llama_oracle / trainer_state.json
linluqiu's picture
Add files using upload-large-folder tool
5b285a8 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9969230769230769,
"eval_steps": 100,
"global_step": 243,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0041025641025641026,
"grad_norm": 86.8332649639965,
"learning_rate": 8e-08,
"loss": 7.2742,
"step": 1
},
{
"epoch": 0.020512820512820513,
"grad_norm": 74.05727804792163,
"learning_rate": 4e-07,
"loss": 6.4673,
"step": 5
},
{
"epoch": 0.041025641025641026,
"grad_norm": 38.97594404855973,
"learning_rate": 8e-07,
"loss": 4.526,
"step": 10
},
{
"epoch": 0.06153846153846154,
"grad_norm": 5.6840773038859655,
"learning_rate": 1.2e-06,
"loss": 1.6883,
"step": 15
},
{
"epoch": 0.08205128205128205,
"grad_norm": 1.3276030305734883,
"learning_rate": 1.6e-06,
"loss": 1.1059,
"step": 20
},
{
"epoch": 0.10256410256410256,
"grad_norm": 2.343635823419092,
"learning_rate": 2e-06,
"loss": 1.1063,
"step": 25
},
{
"epoch": 0.12307692307692308,
"grad_norm": 2.289270560431258,
"learning_rate": 1.9974051702905273e-06,
"loss": 1.0758,
"step": 30
},
{
"epoch": 0.14358974358974358,
"grad_norm": 1.5816159769523035,
"learning_rate": 1.9896341474445524e-06,
"loss": 1.01,
"step": 35
},
{
"epoch": 0.1641025641025641,
"grad_norm": 2.3893777455781637,
"learning_rate": 1.976727260423982e-06,
"loss": 0.9688,
"step": 40
},
{
"epoch": 0.18461538461538463,
"grad_norm": 1.6349839241564788,
"learning_rate": 1.9587514915766123e-06,
"loss": 0.9678,
"step": 45
},
{
"epoch": 0.20512820512820512,
"grad_norm": 3.191279307503814,
"learning_rate": 1.935800129020554e-06,
"loss": 0.9354,
"step": 50
},
{
"epoch": 0.22564102564102564,
"grad_norm": 1.6448779128510165,
"learning_rate": 1.907992282510675e-06,
"loss": 0.9385,
"step": 55
},
{
"epoch": 0.24615384615384617,
"grad_norm": 1.9829387005126868,
"learning_rate": 1.8754722652995345e-06,
"loss": 0.9058,
"step": 60
},
{
"epoch": 0.26666666666666666,
"grad_norm": 2.29111150139111,
"learning_rate": 1.8384088452007577e-06,
"loss": 0.8891,
"step": 65
},
{
"epoch": 0.28717948717948716,
"grad_norm": 1.5810061706904381,
"learning_rate": 1.7969943687415575e-06,
"loss": 0.8934,
"step": 70
},
{
"epoch": 0.3076923076923077,
"grad_norm": 3.073136058331183,
"learning_rate": 1.751443762949772e-06,
"loss": 0.9122,
"step": 75
},
{
"epoch": 0.3282051282051282,
"grad_norm": 1.7225381141190559,
"learning_rate": 1.7019934199557866e-06,
"loss": 0.9145,
"step": 80
},
{
"epoch": 0.3487179487179487,
"grad_norm": 1.3310212541816835,
"learning_rate": 1.6488999701978903e-06,
"loss": 0.8261,
"step": 85
},
{
"epoch": 0.36923076923076925,
"grad_norm": 1.6745334659944915,
"learning_rate": 1.5924389505977035e-06,
"loss": 0.8751,
"step": 90
},
{
"epoch": 0.38974358974358975,
"grad_norm": 1.8233491777788193,
"learning_rate": 1.5329033746173974e-06,
"loss": 0.8872,
"step": 95
},
{
"epoch": 0.41025641025641024,
"grad_norm": 1.3524539580904948,
"learning_rate": 1.4706022116196205e-06,
"loss": 0.9192,
"step": 100
},
{
"epoch": 0.4307692307692308,
"grad_norm": 1.2310835465382302,
"learning_rate": 1.4058587834217354e-06,
"loss": 0.8605,
"step": 105
},
{
"epoch": 0.4512820512820513,
"grad_norm": 1.4801328470056587,
"learning_rate": 1.3390090863657047e-06,
"loss": 0.8715,
"step": 110
},
{
"epoch": 0.4717948717948718,
"grad_norm": 1.117288039177199,
"learning_rate": 1.2704000476115078e-06,
"loss": 0.8293,
"step": 115
},
{
"epoch": 0.49230769230769234,
"grad_norm": 2.3897644821265698,
"learning_rate": 1.200387724703341e-06,
"loss": 0.8803,
"step": 120
},
{
"epoch": 0.5128205128205128,
"grad_norm": 1.270827623469446,
"learning_rate": 1.1293354577522264e-06,
"loss": 0.8735,
"step": 125
},
{
"epoch": 0.5333333333333333,
"grad_norm": 1.2548143617737169,
"learning_rate": 1.0576119838245842e-06,
"loss": 0.8359,
"step": 130
},
{
"epoch": 0.5538461538461539,
"grad_norm": 1.184800287371971,
"learning_rate": 9.85589523322443e-07,
"loss": 0.883,
"step": 135
},
{
"epoch": 0.5743589743589743,
"grad_norm": 1.5024665213422632,
"learning_rate": 9.136418482863228e-07,
"loss": 0.85,
"step": 140
},
{
"epoch": 0.5948717948717949,
"grad_norm": 1.0627183442814103,
"learning_rate": 8.42142342645646e-07,
"loss": 0.8154,
"step": 145
},
{
"epoch": 0.6153846153846154,
"grad_norm": 1.7796232405974886,
"learning_rate": 7.714620644833109e-07,
"loss": 0.8369,
"step": 150
},
{
"epoch": 0.6358974358974359,
"grad_norm": 1.3421106550942563,
"learning_rate": 7.019678203706163e-07,
"loss": 0.8171,
"step": 155
},
{
"epoch": 0.6564102564102564,
"grad_norm": 1.1941093068038942,
"learning_rate": 6.340202617660841e-07,
"loss": 0.8119,
"step": 160
},
{
"epoch": 0.676923076923077,
"grad_norm": 0.910589742578592,
"learning_rate": 5.679720133572206e-07,
"loss": 0.7591,
"step": 165
},
{
"epoch": 0.6974358974358974,
"grad_norm": 1.1777506504426953,
"learning_rate": 5.041658430584852e-07,
"loss": 0.7912,
"step": 170
},
{
"epoch": 0.717948717948718,
"grad_norm": 1.0581504812865925,
"learning_rate": 4.429328831625565e-07,
"loss": 0.8096,
"step": 175
},
{
"epoch": 0.7384615384615385,
"grad_norm": 0.9582774191881785,
"learning_rate": 3.8459091187650726e-07,
"loss": 0.827,
"step": 180
},
{
"epoch": 0.7589743589743589,
"grad_norm": 1.1383169940367925,
"learning_rate": 3.294427041611425e-07,
"loss": 0.7893,
"step": 185
},
{
"epoch": 0.7794871794871795,
"grad_norm": 1.6396416587696268,
"learning_rate": 2.777744604320705e-07,
"loss": 0.7929,
"step": 190
},
{
"epoch": 0.8,
"grad_norm": 1.2294302106292587,
"learning_rate": 2.2985432127701941e-07,
"loss": 0.7893,
"step": 195
},
{
"epoch": 0.8205128205128205,
"grad_norm": 1.8335365896381306,
"learning_rate": 1.8593097589751316e-07,
"loss": 0.7996,
"step": 200
},
{
"epoch": 0.841025641025641,
"grad_norm": 1.2025469183045263,
"learning_rate": 1.4623237149661137e-07,
"loss": 0.7809,
"step": 205
},
{
"epoch": 0.8615384615384616,
"grad_norm": 1.1758082969534152,
"learning_rate": 1.1096453031056264e-07,
"loss": 0.7489,
"step": 210
},
{
"epoch": 0.882051282051282,
"grad_norm": 0.9606044331683303,
"learning_rate": 8.031048042356392e-08,
"loss": 0.7648,
"step": 215
},
{
"epoch": 0.9025641025641026,
"grad_norm": 1.1443378986807335,
"learning_rate": 5.442930591433992e-08,
"loss": 0.7949,
"step": 220
},
{
"epoch": 0.9230769230769231,
"grad_norm": 1.0819245011051657,
"learning_rate": 3.345532126395578e-08,
"loss": 0.7707,
"step": 225
},
{
"epoch": 0.9435897435897436,
"grad_norm": 1.1813637949996902,
"learning_rate": 1.7497374309405344e-08,
"loss": 0.7604,
"step": 230
},
{
"epoch": 0.9641025641025641,
"grad_norm": 1.057572311636513,
"learning_rate": 6.6382813604083375e-09,
"loss": 0.7773,
"step": 235
},
{
"epoch": 0.9846153846153847,
"grad_norm": 1.1501667694643198,
"learning_rate": 9.343974109685682e-10,
"loss": 0.7849,
"step": 240
},
{
"epoch": 0.9969230769230769,
"step": 243,
"total_flos": 33199876669440.0,
"train_loss": 1.074208640267329,
"train_runtime": 3530.8079,
"train_samples_per_second": 8.837,
"train_steps_per_second": 0.069
}
],
"logging_steps": 5,
"max_steps": 243,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 1000,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 33199876669440.0,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}