{ "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": 0.7035078413166893, "forget_Q_A_Prob": 0.8804745058715343, "forget_Q_A_ROUGE": 0.8224148587081459, "forget_truth_ratio": 0.4751472517287268, "model_utility": 0.5986573345007462, "privleak": -99.33374998013325, "step": 0 }, { "epoch": 0.2, "grad_norm": 1176.0, "learning_rate": 1.6000000000000001e-06, "loss": 55.8963, "step": 5 }, { "epoch": 0.4, "grad_norm": 1064.0, "learning_rate": 3.6000000000000003e-06, "loss": 55.7682, "step": 10 }, { "epoch": 0.6, "grad_norm": 1184.0, "learning_rate": 5.600000000000001e-06, "loss": 55.3329, "step": 15 }, { "epoch": 0.8, "grad_norm": 1168.0, "learning_rate": 7.600000000000001e-06, "loss": 53.5387, "step": 20 }, { "epoch": 1.0, "grad_norm": 1144.0, "learning_rate": 9.600000000000001e-06, "loss": 47.16, "step": 25 }, { "epoch": 1.0, "extraction_strength": 0.4727573242328045, "forget_Q_A_Prob": 0.7682410891354085, "forget_Q_A_ROUGE": 0.6679393669272393, "forget_truth_ratio": 0.5215462041054256, "model_utility": 0.5802270741620733, "privleak": -98.86124998022777, "step": 25 }, { "epoch": 1.2, "grad_norm": 732.0, "learning_rate": 9.822222222222223e-06, "loss": 32.0906, "step": 30 }, { "epoch": 1.4, "grad_norm": 592.0, "learning_rate": 9.600000000000001e-06, "loss": 24.9835, "step": 35 }, { "epoch": 1.6, "grad_norm": 400.0, "learning_rate": 9.377777777777779e-06, "loss": 15.8103, "step": 40 }, { "epoch": 1.8, "grad_norm": 166.0, "learning_rate": 9.155555555555557e-06, "loss": 11.057, "step": 45 }, { "epoch": 2.0, "grad_norm": 108.0, "learning_rate": 8.933333333333333e-06, "loss": 9.1845, "step": 50 }, { "epoch": 2.0, "extraction_strength": 0.08614304082608436, "forget_Q_A_Prob": 0.23828835071995855, "forget_Q_A_ROUGE": 0.266978506986982, "forget_truth_ratio": 0.6758352799741583, "model_utility": 0.365025395428697, "privleak": -75.04874998499025, "step": 50 }, { "epoch": 2.2, "grad_norm": 120.5, "learning_rate": 8.711111111111111e-06, "loss": 9.094, "step": 55 }, { "epoch": 2.4, "grad_norm": 202.0, "learning_rate": 8.48888888888889e-06, "loss": 7.5936, "step": 60 }, { "epoch": 2.6, "grad_norm": 82.0, "learning_rate": 8.266666666666667e-06, "loss": 6.9892, "step": 65 }, { "epoch": 2.8, "grad_norm": 84.0, "learning_rate": 8.044444444444444e-06, "loss": 6.5445, "step": 70 }, { "epoch": 3.0, "grad_norm": 89.5, "learning_rate": 7.822222222222224e-06, "loss": 6.2284, "step": 75 }, { "epoch": 3.0, "extraction_strength": 0.09391103318955707, "forget_Q_A_Prob": 0.22939431543461977, "forget_Q_A_ROUGE": 0.19464342599083026, "forget_truth_ratio": 0.6668836791633079, "model_utility": 0.34932386649395714, "privleak": -65.46124998690773, "step": 75 }, { "epoch": 3.2, "grad_norm": 129.0, "learning_rate": 7.600000000000001e-06, "loss": 5.9146, "step": 80 }, { "epoch": 3.4, "grad_norm": 76.5, "learning_rate": 7.377777777777778e-06, "loss": 5.5015, "step": 85 }, { "epoch": 3.6, "grad_norm": 104.0, "learning_rate": 7.155555555555556e-06, "loss": 5.37, "step": 90 }, { "epoch": 3.8, "grad_norm": 91.5, "learning_rate": 6.9333333333333344e-06, "loss": 4.9139, "step": 95 }, { "epoch": 4.0, "grad_norm": 118.5, "learning_rate": 6.711111111111111e-06, "loss": 4.9409, "step": 100 }, { "epoch": 4.0, "extraction_strength": 0.09873176679531875, "forget_Q_A_Prob": 0.1875428356032353, "forget_Q_A_ROUGE": 0.22033509820856984, "forget_truth_ratio": 0.6531148071926944, "model_utility": 0.43717298699454965, "privleak": -38.19124999236173, "step": 100 }, { "epoch": 4.2, "grad_norm": 175.0, "learning_rate": 6.488888888888889e-06, "loss": 4.2452, "step": 105 }, { "epoch": 4.4, "grad_norm": 144.0, "learning_rate": 6.266666666666668e-06, "loss": 4.0208, "step": 110 }, { "epoch": 4.6, "grad_norm": 173.0, "learning_rate": 6.044444444444445e-06, "loss": 3.709, "step": 115 }, { "epoch": 4.8, "grad_norm": 228.0, "learning_rate": 5.822222222222223e-06, "loss": 3.4964, "step": 120 }, { "epoch": 5.0, "grad_norm": 220.0, "learning_rate": 5.600000000000001e-06, "loss": 3.3359, "step": 125 }, { "epoch": 5.0, "extraction_strength": 0.09858533284264409, "forget_Q_A_Prob": 0.15545873703609686, "forget_Q_A_ROUGE": 0.2513310748335476, "forget_truth_ratio": 0.6237154687561808, "model_utility": 0.49362893415678766, "privleak": -13.094999997381013, "step": 125 }, { "epoch": 5.2, "grad_norm": 140.0, "learning_rate": 5.3777777777777784e-06, "loss": 2.7955, "step": 130 }, { "epoch": 5.4, "grad_norm": 146.0, "learning_rate": 5.155555555555556e-06, "loss": 2.5476, "step": 135 }, { "epoch": 5.6, "grad_norm": 147.0, "learning_rate": 4.933333333333334e-06, "loss": 2.3843, "step": 140 }, { "epoch": 5.8, "grad_norm": 233.0, "learning_rate": 4.711111111111111e-06, "loss": 2.4493, "step": 145 }, { "epoch": 6.0, "grad_norm": 173.0, "learning_rate": 4.488888888888889e-06, "loss": 2.2127, "step": 150 }, { "epoch": 6.0, "extraction_strength": 0.10802013307923694, "forget_Q_A_Prob": 0.15392853119468783, "forget_Q_A_ROUGE": 0.27986895187799854, "forget_truth_ratio": 0.5806421898784745, "model_utility": 0.5353537708875081, "privleak": -21.444999995711004, "step": 150 }, { "epoch": 6.2, "grad_norm": 215.0, "learning_rate": 4.266666666666668e-06, "loss": 2.1857, "step": 155 }, { "epoch": 6.4, "grad_norm": 105.5, "learning_rate": 4.044444444444445e-06, "loss": 1.7586, "step": 160 }, { "epoch": 6.6, "grad_norm": 116.0, "learning_rate": 3.8222222222222224e-06, "loss": 1.9203, "step": 165 }, { "epoch": 6.8, "grad_norm": 116.0, "learning_rate": 3.6000000000000003e-06, "loss": 1.8145, "step": 170 }, { "epoch": 7.0, "grad_norm": 153.0, "learning_rate": 3.377777777777778e-06, "loss": 1.8083, "step": 175 }, { "epoch": 7.0, "extraction_strength": 0.11600868539246466, "forget_Q_A_Prob": 0.14874292317486834, "forget_Q_A_ROUGE": 0.2956713274905, "forget_truth_ratio": 0.5689891586427206, "model_utility": 0.5515531045256538, "privleak": -15.800624996839888, "step": 175 }, { "epoch": 7.2, "grad_norm": 126.5, "learning_rate": 3.1555555555555555e-06, "loss": 1.5654, "step": 180 }, { "epoch": 7.4, "grad_norm": 132.0, "learning_rate": 2.9333333333333338e-06, "loss": 1.6857, "step": 185 }, { "epoch": 7.6, "grad_norm": 185.0, "learning_rate": 2.7111111111111116e-06, "loss": 1.583, "step": 190 }, { "epoch": 7.8, "grad_norm": 214.0, "learning_rate": 2.488888888888889e-06, "loss": 1.6843, "step": 195 }, { "epoch": 8.0, "grad_norm": 192.0, "learning_rate": 2.266666666666667e-06, "loss": 1.5761, "step": 200 }, { "epoch": 8.0, "extraction_strength": 0.11815252416178619, "forget_Q_A_Prob": 0.14210461506576394, "forget_Q_A_ROUGE": 0.2970728548646609, "forget_truth_ratio": 0.5614269492250329, "model_utility": 0.5591748848281174, "privleak": -17.166249996566755, "step": 200 }, { "epoch": 8.2, "grad_norm": 205.0, "learning_rate": 2.0444444444444447e-06, "loss": 1.3298, "step": 205 }, { "epoch": 8.4, "grad_norm": 114.0, "learning_rate": 1.8222222222222225e-06, "loss": 1.4877, "step": 210 }, { "epoch": 8.6, "grad_norm": 169.0, "learning_rate": 1.6000000000000001e-06, "loss": 1.5678, "step": 215 }, { "epoch": 8.8, "grad_norm": 144.0, "learning_rate": 1.377777777777778e-06, "loss": 1.4845, "step": 220 }, { "epoch": 9.0, "grad_norm": 155.0, "learning_rate": 1.1555555555555556e-06, "loss": 1.594, "step": 225 }, { "epoch": 9.0, "extraction_strength": 0.11882007020428687, "forget_Q_A_Prob": 0.13278577822376975, "forget_Q_A_ROUGE": 0.30219775268292687, "forget_truth_ratio": 0.559432289746286, "model_utility": 0.5593146400206448, "privleak": -14.034999997192998, "step": 225 }, { "epoch": 9.2, "grad_norm": 188.0, "learning_rate": 9.333333333333334e-07, "loss": 1.3878, "step": 230 }, { "epoch": 9.4, "grad_norm": 120.5, "learning_rate": 7.111111111111112e-07, "loss": 1.4716, "step": 235 }, { "epoch": 9.6, "grad_norm": 148.0, "learning_rate": 4.88888888888889e-07, "loss": 1.4323, "step": 240 }, { "epoch": 9.8, "grad_norm": 163.0, "learning_rate": 2.666666666666667e-07, "loss": 1.5144, "step": 245 }, { "epoch": 10.0, "grad_norm": 136.0, "learning_rate": 4.444444444444445e-08, "loss": 1.5361, "step": 250 }, { "epoch": 10.0, "extraction_strength": 0.11907237943929123, "forget_Q_A_Prob": 0.13566936177085154, "forget_Q_A_ROUGE": 0.3076188844200944, "forget_truth_ratio": 0.5612359669808998, "model_utility": 0.5599261205924377, "privleak": -13.192499997361507, "step": 250 }, { "epoch": 10.0, "step": 250, "total_flos": 0.0, "train_loss": 9.749949935913087, "train_runtime": 1634.0475, "train_samples_per_second": 2.448, "train_steps_per_second": 0.153 } ], "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 }