mikhail-panzo's picture
Training in progress, step 4500, checkpoint
c28d10d verified
raw
history blame
17.8 kB
{
"best_metric": 0.3783535957336426,
"best_model_checkpoint": "mikhail_panzo/zlm_b128_le5_s12000/checkpoint-4500",
"epoch": 7.542198952879581,
"eval_steps": 500,
"global_step": 4500,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.08376963350785341,
"grad_norm": 6.650441646575928,
"learning_rate": 2.4500000000000004e-07,
"loss": 1.1088,
"step": 50
},
{
"epoch": 0.16753926701570682,
"grad_norm": 13.608625411987305,
"learning_rate": 4.95e-07,
"loss": 1.1077,
"step": 100
},
{
"epoch": 0.2513089005235602,
"grad_norm": 2.7265217304229736,
"learning_rate": 7.450000000000001e-07,
"loss": 0.9759,
"step": 150
},
{
"epoch": 0.33507853403141363,
"grad_norm": 5.083759307861328,
"learning_rate": 9.950000000000002e-07,
"loss": 0.8705,
"step": 200
},
{
"epoch": 0.418848167539267,
"grad_norm": 1.8121678829193115,
"learning_rate": 1.2450000000000002e-06,
"loss": 0.8185,
"step": 250
},
{
"epoch": 0.5026178010471204,
"grad_norm": 2.275899648666382,
"learning_rate": 1.495e-06,
"loss": 0.784,
"step": 300
},
{
"epoch": 0.5863874345549738,
"grad_norm": 2.2861170768737793,
"learning_rate": 1.745e-06,
"loss": 0.7545,
"step": 350
},
{
"epoch": 0.6701570680628273,
"grad_norm": 3.1991238594055176,
"learning_rate": 1.9950000000000004e-06,
"loss": 0.7227,
"step": 400
},
{
"epoch": 0.7539267015706806,
"grad_norm": 1.843092679977417,
"learning_rate": 2.245e-06,
"loss": 0.6829,
"step": 450
},
{
"epoch": 0.837696335078534,
"grad_norm": 1.4700403213500977,
"learning_rate": 2.4950000000000003e-06,
"loss": 0.6644,
"step": 500
},
{
"epoch": 0.837696335078534,
"eval_loss": 0.5727962851524353,
"eval_runtime": 293.5061,
"eval_samples_per_second": 28.923,
"eval_steps_per_second": 3.618,
"step": 500
},
{
"epoch": 0.9214659685863874,
"grad_norm": 1.4362367391586304,
"learning_rate": 2.7450000000000004e-06,
"loss": 0.6487,
"step": 550
},
{
"epoch": 1.0052356020942408,
"grad_norm": 1.8555901050567627,
"learning_rate": 2.995e-06,
"loss": 0.6301,
"step": 600
},
{
"epoch": 1.0890052356020943,
"grad_norm": 1.9496757984161377,
"learning_rate": 3.2450000000000003e-06,
"loss": 0.6237,
"step": 650
},
{
"epoch": 1.1727748691099475,
"grad_norm": 3.0705084800720215,
"learning_rate": 3.495e-06,
"loss": 0.6156,
"step": 700
},
{
"epoch": 1.256544502617801,
"grad_norm": 2.2215688228607178,
"learning_rate": 3.745e-06,
"loss": 0.5945,
"step": 750
},
{
"epoch": 1.3403141361256545,
"grad_norm": 2.021375894546509,
"learning_rate": 3.995000000000001e-06,
"loss": 0.5891,
"step": 800
},
{
"epoch": 1.4240837696335078,
"grad_norm": 2.0354769229888916,
"learning_rate": 4.245e-06,
"loss": 0.5781,
"step": 850
},
{
"epoch": 1.5078534031413613,
"grad_norm": 1.8674426078796387,
"learning_rate": 4.495e-06,
"loss": 0.5677,
"step": 900
},
{
"epoch": 1.5916230366492146,
"grad_norm": 2.9264817237854004,
"learning_rate": 4.745e-06,
"loss": 0.5576,
"step": 950
},
{
"epoch": 1.675392670157068,
"grad_norm": 1.9513416290283203,
"learning_rate": 4.9950000000000005e-06,
"loss": 0.5594,
"step": 1000
},
{
"epoch": 1.675392670157068,
"eval_loss": 0.48248979449272156,
"eval_runtime": 294.2627,
"eval_samples_per_second": 28.848,
"eval_steps_per_second": 3.609,
"step": 1000
},
{
"epoch": 1.7591623036649215,
"grad_norm": 1.934941291809082,
"learning_rate": 5.245e-06,
"loss": 0.5571,
"step": 1050
},
{
"epoch": 1.8429319371727748,
"grad_norm": 1.7669196128845215,
"learning_rate": 5.495000000000001e-06,
"loss": 0.5477,
"step": 1100
},
{
"epoch": 1.9267015706806283,
"grad_norm": 3.954806089401245,
"learning_rate": 5.745000000000001e-06,
"loss": 0.5294,
"step": 1150
},
{
"epoch": 2.0104712041884816,
"grad_norm": 2.8323569297790527,
"learning_rate": 5.995000000000001e-06,
"loss": 0.5261,
"step": 1200
},
{
"epoch": 2.094240837696335,
"grad_norm": 2.193530797958374,
"learning_rate": 6.245000000000001e-06,
"loss": 0.5262,
"step": 1250
},
{
"epoch": 2.1780104712041886,
"grad_norm": 1.5814998149871826,
"learning_rate": 6.4950000000000005e-06,
"loss": 0.5204,
"step": 1300
},
{
"epoch": 2.261780104712042,
"grad_norm": 5.6610918045043945,
"learning_rate": 6.745000000000001e-06,
"loss": 0.5157,
"step": 1350
},
{
"epoch": 2.345549738219895,
"grad_norm": 9.156899452209473,
"learning_rate": 6.995000000000001e-06,
"loss": 0.5178,
"step": 1400
},
{
"epoch": 2.4293193717277486,
"grad_norm": 2.3170149326324463,
"learning_rate": 7.245000000000001e-06,
"loss": 0.5163,
"step": 1450
},
{
"epoch": 2.513089005235602,
"grad_norm": 2.1565325260162354,
"learning_rate": 7.495000000000001e-06,
"loss": 0.5042,
"step": 1500
},
{
"epoch": 2.513089005235602,
"eval_loss": 0.4464746415615082,
"eval_runtime": 293.8295,
"eval_samples_per_second": 28.891,
"eval_steps_per_second": 3.614,
"step": 1500
},
{
"epoch": 2.5997905759162303,
"grad_norm": 2.4758172035217285,
"learning_rate": 7.745e-06,
"loss": 0.5031,
"step": 1550
},
{
"epoch": 2.683560209424084,
"grad_norm": 2.1877381801605225,
"learning_rate": 7.990000000000001e-06,
"loss": 0.4994,
"step": 1600
},
{
"epoch": 2.7673298429319373,
"grad_norm": 2.4486210346221924,
"learning_rate": 8.24e-06,
"loss": 0.4965,
"step": 1650
},
{
"epoch": 2.8510994764397903,
"grad_norm": 2.596200704574585,
"learning_rate": 8.49e-06,
"loss": 0.4988,
"step": 1700
},
{
"epoch": 2.934869109947644,
"grad_norm": 1.7787096500396729,
"learning_rate": 8.740000000000001e-06,
"loss": 0.4947,
"step": 1750
},
{
"epoch": 3.0186387434554973,
"grad_norm": 2.09403133392334,
"learning_rate": 8.99e-06,
"loss": 0.4845,
"step": 1800
},
{
"epoch": 3.102408376963351,
"grad_norm": 2.0056636333465576,
"learning_rate": 9.240000000000001e-06,
"loss": 0.4833,
"step": 1850
},
{
"epoch": 3.1861780104712043,
"grad_norm": 3.200199842453003,
"learning_rate": 9.49e-06,
"loss": 0.484,
"step": 1900
},
{
"epoch": 3.269947643979058,
"grad_norm": 2.5462379455566406,
"learning_rate": 9.74e-06,
"loss": 0.4726,
"step": 1950
},
{
"epoch": 3.353717277486911,
"grad_norm": 1.3994622230529785,
"learning_rate": 9.990000000000001e-06,
"loss": 0.4795,
"step": 2000
},
{
"epoch": 3.353717277486911,
"eval_loss": 0.42617055773735046,
"eval_runtime": 269.3627,
"eval_samples_per_second": 31.515,
"eval_steps_per_second": 3.943,
"step": 2000
},
{
"epoch": 3.4374869109947643,
"grad_norm": 1.8979076147079468,
"learning_rate": 9.952e-06,
"loss": 0.4792,
"step": 2050
},
{
"epoch": 3.521256544502618,
"grad_norm": 1.4493324756622314,
"learning_rate": 9.902000000000001e-06,
"loss": 0.4751,
"step": 2100
},
{
"epoch": 3.6050261780104713,
"grad_norm": 3.3374929428100586,
"learning_rate": 9.852e-06,
"loss": 0.4736,
"step": 2150
},
{
"epoch": 3.6887958115183244,
"grad_norm": 3.5062992572784424,
"learning_rate": 9.802e-06,
"loss": 0.4675,
"step": 2200
},
{
"epoch": 3.772565445026178,
"grad_norm": 2.250505208969116,
"learning_rate": 9.752e-06,
"loss": 0.4611,
"step": 2250
},
{
"epoch": 3.8563350785340313,
"grad_norm": 1.793270468711853,
"learning_rate": 9.702e-06,
"loss": 0.4605,
"step": 2300
},
{
"epoch": 3.940104712041885,
"grad_norm": 1.663677453994751,
"learning_rate": 9.652e-06,
"loss": 0.4693,
"step": 2350
},
{
"epoch": 4.023874345549738,
"grad_norm": 2.1321282386779785,
"learning_rate": 9.602e-06,
"loss": 0.4587,
"step": 2400
},
{
"epoch": 4.107643979057592,
"grad_norm": 1.7361410856246948,
"learning_rate": 9.552000000000001e-06,
"loss": 0.4611,
"step": 2450
},
{
"epoch": 4.191413612565445,
"grad_norm": 2.167386770248413,
"learning_rate": 9.502000000000002e-06,
"loss": 0.455,
"step": 2500
},
{
"epoch": 4.191413612565445,
"eval_loss": 0.40905508399009705,
"eval_runtime": 269.8231,
"eval_samples_per_second": 31.461,
"eval_steps_per_second": 3.936,
"step": 2500
},
{
"epoch": 4.275183246073299,
"grad_norm": 1.430746078491211,
"learning_rate": 9.452000000000002e-06,
"loss": 0.4581,
"step": 2550
},
{
"epoch": 4.358952879581151,
"grad_norm": 2.1168527603149414,
"learning_rate": 9.402e-06,
"loss": 0.4516,
"step": 2600
},
{
"epoch": 4.442722513089005,
"grad_norm": 2.1330721378326416,
"learning_rate": 9.353000000000002e-06,
"loss": 0.4505,
"step": 2650
},
{
"epoch": 4.526492146596858,
"grad_norm": 1.274557113647461,
"learning_rate": 9.303e-06,
"loss": 0.4575,
"step": 2700
},
{
"epoch": 4.610261780104712,
"grad_norm": 1.835204839706421,
"learning_rate": 9.253000000000001e-06,
"loss": 0.4491,
"step": 2750
},
{
"epoch": 4.694031413612565,
"grad_norm": 2.0255746841430664,
"learning_rate": 9.203000000000002e-06,
"loss": 0.4484,
"step": 2800
},
{
"epoch": 4.777801047120419,
"grad_norm": 2.4793522357940674,
"learning_rate": 9.153e-06,
"loss": 0.4489,
"step": 2850
},
{
"epoch": 4.861570680628272,
"grad_norm": 1.7192201614379883,
"learning_rate": 9.103e-06,
"loss": 0.4502,
"step": 2900
},
{
"epoch": 4.945340314136126,
"grad_norm": 1.9378846883773804,
"learning_rate": 9.053000000000001e-06,
"loss": 0.4483,
"step": 2950
},
{
"epoch": 5.029109947643979,
"grad_norm": 2.4576103687286377,
"learning_rate": 9.003e-06,
"loss": 0.4474,
"step": 3000
},
{
"epoch": 5.029109947643979,
"eval_loss": 0.39903518557548523,
"eval_runtime": 269.2955,
"eval_samples_per_second": 31.523,
"eval_steps_per_second": 3.944,
"step": 3000
},
{
"epoch": 5.112879581151833,
"grad_norm": 1.4718371629714966,
"learning_rate": 8.953e-06,
"loss": 0.4428,
"step": 3050
},
{
"epoch": 5.196649214659685,
"grad_norm": 2.4222004413604736,
"learning_rate": 8.903000000000001e-06,
"loss": 0.4417,
"step": 3100
},
{
"epoch": 5.280418848167539,
"grad_norm": 1.5284855365753174,
"learning_rate": 8.853e-06,
"loss": 0.4395,
"step": 3150
},
{
"epoch": 5.364188481675392,
"grad_norm": 1.7470890283584595,
"learning_rate": 8.803e-06,
"loss": 0.4359,
"step": 3200
},
{
"epoch": 5.447958115183246,
"grad_norm": 2.1919684410095215,
"learning_rate": 8.753e-06,
"loss": 0.4389,
"step": 3250
},
{
"epoch": 5.531727748691099,
"grad_norm": 1.6179628372192383,
"learning_rate": 8.703e-06,
"loss": 0.4423,
"step": 3300
},
{
"epoch": 5.615497382198953,
"grad_norm": 1.420087218284607,
"learning_rate": 8.653e-06,
"loss": 0.4324,
"step": 3350
},
{
"epoch": 5.699267015706806,
"grad_norm": 1.6972101926803589,
"learning_rate": 8.603e-06,
"loss": 0.4327,
"step": 3400
},
{
"epoch": 5.78303664921466,
"grad_norm": 8.797286033630371,
"learning_rate": 8.553000000000001e-06,
"loss": 0.4347,
"step": 3450
},
{
"epoch": 5.866806282722513,
"grad_norm": 1.5441597700119019,
"learning_rate": 8.503e-06,
"loss": 0.4342,
"step": 3500
},
{
"epoch": 5.866806282722513,
"eval_loss": 0.39290204644203186,
"eval_runtime": 272.1215,
"eval_samples_per_second": 31.196,
"eval_steps_per_second": 3.903,
"step": 3500
},
{
"epoch": 5.950575916230367,
"grad_norm": 1.957065224647522,
"learning_rate": 8.453000000000002e-06,
"loss": 0.4288,
"step": 3550
},
{
"epoch": 6.0343455497382195,
"grad_norm": 1.3891559839248657,
"learning_rate": 8.403e-06,
"loss": 0.429,
"step": 3600
},
{
"epoch": 6.118115183246073,
"grad_norm": 1.6092897653579712,
"learning_rate": 8.353000000000001e-06,
"loss": 0.431,
"step": 3650
},
{
"epoch": 6.201884816753926,
"grad_norm": 1.5566798448562622,
"learning_rate": 8.303000000000002e-06,
"loss": 0.428,
"step": 3700
},
{
"epoch": 6.28565445026178,
"grad_norm": 2.70001220703125,
"learning_rate": 8.253e-06,
"loss": 0.433,
"step": 3750
},
{
"epoch": 6.369424083769633,
"grad_norm": 2.6352884769439697,
"learning_rate": 8.203000000000001e-06,
"loss": 0.431,
"step": 3800
},
{
"epoch": 6.453193717277487,
"grad_norm": 2.049609422683716,
"learning_rate": 8.153000000000001e-06,
"loss": 0.4279,
"step": 3850
},
{
"epoch": 6.53696335078534,
"grad_norm": 1.422606110572815,
"learning_rate": 8.103e-06,
"loss": 0.4266,
"step": 3900
},
{
"epoch": 6.620732984293194,
"grad_norm": 1.659340739250183,
"learning_rate": 8.053e-06,
"loss": 0.4293,
"step": 3950
},
{
"epoch": 6.704502617801047,
"grad_norm": 1.6214194297790527,
"learning_rate": 8.003000000000001e-06,
"loss": 0.4258,
"step": 4000
},
{
"epoch": 6.704502617801047,
"eval_loss": 0.385530561208725,
"eval_runtime": 270.5177,
"eval_samples_per_second": 31.381,
"eval_steps_per_second": 3.926,
"step": 4000
},
{
"epoch": 6.788272251308901,
"grad_norm": 2.0542335510253906,
"learning_rate": 7.953e-06,
"loss": 0.4232,
"step": 4050
},
{
"epoch": 6.872041884816754,
"grad_norm": 2.0001602172851562,
"learning_rate": 7.903e-06,
"loss": 0.4201,
"step": 4100
},
{
"epoch": 6.955811518324607,
"grad_norm": 1.5019524097442627,
"learning_rate": 7.853000000000001e-06,
"loss": 0.4202,
"step": 4150
},
{
"epoch": 7.0395811518324605,
"grad_norm": 1.5901544094085693,
"learning_rate": 7.803000000000001e-06,
"loss": 0.4214,
"step": 4200
},
{
"epoch": 7.123350785340314,
"grad_norm": 1.3859797716140747,
"learning_rate": 7.753e-06,
"loss": 0.4268,
"step": 4250
},
{
"epoch": 7.2071204188481675,
"grad_norm": 1.6223664283752441,
"learning_rate": 7.703e-06,
"loss": 0.4163,
"step": 4300
},
{
"epoch": 7.290890052356021,
"grad_norm": 1.9755197763442993,
"learning_rate": 7.653000000000001e-06,
"loss": 0.421,
"step": 4350
},
{
"epoch": 7.374659685863874,
"grad_norm": 1.9882097244262695,
"learning_rate": 7.603000000000001e-06,
"loss": 0.4215,
"step": 4400
},
{
"epoch": 7.458429319371728,
"grad_norm": 1.4274579286575317,
"learning_rate": 7.553e-06,
"loss": 0.424,
"step": 4450
},
{
"epoch": 7.542198952879581,
"grad_norm": 1.5370174646377563,
"learning_rate": 7.503e-06,
"loss": 0.4175,
"step": 4500
},
{
"epoch": 7.542198952879581,
"eval_loss": 0.3783535957336426,
"eval_runtime": 264.9436,
"eval_samples_per_second": 32.041,
"eval_steps_per_second": 4.008,
"step": 4500
}
],
"logging_steps": 50,
"max_steps": 12000,
"num_input_tokens_seen": 0,
"num_train_epochs": 21,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 8.073028933214573e+16,
"train_batch_size": 16,
"trial_name": null,
"trial_params": null
}