targets_mean": 3861.1,
|
|
"valid_targets_min": 2219
|
|
},
|
|
{
|
|
"epoch": 6.819591836734694,
|
|
"grad_norm": 0.48169611662455075,
|
|
"learning_rate": 8.299187969385226e-08,
|
|
"loss": 0.1685,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07968790829181671,
|
|
"step": 4180,
|
|
"valid_targets_mean": 4887.4,
|
|
"valid_targets_min": 2593
|
|
},
|
|
{
|
|
"epoch": 6.8277551020408165,
|
|
"grad_norm": 0.4578362029303609,
|
|
"learning_rate": 7.575186887424757e-08,
|
|
"loss": 0.1696,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08985200524330139,
|
|
"step": 4185,
|
|
"valid_targets_mean": 7675.6,
|
|
"valid_targets_min": 2966
|
|
},
|
|
{
|
|
"epoch": 6.835918367346939,
|
|
"grad_norm": 0.5139405442593589,
|
|
"learning_rate": 6.884163591313941e-08,
|
|
"loss": 0.162,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08233918249607086,
|
|
"step": 4190,
|
|
"valid_targets_mean": 4719.0,
|
|
"valid_targets_min": 1192
|
|
},
|
|
{
|
|
"epoch": 6.844081632653062,
|
|
"grad_norm": 0.5600721868850346,
|
|
"learning_rate": 6.226129518582369e-08,
|
|
"loss": 0.1722,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08688675612211227,
|
|
"step": 4195,
|
|
"valid_targets_mean": 4590.5,
|
|
"valid_targets_min": 2579
|
|
},
|
|
{
|
|
"epoch": 6.852244897959183,
|
|
"grad_norm": 0.4247434382218982,
|
|
"learning_rate": 5.6010955607366245e-08,
|
|
"loss": 0.1681,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05935890972614288,
|
|
"step": 4200,
|
|
"valid_targets_mean": 3941.9,
|
|
"valid_targets_min": 2516
|
|
},
|
|
{
|
|
"epoch": 6.860408163265306,
|
|
"grad_norm": 0.6070505109956861,
|
|
"learning_rate": 5.009072063077769e-08,
|
|
"loss": 0.185,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06743767112493515,
|
|
"step": 4205,
|
|
"valid_targets_mean": 3737.9,
|
|
"valid_targets_min": 758
|
|
},
|
|
{
|
|
"epoch": 6.868571428571428,
|
|
"grad_norm": 0.5891898407341091,
|
|
"learning_rate": 4.450068824532583e-08,
|
|
"loss": 0.1779,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08612845838069916,
|
|
"step": 4210,
|
|
"valid_targets_mean": 4001.5,
|
|
"valid_targets_min": 191
|
|
},
|
|
{
|
|
"epoch": 6.876734693877551,
|
|
"grad_norm": 0.49097204196746064,
|
|
"learning_rate": 3.924095097489922e-08,
|
|
"loss": 0.173,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09206587821245193,
|
|
"step": 4215,
|
|
"valid_targets_mean": 5461.0,
|
|
"valid_targets_min": 3681
|
|
},
|
|
{
|
|
"epoch": 6.884897959183673,
|
|
"grad_norm": 0.5234098272363491,
|
|
"learning_rate": 3.4311595876479475e-08,
|
|
"loss": 0.1761,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07597896456718445,
|
|
"step": 4220,
|
|
"valid_targets_mean": 4421.2,
|
|
"valid_targets_min": 1266
|
|
},
|
|
{
|
|
"epoch": 6.893061224489796,
|
|
"grad_norm": 0.6033382656401394,
|
|
"learning_rate": 2.971270453869579e-08,
|
|
"loss": 0.1733,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08271235227584839,
|
|
"step": 4225,
|
|
"valid_targets_mean": 3503.1,
|
|
"valid_targets_min": 1923
|
|
},
|
|
{
|
|
"epoch": 6.9012244897959185,
|
|
"grad_norm": 0.6792301772472162,
|
|
"learning_rate": 2.5444353080485983e-08,
|
|
"loss": 0.1784,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09355944395065308,
|
|
"step": 4230,
|
|
"valid_targets_mean": 6051.4,
|
|
"valid_targets_min": 2813
|
|
},
|
|
{
|
|
"epoch": 6.909387755102041,
|
|
"grad_norm": 0.48141128673204825,
|
|
"learning_rate": 2.150661214982641e-08,
|
|
"loss": 0.1679,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09260822832584381,
|
|
"step": 4235,
|
|
"valid_targets_mean": 5953.1,
|
|
"valid_targets_min": 3984
|
|
},
|
|
{
|
|
"epoch": 6.917551020408164,
|
|
"grad_norm": 0.5441935587073681,
|
|
"learning_rate": 1.7899546922564016e-08,
|
|
"loss": 0.182,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0769985094666481,
|
|
"step": 4240,
|
|
"valid_targets_mean": 3709.2,
|
|
"valid_targets_min": 1469
|
|
},
|
|
{
|
|
"epoch": 6.925714285714285,
|
|
"grad_norm": 0.5536238988411614,
|
|
"learning_rate": 1.4623217101339405e-08,
|
|
"loss": 0.1655,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07819455862045288,
|
|
"step": 4245,
|
|
"valid_targets_mean": 3355.0,
|
|
"valid_targets_min": 160
|
|
},
|
|
{
|
|
"epoch": 6.933877551020408,
|
|
"grad_norm": 0.5080659448282431,
|
|
"learning_rate": 1.1677676914603197e-08,
|
|
"loss": 0.1645,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07104445993900299,
|
|
"step": 4250,
|
|
"valid_targets_mean": 4236.8,
|
|
"valid_targets_min": 2233
|
|
},
|
|
{
|
|
"epoch": 6.94204081632653,
|
|
"grad_norm": 0.5853986089854679,
|
|
"learning_rate": 9.06297511570564e-09,
|
|
"loss": 0.178,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1054919958114624,
|
|
"step": 4255,
|
|
"valid_targets_mean": 4447.1,
|
|
"valid_targets_min": 1903
|
|
},
|
|
{
|
|
"epoch": 6.950204081632653,
|
|
"grad_norm": 0.5012639100276955,
|
|
"learning_rate": 6.779154982103908e-09,
|
|
"loss": 0.1617,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07082746922969818,
|
|
"step": 4260,
|
|
"valid_targets_mean": 4263.4,
|
|
"valid_targets_min": 1060
|
|
},
|
|
{
|
|
"epoch": 6.9583673469387755,
|
|
"grad_norm": 0.5178754273320104,
|
|
"learning_rate": 4.8262543146382345e-09,
|
|
"loss": 0.1806,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07891260087490082,
|
|
"step": 4265,
|
|
"valid_targets_mean": 4791.5,
|
|
"valid_targets_min": 2335
|
|
},
|
|
{
|
|
"epoch": 6.966530612244898,
|
|
"grad_norm": 0.4976765498227958,
|
|
"learning_rate": 3.2043054369057523e-09,
|
|
"loss": 0.1783,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10171708464622498,
|
|
"step": 4270,
|
|
"valid_targets_mean": 5939.1,
|
|
"valid_targets_min": 4245
|
|
},
|
|
{
|
|
"epoch": 6.9746938775510205,
|
|
"grad_norm": 0.511741522266631,
|
|
"learning_rate": 1.9133351947320246e-09,
|
|
"loss": 0.1772,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08228804171085358,
|
|
"step": 4275,
|
|
"valid_targets_mean": 3950.4,
|
|
"valid_targets_min": 3062
|
|
},
|
|
{
|
|
"epoch": 6.982857142857143,
|
|
"grad_norm": 0.580838054903444,
|
|
"learning_rate": 9.53364955715852e-10,
|
|
"loss": 0.1724,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08037175238132477,
|
|
"step": 4280,
|
|
"valid_targets_mean": 3290.5,
|
|
"valid_targets_min": 1396
|
|
},
|
|
{
|
|
"epoch": 6.991020408163266,
|
|
"grad_norm": 0.5444763889021368,
|
|
"learning_rate": 3.244106088873267e-10,
|
|
"loss": 0.1788,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08252573758363724,
|
|
"step": 4285,
|
|
"valid_targets_mean": 3990.9,
|
|
"valid_targets_min": 1991
|
|
},
|
|
{
|
|
"epoch": 6.999183673469387,
|
|
"grad_norm": 0.5481609381293677,
|
|
"learning_rate": 2.648256443471553e-11,
|
|
"loss": 0.1759,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0896102786064148,
|
|
"step": 4290,
|
|
"valid_targets_mean": 4104.1,
|
|
"valid_targets_min": 2246
|
|
},
|
|
{
|
|
"epoch": 7.0,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15556100010871887,
|
|
"step": 4291,
|
|
"total_flos": 1.820182243006546e+18,
|
|
"train_loss": 0.2264533686425352,
|
|
"train_runtime": 56587.5923,
|
|
"train_samples_per_second": 1.212,
|
|
"train_steps_per_second": 0.076,
|
|
"valid_targets_mean": 4983.8,
|
|
"valid_targets_min": 1127
|
|
}
|
|
],
|
|
"logging_steps": 5,
|
|
"max_steps": 4291,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 7,
|
|
"save_steps": 200,
|
|
"stateful_callbacks": {
|
|
"TrainerControl": {
|
|
"args": {
|
|
"should_epoch_stop": false,
|
|
"should_evaluate": false,
|
|
"should_log": false,
|
|
"should_save": true,
|
|
"should_training_stop": true
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 1.820182243006546e+18,
|
|
"train_batch_size": 1,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|