{ "best_metric": 0.6964578628540039, "best_model_checkpoint": "./output_v2/7b_cluster06_Nous-Hermes-llama-2-7b_partitioned_v3_standardized_06/checkpoint-800", "epoch": 3.218390804597701, "global_step": 1400, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.02, "learning_rate": 0.0002, "loss": 0.8662, "step": 10 }, { "epoch": 0.05, "learning_rate": 0.0002, "loss": 0.7634, "step": 20 }, { "epoch": 0.07, "learning_rate": 0.0002, "loss": 0.8485, "step": 30 }, { "epoch": 0.09, "learning_rate": 0.0002, "loss": 0.7866, "step": 40 }, { "epoch": 0.11, "learning_rate": 0.0002, "loss": 0.7199, "step": 50 }, { "epoch": 0.14, "learning_rate": 0.0002, "loss": 0.7387, "step": 60 }, { "epoch": 0.16, "learning_rate": 0.0002, "loss": 0.705, "step": 70 }, { "epoch": 0.18, "learning_rate": 0.0002, "loss": 0.7617, "step": 80 }, { "epoch": 0.21, "learning_rate": 0.0002, "loss": 0.7022, "step": 90 }, { "epoch": 0.23, "learning_rate": 0.0002, "loss": 0.715, "step": 100 }, { "epoch": 0.25, "learning_rate": 0.0002, "loss": 0.6946, "step": 110 }, { "epoch": 0.28, "learning_rate": 0.0002, "loss": 0.7201, "step": 120 }, { "epoch": 0.3, "learning_rate": 0.0002, "loss": 0.6633, "step": 130 }, { "epoch": 0.32, "learning_rate": 0.0002, "loss": 0.687, "step": 140 }, { "epoch": 0.34, "learning_rate": 0.0002, "loss": 0.7582, "step": 150 }, { "epoch": 0.37, "learning_rate": 0.0002, "loss": 0.7141, "step": 160 }, { "epoch": 0.39, "learning_rate": 0.0002, "loss": 0.7852, "step": 170 }, { "epoch": 0.41, "learning_rate": 0.0002, "loss": 0.7228, "step": 180 }, { "epoch": 0.44, "learning_rate": 0.0002, "loss": 0.7682, "step": 190 }, { "epoch": 0.46, "learning_rate": 0.0002, "loss": 0.6408, "step": 200 }, { "epoch": 0.46, "eval_loss": 0.7120537757873535, "eval_runtime": 248.212, "eval_samples_per_second": 4.029, "eval_steps_per_second": 2.014, "step": 200 }, { "epoch": 0.46, "mmlu_eval_accuracy": 0.47641762588949127, "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727, "mmlu_eval_accuracy_anatomy": 0.5714285714285714, "mmlu_eval_accuracy_astronomy": 0.4375, "mmlu_eval_accuracy_business_ethics": 0.45454545454545453, "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655, "mmlu_eval_accuracy_college_biology": 0.375, "mmlu_eval_accuracy_college_chemistry": 0.125, "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453, "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182, "mmlu_eval_accuracy_college_medicine": 0.36363636363636365, "mmlu_eval_accuracy_college_physics": 0.45454545454545453, "mmlu_eval_accuracy_computer_security": 0.2727272727272727, "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615, "mmlu_eval_accuracy_econometrics": 0.16666666666666666, "mmlu_eval_accuracy_electrical_engineering": 0.625, "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637, "mmlu_eval_accuracy_formal_logic": 0.2857142857142857, "mmlu_eval_accuracy_global_facts": 0.5, "mmlu_eval_accuracy_high_school_biology": 0.34375, "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091, "mmlu_eval_accuracy_high_school_computer_science": 0.7777777777777778, "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112, "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": 0.3103448275862069, "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464, "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826, "mmlu_eval_accuracy_high_school_psychology": 0.6833333333333333, "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654, "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273, "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384, "mmlu_eval_accuracy_human_aging": 0.6956521739130435, "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667, "mmlu_eval_accuracy_international_law": 0.8461538461538461, "mmlu_eval_accuracy_jurisprudence": 0.5454545454545454, "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556, "mmlu_eval_accuracy_machine_learning": 0.2727272727272727, "mmlu_eval_accuracy_management": 0.7272727272727273, "mmlu_eval_accuracy_marketing": 0.76, "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": 0.5757575757575758, "mmlu_eval_accuracy_philosophy": 0.4117647058823529, "mmlu_eval_accuracy_prehistory": 0.4, "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613, "mmlu_eval_accuracy_professional_law": 0.35294117647058826, "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": 0.6818181818181818, "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364, "mmlu_eval_accuracy_virology": 0.4444444444444444, "mmlu_eval_accuracy_world_religions": 0.7368421052631579, "mmlu_loss": 1.048742408132273, "step": 200 }, { "epoch": 0.48, "learning_rate": 0.0002, "loss": 0.7296, "step": 210 }, { "epoch": 0.51, "learning_rate": 0.0002, "loss": 0.7013, "step": 220 }, { "epoch": 0.53, "learning_rate": 0.0002, "loss": 0.7162, "step": 230 }, { "epoch": 0.55, "learning_rate": 0.0002, "loss": 0.6838, "step": 240 }, { "epoch": 0.57, "learning_rate": 0.0002, "loss": 0.7733, "step": 250 }, { "epoch": 0.6, "learning_rate": 0.0002, "loss": 0.6866, "step": 260 }, { "epoch": 0.62, "learning_rate": 0.0002, "loss": 0.7496, "step": 270 }, { "epoch": 0.64, "learning_rate": 0.0002, "loss": 0.7202, "step": 280 }, { "epoch": 0.67, "learning_rate": 0.0002, "loss": 0.6604, "step": 290 }, { "epoch": 0.69, "learning_rate": 0.0002, "loss": 0.6567, "step": 300 }, { "epoch": 0.71, "learning_rate": 0.0002, "loss": 0.7013, "step": 310 }, { "epoch": 0.74, "learning_rate": 0.0002, "loss": 0.7217, "step": 320 }, { "epoch": 0.76, "learning_rate": 0.0002, "loss": 0.7105, "step": 330 }, { "epoch": 0.78, "learning_rate": 0.0002, "loss": 0.7674, "step": 340 }, { "epoch": 0.8, "learning_rate": 0.0002, "loss": 0.709, "step": 350 }, { "epoch": 0.83, "learning_rate": 0.0002, "loss": 0.6924, "step": 360 }, { "epoch": 0.85, "learning_rate": 0.0002, "loss": 0.7067, "step": 370 }, { "epoch": 0.87, "learning_rate": 0.0002, "loss": 0.7528, "step": 380 }, { "epoch": 0.9, "learning_rate": 0.0002, "loss": 0.6967, "step": 390 }, { "epoch": 0.92, "learning_rate": 0.0002, "loss": 0.7424, "step": 400 }, { "epoch": 0.92, "eval_loss": 0.7028327584266663, "eval_runtime": 248.3077, "eval_samples_per_second": 4.027, "eval_steps_per_second": 2.014, "step": 400 }, { "epoch": 0.92, "mmlu_eval_accuracy": 0.4534692920503279, "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182, "mmlu_eval_accuracy_anatomy": 0.5714285714285714, "mmlu_eval_accuracy_astronomy": 0.375, "mmlu_eval_accuracy_business_ethics": 0.5454545454545454, "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552, "mmlu_eval_accuracy_college_biology": 0.375, "mmlu_eval_accuracy_college_chemistry": 0.125, "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": 0.34615384615384615, "mmlu_eval_accuracy_econometrics": 0.16666666666666666, "mmlu_eval_accuracy_electrical_engineering": 0.5, "mmlu_eval_accuracy_elementary_mathematics": 0.2926829268292683, "mmlu_eval_accuracy_formal_logic": 0.2857142857142857, "mmlu_eval_accuracy_global_facts": 0.5, "mmlu_eval_accuracy_high_school_biology": 0.34375, "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": 0.7272727272727273, "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666, "mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723, "mmlu_eval_accuracy_high_school_mathematics": 0.3103448275862069, "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231, "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826, "mmlu_eval_accuracy_high_school_psychology": 0.7, "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913, "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364, "mmlu_eval_accuracy_high_school_world_history": 0.5, "mmlu_eval_accuracy_human_aging": 0.6956521739130435, "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": 0.7272727272727273, "mmlu_eval_accuracy_marketing": 0.72, "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273, "mmlu_eval_accuracy_miscellaneous": 0.686046511627907, "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576, "mmlu_eval_accuracy_moral_scenarios": 0.24, "mmlu_eval_accuracy_nutrition": 0.6060606060606061, "mmlu_eval_accuracy_philosophy": 0.4117647058823529, "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": 0.4057971014492754, "mmlu_eval_accuracy_public_relations": 0.4166666666666667, "mmlu_eval_accuracy_security_studies": 0.48148148148148145, "mmlu_eval_accuracy_sociology": 0.7727272727272727, "mmlu_eval_accuracy_us_foreign_policy": 0.45454545454545453, "mmlu_eval_accuracy_virology": 0.3888888888888889, "mmlu_eval_accuracy_world_religions": 0.7368421052631579, "mmlu_loss": 1.034321005696883, "step": 400 }, { "epoch": 0.94, "learning_rate": 0.0002, "loss": 0.722, "step": 410 }, { "epoch": 0.97, "learning_rate": 0.0002, "loss": 0.7236, "step": 420 }, { "epoch": 0.99, "learning_rate": 0.0002, "loss": 0.7075, "step": 430 }, { "epoch": 1.01, "learning_rate": 0.0002, "loss": 0.6595, "step": 440 }, { "epoch": 1.03, "learning_rate": 0.0002, "loss": 0.6631, "step": 450 }, { "epoch": 1.06, "learning_rate": 0.0002, "loss": 0.6957, "step": 460 }, { "epoch": 1.08, "learning_rate": 0.0002, "loss": 0.6092, "step": 470 }, { "epoch": 1.1, "learning_rate": 0.0002, "loss": 0.6014, "step": 480 }, { "epoch": 1.13, "learning_rate": 0.0002, "loss": 0.6629, "step": 490 }, { "epoch": 1.15, "learning_rate": 0.0002, "loss": 0.6606, "step": 500 }, { "epoch": 1.17, "learning_rate": 0.0002, "loss": 0.6623, "step": 510 }, { "epoch": 1.2, "learning_rate": 0.0002, "loss": 0.6528, "step": 520 }, { "epoch": 1.22, "learning_rate": 0.0002, "loss": 0.619, "step": 530 }, { "epoch": 1.24, "learning_rate": 0.0002, "loss": 0.6822, "step": 540 }, { "epoch": 1.26, "learning_rate": 0.0002, "loss": 0.6897, "step": 550 }, { "epoch": 1.29, "learning_rate": 0.0002, "loss": 0.6194, "step": 560 }, { "epoch": 1.31, "learning_rate": 0.0002, "loss": 0.6854, "step": 570 }, { "epoch": 1.33, "learning_rate": 0.0002, "loss": 0.6408, "step": 580 }, { "epoch": 1.36, "learning_rate": 0.0002, "loss": 0.6705, "step": 590 }, { "epoch": 1.38, "learning_rate": 0.0002, "loss": 0.6806, "step": 600 }, { "epoch": 1.38, "eval_loss": 0.7048470377922058, "eval_runtime": 248.1356, "eval_samples_per_second": 4.03, "eval_steps_per_second": 2.015, "step": 600 }, { "epoch": 1.38, "mmlu_eval_accuracy": 0.45527626426304835, "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182, "mmlu_eval_accuracy_anatomy": 0.5714285714285714, "mmlu_eval_accuracy_astronomy": 0.375, "mmlu_eval_accuracy_business_ethics": 0.5454545454545454, "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655, "mmlu_eval_accuracy_college_biology": 0.375, "mmlu_eval_accuracy_college_chemistry": 0.125, "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727, "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182, "mmlu_eval_accuracy_college_medicine": 0.4090909090909091, "mmlu_eval_accuracy_college_physics": 0.45454545454545453, "mmlu_eval_accuracy_computer_security": 0.36363636363636365, "mmlu_eval_accuracy_conceptual_physics": 0.5, "mmlu_eval_accuracy_econometrics": 0.16666666666666666, "mmlu_eval_accuracy_electrical_engineering": 0.4375, "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073, "mmlu_eval_accuracy_formal_logic": 0.2857142857142857, "mmlu_eval_accuracy_global_facts": 0.5, "mmlu_eval_accuracy_high_school_biology": 0.4375, "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453, "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556, "mmlu_eval_accuracy_high_school_european_history": 0.5, "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727, "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191, "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256, "mmlu_eval_accuracy_high_school_mathematics": 0.3103448275862069, "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231, "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354, "mmlu_eval_accuracy_high_school_psychology": 0.7, "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654, "mmlu_eval_accuracy_high_school_us_history": 0.5909090909090909, "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384, "mmlu_eval_accuracy_human_aging": 0.6956521739130435, "mmlu_eval_accuracy_human_sexuality": 0.5, "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": 0.2727272727272727, "mmlu_eval_accuracy_management": 0.6363636363636364, "mmlu_eval_accuracy_marketing": 0.76, "mmlu_eval_accuracy_medical_genetics": 0.6363636363636364, "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628, "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316, "mmlu_eval_accuracy_moral_scenarios": 0.24, "mmlu_eval_accuracy_nutrition": 0.5757575757575758, "mmlu_eval_accuracy_philosophy": 0.38235294117647056, "mmlu_eval_accuracy_prehistory": 0.5142857142857142, "mmlu_eval_accuracy_professional_accounting": 0.22580645161290322, "mmlu_eval_accuracy_professional_law": 0.31176470588235294, "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744, "mmlu_eval_accuracy_professional_psychology": 0.37681159420289856, "mmlu_eval_accuracy_public_relations": 0.4166666666666667, "mmlu_eval_accuracy_security_studies": 0.48148148148148145, "mmlu_eval_accuracy_sociology": 0.6818181818181818, "mmlu_eval_accuracy_us_foreign_policy": 0.45454545454545453, "mmlu_eval_accuracy_virology": 0.5, "mmlu_eval_accuracy_world_religions": 0.7894736842105263, "mmlu_loss": 0.8697303273707784, "step": 600 }, { "epoch": 1.4, "learning_rate": 0.0002, "loss": 0.6112, "step": 610 }, { "epoch": 1.43, "learning_rate": 0.0002, "loss": 0.6605, "step": 620 }, { "epoch": 1.45, "learning_rate": 0.0002, "loss": 0.6882, "step": 630 }, { "epoch": 1.47, "learning_rate": 0.0002, "loss": 0.6504, "step": 640 }, { "epoch": 1.49, "learning_rate": 0.0002, "loss": 0.6036, "step": 650 }, { "epoch": 1.52, "learning_rate": 0.0002, "loss": 0.5717, "step": 660 }, { "epoch": 1.54, "learning_rate": 0.0002, "loss": 0.669, "step": 670 }, { "epoch": 1.56, "learning_rate": 0.0002, "loss": 0.6218, "step": 680 }, { "epoch": 1.59, "learning_rate": 0.0002, "loss": 0.6473, "step": 690 }, { "epoch": 1.61, "learning_rate": 0.0002, "loss": 0.661, "step": 700 }, { "epoch": 1.63, "learning_rate": 0.0002, "loss": 0.6366, "step": 710 }, { "epoch": 1.66, "learning_rate": 0.0002, "loss": 0.6217, "step": 720 }, { "epoch": 1.68, "learning_rate": 0.0002, "loss": 0.6534, "step": 730 }, { "epoch": 1.7, "learning_rate": 0.0002, "loss": 0.6491, "step": 740 }, { "epoch": 1.72, "learning_rate": 0.0002, "loss": 0.6436, "step": 750 }, { "epoch": 1.75, "learning_rate": 0.0002, "loss": 0.6816, "step": 760 }, { "epoch": 1.77, "learning_rate": 0.0002, "loss": 0.6326, "step": 770 }, { "epoch": 1.79, "learning_rate": 0.0002, "loss": 0.6431, "step": 780 }, { "epoch": 1.82, "learning_rate": 0.0002, "loss": 0.6536, "step": 790 }, { "epoch": 1.84, "learning_rate": 0.0002, "loss": 0.659, "step": 800 }, { "epoch": 1.84, "eval_loss": 0.6964578628540039, "eval_runtime": 248.2292, "eval_samples_per_second": 4.029, "eval_steps_per_second": 2.014, "step": 800 }, { "epoch": 1.84, "mmlu_eval_accuracy": 0.46974001380312136, "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727, "mmlu_eval_accuracy_anatomy": 0.5714285714285714, "mmlu_eval_accuracy_astronomy": 0.3125, "mmlu_eval_accuracy_business_ethics": 0.5454545454545454, "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552, "mmlu_eval_accuracy_college_biology": 0.4375, "mmlu_eval_accuracy_college_chemistry": 0.125, "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365, "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182, "mmlu_eval_accuracy_college_medicine": 0.4090909090909091, "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": 0.5, "mmlu_eval_accuracy_elementary_mathematics": 0.2682926829268293, "mmlu_eval_accuracy_formal_logic": 0.2857142857142857, "mmlu_eval_accuracy_global_facts": 0.5, "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": 0.8181818181818182, "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191, "mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723, "mmlu_eval_accuracy_high_school_mathematics": 0.3103448275862069, "mmlu_eval_accuracy_high_school_microeconomics": 0.5, "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826, "mmlu_eval_accuracy_high_school_psychology": 0.7166666666666667, "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173, "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818, "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384, "mmlu_eval_accuracy_human_aging": 0.6521739130434783, "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667, "mmlu_eval_accuracy_international_law": 0.9230769230769231, "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453, "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556, "mmlu_eval_accuracy_machine_learning": 0.36363636363636365, "mmlu_eval_accuracy_management": 0.7272727272727273, "mmlu_eval_accuracy_marketing": 0.68, "mmlu_eval_accuracy_medical_genetics": 0.6363636363636364, "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628, "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576, "mmlu_eval_accuracy_moral_scenarios": 0.24, "mmlu_eval_accuracy_nutrition": 0.6363636363636364, "mmlu_eval_accuracy_philosophy": 0.4117647058823529, "mmlu_eval_accuracy_prehistory": 0.4857142857142857, "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225, "mmlu_eval_accuracy_professional_law": 0.3058823529411765, "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744, "mmlu_eval_accuracy_professional_psychology": 0.4057971014492754, "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": 0.45454545454545453, "mmlu_eval_accuracy_virology": 0.3888888888888889, "mmlu_eval_accuracy_world_religions": 0.7368421052631579, "mmlu_loss": 1.069855675548704, "step": 800 }, { "epoch": 1.86, "learning_rate": 0.0002, "loss": 0.6798, "step": 810 }, { "epoch": 1.89, "learning_rate": 0.0002, "loss": 0.6485, "step": 820 }, { "epoch": 1.91, "learning_rate": 0.0002, "loss": 0.6419, "step": 830 }, { "epoch": 1.93, "learning_rate": 0.0002, "loss": 0.6528, "step": 840 }, { "epoch": 1.95, "learning_rate": 0.0002, "loss": 0.6674, "step": 850 }, { "epoch": 1.98, "learning_rate": 0.0002, "loss": 0.6487, "step": 860 }, { "epoch": 2.0, "learning_rate": 0.0002, "loss": 0.6742, "step": 870 }, { "epoch": 2.02, "learning_rate": 0.0002, "loss": 0.5303, "step": 880 }, { "epoch": 2.05, "learning_rate": 0.0002, "loss": 0.5264, "step": 890 }, { "epoch": 2.07, "learning_rate": 0.0002, "loss": 0.5578, "step": 900 }, { "epoch": 2.09, "learning_rate": 0.0002, "loss": 0.5399, "step": 910 }, { "epoch": 2.11, "learning_rate": 0.0002, "loss": 0.6028, "step": 920 }, { "epoch": 2.14, "learning_rate": 0.0002, "loss": 0.5292, "step": 930 }, { "epoch": 2.16, "learning_rate": 0.0002, "loss": 0.5715, "step": 940 }, { "epoch": 2.18, "learning_rate": 0.0002, "loss": 0.514, "step": 950 }, { "epoch": 2.21, "learning_rate": 0.0002, "loss": 0.5381, "step": 960 }, { "epoch": 2.23, "learning_rate": 0.0002, "loss": 0.5259, "step": 970 }, { "epoch": 2.25, "learning_rate": 0.0002, "loss": 0.5476, "step": 980 }, { "epoch": 2.28, "learning_rate": 0.0002, "loss": 0.5369, "step": 990 }, { "epoch": 2.3, "learning_rate": 0.0002, "loss": 0.5541, "step": 1000 }, { "epoch": 2.3, "eval_loss": 0.7225061058998108, "eval_runtime": 248.2663, "eval_samples_per_second": 4.028, "eval_steps_per_second": 2.014, "step": 1000 }, { "epoch": 2.3, "mmlu_eval_accuracy": 0.46504233740978407, "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727, "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": 0.4375, "mmlu_eval_accuracy_college_chemistry": 0.125, "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365, "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182, "mmlu_eval_accuracy_college_medicine": 0.36363636363636365, "mmlu_eval_accuracy_college_physics": 0.45454545454545453, "mmlu_eval_accuracy_computer_security": 0.2727272727272727, "mmlu_eval_accuracy_conceptual_physics": 0.3076923076923077, "mmlu_eval_accuracy_econometrics": 0.16666666666666666, "mmlu_eval_accuracy_electrical_engineering": 0.5, "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073, "mmlu_eval_accuracy_formal_logic": 0.21428571428571427, "mmlu_eval_accuracy_global_facts": 0.4, "mmlu_eval_accuracy_high_school_biology": 0.34375, "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091, "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666, "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112, "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636, "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191, "mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723, "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724, "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464, "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354, "mmlu_eval_accuracy_high_school_psychology": 0.7, "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173, "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273, "mmlu_eval_accuracy_high_school_world_history": 0.6538461538461539, "mmlu_eval_accuracy_human_aging": 0.6521739130434783, "mmlu_eval_accuracy_human_sexuality": 0.5, "mmlu_eval_accuracy_international_law": 0.9230769230769231, "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453, "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556, "mmlu_eval_accuracy_machine_learning": 0.2727272727272727, "mmlu_eval_accuracy_management": 0.6363636363636364, "mmlu_eval_accuracy_marketing": 0.72, "mmlu_eval_accuracy_medical_genetics": 0.6363636363636364, "mmlu_eval_accuracy_miscellaneous": 0.686046511627907, "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316, "mmlu_eval_accuracy_moral_scenarios": 0.24, "mmlu_eval_accuracy_nutrition": 0.6060606060606061, "mmlu_eval_accuracy_philosophy": 0.35294117647058826, "mmlu_eval_accuracy_prehistory": 0.42857142857142855, "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225, "mmlu_eval_accuracy_professional_law": 0.3588235294117647, "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744, "mmlu_eval_accuracy_professional_psychology": 0.37681159420289856, "mmlu_eval_accuracy_public_relations": 0.5, "mmlu_eval_accuracy_security_studies": 0.48148148148148145, "mmlu_eval_accuracy_sociology": 0.6818181818181818, "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364, "mmlu_eval_accuracy_virology": 0.4444444444444444, "mmlu_eval_accuracy_world_religions": 0.7894736842105263, "mmlu_loss": 1.1883746445879302, "step": 1000 }, { "epoch": 2.32, "learning_rate": 0.0002, "loss": 0.5264, "step": 1010 }, { "epoch": 2.34, "learning_rate": 0.0002, "loss": 0.5494, "step": 1020 }, { "epoch": 2.37, "learning_rate": 0.0002, "loss": 0.5728, "step": 1030 }, { "epoch": 2.39, "learning_rate": 0.0002, "loss": 0.5291, "step": 1040 }, { "epoch": 2.41, "learning_rate": 0.0002, "loss": 0.523, "step": 1050 }, { "epoch": 2.44, "learning_rate": 0.0002, "loss": 0.5893, "step": 1060 }, { "epoch": 2.46, "learning_rate": 0.0002, "loss": 0.5839, "step": 1070 }, { "epoch": 2.48, "learning_rate": 0.0002, "loss": 0.5653, "step": 1080 }, { "epoch": 2.51, "learning_rate": 0.0002, "loss": 0.5518, "step": 1090 }, { "epoch": 2.53, "learning_rate": 0.0002, "loss": 0.5497, "step": 1100 }, { "epoch": 2.55, "learning_rate": 0.0002, "loss": 0.5789, "step": 1110 }, { "epoch": 2.57, "learning_rate": 0.0002, "loss": 0.5358, "step": 1120 }, { "epoch": 2.6, "learning_rate": 0.0002, "loss": 0.5576, "step": 1130 }, { "epoch": 2.62, "learning_rate": 0.0002, "loss": 0.5015, "step": 1140 }, { "epoch": 2.64, "learning_rate": 0.0002, "loss": 0.5494, "step": 1150 }, { "epoch": 2.67, "learning_rate": 0.0002, "loss": 0.5482, "step": 1160 }, { "epoch": 2.69, "learning_rate": 0.0002, "loss": 0.5882, "step": 1170 }, { "epoch": 2.71, "learning_rate": 0.0002, "loss": 0.5525, "step": 1180 }, { "epoch": 2.74, "learning_rate": 0.0002, "loss": 0.5455, "step": 1190 }, { "epoch": 2.76, "learning_rate": 0.0002, "loss": 0.5813, "step": 1200 }, { "epoch": 2.76, "eval_loss": 0.7202425003051758, "eval_runtime": 248.3045, "eval_samples_per_second": 4.027, "eval_steps_per_second": 2.014, "step": 1200 }, { "epoch": 2.76, "mmlu_eval_accuracy": 0.4683330314670407, "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727, "mmlu_eval_accuracy_anatomy": 0.5714285714285714, "mmlu_eval_accuracy_astronomy": 0.3125, "mmlu_eval_accuracy_business_ethics": 0.45454545454545453, "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552, "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": 0.5, "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": 0.4375, "mmlu_eval_accuracy_elementary_mathematics": 0.2682926829268293, "mmlu_eval_accuracy_formal_logic": 0.21428571428571427, "mmlu_eval_accuracy_global_facts": 0.4, "mmlu_eval_accuracy_high_school_biology": 0.375, "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091, "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666, "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556, "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182, "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714, "mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723, "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276, "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464, "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882, "mmlu_eval_accuracy_high_school_psychology": 0.8, "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173, "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273, "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": 0.9230769230769231, "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365, "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556, "mmlu_eval_accuracy_machine_learning": 0.2727272727272727, "mmlu_eval_accuracy_management": 0.7272727272727273, "mmlu_eval_accuracy_marketing": 0.72, "mmlu_eval_accuracy_medical_genetics": 0.6363636363636364, "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186, "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576, "mmlu_eval_accuracy_moral_scenarios": 0.24, "mmlu_eval_accuracy_nutrition": 0.6363636363636364, "mmlu_eval_accuracy_philosophy": 0.4117647058823529, "mmlu_eval_accuracy_prehistory": 0.4857142857142857, "mmlu_eval_accuracy_professional_accounting": 0.1935483870967742, "mmlu_eval_accuracy_professional_law": 0.3058823529411765, "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744, "mmlu_eval_accuracy_professional_psychology": 0.4057971014492754, "mmlu_eval_accuracy_public_relations": 0.4166666666666667, "mmlu_eval_accuracy_security_studies": 0.48148148148148145, "mmlu_eval_accuracy_sociology": 0.7272727272727273, "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182, "mmlu_eval_accuracy_virology": 0.5, "mmlu_eval_accuracy_world_religions": 0.7368421052631579, "mmlu_loss": 1.1609547404528597, "step": 1200 }, { "epoch": 2.78, "learning_rate": 0.0002, "loss": 0.5819, "step": 1210 }, { "epoch": 2.8, "learning_rate": 0.0002, "loss": 0.5756, "step": 1220 }, { "epoch": 2.83, "learning_rate": 0.0002, "loss": 0.5345, "step": 1230 }, { "epoch": 2.85, "learning_rate": 0.0002, "loss": 0.5642, "step": 1240 }, { "epoch": 2.87, "learning_rate": 0.0002, "loss": 0.5226, "step": 1250 }, { "epoch": 2.9, "learning_rate": 0.0002, "loss": 0.5812, "step": 1260 }, { "epoch": 2.92, "learning_rate": 0.0002, "loss": 0.5701, "step": 1270 }, { "epoch": 2.94, "learning_rate": 0.0002, "loss": 0.5534, "step": 1280 }, { "epoch": 2.97, "learning_rate": 0.0002, "loss": 0.526, "step": 1290 }, { "epoch": 2.99, "learning_rate": 0.0002, "loss": 0.5366, "step": 1300 }, { "epoch": 3.01, "learning_rate": 0.0002, "loss": 0.4988, "step": 1310 }, { "epoch": 3.03, "learning_rate": 0.0002, "loss": 0.4567, "step": 1320 }, { "epoch": 3.06, "learning_rate": 0.0002, "loss": 0.4366, "step": 1330 }, { "epoch": 3.08, "learning_rate": 0.0002, "loss": 0.4387, "step": 1340 }, { "epoch": 3.1, "learning_rate": 0.0002, "loss": 0.3887, "step": 1350 }, { "epoch": 3.13, "learning_rate": 0.0002, "loss": 0.4344, "step": 1360 }, { "epoch": 3.15, "learning_rate": 0.0002, "loss": 0.4543, "step": 1370 }, { "epoch": 3.17, "learning_rate": 0.0002, "loss": 0.4104, "step": 1380 }, { "epoch": 3.2, "learning_rate": 0.0002, "loss": 0.4564, "step": 1390 }, { "epoch": 3.22, "learning_rate": 0.0002, "loss": 0.4217, "step": 1400 }, { "epoch": 3.22, "eval_loss": 0.7734031081199646, "eval_runtime": 248.2207, "eval_samples_per_second": 4.029, "eval_steps_per_second": 2.014, "step": 1400 }, { "epoch": 3.22, "mmlu_eval_accuracy": 0.4660104484440945, "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727, "mmlu_eval_accuracy_anatomy": 0.5714285714285714, "mmlu_eval_accuracy_astronomy": 0.375, "mmlu_eval_accuracy_business_ethics": 0.5454545454545454, "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655, "mmlu_eval_accuracy_college_biology": 0.375, "mmlu_eval_accuracy_college_chemistry": 0.125, "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365, "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182, "mmlu_eval_accuracy_college_medicine": 0.45454545454545453, "mmlu_eval_accuracy_college_physics": 0.45454545454545453, "mmlu_eval_accuracy_computer_security": 0.2727272727272727, "mmlu_eval_accuracy_conceptual_physics": 0.3076923076923077, "mmlu_eval_accuracy_econometrics": 0.16666666666666666, "mmlu_eval_accuracy_electrical_engineering": 0.4375, "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073, "mmlu_eval_accuracy_formal_logic": 0.21428571428571427, "mmlu_eval_accuracy_global_facts": 0.4, "mmlu_eval_accuracy_high_school_biology": 0.34375, "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453, "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556, "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556, "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727, "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666, "mmlu_eval_accuracy_high_school_macroeconomics": 0.27906976744186046, "mmlu_eval_accuracy_high_school_mathematics": 0.3448275862068966, "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231, "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354, "mmlu_eval_accuracy_high_school_psychology": 0.7333333333333333, "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173, "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273, "mmlu_eval_accuracy_high_school_world_history": 0.6538461538461539, "mmlu_eval_accuracy_human_aging": 0.6956521739130435, "mmlu_eval_accuracy_human_sexuality": 0.5, "mmlu_eval_accuracy_international_law": 0.8461538461538461, "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365, "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556, "mmlu_eval_accuracy_machine_learning": 0.2727272727272727, "mmlu_eval_accuracy_management": 0.6363636363636364, "mmlu_eval_accuracy_marketing": 0.72, "mmlu_eval_accuracy_medical_genetics": 0.6363636363636364, "mmlu_eval_accuracy_miscellaneous": 0.6976744186046512, "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316, "mmlu_eval_accuracy_moral_scenarios": 0.24, "mmlu_eval_accuracy_nutrition": 0.6060606060606061, "mmlu_eval_accuracy_philosophy": 0.38235294117647056, "mmlu_eval_accuracy_prehistory": 0.4, "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194, "mmlu_eval_accuracy_professional_law": 0.3411764705882353, "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744, "mmlu_eval_accuracy_professional_psychology": 0.4057971014492754, "mmlu_eval_accuracy_public_relations": 0.5, "mmlu_eval_accuracy_security_studies": 0.4074074074074074, "mmlu_eval_accuracy_sociology": 0.6818181818181818, "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273, "mmlu_eval_accuracy_virology": 0.5, "mmlu_eval_accuracy_world_religions": 0.7894736842105263, "mmlu_loss": 1.0775621755841351, "step": 1400 } ], "max_steps": 5000, "num_train_epochs": 12, "total_flos": 3.3098129260712755e+17, "trial_name": null, "trial_params": null }