| { | |
| "best_metric": 1.077925443649292, | |
| "best_model_checkpoint": "./outputs/instruct-lora-8b-aplly_chat_template-capital/checkpoint-560", | |
| "epoch": 1.0169348010160881, | |
| "eval_steps": 20, | |
| "global_step": 600, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.001693480101608806, | |
| "eval_loss": 1.4109762907028198, | |
| "eval_runtime": 44.8838, | |
| "eval_samples_per_second": 23.394, | |
| "eval_steps_per_second": 5.86, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.03386960203217612, | |
| "grad_norm": 0.7916573882102966, | |
| "learning_rate": 3.3898305084745763e-06, | |
| "loss": 1.4877, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03386960203217612, | |
| "eval_loss": 1.4077372550964355, | |
| "eval_runtime": 46.2342, | |
| "eval_samples_per_second": 22.71, | |
| "eval_steps_per_second": 5.688, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06773920406435224, | |
| "grad_norm": 0.7490960359573364, | |
| "learning_rate": 6.779661016949153e-06, | |
| "loss": 1.4093, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.06773920406435224, | |
| "eval_loss": 1.3683991432189941, | |
| "eval_runtime": 46.9004, | |
| "eval_samples_per_second": 22.388, | |
| "eval_steps_per_second": 5.608, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.10160880609652836, | |
| "grad_norm": 0.7929967641830444, | |
| "learning_rate": 1.016949152542373e-05, | |
| "loss": 1.3575, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.10160880609652836, | |
| "eval_loss": 1.2657679319381714, | |
| "eval_runtime": 48.1658, | |
| "eval_samples_per_second": 21.8, | |
| "eval_steps_per_second": 5.46, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.1354784081287045, | |
| "grad_norm": 0.8403828740119934, | |
| "learning_rate": 1.3559322033898305e-05, | |
| "loss": 1.2616, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1354784081287045, | |
| "eval_loss": 1.2246084213256836, | |
| "eval_runtime": 44.8525, | |
| "eval_samples_per_second": 23.41, | |
| "eval_steps_per_second": 5.864, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1693480101608806, | |
| "grad_norm": 1.071118950843811, | |
| "learning_rate": 1.694915254237288e-05, | |
| "loss": 1.2205, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.1693480101608806, | |
| "eval_loss": 1.2001967430114746, | |
| "eval_runtime": 44.851, | |
| "eval_samples_per_second": 23.411, | |
| "eval_steps_per_second": 5.864, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.20321761219305673, | |
| "grad_norm": 1.0767000913619995, | |
| "learning_rate": 2.033898305084746e-05, | |
| "loss": 1.2187, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.20321761219305673, | |
| "eval_loss": 1.183129906654358, | |
| "eval_runtime": 45.0629, | |
| "eval_samples_per_second": 23.301, | |
| "eval_steps_per_second": 5.836, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.23708721422523285, | |
| "grad_norm": 1.061275839805603, | |
| "learning_rate": 2.3728813559322036e-05, | |
| "loss": 1.1836, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.23708721422523285, | |
| "eval_loss": 1.1728578805923462, | |
| "eval_runtime": 44.9354, | |
| "eval_samples_per_second": 23.367, | |
| "eval_steps_per_second": 5.853, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.270956816257409, | |
| "grad_norm": 1.1500962972640991, | |
| "learning_rate": 2.711864406779661e-05, | |
| "loss": 1.1149, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.270956816257409, | |
| "eval_loss": 1.1615699529647827, | |
| "eval_runtime": 44.8412, | |
| "eval_samples_per_second": 23.416, | |
| "eval_steps_per_second": 5.865, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.3048264182895851, | |
| "grad_norm": 1.274034023284912, | |
| "learning_rate": 2.9999737474980266e-05, | |
| "loss": 1.1049, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.3048264182895851, | |
| "eval_loss": 1.151460886001587, | |
| "eval_runtime": 44.8418, | |
| "eval_samples_per_second": 23.416, | |
| "eval_steps_per_second": 5.865, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.3386960203217612, | |
| "grad_norm": 1.2562557458877563, | |
| "learning_rate": 2.998457196315866e-05, | |
| "loss": 1.1254, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.3386960203217612, | |
| "eval_loss": 1.1418647766113281, | |
| "eval_runtime": 44.8753, | |
| "eval_samples_per_second": 23.398, | |
| "eval_steps_per_second": 5.861, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.37256562235393736, | |
| "grad_norm": 1.3587268590927124, | |
| "learning_rate": 2.9946097849501546e-05, | |
| "loss": 1.1043, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.37256562235393736, | |
| "eval_loss": 1.1385948657989502, | |
| "eval_runtime": 49.1418, | |
| "eval_samples_per_second": 21.367, | |
| "eval_steps_per_second": 5.352, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.40643522438611346, | |
| "grad_norm": 1.3932058811187744, | |
| "learning_rate": 2.988437498074987e-05, | |
| "loss": 1.1076, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.40643522438611346, | |
| "eval_loss": 1.1299232244491577, | |
| "eval_runtime": 47.3037, | |
| "eval_samples_per_second": 22.197, | |
| "eval_steps_per_second": 5.56, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.4403048264182896, | |
| "grad_norm": 1.4880738258361816, | |
| "learning_rate": 2.9799499367238472e-05, | |
| "loss": 1.1307, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.4403048264182896, | |
| "eval_loss": 1.1247954368591309, | |
| "eval_runtime": 44.8376, | |
| "eval_samples_per_second": 23.418, | |
| "eval_steps_per_second": 5.866, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.4741744284504657, | |
| "grad_norm": 1.3387629985809326, | |
| "learning_rate": 2.969160303355143e-05, | |
| "loss": 1.0895, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.4741744284504657, | |
| "eval_loss": 1.1187076568603516, | |
| "eval_runtime": 44.9374, | |
| "eval_samples_per_second": 23.366, | |
| "eval_steps_per_second": 5.853, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.5080440304826418, | |
| "grad_norm": 1.467492699623108, | |
| "learning_rate": 2.95608538131569e-05, | |
| "loss": 1.0912, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.5080440304826418, | |
| "eval_loss": 1.1132631301879883, | |
| "eval_runtime": 47.2897, | |
| "eval_samples_per_second": 22.204, | |
| "eval_steps_per_second": 5.561, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.541913632514818, | |
| "grad_norm": 1.5851348638534546, | |
| "learning_rate": 2.940745508734104e-05, | |
| "loss": 1.0932, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.541913632514818, | |
| "eval_loss": 1.1109942197799683, | |
| "eval_runtime": 45.5061, | |
| "eval_samples_per_second": 23.074, | |
| "eval_steps_per_second": 5.779, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.5757832345469941, | |
| "grad_norm": 1.569037914276123, | |
| "learning_rate": 2.9231645468847078e-05, | |
| "loss": 1.113, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.5757832345469941, | |
| "eval_loss": 1.1062214374542236, | |
| "eval_runtime": 44.8719, | |
| "eval_samples_per_second": 23.4, | |
| "eval_steps_per_second": 5.861, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.6096528365791702, | |
| "grad_norm": 1.525785207748413, | |
| "learning_rate": 2.903369843071157e-05, | |
| "loss": 1.0623, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.6096528365791702, | |
| "eval_loss": 1.1012595891952515, | |
| "eval_runtime": 48.3892, | |
| "eval_samples_per_second": 21.699, | |
| "eval_steps_per_second": 5.435, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.6435224386113463, | |
| "grad_norm": 1.4866666793823242, | |
| "learning_rate": 2.881392188087528e-05, | |
| "loss": 1.0934, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.6435224386113463, | |
| "eval_loss": 1.100132942199707, | |
| "eval_runtime": 44.8502, | |
| "eval_samples_per_second": 23.411, | |
| "eval_steps_per_second": 5.864, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.6773920406435224, | |
| "grad_norm": 1.5744224786758423, | |
| "learning_rate": 2.8572657683230322e-05, | |
| "loss": 1.0663, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.6773920406435224, | |
| "eval_loss": 1.0957869291305542, | |
| "eval_runtime": 44.8474, | |
| "eval_samples_per_second": 23.413, | |
| "eval_steps_per_second": 5.864, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.7112616426756986, | |
| "grad_norm": 1.4972140789031982, | |
| "learning_rate": 2.8310281125848574e-05, | |
| "loss": 1.081, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.7112616426756986, | |
| "eval_loss": 1.0911184549331665, | |
| "eval_runtime": 44.9252, | |
| "eval_samples_per_second": 23.372, | |
| "eval_steps_per_second": 5.854, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.7451312447078747, | |
| "grad_norm": 1.7231398820877075, | |
| "learning_rate": 2.80272003372186e-05, | |
| "loss": 1.0572, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.7451312447078747, | |
| "eval_loss": 1.089966058731079, | |
| "eval_runtime": 44.8357, | |
| "eval_samples_per_second": 23.419, | |
| "eval_steps_per_second": 5.866, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.7790008467400508, | |
| "grad_norm": 1.5352925062179565, | |
| "learning_rate": 2.7723855651399027e-05, | |
| "loss": 1.0459, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.7790008467400508, | |
| "eval_loss": 1.084614872932434, | |
| "eval_runtime": 44.8258, | |
| "eval_samples_per_second": 23.424, | |
| "eval_steps_per_second": 5.867, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.8128704487722269, | |
| "grad_norm": 1.5007786750793457, | |
| "learning_rate": 2.7400718923076004e-05, | |
| "loss": 1.0621, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.8128704487722269, | |
| "eval_loss": 1.0817887783050537, | |
| "eval_runtime": 44.8244, | |
| "eval_samples_per_second": 23.425, | |
| "eval_steps_per_second": 5.867, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.8467400508044031, | |
| "grad_norm": 1.6158475875854492, | |
| "learning_rate": 2.7058292793590064e-05, | |
| "loss": 1.0584, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.8467400508044031, | |
| "eval_loss": 1.0819875001907349, | |
| "eval_runtime": 47.2009, | |
| "eval_samples_per_second": 22.245, | |
| "eval_steps_per_second": 5.572, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.8806096528365792, | |
| "grad_norm": 1.740872859954834, | |
| "learning_rate": 2.6697109909074174e-05, | |
| "loss": 1.0393, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.8806096528365792, | |
| "eval_loss": 1.0836067199707031, | |
| "eval_runtime": 44.9083, | |
| "eval_samples_per_second": 23.381, | |
| "eval_steps_per_second": 5.856, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.9144792548687553, | |
| "grad_norm": 1.5816829204559326, | |
| "learning_rate": 2.6317732091919095e-05, | |
| "loss": 1.0342, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.9144792548687553, | |
| "eval_loss": 1.0809314250946045, | |
| "eval_runtime": 45.5743, | |
| "eval_samples_per_second": 23.039, | |
| "eval_steps_per_second": 5.771, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.9483488569009314, | |
| "grad_norm": 1.5456470251083374, | |
| "learning_rate": 2.5920749466854923e-05, | |
| "loss": 1.0132, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.9483488569009314, | |
| "eval_loss": 1.077925443649292, | |
| "eval_runtime": 47.3001, | |
| "eval_samples_per_second": 22.199, | |
| "eval_steps_per_second": 5.56, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.9822184589331076, | |
| "grad_norm": 1.646410584449768, | |
| "learning_rate": 2.550677954300811e-05, | |
| "loss": 1.0256, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.9822184589331076, | |
| "eval_loss": 1.077932596206665, | |
| "eval_runtime": 48.3066, | |
| "eval_samples_per_second": 21.736, | |
| "eval_steps_per_second": 5.444, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.0169348010160881, | |
| "grad_norm": 1.9027409553527832, | |
| "learning_rate": 2.5076466253361893e-05, | |
| "loss": 1.0651, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.0169348010160881, | |
| "eval_loss": 1.0789446830749512, | |
| "eval_runtime": 44.8937, | |
| "eval_samples_per_second": 23.389, | |
| "eval_steps_per_second": 5.858, | |
| "step": 600 | |
| } | |
| ], | |
| "logging_steps": 20, | |
| "max_steps": 1770, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 200, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 3, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 2 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.462097548054364e+17, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |