Files
tofu_Llama-3.2-1B-Instruct_…/trainer_state.json

504 lines
12 KiB
JSON
Raw Permalink Normal View History

{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 10.0,
"eval_steps": 500,
"global_step": 250,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0,
"extraction_strength": 0.05119585323242784,
"forget_Q_A_Prob": 0.10887884757248685,
"forget_Q_A_ROUGE": 0.27890823225307393,
"forget_truth_ratio": 0.6448834990772232,
"model_utility": 0.2185219733034594,
"privleak": -47.96999999040602,
"step": 0
},
{
"epoch": 0.2,
"grad_norm": 660.0,
"learning_rate": 1.6000000000000001e-06,
"loss": 9.7906,
"step": 5
},
{
"epoch": 0.4,
"grad_norm": 580.0,
"learning_rate": 3.6000000000000003e-06,
"loss": 9.7614,
"step": 10
},
{
"epoch": 0.6,
"grad_norm": 592.0,
"learning_rate": 5.600000000000001e-06,
"loss": 9.9195,
"step": 15
},
{
"epoch": 0.8,
"grad_norm": 664.0,
"learning_rate": 7.600000000000001e-06,
"loss": 9.7584,
"step": 20
},
{
"epoch": 1.0,
"grad_norm": 426.0,
"learning_rate": 9.600000000000001e-06,
"loss": 8.8246,
"step": 25
},
{
"epoch": 1.0,
"extraction_strength": 0.062394619827353906,
"forget_Q_A_Prob": 0.20349505057558417,
"forget_Q_A_ROUGE": 0.3363523902144183,
"forget_truth_ratio": 0.6414198592494345,
"model_utility": 0.27790674597563475,
"privleak": -66.64374998667122,
"step": 25
},
{
"epoch": 1.2,
"grad_norm": 225.0,
"learning_rate": 9.822222222222223e-06,
"loss": 6.6271,
"step": 30
},
{
"epoch": 1.4,
"grad_norm": 155.0,
"learning_rate": 9.600000000000001e-06,
"loss": 5.4214,
"step": 35
},
{
"epoch": 1.6,
"grad_norm": 197.0,
"learning_rate": 9.377777777777779e-06,
"loss": 5.5142,
"step": 40
},
{
"epoch": 1.8,
"grad_norm": 134.0,
"learning_rate": 9.155555555555557e-06,
"loss": 4.6586,
"step": 45
},
{
"epoch": 2.0,
"grad_norm": 155.0,
"learning_rate": 8.933333333333333e-06,
"loss": 4.6746,
"step": 50
},
{
"epoch": 2.0,
"extraction_strength": 0.08291476394201429,
"forget_Q_A_Prob": 0.3349009375087917,
"forget_Q_A_ROUGE": 0.38400245016174755,
"forget_truth_ratio": 0.6230931944181684,
"model_utility": 0.3793508809768731,
"privleak": -82.57624998348474,
"step": 50
},
{
"epoch": 2.2,
"grad_norm": 139.0,
"learning_rate": 8.711111111111111e-06,
"loss": 4.7704,
"step": 55
},
{
"epoch": 2.4,
"grad_norm": 130.0,
"learning_rate": 8.48888888888889e-06,
"loss": 4.5076,
"step": 60
},
{
"epoch": 2.6,
"grad_norm": 150.0,
"learning_rate": 8.266666666666667e-06,
"loss": 4.6811,
"step": 65
},
{
"epoch": 2.8,
"grad_norm": 178.0,
"learning_rate": 8.044444444444444e-06,
"loss": 4.8506,
"step": 70
},
{
"epoch": 3.0,
"grad_norm": 143.0,
"learning_rate": 7.822222222222224e-06,
"loss": 4.5449,
"step": 75
},
{
"epoch": 3.0,
"extraction_strength": 0.08463917597529186,
"forget_Q_A_Prob": 0.34760695200413466,
"forget_Q_A_ROUGE": 0.38394973457576287,
"forget_truth_ratio": 0.6189158593014839,
"model_utility": 0.38425409126783,
"privleak": -83.71749998325649,
"step": 75
},
{
"epoch": 3.2,
"grad_norm": 139.0,
"learning_rate": 7.600000000000001e-06,
"loss": 4.6021,
"step": 80
},
{
"epoch": 3.4,
"grad_norm": 145.0,
"learning_rate": 7.377777777777778e-06,
"loss": 4.5443,
"step": 85
},
{
"epoch": 3.6,
"grad_norm": 164.0,
"learning_rate": 7.155555555555556e-06,
"loss": 4.4261,
"step": 90
},
{
"epoch": 3.8,
"grad_norm": 125.5,
"learning_rate": 6.9333333333333344e-06,
"loss": 4.6176,
"step": 95
},
{
"epoch": 4.0,
"grad_norm": 141.0,
"learning_rate": 6.711111111111111e-06,
"loss": 4.3231,
"step": 100
},
{
"epoch": 4.0,
"extraction_strength": 0.08717998187117214,
"forget_Q_A_Prob": 0.35111760402098297,
"forget_Q_A_ROUGE": 0.3848115314453001,
"forget_truth_ratio": 0.6181410546581807,
"model_utility": 0.38945632253152707,
"privleak": -83.93999998321199,
"step": 100
},
{
"epoch": 4.2,
"grad_norm": 134.0,
"learning_rate": 6.488888888888889e-06,
"loss": 4.4801,
"step": 105
},
{
"epoch": 4.4,
"grad_norm": 128.0,
"learning_rate": 6.266666666666668e-06,
"loss": 4.3562,
"step": 110
},
{
"epoch": 4.6,
"grad_norm": 142.0,
"learning_rate": 6.044444444444445e-06,
"loss": 4.4758,
"step": 115
},
{
"epoch": 4.8,
"grad_norm": 127.5,
"learning_rate": 5.822222222222223e-06,
"loss": 4.3376,
"step": 120
},
{
"epoch": 5.0,
"grad_norm": 131.0,
"learning_rate": 5.600000000000001e-06,
"loss": 4.3354,
"step": 125
},
{
"epoch": 5.0,
"extraction_strength": 0.08658014531492801,
"forget_Q_A_Prob": 0.3523250863514841,
"forget_Q_A_ROUGE": 0.3850742811515734,
"forget_truth_ratio": 0.6182047623953388,
"model_utility": 0.3867871595048101,
"privleak": -84.0524999831895,
"step": 125
},
{
"epoch": 5.2,
"grad_norm": 127.0,
"learning_rate": 5.3777777777777784e-06,
"loss": 4.415,
"step": 130
},
{
"epoch": 5.4,
"grad_norm": 119.0,
"learning_rate": 5.155555555555556e-06,
"loss": 4.3584,
"step": 135
},
{
"epoch": 5.6,
"grad_norm": 132.0,
"learning_rate": 4.933333333333334e-06,
"loss": 4.3697,
"step": 140
},
{
"epoch": 5.8,
"grad_norm": 163.0,
"learning_rate": 4.711111111111111e-06,
"loss": 4.3684,
"step": 145
},
{
"epoch": 6.0,
"grad_norm": 122.5,
"learning_rate": 4.488888888888889e-06,
"loss": 4.2881,
"step": 150
},
{
"epoch": 6.0,
"extraction_strength": 0.08737657392397029,
"forget_Q_A_Prob": 0.35286230804398655,
"forget_Q_A_ROUGE": 0.38548027127365925,
"forget_truth_ratio": 0.6177007290305743,
"model_utility": 0.389570021337874,
"privleak": -84.09624998318074,
"step": 150
},
{
"epoch": 6.2,
"grad_norm": 142.0,
"learning_rate": 4.266666666666668e-06,
"loss": 4.2205,
"step": 155
},
{
"epoch": 6.4,
"grad_norm": 142.0,
"learning_rate": 4.044444444444445e-06,
"loss": 4.3014,
"step": 160
},
{
"epoch": 6.6,
"grad_norm": 123.5,
"learning_rate": 3.8222222222222224e-06,
"loss": 4.3313,
"step": 165
},
{
"epoch": 6.8,
"grad_norm": 144.0,
"learning_rate": 3.6000000000000003e-06,
"loss": 4.4139,
"step": 170
},
{
"epoch": 7.0,
"grad_norm": 156.0,
"learning_rate": 3.377777777777778e-06,
"loss": 4.3248,
"step": 175
},
{
"epoch": 7.0,
"extraction_strength": 0.08715686518263624,
"forget_Q_A_Prob": 0.3529359891824424,
"forget_Q_A_ROUGE": 0.3858334433873904,
"forget_truth_ratio": 0.61761660709215,
"model_utility": 0.38596120600351796,
"privleak": -84.109999983178,
"step": 175
},
{
"epoch": 7.2,
"grad_norm": 128.0,
"learning_rate": 3.1555555555555555e-06,
"loss": 4.3152,
"step": 180
},
{
"epoch": 7.4,
"grad_norm": 120.0,
"learning_rate": 2.9333333333333338e-06,
"loss": 4.322,
"step": 185
},
{
"epoch": 7.6,
"grad_norm": 115.0,
"learning_rate": 2.7111111111111116e-06,
"loss": 4.3278,
"step": 190
},
{
"epoch": 7.8,
"grad_norm": 143.0,
"learning_rate": 2.488888888888889e-06,
"loss": 4.2159,
"step": 195
},
{
"epoch": 8.0,
"grad_norm": 132.0,
"learning_rate": 2.266666666666667e-06,
"loss": 4.2292,
"step": 200
},
{
"epoch": 8.0,
"extraction_strength": 0.0845172504995537,
"forget_Q_A_Prob": 0.3530563661269844,
"forget_Q_A_ROUGE": 0.3858238347046325,
"forget_truth_ratio": 0.6176789569373814,
"model_utility": 0.38865403947163873,
"privleak": -84.05874998318824,
"step": 200
},
{
"epoch": 8.2,
"grad_norm": 118.5,
"learning_rate": 2.0444444444444447e-06,
"loss": 4.1904,
"step": 205
},
{
"epoch": 8.4,
"grad_norm": 124.5,
"learning_rate": 1.8222222222222225e-06,
"loss": 4.223,
"step": 210
},
{
"epoch": 8.6,
"grad_norm": 143.0,
"learning_rate": 1.6000000000000001e-06,
"loss": 4.447,
"step": 215
},
{
"epoch": 8.8,
"grad_norm": 143.0,
"learning_rate": 1.377777777777778e-06,
"loss": 4.4513,
"step": 220
},
{
"epoch": 9.0,
"grad_norm": 126.0,
"learning_rate": 1.1555555555555556e-06,
"loss": 4.323,
"step": 225
},
{
"epoch": 9.0,
"extraction_strength": 0.08690805565882671,
"forget_Q_A_Prob": 0.3531488827429712,
"forget_Q_A_ROUGE": 0.38595172349337675,
"forget_truth_ratio": 0.6176135041820996,
"model_utility": 0.38656524448187823,
"privleak": -84.14999998316999,
"step": 225
},
{
"epoch": 9.2,
"grad_norm": 119.5,
"learning_rate": 9.333333333333334e-07,
"loss": 3.9312,
"step": 230
},
{
"epoch": 9.4,
"grad_norm": 134.0,
"learning_rate": 7.111111111111112e-07,
"loss": 4.3145,
"step": 235
},
{
"epoch": 9.6,
"grad_norm": 139.0,
"learning_rate": 4.88888888888889e-07,
"loss": 4.4723,
"step": 240
},
{
"epoch": 9.8,
"grad_norm": 120.5,
"learning_rate": 2.666666666666667e-07,
"loss": 4.4162,
"step": 245
},
{
"epoch": 10.0,
"grad_norm": 144.0,
"learning_rate": 4.444444444444445e-08,
"loss": 4.4783,
"step": 250
},
{
"epoch": 10.0,
"extraction_strength": 0.08670724689551795,
"forget_Q_A_Prob": 0.3530087990872562,
"forget_Q_A_ROUGE": 0.3857650261516718,
"forget_truth_ratio": 0.6175223364978973,
"model_utility": 0.3884310780989943,
"privleak": -84.10499998317901,
"step": 250
},
{
"epoch": 10.0,
"step": 250,
"total_flos": 0.0,
"train_loss": 5.017040328979492,
"train_runtime": 3681.7906,
"train_samples_per_second": 1.086,
"train_steps_per_second": 0.068
}
],
"logging_steps": 5,
"max_steps": 250,
"num_input_tokens_seen": 0,
"num_train_epochs": 10,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": false,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}