Nous-Hermes-llama-2-7b_7b_cluster021_partitioned_v3_standardized_021
/
checkpoint-1000
/trainer_state.json
{ | |
"best_metric": 0.5762849450111389, | |
"best_model_checkpoint": "./output_v2/7b_cluster021_Nous-Hermes-llama-2-7b_partitioned_v3_standardized_021/checkpoint-800", | |
"epoch": 1.0196278358399185, | |
"global_step": 1000, | |
"is_hyper_param_search": false, | |
"is_local_process_zero": true, | |
"is_world_process_zero": true, | |
"log_history": [ | |
{ | |
"epoch": 0.01, | |
"learning_rate": 0.0002, | |
"loss": 0.6669, | |
"step": 10 | |
}, | |
{ | |
"epoch": 0.02, | |
"learning_rate": 0.0002, | |
"loss": 0.6301, | |
"step": 20 | |
}, | |
{ | |
"epoch": 0.03, | |
"learning_rate": 0.0002, | |
"loss": 0.7049, | |
"step": 30 | |
}, | |
{ | |
"epoch": 0.04, | |
"learning_rate": 0.0002, | |
"loss": 0.676, | |
"step": 40 | |
}, | |
{ | |
"epoch": 0.05, | |
"learning_rate": 0.0002, | |
"loss": 0.6972, | |
"step": 50 | |
}, | |
{ | |
"epoch": 0.06, | |
"learning_rate": 0.0002, | |
"loss": 0.7754, | |
"step": 60 | |
}, | |
{ | |
"epoch": 0.07, | |
"learning_rate": 0.0002, | |
"loss": 0.7056, | |
"step": 70 | |
}, | |
{ | |
"epoch": 0.08, | |
"learning_rate": 0.0002, | |
"loss": 0.7018, | |
"step": 80 | |
}, | |
{ | |
"epoch": 0.09, | |
"learning_rate": 0.0002, | |
"loss": 0.6906, | |
"step": 90 | |
}, | |
{ | |
"epoch": 0.1, | |
"learning_rate": 0.0002, | |
"loss": 0.6263, | |
"step": 100 | |
}, | |
{ | |
"epoch": 0.11, | |
"learning_rate": 0.0002, | |
"loss": 0.6163, | |
"step": 110 | |
}, | |
{ | |
"epoch": 0.12, | |
"learning_rate": 0.0002, | |
"loss": 0.6824, | |
"step": 120 | |
}, | |
{ | |
"epoch": 0.13, | |
"learning_rate": 0.0002, | |
"loss": 0.5913, | |
"step": 130 | |
}, | |
{ | |
"epoch": 0.14, | |
"learning_rate": 0.0002, | |
"loss": 0.5903, | |
"step": 140 | |
}, | |
{ | |
"epoch": 0.15, | |
"learning_rate": 0.0002, | |
"loss": 0.6851, | |
"step": 150 | |
}, | |
{ | |
"epoch": 0.16, | |
"learning_rate": 0.0002, | |
"loss": 0.5648, | |
"step": 160 | |
}, | |
{ | |
"epoch": 0.17, | |
"learning_rate": 0.0002, | |
"loss": 0.5615, | |
"step": 170 | |
}, | |
{ | |
"epoch": 0.18, | |
"learning_rate": 0.0002, | |
"loss": 0.5532, | |
"step": 180 | |
}, | |
{ | |
"epoch": 0.19, | |
"learning_rate": 0.0002, | |
"loss": 0.5662, | |
"step": 190 | |
}, | |
{ | |
"epoch": 0.2, | |
"learning_rate": 0.0002, | |
"loss": 0.6302, | |
"step": 200 | |
}, | |
{ | |
"epoch": 0.2, | |
"eval_loss": 0.6043372750282288, | |
"eval_runtime": 155.7668, | |
"eval_samples_per_second": 6.42, | |
"eval_steps_per_second": 3.21, | |
"step": 200 | |
}, | |
{ | |
"epoch": 0.2, | |
"mmlu_eval_accuracy": 0.45742535599141315, | |
"mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727, | |
"mmlu_eval_accuracy_anatomy": 0.6428571428571429, | |
"mmlu_eval_accuracy_astronomy": 0.375, | |
"mmlu_eval_accuracy_business_ethics": 0.45454545454545453, | |
"mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586, | |
"mmlu_eval_accuracy_college_biology": 0.4375, | |
"mmlu_eval_accuracy_college_chemistry": 0.25, | |
"mmlu_eval_accuracy_college_computer_science": 0.45454545454545453, | |
"mmlu_eval_accuracy_college_mathematics": 0.18181818181818182, | |
"mmlu_eval_accuracy_college_medicine": 0.2727272727272727, | |
"mmlu_eval_accuracy_college_physics": 0.45454545454545453, | |
"mmlu_eval_accuracy_computer_security": 0.2727272727272727, | |
"mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231, | |
"mmlu_eval_accuracy_econometrics": 0.16666666666666666, | |
"mmlu_eval_accuracy_electrical_engineering": 0.375, | |
"mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536, | |
"mmlu_eval_accuracy_formal_logic": 0.21428571428571427, | |
"mmlu_eval_accuracy_global_facts": 0.3, | |
"mmlu_eval_accuracy_high_school_biology": 0.375, | |
"mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091, | |
"mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556, | |
"mmlu_eval_accuracy_high_school_european_history": 0.5, | |
"mmlu_eval_accuracy_high_school_geography": 0.8181818181818182, | |
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666, | |
"mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723, | |
"mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276, | |
"mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231, | |
"mmlu_eval_accuracy_high_school_physics": 0.35294117647058826, | |
"mmlu_eval_accuracy_high_school_psychology": 0.7333333333333333, | |
"mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913, | |
"mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273, | |
"mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384, | |
"mmlu_eval_accuracy_human_aging": 0.7391304347826086, | |
"mmlu_eval_accuracy_human_sexuality": 0.3333333333333333, | |
"mmlu_eval_accuracy_international_law": 0.8461538461538461, | |
"mmlu_eval_accuracy_jurisprudence": 0.2727272727272727, | |
"mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556, | |
"mmlu_eval_accuracy_machine_learning": 0.2727272727272727, | |
"mmlu_eval_accuracy_management": 0.5454545454545454, | |
"mmlu_eval_accuracy_marketing": 0.68, | |
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273, | |
"mmlu_eval_accuracy_miscellaneous": 0.6511627906976745, | |
"mmlu_eval_accuracy_moral_disputes": 0.42105263157894735, | |
"mmlu_eval_accuracy_moral_scenarios": 0.24, | |
"mmlu_eval_accuracy_nutrition": 0.48484848484848486, | |
"mmlu_eval_accuracy_philosophy": 0.5, | |
"mmlu_eval_accuracy_prehistory": 0.4857142857142857, | |
"mmlu_eval_accuracy_professional_accounting": 0.2903225806451613, | |
"mmlu_eval_accuracy_professional_law": 0.3352941176470588, | |
"mmlu_eval_accuracy_professional_medicine": 0.41935483870967744, | |
"mmlu_eval_accuracy_professional_psychology": 0.4057971014492754, | |
"mmlu_eval_accuracy_public_relations": 0.6666666666666666, | |
"mmlu_eval_accuracy_security_studies": 0.48148148148148145, | |
"mmlu_eval_accuracy_sociology": 0.5909090909090909, | |
"mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364, | |
"mmlu_eval_accuracy_virology": 0.5, | |
"mmlu_eval_accuracy_world_religions": 0.7368421052631579, | |
"mmlu_loss": 0.8281081981918053, | |
"step": 200 | |
}, | |
{ | |
"epoch": 0.21, | |
"learning_rate": 0.0002, | |
"loss": 0.6021, | |
"step": 210 | |
}, | |
{ | |
"epoch": 0.22, | |
"learning_rate": 0.0002, | |
"loss": 0.6446, | |
"step": 220 | |
}, | |
{ | |
"epoch": 0.23, | |
"learning_rate": 0.0002, | |
"loss": 0.5811, | |
"step": 230 | |
}, | |
{ | |
"epoch": 0.24, | |
"learning_rate": 0.0002, | |
"loss": 0.5666, | |
"step": 240 | |
}, | |
{ | |
"epoch": 0.25, | |
"learning_rate": 0.0002, | |
"loss": 0.6179, | |
"step": 250 | |
}, | |
{ | |
"epoch": 0.27, | |
"learning_rate": 0.0002, | |
"loss": 0.6227, | |
"step": 260 | |
}, | |
{ | |
"epoch": 0.28, | |
"learning_rate": 0.0002, | |
"loss": 0.6022, | |
"step": 270 | |
}, | |
{ | |
"epoch": 0.29, | |
"learning_rate": 0.0002, | |
"loss": 0.6208, | |
"step": 280 | |
}, | |
{ | |
"epoch": 0.3, | |
"learning_rate": 0.0002, | |
"loss": 0.6157, | |
"step": 290 | |
}, | |
{ | |
"epoch": 0.31, | |
"learning_rate": 0.0002, | |
"loss": 0.6555, | |
"step": 300 | |
}, | |
{ | |
"epoch": 0.32, | |
"learning_rate": 0.0002, | |
"loss": 0.6757, | |
"step": 310 | |
}, | |
{ | |
"epoch": 0.33, | |
"learning_rate": 0.0002, | |
"loss": 0.5579, | |
"step": 320 | |
}, | |
{ | |
"epoch": 0.34, | |
"learning_rate": 0.0002, | |
"loss": 0.6288, | |
"step": 330 | |
}, | |
{ | |
"epoch": 0.35, | |
"learning_rate": 0.0002, | |
"loss": 0.6245, | |
"step": 340 | |
}, | |
{ | |
"epoch": 0.36, | |
"learning_rate": 0.0002, | |
"loss": 0.6182, | |
"step": 350 | |
}, | |
{ | |
"epoch": 0.37, | |
"learning_rate": 0.0002, | |
"loss": 0.596, | |
"step": 360 | |
}, | |
{ | |
"epoch": 0.38, | |
"learning_rate": 0.0002, | |
"loss": 0.6773, | |
"step": 370 | |
}, | |
{ | |
"epoch": 0.39, | |
"learning_rate": 0.0002, | |
"loss": 0.6496, | |
"step": 380 | |
}, | |
{ | |
"epoch": 0.4, | |
"learning_rate": 0.0002, | |
"loss": 0.7691, | |
"step": 390 | |
}, | |
{ | |
"epoch": 0.41, | |
"learning_rate": 0.0002, | |
"loss": 0.6709, | |
"step": 400 | |
}, | |
{ | |
"epoch": 0.41, | |
"eval_loss": 0.610517680644989, | |
"eval_runtime": 155.7719, | |
"eval_samples_per_second": 6.42, | |
"eval_steps_per_second": 3.21, | |
"step": 400 | |
}, | |
{ | |
"epoch": 0.41, | |
"mmlu_eval_accuracy": 0.46450897165748284, | |
"mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365, | |
"mmlu_eval_accuracy_anatomy": 0.5714285714285714, | |
"mmlu_eval_accuracy_astronomy": 0.375, | |
"mmlu_eval_accuracy_business_ethics": 0.45454545454545453, | |
"mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655, | |
"mmlu_eval_accuracy_college_biology": 0.5, | |
"mmlu_eval_accuracy_college_chemistry": 0.125, | |
"mmlu_eval_accuracy_college_computer_science": 0.36363636363636365, | |
"mmlu_eval_accuracy_college_mathematics": 0.2727272727272727, | |
"mmlu_eval_accuracy_college_medicine": 0.2727272727272727, | |
"mmlu_eval_accuracy_college_physics": 0.45454545454545453, | |
"mmlu_eval_accuracy_computer_security": 0.45454545454545453, | |
"mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231, | |
"mmlu_eval_accuracy_econometrics": 0.16666666666666666, | |
"mmlu_eval_accuracy_electrical_engineering": 0.3125, | |
"mmlu_eval_accuracy_elementary_mathematics": 0.43902439024390244, | |
"mmlu_eval_accuracy_formal_logic": 0.2857142857142857, | |
"mmlu_eval_accuracy_global_facts": 0.6, | |
"mmlu_eval_accuracy_high_school_biology": 0.40625, | |
"mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091, | |
"mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444, | |
"mmlu_eval_accuracy_high_school_european_history": 0.5, | |
"mmlu_eval_accuracy_high_school_geography": 0.8181818181818182, | |
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666, | |
"mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256, | |
"mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276, | |
"mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231, | |
"mmlu_eval_accuracy_high_school_physics": 0.35294117647058826, | |
"mmlu_eval_accuracy_high_school_psychology": 0.75, | |
"mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173, | |
"mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364, | |
"mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769, | |
"mmlu_eval_accuracy_human_aging": 0.6086956521739131, | |
"mmlu_eval_accuracy_human_sexuality": 0.4166666666666667, | |
"mmlu_eval_accuracy_international_law": 0.7692307692307693, | |
"mmlu_eval_accuracy_jurisprudence": 0.36363636363636365, | |
"mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556, | |
"mmlu_eval_accuracy_machine_learning": 0.18181818181818182, | |
"mmlu_eval_accuracy_management": 0.5454545454545454, | |
"mmlu_eval_accuracy_marketing": 0.8, | |
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273, | |
"mmlu_eval_accuracy_miscellaneous": 0.6511627906976745, | |
"mmlu_eval_accuracy_moral_disputes": 0.47368421052631576, | |
"mmlu_eval_accuracy_moral_scenarios": 0.24, | |
"mmlu_eval_accuracy_nutrition": 0.5151515151515151, | |
"mmlu_eval_accuracy_philosophy": 0.47058823529411764, | |
"mmlu_eval_accuracy_prehistory": 0.4857142857142857, | |
"mmlu_eval_accuracy_professional_accounting": 0.2903225806451613, | |
"mmlu_eval_accuracy_professional_law": 0.3411764705882353, | |
"mmlu_eval_accuracy_professional_medicine": 0.45161290322580644, | |
"mmlu_eval_accuracy_professional_psychology": 0.42028985507246375, | |
"mmlu_eval_accuracy_public_relations": 0.5833333333333334, | |
"mmlu_eval_accuracy_security_studies": 0.5185185185185185, | |
"mmlu_eval_accuracy_sociology": 0.6818181818181818, | |
"mmlu_eval_accuracy_us_foreign_policy": 0.45454545454545453, | |
"mmlu_eval_accuracy_virology": 0.3888888888888889, | |
"mmlu_eval_accuracy_world_religions": 0.7368421052631579, | |
"mmlu_loss": 0.9209539828428089, | |
"step": 400 | |
}, | |
{ | |
"epoch": 0.42, | |
"learning_rate": 0.0002, | |
"loss": 0.6442, | |
"step": 410 | |
}, | |
{ | |
"epoch": 0.43, | |
"learning_rate": 0.0002, | |
"loss": 0.5714, | |
"step": 420 | |
}, | |
{ | |
"epoch": 0.44, | |
"learning_rate": 0.0002, | |
"loss": 0.651, | |
"step": 430 | |
}, | |
{ | |
"epoch": 0.45, | |
"learning_rate": 0.0002, | |
"loss": 0.5631, | |
"step": 440 | |
}, | |
{ | |
"epoch": 0.46, | |
"learning_rate": 0.0002, | |
"loss": 0.6654, | |
"step": 450 | |
}, | |
{ | |
"epoch": 0.47, | |
"learning_rate": 0.0002, | |
"loss": 0.6274, | |
"step": 460 | |
}, | |
{ | |
"epoch": 0.48, | |
"learning_rate": 0.0002, | |
"loss": 0.6068, | |
"step": 470 | |
}, | |
{ | |
"epoch": 0.49, | |
"learning_rate": 0.0002, | |
"loss": 0.6132, | |
"step": 480 | |
}, | |
{ | |
"epoch": 0.5, | |
"learning_rate": 0.0002, | |
"loss": 0.5813, | |
"step": 490 | |
}, | |
{ | |
"epoch": 0.51, | |
"learning_rate": 0.0002, | |
"loss": 0.5631, | |
"step": 500 | |
}, | |
{ | |
"epoch": 0.52, | |
"learning_rate": 0.0002, | |
"loss": 0.6225, | |
"step": 510 | |
}, | |
{ | |
"epoch": 0.53, | |
"learning_rate": 0.0002, | |
"loss": 0.6485, | |
"step": 520 | |
}, | |
{ | |
"epoch": 0.54, | |
"learning_rate": 0.0002, | |
"loss": 0.6288, | |
"step": 530 | |
}, | |
{ | |
"epoch": 0.55, | |
"learning_rate": 0.0002, | |
"loss": 0.6293, | |
"step": 540 | |
}, | |
{ | |
"epoch": 0.56, | |
"learning_rate": 0.0002, | |
"loss": 0.6834, | |
"step": 550 | |
}, | |
{ | |
"epoch": 0.57, | |
"learning_rate": 0.0002, | |
"loss": 0.5742, | |
"step": 560 | |
}, | |
{ | |
"epoch": 0.58, | |
"learning_rate": 0.0002, | |
"loss": 0.6218, | |
"step": 570 | |
}, | |
{ | |
"epoch": 0.59, | |
"learning_rate": 0.0002, | |
"loss": 0.6191, | |
"step": 580 | |
}, | |
{ | |
"epoch": 0.6, | |
"learning_rate": 0.0002, | |
"loss": 0.6053, | |
"step": 590 | |
}, | |
{ | |
"epoch": 0.61, | |
"learning_rate": 0.0002, | |
"loss": 0.5944, | |
"step": 600 | |
}, | |
{ | |
"epoch": 0.61, | |
"eval_loss": 0.5885298252105713, | |
"eval_runtime": 155.667, | |
"eval_samples_per_second": 6.424, | |
"eval_steps_per_second": 3.212, | |
"step": 600 | |
}, | |
{ | |
"epoch": 0.61, | |
"mmlu_eval_accuracy": 0.45491265360439226, | |
"mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727, | |
"mmlu_eval_accuracy_anatomy": 0.5714285714285714, | |
"mmlu_eval_accuracy_astronomy": 0.5, | |
"mmlu_eval_accuracy_business_ethics": 0.45454545454545453, | |
"mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655, | |
"mmlu_eval_accuracy_college_biology": 0.4375, | |
"mmlu_eval_accuracy_college_chemistry": 0.25, | |
"mmlu_eval_accuracy_college_computer_science": 0.36363636363636365, | |
"mmlu_eval_accuracy_college_mathematics": 0.18181818181818182, | |
"mmlu_eval_accuracy_college_medicine": 0.2727272727272727, | |
"mmlu_eval_accuracy_college_physics": 0.45454545454545453, | |
"mmlu_eval_accuracy_computer_security": 0.36363636363636365, | |
"mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156, | |
"mmlu_eval_accuracy_econometrics": 0.16666666666666666, | |
"mmlu_eval_accuracy_electrical_engineering": 0.4375, | |
"mmlu_eval_accuracy_elementary_mathematics": 0.4146341463414634, | |
"mmlu_eval_accuracy_formal_logic": 0.35714285714285715, | |
"mmlu_eval_accuracy_global_facts": 0.5, | |
"mmlu_eval_accuracy_high_school_biology": 0.3125, | |
"mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091, | |
"mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444, | |
"mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556, | |
"mmlu_eval_accuracy_high_school_geography": 0.7727272727272727, | |
"mmlu_eval_accuracy_high_school_government_and_politics": 0.7142857142857143, | |
"mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723, | |
"mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793, | |
"mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464, | |
"mmlu_eval_accuracy_high_school_physics": 0.29411764705882354, | |
"mmlu_eval_accuracy_high_school_psychology": 0.6666666666666666, | |
"mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654, | |
"mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909, | |
"mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769, | |
"mmlu_eval_accuracy_human_aging": 0.5652173913043478, | |
"mmlu_eval_accuracy_human_sexuality": 0.3333333333333333, | |
"mmlu_eval_accuracy_international_law": 0.7692307692307693, | |
"mmlu_eval_accuracy_jurisprudence": 0.2727272727272727, | |
"mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556, | |
"mmlu_eval_accuracy_machine_learning": 0.18181818181818182, | |
"mmlu_eval_accuracy_management": 0.5454545454545454, | |
"mmlu_eval_accuracy_marketing": 0.72, | |
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273, | |
"mmlu_eval_accuracy_miscellaneous": 0.6744186046511628, | |
"mmlu_eval_accuracy_moral_disputes": 0.4473684210526316, | |
"mmlu_eval_accuracy_moral_scenarios": 0.24, | |
"mmlu_eval_accuracy_nutrition": 0.5151515151515151, | |
"mmlu_eval_accuracy_philosophy": 0.5294117647058824, | |
"mmlu_eval_accuracy_prehistory": 0.5142857142857142, | |
"mmlu_eval_accuracy_professional_accounting": 0.2903225806451613, | |
"mmlu_eval_accuracy_professional_law": 0.3235294117647059, | |
"mmlu_eval_accuracy_professional_medicine": 0.41935483870967744, | |
"mmlu_eval_accuracy_professional_psychology": 0.391304347826087, | |
"mmlu_eval_accuracy_public_relations": 0.5, | |
"mmlu_eval_accuracy_security_studies": 0.5185185185185185, | |
"mmlu_eval_accuracy_sociology": 0.5909090909090909, | |
"mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454, | |
"mmlu_eval_accuracy_virology": 0.5555555555555556, | |
"mmlu_eval_accuracy_world_religions": 0.7368421052631579, | |
"mmlu_loss": 0.8377707550501543, | |
"step": 600 | |
}, | |
{ | |
"epoch": 0.62, | |
"learning_rate": 0.0002, | |
"loss": 0.6014, | |
"step": 610 | |
}, | |
{ | |
"epoch": 0.63, | |
"learning_rate": 0.0002, | |
"loss": 0.6538, | |
"step": 620 | |
}, | |
{ | |
"epoch": 0.64, | |
"learning_rate": 0.0002, | |
"loss": 0.6049, | |
"step": 630 | |
}, | |
{ | |
"epoch": 0.65, | |
"learning_rate": 0.0002, | |
"loss": 0.6813, | |
"step": 640 | |
}, | |
{ | |
"epoch": 0.66, | |
"learning_rate": 0.0002, | |
"loss": 0.6082, | |
"step": 650 | |
}, | |
{ | |
"epoch": 0.67, | |
"learning_rate": 0.0002, | |
"loss": 0.592, | |
"step": 660 | |
}, | |
{ | |
"epoch": 0.68, | |
"learning_rate": 0.0002, | |
"loss": 0.6467, | |
"step": 670 | |
}, | |
{ | |
"epoch": 0.69, | |
"learning_rate": 0.0002, | |
"loss": 0.6501, | |
"step": 680 | |
}, | |
{ | |
"epoch": 0.7, | |
"learning_rate": 0.0002, | |
"loss": 0.5768, | |
"step": 690 | |
}, | |
{ | |
"epoch": 0.71, | |
"learning_rate": 0.0002, | |
"loss": 0.5692, | |
"step": 700 | |
}, | |
{ | |
"epoch": 0.72, | |
"learning_rate": 0.0002, | |
"loss": 0.6291, | |
"step": 710 | |
}, | |
{ | |
"epoch": 0.73, | |
"learning_rate": 0.0002, | |
"loss": 0.5957, | |
"step": 720 | |
}, | |
{ | |
"epoch": 0.74, | |
"learning_rate": 0.0002, | |
"loss": 0.7257, | |
"step": 730 | |
}, | |
{ | |
"epoch": 0.75, | |
"learning_rate": 0.0002, | |
"loss": 0.6799, | |
"step": 740 | |
}, | |
{ | |
"epoch": 0.76, | |
"learning_rate": 0.0002, | |
"loss": 0.5645, | |
"step": 750 | |
}, | |
{ | |
"epoch": 0.77, | |
"learning_rate": 0.0002, | |
"loss": 0.6182, | |
"step": 760 | |
}, | |
{ | |
"epoch": 0.79, | |
"learning_rate": 0.0002, | |
"loss": 0.5585, | |
"step": 770 | |
}, | |
{ | |
"epoch": 0.8, | |
"learning_rate": 0.0002, | |
"loss": 0.6406, | |
"step": 780 | |
}, | |
{ | |
"epoch": 0.81, | |
"learning_rate": 0.0002, | |
"loss": 0.6492, | |
"step": 790 | |
}, | |
{ | |
"epoch": 0.82, | |
"learning_rate": 0.0002, | |
"loss": 0.6522, | |
"step": 800 | |
}, | |
{ | |
"epoch": 0.82, | |
"eval_loss": 0.5762849450111389, | |
"eval_runtime": 155.9758, | |
"eval_samples_per_second": 6.411, | |
"eval_steps_per_second": 3.206, | |
"step": 800 | |
}, | |
{ | |
"epoch": 0.82, | |
"mmlu_eval_accuracy": 0.44603027709964993, | |
"mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365, | |
"mmlu_eval_accuracy_anatomy": 0.5, | |
"mmlu_eval_accuracy_astronomy": 0.375, | |
"mmlu_eval_accuracy_business_ethics": 0.5454545454545454, | |
"mmlu_eval_accuracy_clinical_knowledge": 0.3793103448275862, | |
"mmlu_eval_accuracy_college_biology": 0.3125, | |
"mmlu_eval_accuracy_college_chemistry": 0.125, | |
"mmlu_eval_accuracy_college_computer_science": 0.5454545454545454, | |
"mmlu_eval_accuracy_college_mathematics": 0.18181818181818182, | |
"mmlu_eval_accuracy_college_medicine": 0.3181818181818182, | |
"mmlu_eval_accuracy_college_physics": 0.5454545454545454, | |
"mmlu_eval_accuracy_computer_security": 0.2727272727272727, | |
"mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464, | |
"mmlu_eval_accuracy_econometrics": 0.16666666666666666, | |
"mmlu_eval_accuracy_electrical_engineering": 0.375, | |
"mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244, | |
"mmlu_eval_accuracy_formal_logic": 0.42857142857142855, | |
"mmlu_eval_accuracy_global_facts": 0.3, | |
"mmlu_eval_accuracy_high_school_biology": 0.34375, | |
"mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091, | |
"mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556, | |
"mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556, | |
"mmlu_eval_accuracy_high_school_geography": 0.7272727272727273, | |
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191, | |
"mmlu_eval_accuracy_high_school_macroeconomics": 0.3023255813953488, | |
"mmlu_eval_accuracy_high_school_mathematics": 0.3793103448275862, | |
"mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464, | |
"mmlu_eval_accuracy_high_school_physics": 0.35294117647058826, | |
"mmlu_eval_accuracy_high_school_psychology": 0.6666666666666666, | |
"mmlu_eval_accuracy_high_school_statistics": 0.21739130434782608, | |
"mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818, | |
"mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769, | |
"mmlu_eval_accuracy_human_aging": 0.6521739130434783, | |
"mmlu_eval_accuracy_human_sexuality": 0.25, | |
"mmlu_eval_accuracy_international_law": 0.8461538461538461, | |
"mmlu_eval_accuracy_jurisprudence": 0.2727272727272727, | |
"mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556, | |
"mmlu_eval_accuracy_machine_learning": 0.18181818181818182, | |
"mmlu_eval_accuracy_management": 0.45454545454545453, | |
"mmlu_eval_accuracy_marketing": 0.68, | |
"mmlu_eval_accuracy_medical_genetics": 0.6363636363636364, | |
"mmlu_eval_accuracy_miscellaneous": 0.6511627906976745, | |
"mmlu_eval_accuracy_moral_disputes": 0.47368421052631576, | |
"mmlu_eval_accuracy_moral_scenarios": 0.24, | |
"mmlu_eval_accuracy_nutrition": 0.48484848484848486, | |
"mmlu_eval_accuracy_philosophy": 0.5, | |
"mmlu_eval_accuracy_prehistory": 0.4, | |
"mmlu_eval_accuracy_professional_accounting": 0.3225806451612903, | |
"mmlu_eval_accuracy_professional_law": 0.32941176470588235, | |
"mmlu_eval_accuracy_professional_medicine": 0.41935483870967744, | |
"mmlu_eval_accuracy_professional_psychology": 0.391304347826087, | |
"mmlu_eval_accuracy_public_relations": 0.5, | |
"mmlu_eval_accuracy_security_studies": 0.48148148148148145, | |
"mmlu_eval_accuracy_sociology": 0.5454545454545454, | |
"mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364, | |
"mmlu_eval_accuracy_virology": 0.5, | |
"mmlu_eval_accuracy_world_religions": 0.7368421052631579, | |
"mmlu_loss": 0.9514536509003403, | |
"step": 800 | |
}, | |
{ | |
"epoch": 0.83, | |
"learning_rate": 0.0002, | |
"loss": 0.5656, | |
"step": 810 | |
}, | |
{ | |
"epoch": 0.84, | |
"learning_rate": 0.0002, | |
"loss": 0.6464, | |
"step": 820 | |
}, | |
{ | |
"epoch": 0.85, | |
"learning_rate": 0.0002, | |
"loss": 0.6604, | |
"step": 830 | |
}, | |
{ | |
"epoch": 0.86, | |
"learning_rate": 0.0002, | |
"loss": 0.6033, | |
"step": 840 | |
}, | |
{ | |
"epoch": 0.87, | |
"learning_rate": 0.0002, | |
"loss": 0.6011, | |
"step": 850 | |
}, | |
{ | |
"epoch": 0.88, | |
"learning_rate": 0.0002, | |
"loss": 0.5798, | |
"step": 860 | |
}, | |
{ | |
"epoch": 0.89, | |
"learning_rate": 0.0002, | |
"loss": 0.6034, | |
"step": 870 | |
}, | |
{ | |
"epoch": 0.9, | |
"learning_rate": 0.0002, | |
"loss": 0.6168, | |
"step": 880 | |
}, | |
{ | |
"epoch": 0.91, | |
"learning_rate": 0.0002, | |
"loss": 0.6036, | |
"step": 890 | |
}, | |
{ | |
"epoch": 0.92, | |
"learning_rate": 0.0002, | |
"loss": 0.677, | |
"step": 900 | |
}, | |
{ | |
"epoch": 0.93, | |
"learning_rate": 0.0002, | |
"loss": 0.6159, | |
"step": 910 | |
}, | |
{ | |
"epoch": 0.94, | |
"learning_rate": 0.0002, | |
"loss": 0.6319, | |
"step": 920 | |
}, | |
{ | |
"epoch": 0.95, | |
"learning_rate": 0.0002, | |
"loss": 0.6347, | |
"step": 930 | |
}, | |
{ | |
"epoch": 0.96, | |
"learning_rate": 0.0002, | |
"loss": 0.5835, | |
"step": 940 | |
}, | |
{ | |
"epoch": 0.97, | |
"learning_rate": 0.0002, | |
"loss": 0.6045, | |
"step": 950 | |
}, | |
{ | |
"epoch": 0.98, | |
"learning_rate": 0.0002, | |
"loss": 0.6415, | |
"step": 960 | |
}, | |
{ | |
"epoch": 0.99, | |
"learning_rate": 0.0002, | |
"loss": 0.6335, | |
"step": 970 | |
}, | |
{ | |
"epoch": 1.0, | |
"learning_rate": 0.0002, | |
"loss": 0.6833, | |
"step": 980 | |
}, | |
{ | |
"epoch": 1.01, | |
"learning_rate": 0.0002, | |
"loss": 0.5257, | |
"step": 990 | |
}, | |
{ | |
"epoch": 1.02, | |
"learning_rate": 0.0002, | |
"loss": 0.5649, | |
"step": 1000 | |
}, | |
{ | |
"epoch": 1.02, | |
"eval_loss": 0.576836109161377, | |
"eval_runtime": 156.2992, | |
"eval_samples_per_second": 6.398, | |
"eval_steps_per_second": 3.199, | |
"step": 1000 | |
}, | |
{ | |
"epoch": 1.02, | |
"mmlu_eval_accuracy": 0.45995155074883465, | |
"mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365, | |
"mmlu_eval_accuracy_anatomy": 0.5714285714285714, | |
"mmlu_eval_accuracy_astronomy": 0.3125, | |
"mmlu_eval_accuracy_business_ethics": 0.45454545454545453, | |
"mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586, | |
"mmlu_eval_accuracy_college_biology": 0.3125, | |
"mmlu_eval_accuracy_college_chemistry": 0.25, | |
"mmlu_eval_accuracy_college_computer_science": 0.36363636363636365, | |
"mmlu_eval_accuracy_college_mathematics": 0.18181818181818182, | |
"mmlu_eval_accuracy_college_medicine": 0.3181818181818182, | |
"mmlu_eval_accuracy_college_physics": 0.45454545454545453, | |
"mmlu_eval_accuracy_computer_security": 0.36363636363636365, | |
"mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231, | |
"mmlu_eval_accuracy_econometrics": 0.16666666666666666, | |
"mmlu_eval_accuracy_electrical_engineering": 0.375, | |
"mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244, | |
"mmlu_eval_accuracy_formal_logic": 0.35714285714285715, | |
"mmlu_eval_accuracy_global_facts": 0.4, | |
"mmlu_eval_accuracy_high_school_biology": 0.40625, | |
"mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091, | |
"mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666, | |
"mmlu_eval_accuracy_high_school_european_history": 0.5, | |
"mmlu_eval_accuracy_high_school_geography": 0.7727272727272727, | |
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666, | |
"mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723, | |
"mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793, | |
"mmlu_eval_accuracy_high_school_microeconomics": 0.5, | |
"mmlu_eval_accuracy_high_school_physics": 0.35294117647058826, | |
"mmlu_eval_accuracy_high_school_psychology": 0.6666666666666666, | |
"mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654, | |
"mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273, | |
"mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769, | |
"mmlu_eval_accuracy_human_aging": 0.6956521739130435, | |
"mmlu_eval_accuracy_human_sexuality": 0.3333333333333333, | |
"mmlu_eval_accuracy_international_law": 0.7692307692307693, | |
"mmlu_eval_accuracy_jurisprudence": 0.36363636363636365, | |
"mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556, | |
"mmlu_eval_accuracy_machine_learning": 0.18181818181818182, | |
"mmlu_eval_accuracy_management": 0.5454545454545454, | |
"mmlu_eval_accuracy_marketing": 0.76, | |
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273, | |
"mmlu_eval_accuracy_miscellaneous": 0.6744186046511628, | |
"mmlu_eval_accuracy_moral_disputes": 0.47368421052631576, | |
"mmlu_eval_accuracy_moral_scenarios": 0.24, | |
"mmlu_eval_accuracy_nutrition": 0.5151515151515151, | |
"mmlu_eval_accuracy_philosophy": 0.4411764705882353, | |
"mmlu_eval_accuracy_prehistory": 0.4857142857142857, | |
"mmlu_eval_accuracy_professional_accounting": 0.25806451612903225, | |
"mmlu_eval_accuracy_professional_law": 0.3352941176470588, | |
"mmlu_eval_accuracy_professional_medicine": 0.45161290322580644, | |
"mmlu_eval_accuracy_professional_psychology": 0.42028985507246375, | |
"mmlu_eval_accuracy_public_relations": 0.5, | |
"mmlu_eval_accuracy_security_studies": 0.48148148148148145, | |
"mmlu_eval_accuracy_sociology": 0.6363636363636364, | |
"mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364, | |
"mmlu_eval_accuracy_virology": 0.4444444444444444, | |
"mmlu_eval_accuracy_world_religions": 0.7368421052631579, | |
"mmlu_loss": 0.9169297016301137, | |
"step": 1000 | |
} | |
], | |
"max_steps": 5000, | |
"num_train_epochs": 6, | |
"total_flos": 1.8202917189124915e+17, | |
"trial_name": null, | |
"trial_params": null | |
} | |