| { | |
| "best_metric": 0.7849208402678758, | |
| "best_model_checkpoint": "tiny-llama/checkpoint-1284", | |
| "epoch": 8.999610743479954, | |
| "eval_steps": 500, | |
| "global_step": 5780, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.614485981308411e-05, | |
| "loss": 1.0444, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.8055770720371804, | |
| "eval_f1_macro": 0.6986433125217109, | |
| "eval_f1_micro": 0.8055770720371804, | |
| "eval_f1_weighted": 0.8014415131779974, | |
| "eval_loss": 0.5968185067176819, | |
| "eval_macro_fpr": 0.017477297659887057, | |
| "eval_macro_sensitivity": 0.6995136011888944, | |
| "eval_macro_specificity": 0.9852356804676051, | |
| "eval_precision": 0.8049997976909469, | |
| "eval_precision_macro": 0.7121994461809698, | |
| "eval_recall": 0.8055770720371804, | |
| "eval_recall_macro": 0.6995136011888944, | |
| "eval_runtime": 1398.4578, | |
| "eval_samples_per_second": 0.923, | |
| "eval_steps_per_second": 0.462, | |
| "eval_weighted_fpr": 0.01694686381743299, | |
| "eval_weighted_sensitivity": 0.8055770720371804, | |
| "eval_weighted_specificity": 0.9729581349768969, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 4.2250778816199374e-05, | |
| "loss": 0.4788, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.8195197521301317, | |
| "eval_f1_macro": 0.7849208402678758, | |
| "eval_f1_micro": 0.8195197521301317, | |
| "eval_f1_weighted": 0.8172221848923289, | |
| "eval_loss": 0.696565568447113, | |
| "eval_macro_fpr": 0.016141504898788652, | |
| "eval_macro_sensitivity": 0.782528240704334, | |
| "eval_macro_specificity": 0.9863319896705219, | |
| "eval_precision": 0.8221579556499354, | |
| "eval_precision_macro": 0.8091762597974816, | |
| "eval_recall": 0.8195197521301317, | |
| "eval_recall_macro": 0.782528240704334, | |
| "eval_runtime": 1392.2226, | |
| "eval_samples_per_second": 0.927, | |
| "eval_steps_per_second": 0.464, | |
| "eval_weighted_fpr": 0.01548687271518777, | |
| "eval_weighted_sensitivity": 0.8195197521301317, | |
| "eval_weighted_specificity": 0.9754600929277001, | |
| "step": 1284 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 3.836448598130841e-05, | |
| "loss": 0.3354, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.8326878388845856, | |
| "eval_f1_macro": 0.7742114597713703, | |
| "eval_f1_micro": 0.8326878388845856, | |
| "eval_f1_weighted": 0.8281979563082781, | |
| "eval_loss": 0.8046467304229736, | |
| "eval_macro_fpr": 0.014819935806968648, | |
| "eval_macro_sensitivity": 0.7581833318977771, | |
| "eval_macro_specificity": 0.9872339538797621, | |
| "eval_precision": 0.8276269450338689, | |
| "eval_precision_macro": 0.8057581057880366, | |
| "eval_recall": 0.8326878388845856, | |
| "eval_recall_macro": 0.7581833318977771, | |
| "eval_runtime": 1396.4981, | |
| "eval_samples_per_second": 0.924, | |
| "eval_steps_per_second": 0.463, | |
| "eval_weighted_fpr": 0.01414908947988995, | |
| "eval_weighted_sensitivity": 0.8326878388845856, | |
| "eval_weighted_specificity": 0.9758214693118433, | |
| "step": 1926 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 3.4470404984423676e-05, | |
| "loss": 0.2132, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 3.057632398753894e-05, | |
| "loss": 0.0571, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.8264910921766073, | |
| "eval_f1_macro": 0.7690440255949657, | |
| "eval_f1_micro": 0.8264910921766073, | |
| "eval_f1_weighted": 0.8261718999247398, | |
| "eval_loss": 1.114306926727295, | |
| "eval_macro_fpr": 0.015230795356958942, | |
| "eval_macro_sensitivity": 0.7762905253241136, | |
| "eval_macro_specificity": 0.9869101098294825, | |
| "eval_precision": 0.8312136257886843, | |
| "eval_precision_macro": 0.7904073004865061, | |
| "eval_recall": 0.8264910921766073, | |
| "eval_recall_macro": 0.7762905253241136, | |
| "eval_runtime": 1399.2136, | |
| "eval_samples_per_second": 0.923, | |
| "eval_steps_per_second": 0.462, | |
| "eval_weighted_fpr": 0.014773776546629732, | |
| "eval_weighted_sensitivity": 0.8264910921766073, | |
| "eval_weighted_specificity": 0.9771605552656325, | |
| "step": 2569 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 2.6682242990654205e-05, | |
| "loss": 0.0187, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.8319132455460883, | |
| "eval_f1_macro": 0.7637547171369754, | |
| "eval_f1_micro": 0.8319132455460883, | |
| "eval_f1_weighted": 0.8302763212360192, | |
| "eval_loss": 1.110449194908142, | |
| "eval_macro_fpr": 0.014862942310551715, | |
| "eval_macro_sensitivity": 0.7723638117326753, | |
| "eval_macro_specificity": 0.9872593002552416, | |
| "eval_precision": 0.8316344032874086, | |
| "eval_precision_macro": 0.7744559231287563, | |
| "eval_recall": 0.8319132455460883, | |
| "eval_recall_macro": 0.7723638117326753, | |
| "eval_runtime": 1397.1179, | |
| "eval_samples_per_second": 0.924, | |
| "eval_steps_per_second": 0.462, | |
| "eval_weighted_fpr": 0.014226709499770536, | |
| "eval_weighted_sensitivity": 0.8319132455460883, | |
| "eval_weighted_specificity": 0.9769762582825359, | |
| "step": 3211 | |
| }, | |
| { | |
| "epoch": 5.45, | |
| "learning_rate": 2.2788161993769472e-05, | |
| "loss": 0.0071, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.8241673121611154, | |
| "eval_f1_macro": 0.745099817904837, | |
| "eval_f1_micro": 0.8241673121611154, | |
| "eval_f1_weighted": 0.8209070583364038, | |
| "eval_loss": 1.144503116607666, | |
| "eval_macro_fpr": 0.01570185641881881, | |
| "eval_macro_sensitivity": 0.738352723089731, | |
| "eval_macro_specificity": 0.9866424210438552, | |
| "eval_precision": 0.8210083679630991, | |
| "eval_precision_macro": 0.7683618012500858, | |
| "eval_recall": 0.8241673121611154, | |
| "eval_recall_macro": 0.738352723089731, | |
| "eval_runtime": 1397.6847, | |
| "eval_samples_per_second": 0.924, | |
| "eval_steps_per_second": 0.462, | |
| "eval_weighted_fpr": 0.015010249289162203, | |
| "eval_weighted_sensitivity": 0.8241673121611154, | |
| "eval_weighted_specificity": 0.9754690034967137, | |
| "step": 3853 | |
| }, | |
| { | |
| "epoch": 6.23, | |
| "learning_rate": 1.8894080996884735e-05, | |
| "loss": 0.0002, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.8326878388845856, | |
| "eval_f1_macro": 0.7617153787626418, | |
| "eval_f1_micro": 0.8326878388845856, | |
| "eval_f1_weighted": 0.8292501117913054, | |
| "eval_loss": 1.2032252550125122, | |
| "eval_macro_fpr": 0.014823835936918472, | |
| "eval_macro_sensitivity": 0.752929076567028, | |
| "eval_macro_specificity": 0.9872777146881934, | |
| "eval_precision": 0.8301989992809635, | |
| "eval_precision_macro": 0.7985351753450873, | |
| "eval_recall": 0.8326878388845856, | |
| "eval_recall_macro": 0.752929076567028, | |
| "eval_runtime": 1403.2825, | |
| "eval_samples_per_second": 0.92, | |
| "eval_steps_per_second": 0.46, | |
| "eval_weighted_fpr": 0.01414908947988995, | |
| "eval_weighted_sensitivity": 0.8326878388845856, | |
| "eval_weighted_specificity": 0.9764778814383137, | |
| "step": 4495 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 1.5e-05, | |
| "loss": 0.0028, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 7.79, | |
| "learning_rate": 1.1105919003115265e-05, | |
| "loss": 0.0028, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.82571649883811, | |
| "eval_f1_macro": 0.755227710037701, | |
| "eval_f1_micro": 0.82571649883811, | |
| "eval_f1_weighted": 0.8229426380981488, | |
| "eval_loss": 1.1918259859085083, | |
| "eval_macro_fpr": 0.015546599693760165, | |
| "eval_macro_sensitivity": 0.7493334411200354, | |
| "eval_macro_specificity": 0.986754396380849, | |
| "eval_precision": 0.8225901566549486, | |
| "eval_precision_macro": 0.7737967354767298, | |
| "eval_recall": 0.82571649883811, | |
| "eval_recall_macro": 0.7493334411200354, | |
| "eval_runtime": 1398.648, | |
| "eval_samples_per_second": 0.923, | |
| "eval_steps_per_second": 0.462, | |
| "eval_weighted_fpr": 0.01485246550927454, | |
| "eval_weighted_sensitivity": 0.82571649883811, | |
| "eval_weighted_specificity": 0.975599446874626, | |
| "step": 5138 | |
| }, | |
| { | |
| "epoch": 8.56, | |
| "learning_rate": 7.21183800623053e-06, | |
| "loss": 0.0, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.831138652207591, | |
| "eval_f1_macro": 0.7592477157964134, | |
| "eval_f1_micro": 0.831138652207591, | |
| "eval_f1_weighted": 0.8276161418464294, | |
| "eval_loss": 1.21809720993042, | |
| "eval_macro_fpr": 0.014976927943039621, | |
| "eval_macro_sensitivity": 0.7521526164801944, | |
| "eval_macro_specificity": 0.987168901409934, | |
| "eval_precision": 0.8285866244573639, | |
| "eval_precision_macro": 0.7935003071971033, | |
| "eval_recall": 0.831138652207591, | |
| "eval_recall_macro": 0.7521526164801944, | |
| "eval_runtime": 1400.0048, | |
| "eval_samples_per_second": 0.922, | |
| "eval_steps_per_second": 0.461, | |
| "eval_weighted_fpr": 0.014304461942257218, | |
| "eval_weighted_sensitivity": 0.831138652207591, | |
| "eval_weighted_specificity": 0.9763948689414159, | |
| "step": 5780 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 6420, | |
| "num_train_epochs": 10, | |
| "save_steps": 500, | |
| "total_flos": 5.577968189272228e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |