sdar_4b_random_mask-final / trainer_state.json
autoprogrammer's picture
Upload sdar_4b_random_mask-final
d95f204 verified
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"eval_steps": 500,
"global_step": 351,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.042735042735042736,
"grad_norm": 4.884984970092773,
"learning_rate": 3.6363636363636366e-06,
"loss": 0.2805,
"step": 5
},
{
"epoch": 0.08547008547008547,
"grad_norm": 3.8989036083221436,
"learning_rate": 8.181818181818183e-06,
"loss": 0.2371,
"step": 10
},
{
"epoch": 0.1282051282051282,
"grad_norm": 3.5579559803009033,
"learning_rate": 9.998079135987437e-06,
"loss": 0.243,
"step": 15
},
{
"epoch": 0.17094017094017094,
"grad_norm": 4.4413347244262695,
"learning_rate": 9.98634586692894e-06,
"loss": 0.2425,
"step": 20
},
{
"epoch": 0.21367521367521367,
"grad_norm": 4.2704386711120605,
"learning_rate": 9.963971484502247e-06,
"loss": 0.2579,
"step": 25
},
{
"epoch": 0.2564102564102564,
"grad_norm": 3.503924608230591,
"learning_rate": 9.931003736767013e-06,
"loss": 0.241,
"step": 30
},
{
"epoch": 0.29914529914529914,
"grad_norm": 3.716705083847046,
"learning_rate": 9.887512978558329e-06,
"loss": 0.2511,
"step": 35
},
{
"epoch": 0.3418803418803419,
"grad_norm": 3.202826738357544,
"learning_rate": 9.833592021345938e-06,
"loss": 0.2389,
"step": 40
},
{
"epoch": 0.38461538461538464,
"grad_norm": 4.339502811431885,
"learning_rate": 9.76935593516989e-06,
"loss": 0.258,
"step": 45
},
{
"epoch": 0.42735042735042733,
"grad_norm": 3.6700973510742188,
"learning_rate": 9.694941803075285e-06,
"loss": 0.2323,
"step": 50
},
{
"epoch": 0.4700854700854701,
"grad_norm": 3.2889957427978516,
"learning_rate": 9.610508428570122e-06,
"loss": 0.2685,
"step": 55
},
{
"epoch": 0.5128205128205128,
"grad_norm": 3.249929666519165,
"learning_rate": 9.516235996730645e-06,
"loss": 0.2455,
"step": 60
},
{
"epoch": 0.5555555555555556,
"grad_norm": 3.139861822128296,
"learning_rate": 9.41232568967728e-06,
"loss": 0.2509,
"step": 65
},
{
"epoch": 0.5982905982905983,
"grad_norm": 3.1881346702575684,
"learning_rate": 9.298999257241862e-06,
"loss": 0.2322,
"step": 70
},
{
"epoch": 0.6410256410256411,
"grad_norm": 3.5317623615264893,
"learning_rate": 9.176498543742328e-06,
"loss": 0.2334,
"step": 75
},
{
"epoch": 0.6837606837606838,
"grad_norm": 3.0199837684631348,
"learning_rate": 9.045084971874738e-06,
"loss": 0.2331,
"step": 80
},
{
"epoch": 0.7264957264957265,
"grad_norm": 3.8516860008239746,
"learning_rate": 8.905038984824079e-06,
"loss": 0.252,
"step": 85
},
{
"epoch": 0.7692307692307693,
"grad_norm": 3.244687795639038,
"learning_rate": 8.756659447784367e-06,
"loss": 0.2558,
"step": 90
},
{
"epoch": 0.811965811965812,
"grad_norm": 4.6248884201049805,
"learning_rate": 8.600263010165275e-06,
"loss": 0.2455,
"step": 95
},
{
"epoch": 0.8547008547008547,
"grad_norm": 2.914029121398926,
"learning_rate": 8.436183429846314e-06,
"loss": 0.2539,
"step": 100
},
{
"epoch": 0.8974358974358975,
"grad_norm": 3.7247676849365234,
"learning_rate": 8.264770860920722e-06,
"loss": 0.2604,
"step": 105
},
{
"epoch": 0.9401709401709402,
"grad_norm": 3.5662894248962402,
"learning_rate": 8.086391106448965e-06,
"loss": 0.2432,
"step": 110
},
{
"epoch": 0.9829059829059829,
"grad_norm": 3.140153408050537,
"learning_rate": 7.90142483781658e-06,
"loss": 0.2617,
"step": 115
},
{
"epoch": 1.0256410256410255,
"grad_norm": 1.6056519746780396,
"learning_rate": 7.710266782362248e-06,
"loss": 0.1536,
"step": 120
},
{
"epoch": 1.0683760683760684,
"grad_norm": 1.6721214056015015,
"learning_rate": 7.513324881009769e-06,
"loss": 0.0775,
"step": 125
},
{
"epoch": 1.1111111111111112,
"grad_norm": 2.891897678375244,
"learning_rate": 7.311019417701567e-06,
"loss": 0.0733,
"step": 130
},
{
"epoch": 1.1538461538461537,
"grad_norm": 3.5481584072113037,
"learning_rate": 7.103782122491577e-06,
"loss": 0.0884,
"step": 135
},
{
"epoch": 1.1965811965811965,
"grad_norm": 2.4732353687286377,
"learning_rate": 6.892055250211552e-06,
"loss": 0.0797,
"step": 140
},
{
"epoch": 1.2393162393162394,
"grad_norm": 3.0785977840423584,
"learning_rate": 6.67629063667697e-06,
"loss": 0.0911,
"step": 145
},
{
"epoch": 1.282051282051282,
"grad_norm": 2.6157419681549072,
"learning_rate": 6.456948734446624e-06,
"loss": 0.0856,
"step": 150
},
{
"epoch": 1.3247863247863247,
"grad_norm": 1.851462721824646,
"learning_rate": 6.234497630193666e-06,
"loss": 0.0841,
"step": 155
},
{
"epoch": 1.3675213675213675,
"grad_norm": 2.2886569499969482,
"learning_rate": 6.009412045785051e-06,
"loss": 0.0795,
"step": 160
},
{
"epoch": 1.4102564102564101,
"grad_norm": 2.482672691345215,
"learning_rate": 5.782172325201155e-06,
"loss": 0.0769,
"step": 165
},
{
"epoch": 1.452991452991453,
"grad_norm": 2.409162759780884,
"learning_rate": 5.553263409457504e-06,
"loss": 0.0756,
"step": 170
},
{
"epoch": 1.4957264957264957,
"grad_norm": 2.8806612491607666,
"learning_rate": 5.323173801716222e-06,
"loss": 0.0852,
"step": 175
},
{
"epoch": 1.5384615384615383,
"grad_norm": 2.2364001274108887,
"learning_rate": 5.09239452479565e-06,
"loss": 0.0762,
"step": 180
},
{
"epoch": 1.5811965811965814,
"grad_norm": 1.914794921875,
"learning_rate": 4.861418073302919e-06,
"loss": 0.0762,
"step": 185
},
{
"epoch": 1.623931623931624,
"grad_norm": 2.0296647548675537,
"learning_rate": 4.630737362625631e-06,
"loss": 0.0763,
"step": 190
},
{
"epoch": 1.6666666666666665,
"grad_norm": 2.2641067504882812,
"learning_rate": 4.400844677025585e-06,
"loss": 0.0718,
"step": 195
},
{
"epoch": 1.7094017094017095,
"grad_norm": 2.0905230045318604,
"learning_rate": 4.17223061907935e-06,
"loss": 0.0733,
"step": 200
},
{
"epoch": 1.7521367521367521,
"grad_norm": 2.3458974361419678,
"learning_rate": 3.945383062707652e-06,
"loss": 0.0732,
"step": 205
},
{
"epoch": 1.7948717948717947,
"grad_norm": 2.2764647006988525,
"learning_rate": 3.720786112027822e-06,
"loss": 0.0739,
"step": 210
},
{
"epoch": 1.8376068376068377,
"grad_norm": 2.1260969638824463,
"learning_rate": 3.498919068251237e-06,
"loss": 0.0726,
"step": 215
},
{
"epoch": 1.8803418803418803,
"grad_norm": 3.6211023330688477,
"learning_rate": 3.2802554068303595e-06,
"loss": 0.0678,
"step": 220
},
{
"epoch": 1.9230769230769231,
"grad_norm": 1.942047357559204,
"learning_rate": 3.0652617670382745e-06,
"loss": 0.0652,
"step": 225
},
{
"epoch": 1.965811965811966,
"grad_norm": 2.0708093643188477,
"learning_rate": 2.8543969561369556e-06,
"loss": 0.0704,
"step": 230
},
{
"epoch": 2.0085470085470085,
"grad_norm": 1.3162912130355835,
"learning_rate": 2.648110970259454e-06,
"loss": 0.0586,
"step": 235
},
{
"epoch": 2.051282051282051,
"grad_norm": 1.4112859964370728,
"learning_rate": 2.4468440340954664e-06,
"loss": 0.0223,
"step": 240
},
{
"epoch": 2.094017094017094,
"grad_norm": 0.8296356201171875,
"learning_rate": 2.2510256614296638e-06,
"loss": 0.0203,
"step": 245
},
{
"epoch": 2.1367521367521367,
"grad_norm": 1.1841522455215454,
"learning_rate": 2.061073738537635e-06,
"loss": 0.0193,
"step": 250
},
{
"epoch": 2.1794871794871793,
"grad_norm": 1.3090951442718506,
"learning_rate": 1.8773936323955055e-06,
"loss": 0.0198,
"step": 255
},
{
"epoch": 2.2222222222222223,
"grad_norm": 1.2884639501571655,
"learning_rate": 1.7003773256063882e-06,
"loss": 0.0193,
"step": 260
},
{
"epoch": 2.264957264957265,
"grad_norm": 1.4424365758895874,
"learning_rate": 1.5304025798897521e-06,
"loss": 0.0174,
"step": 265
},
{
"epoch": 2.3076923076923075,
"grad_norm": 1.5278124809265137,
"learning_rate": 1.3678321299188802e-06,
"loss": 0.0159,
"step": 270
},
{
"epoch": 2.3504273504273505,
"grad_norm": 1.0042649507522583,
"learning_rate": 1.213012909226786e-06,
"loss": 0.0198,
"step": 275
},
{
"epoch": 2.393162393162393,
"grad_norm": 1.1120522022247314,
"learning_rate": 1.066275309832584e-06,
"loss": 0.0165,
"step": 280
},
{
"epoch": 2.435897435897436,
"grad_norm": 2.824000358581543,
"learning_rate": 9.279324771682586e-07,
"loss": 0.0173,
"step": 285
},
{
"epoch": 2.4786324786324787,
"grad_norm": 1.0263090133666992,
"learning_rate": 7.98279641810537e-07,
"loss": 0.0191,
"step": 290
},
{
"epoch": 2.5213675213675213,
"grad_norm": 0.8189556002616882,
"learning_rate": 6.775934894439606e-07,
"loss": 0.0184,
"step": 295
},
{
"epoch": 2.564102564102564,
"grad_norm": 1.4359732866287231,
"learning_rate": 5.661315703996905e-07,
"loss": 0.0161,
"step": 300
},
{
"epoch": 2.606837606837607,
"grad_norm": 0.8973503112792969,
"learning_rate": 4.641317500301173e-07,
"loss": 0.0132,
"step": 305
},
{
"epoch": 2.6495726495726495,
"grad_norm": 1.3179709911346436,
"learning_rate": 3.71811701092219e-07,
"loss": 0.018,
"step": 310
},
{
"epoch": 2.6923076923076925,
"grad_norm": 0.7992963194847107,
"learning_rate": 2.893684392229185e-07,
"loss": 0.0179,
"step": 315
},
{
"epoch": 2.735042735042735,
"grad_norm": 1.6558369398117065,
"learning_rate": 2.1697790249779638e-07,
"loss": 0.0186,
"step": 320
},
{
"epoch": 2.7777777777777777,
"grad_norm": 1.3541942834854126,
"learning_rate": 1.547945759703623e-07,
"loss": 0.0204,
"step": 325
},
{
"epoch": 2.8205128205128203,
"grad_norm": 1.408431053161621,
"learning_rate": 1.0295116199317057e-07,
"loss": 0.0213,
"step": 330
},
{
"epoch": 2.8632478632478633,
"grad_norm": 1.0282738208770752,
"learning_rate": 6.15582970243117e-08,
"loss": 0.0196,
"step": 335
},
{
"epoch": 2.905982905982906,
"grad_norm": 1.5794782638549805,
"learning_rate": 3.0704315523631956e-08,
"loss": 0.0178,
"step": 340
},
{
"epoch": 2.948717948717949,
"grad_norm": 1.0120314359664917,
"learning_rate": 1.0455061442548597e-08,
"loss": 0.0226,
"step": 345
},
{
"epoch": 2.9914529914529915,
"grad_norm": 1.4541232585906982,
"learning_rate": 8.537477097364522e-10,
"loss": 0.0168,
"step": 350
},
{
"epoch": 3.0,
"step": 351,
"total_flos": 2.7205837452948275e+17,
"train_loss": 0.11475353346251355,
"train_runtime": 474.0913,
"train_samples_per_second": 47.288,
"train_steps_per_second": 0.74
}
],
"logging_steps": 5,
"max_steps": 351,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 2.7205837452948275e+17,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}