| { | |
| "best_global_step": 328, | |
| "best_metric": 0.6435705423355103, | |
| "best_model_checkpoint": "./britllm_cefr_welsh_allData_DA/fold_3/checkpoint-328", | |
| "epoch": 4.0, | |
| "eval_steps": 500, | |
| "global_step": 328, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 288.0, | |
| "learning_rate": 3.0142372881355933e-05, | |
| "loss": 23.3143, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_A1_f1": 0.7387755102040816, | |
| "eval_A1_precision": 0.6094276094276094, | |
| "eval_A1_recall": 0.9378238341968912, | |
| "eval_A2_f1": 0.6645569620253164, | |
| "eval_A2_precision": 0.6104651162790697, | |
| "eval_A2_recall": 0.7291666666666666, | |
| "eval_B1_f1": 0.5294117647058824, | |
| "eval_B1_precision": 0.6101694915254238, | |
| "eval_B1_recall": 0.4675324675324675, | |
| "eval_B2_f1": 0.6, | |
| "eval_B2_precision": 0.8780487804878049, | |
| "eval_B2_recall": 0.45569620253164556, | |
| "eval_accuracy": 0.6605222734254992, | |
| "eval_loss": 1.0833252668380737, | |
| "eval_runtime": 10.1627, | |
| "eval_samples_per_second": 64.058, | |
| "eval_steps_per_second": 21.353, | |
| "eval_weighted_f1": 0.6470730904506701, | |
| "eval_weighted_precision": 0.70753781441902, | |
| "eval_weighted_recall": 0.6605222734254992, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 44.5, | |
| "learning_rate": 2.013559322033898e-05, | |
| "loss": 12.8134, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_A1_f1": 0.8218390804597702, | |
| "eval_A1_precision": 0.9225806451612903, | |
| "eval_A1_recall": 0.7409326424870466, | |
| "eval_A2_f1": 0.7515527950310559, | |
| "eval_A2_precision": 0.6797752808988764, | |
| "eval_A2_recall": 0.8402777777777778, | |
| "eval_B1_f1": 0.5970149253731343, | |
| "eval_B1_precision": 0.7017543859649122, | |
| "eval_B1_recall": 0.5194805194805194, | |
| "eval_B2_f1": 0.8152610441767069, | |
| "eval_B2_precision": 0.7777777777777778, | |
| "eval_B2_recall": 0.8565400843881856, | |
| "eval_accuracy": 0.7788018433179723, | |
| "eval_loss": 0.8353127241134644, | |
| "eval_runtime": 9.9698, | |
| "eval_samples_per_second": 65.297, | |
| "eval_steps_per_second": 21.766, | |
| "eval_weighted_f1": 0.7773050103484156, | |
| "eval_weighted_precision": 0.7900370599357893, | |
| "eval_weighted_recall": 0.7788018433179723, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 120.0, | |
| "learning_rate": 1.0128813559322034e-05, | |
| "loss": 7.4136, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_A1_f1": 0.9072681704260651, | |
| "eval_A1_precision": 0.8786407766990292, | |
| "eval_A1_recall": 0.9378238341968912, | |
| "eval_A2_f1": 0.8243243243243243, | |
| "eval_A2_precision": 0.8026315789473685, | |
| "eval_A2_recall": 0.8472222222222222, | |
| "eval_B1_f1": 0.7913669064748201, | |
| "eval_B1_precision": 0.8870967741935484, | |
| "eval_B1_recall": 0.7142857142857143, | |
| "eval_B2_f1": 0.8547008547008547, | |
| "eval_B2_precision": 0.8658008658008658, | |
| "eval_B2_recall": 0.8438818565400844, | |
| "eval_accuracy": 0.8571428571428571, | |
| "eval_loss": 0.6585471034049988, | |
| "eval_runtime": 9.9471, | |
| "eval_samples_per_second": 65.446, | |
| "eval_steps_per_second": 21.815, | |
| "eval_weighted_f1": 0.8560749830377833, | |
| "eval_weighted_precision": 0.8581534164040586, | |
| "eval_weighted_recall": 0.8571428571428571, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 11.0, | |
| "learning_rate": 1.2203389830508474e-07, | |
| "loss": 5.9978, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_A1_f1": 0.9090909090909091, | |
| "eval_A1_precision": 0.8866995073891626, | |
| "eval_A1_recall": 0.9326424870466321, | |
| "eval_A2_f1": 0.8178694158075601, | |
| "eval_A2_precision": 0.8095238095238095, | |
| "eval_A2_recall": 0.8263888888888888, | |
| "eval_B1_f1": 0.8, | |
| "eval_B1_precision": 0.8529411764705882, | |
| "eval_B1_recall": 0.7532467532467533, | |
| "eval_B2_f1": 0.8595744680851064, | |
| "eval_B2_precision": 0.8669527896995708, | |
| "eval_B2_recall": 0.8523206751054853, | |
| "eval_accuracy": 0.858678955453149, | |
| "eval_loss": 0.6435705423355103, | |
| "eval_runtime": 10.0772, | |
| "eval_samples_per_second": 64.602, | |
| "eval_steps_per_second": 21.534, | |
| "eval_weighted_f1": 0.8579829343579175, | |
| "eval_weighted_precision": 0.8584465671959608, | |
| "eval_weighted_recall": 0.858678955453149, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "step": 328, | |
| "total_flos": 5835085400309760.0, | |
| "train_loss": 12.384773626560118, | |
| "train_runtime": 1108.4363, | |
| "train_samples_per_second": 9.397, | |
| "train_steps_per_second": 0.296 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_A1_f1": 0.9090909090909091, | |
| "eval_A1_precision": 0.8866995073891626, | |
| "eval_A1_recall": 0.9326424870466321, | |
| "eval_A2_f1": 0.8178694158075601, | |
| "eval_A2_precision": 0.8095238095238095, | |
| "eval_A2_recall": 0.8263888888888888, | |
| "eval_B1_f1": 0.8, | |
| "eval_B1_precision": 0.8529411764705882, | |
| "eval_B1_recall": 0.7532467532467533, | |
| "eval_B2_f1": 0.8595744680851064, | |
| "eval_B2_precision": 0.8669527896995708, | |
| "eval_B2_recall": 0.8523206751054853, | |
| "eval_accuracy": 0.858678955453149, | |
| "eval_loss": 0.6436282992362976, | |
| "eval_runtime": 10.0424, | |
| "eval_samples_per_second": 64.825, | |
| "eval_steps_per_second": 21.608, | |
| "eval_weighted_f1": 0.8579829343579175, | |
| "eval_weighted_precision": 0.8584465671959608, | |
| "eval_weighted_recall": 0.858678955453149, | |
| "step": 328 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 328, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "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": 5835085400309760.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |