{ "best_metric": 0.8871135711669922, "best_model_checkpoint": "./output_v2/7b_cluster04_Nous-Hermes-llama-2-7b_partitioned_v3_standardized_04/checkpoint-1000", "epoch": 0.9047726758651888, "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": 0.9647, "step": 10 }, { "epoch": 0.02, "learning_rate": 0.0002, "loss": 0.9474, "step": 20 }, { "epoch": 0.03, "learning_rate": 0.0002, "loss": 0.9287, "step": 30 }, { "epoch": 0.04, "learning_rate": 0.0002, "loss": 0.8723, "step": 40 }, { "epoch": 0.05, "learning_rate": 0.0002, "loss": 0.8812, "step": 50 }, { "epoch": 0.05, "learning_rate": 0.0002, "loss": 0.8627, "step": 60 }, { "epoch": 0.06, "learning_rate": 0.0002, "loss": 0.8893, "step": 70 }, { "epoch": 0.07, "learning_rate": 0.0002, "loss": 0.8793, "step": 80 }, { "epoch": 0.08, "learning_rate": 0.0002, "loss": 0.8778, "step": 90 }, { "epoch": 0.09, "learning_rate": 0.0002, "loss": 0.8809, "step": 100 }, { "epoch": 0.1, "learning_rate": 0.0002, "loss": 0.8534, "step": 110 }, { "epoch": 0.11, "learning_rate": 0.0002, "loss": 0.8566, "step": 120 }, { "epoch": 0.12, "learning_rate": 0.0002, "loss": 0.8979, "step": 130 }, { "epoch": 0.13, "learning_rate": 0.0002, "loss": 0.8919, "step": 140 }, { "epoch": 0.14, "learning_rate": 0.0002, "loss": 0.8435, "step": 150 }, { "epoch": 0.14, "learning_rate": 0.0002, "loss": 0.8764, "step": 160 }, { "epoch": 0.15, "learning_rate": 0.0002, "loss": 0.8806, "step": 170 }, { "epoch": 0.16, "learning_rate": 0.0002, "loss": 0.8178, "step": 180 }, { "epoch": 0.17, "learning_rate": 0.0002, "loss": 0.8821, "step": 190 }, { "epoch": 0.18, "learning_rate": 0.0002, "loss": 0.8603, "step": 200 }, { "epoch": 0.18, "eval_loss": 0.91291344165802, "eval_runtime": 191.1983, "eval_samples_per_second": 5.23, "eval_steps_per_second": 2.615, "step": 200 }, { "epoch": 0.18, "mmlu_eval_accuracy": 0.4726714210605103, "mmlu_eval_accuracy_abstract_algebra": 0.09090909090909091, "mmlu_eval_accuracy_anatomy": 0.5714285714285714, "mmlu_eval_accuracy_astronomy": 0.4375, "mmlu_eval_accuracy_business_ethics": 0.6363636363636364, "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": 0.45454545454545453, "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": 0.36363636363636365, "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464, "mmlu_eval_accuracy_econometrics": 0.16666666666666666, "mmlu_eval_accuracy_electrical_engineering": 0.4375, "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637, "mmlu_eval_accuracy_formal_logic": 0.21428571428571427, "mmlu_eval_accuracy_global_facts": 0.6, "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": 0.7222222222222222, "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727, "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666, "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256, "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793, "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231, "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826, "mmlu_eval_accuracy_high_school_psychology": 0.6833333333333333, "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173, "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818, "mmlu_eval_accuracy_high_school_world_history": 0.5, "mmlu_eval_accuracy_human_aging": 0.6521739130434783, "mmlu_eval_accuracy_human_sexuality": 0.5, "mmlu_eval_accuracy_international_law": 0.6923076923076923, "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727, "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556, "mmlu_eval_accuracy_machine_learning": 0.2727272727272727, "mmlu_eval_accuracy_management": 0.8181818181818182, "mmlu_eval_accuracy_marketing": 0.8, "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273, "mmlu_eval_accuracy_miscellaneous": 0.686046511627907, "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316, "mmlu_eval_accuracy_moral_scenarios": 0.24, "mmlu_eval_accuracy_nutrition": 0.5757575757575758, "mmlu_eval_accuracy_philosophy": 0.5, "mmlu_eval_accuracy_prehistory": 0.45714285714285713, "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225, "mmlu_eval_accuracy_professional_law": 0.3588235294117647, "mmlu_eval_accuracy_professional_medicine": 0.3870967741935484, "mmlu_eval_accuracy_professional_psychology": 0.391304347826087, "mmlu_eval_accuracy_public_relations": 0.5, "mmlu_eval_accuracy_security_studies": 0.5925925925925926, "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": 0.6842105263157895, "mmlu_loss": 1.0958761447948202, "step": 200 }, { "epoch": 0.19, "learning_rate": 0.0002, "loss": 0.8708, "step": 210 }, { "epoch": 0.2, "learning_rate": 0.0002, "loss": 0.8801, "step": 220 }, { "epoch": 0.21, "learning_rate": 0.0002, "loss": 0.8636, "step": 230 }, { "epoch": 0.22, "learning_rate": 0.0002, "loss": 0.8371, "step": 240 }, { "epoch": 0.23, "learning_rate": 0.0002, "loss": 0.8593, "step": 250 }, { "epoch": 0.24, "learning_rate": 0.0002, "loss": 0.8254, "step": 260 }, { "epoch": 0.24, "learning_rate": 0.0002, "loss": 0.8625, "step": 270 }, { "epoch": 0.25, "learning_rate": 0.0002, "loss": 0.8968, "step": 280 }, { "epoch": 0.26, "learning_rate": 0.0002, "loss": 0.8849, "step": 290 }, { "epoch": 0.27, "learning_rate": 0.0002, "loss": 0.8063, "step": 300 }, { "epoch": 0.28, "learning_rate": 0.0002, "loss": 0.8439, "step": 310 }, { "epoch": 0.29, "learning_rate": 0.0002, "loss": 0.8491, "step": 320 }, { "epoch": 0.3, "learning_rate": 0.0002, "loss": 0.8443, "step": 330 }, { "epoch": 0.31, "learning_rate": 0.0002, "loss": 0.8588, "step": 340 }, { "epoch": 0.32, "learning_rate": 0.0002, "loss": 0.8509, "step": 350 }, { "epoch": 0.33, "learning_rate": 0.0002, "loss": 0.8041, "step": 360 }, { "epoch": 0.33, "learning_rate": 0.0002, "loss": 0.8418, "step": 370 }, { "epoch": 0.34, "learning_rate": 0.0002, "loss": 0.8496, "step": 380 }, { "epoch": 0.35, "learning_rate": 0.0002, "loss": 0.8175, "step": 390 }, { "epoch": 0.36, "learning_rate": 0.0002, "loss": 0.8622, "step": 400 }, { "epoch": 0.36, "eval_loss": 0.9020043015480042, "eval_runtime": 191.4718, "eval_samples_per_second": 5.223, "eval_steps_per_second": 2.611, "step": 400 }, { "epoch": 0.36, "mmlu_eval_accuracy": 0.4652294198230619, "mmlu_eval_accuracy_abstract_algebra": 0.09090909090909091, "mmlu_eval_accuracy_anatomy": 0.5714285714285714, "mmlu_eval_accuracy_astronomy": 0.4375, "mmlu_eval_accuracy_business_ethics": 0.6363636363636364, "mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586, "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": 0.3181818181818182, "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": 0.4375, "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073, "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": 0.7777777777777778, "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": 0.3488372093023256, "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276, "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": 0.2608695652173913, "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364, "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": 0.7692307692307693, "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": 0.76, "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": 0.4411764705882353, "mmlu_eval_accuracy_prehistory": 0.42857142857142855, "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613, "mmlu_eval_accuracy_professional_law": 0.3411764705882353, "mmlu_eval_accuracy_professional_medicine": 0.3870967741935484, "mmlu_eval_accuracy_professional_psychology": 0.391304347826087, "mmlu_eval_accuracy_public_relations": 0.5, "mmlu_eval_accuracy_security_studies": 0.5555555555555556, "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": 0.6842105263157895, "mmlu_loss": 1.1965837302975182, "step": 400 }, { "epoch": 0.37, "learning_rate": 0.0002, "loss": 0.8675, "step": 410 }, { "epoch": 0.38, "learning_rate": 0.0002, "loss": 0.853, "step": 420 }, { "epoch": 0.39, "learning_rate": 0.0002, "loss": 0.8376, "step": 430 }, { "epoch": 0.4, "learning_rate": 0.0002, "loss": 0.8217, "step": 440 }, { "epoch": 0.41, "learning_rate": 0.0002, "loss": 0.8219, "step": 450 }, { "epoch": 0.42, "learning_rate": 0.0002, "loss": 0.8451, "step": 460 }, { "epoch": 0.43, "learning_rate": 0.0002, "loss": 0.8744, "step": 470 }, { "epoch": 0.43, "learning_rate": 0.0002, "loss": 0.8496, "step": 480 }, { "epoch": 0.44, "learning_rate": 0.0002, "loss": 0.8314, "step": 490 }, { "epoch": 0.45, "learning_rate": 0.0002, "loss": 0.8265, "step": 500 }, { "epoch": 0.46, "learning_rate": 0.0002, "loss": 0.8256, "step": 510 }, { "epoch": 0.47, "learning_rate": 0.0002, "loss": 0.8259, "step": 520 }, { "epoch": 0.48, "learning_rate": 0.0002, "loss": 0.8525, "step": 530 }, { "epoch": 0.49, "learning_rate": 0.0002, "loss": 0.8211, "step": 540 }, { "epoch": 0.5, "learning_rate": 0.0002, "loss": 0.833, "step": 550 }, { "epoch": 0.51, "learning_rate": 0.0002, "loss": 0.8733, "step": 560 }, { "epoch": 0.52, "learning_rate": 0.0002, "loss": 0.8583, "step": 570 }, { "epoch": 0.52, "learning_rate": 0.0002, "loss": 0.8137, "step": 580 }, { "epoch": 0.53, "learning_rate": 0.0002, "loss": 0.8369, "step": 590 }, { "epoch": 0.54, "learning_rate": 0.0002, "loss": 0.8416, "step": 600 }, { "epoch": 0.54, "eval_loss": 0.8950903415679932, "eval_runtime": 190.9856, "eval_samples_per_second": 5.236, "eval_steps_per_second": 2.618, "step": 600 }, { "epoch": 0.54, "mmlu_eval_accuracy": 0.46404769451684064, "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": 0.4375, "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": 0.2727272727272727, "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": 0.4375, "mmlu_eval_accuracy_elementary_mathematics": 0.24390243902439024, "mmlu_eval_accuracy_formal_logic": 0.35714285714285715, "mmlu_eval_accuracy_global_facts": 0.5, "mmlu_eval_accuracy_high_school_biology": 0.375, "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365, "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666, "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556, "mmlu_eval_accuracy_high_school_geography": 0.6818181818181818, "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": 0.46153846153846156, "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826, "mmlu_eval_accuracy_high_school_psychology": 0.7333333333333333, "mmlu_eval_accuracy_high_school_statistics": 0.21739130434782608, "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": 0.8461538461538461, "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453, "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556, "mmlu_eval_accuracy_machine_learning": 0.18181818181818182, "mmlu_eval_accuracy_management": 0.7272727272727273, "mmlu_eval_accuracy_marketing": 0.8, "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273, "mmlu_eval_accuracy_miscellaneous": 0.6976744186046512, "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576, "mmlu_eval_accuracy_moral_scenarios": 0.24, "mmlu_eval_accuracy_nutrition": 0.6060606060606061, "mmlu_eval_accuracy_philosophy": 0.4411764705882353, "mmlu_eval_accuracy_prehistory": 0.4857142857142857, "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225, "mmlu_eval_accuracy_professional_law": 0.3411764705882353, "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": 0.5925925925925926, "mmlu_eval_accuracy_sociology": 0.6818181818181818, "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454, "mmlu_eval_accuracy_virology": 0.4444444444444444, "mmlu_eval_accuracy_world_religions": 0.7368421052631579, "mmlu_loss": 1.0818409637667181, "step": 600 }, { "epoch": 0.55, "learning_rate": 0.0002, "loss": 0.8233, "step": 610 }, { "epoch": 0.56, "learning_rate": 0.0002, "loss": 0.8159, "step": 620 }, { "epoch": 0.57, "learning_rate": 0.0002, "loss": 0.8161, "step": 630 }, { "epoch": 0.58, "learning_rate": 0.0002, "loss": 0.8297, "step": 640 }, { "epoch": 0.59, "learning_rate": 0.0002, "loss": 0.8127, "step": 650 }, { "epoch": 0.6, "learning_rate": 0.0002, "loss": 0.8427, "step": 660 }, { "epoch": 0.61, "learning_rate": 0.0002, "loss": 0.8061, "step": 670 }, { "epoch": 0.62, "learning_rate": 0.0002, "loss": 0.8576, "step": 680 }, { "epoch": 0.62, "learning_rate": 0.0002, "loss": 0.8387, "step": 690 }, { "epoch": 0.63, "learning_rate": 0.0002, "loss": 0.887, "step": 700 }, { "epoch": 0.64, "learning_rate": 0.0002, "loss": 0.8603, "step": 710 }, { "epoch": 0.65, "learning_rate": 0.0002, "loss": 0.831, "step": 720 }, { "epoch": 0.66, "learning_rate": 0.0002, "loss": 0.8268, "step": 730 }, { "epoch": 0.67, "learning_rate": 0.0002, "loss": 0.8922, "step": 740 }, { "epoch": 0.68, "learning_rate": 0.0002, "loss": 0.8329, "step": 750 }, { "epoch": 0.69, "learning_rate": 0.0002, "loss": 0.8571, "step": 760 }, { "epoch": 0.7, "learning_rate": 0.0002, "loss": 0.8538, "step": 770 }, { "epoch": 0.71, "learning_rate": 0.0002, "loss": 0.8543, "step": 780 }, { "epoch": 0.71, "learning_rate": 0.0002, "loss": 0.8591, "step": 790 }, { "epoch": 0.72, "learning_rate": 0.0002, "loss": 0.8743, "step": 800 }, { "epoch": 0.72, "eval_loss": 0.8907580375671387, "eval_runtime": 191.1671, "eval_samples_per_second": 5.231, "eval_steps_per_second": 2.616, "step": 800 }, { "epoch": 0.72, "mmlu_eval_accuracy": 0.4759603492610683, "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727, "mmlu_eval_accuracy_anatomy": 0.5714285714285714, "mmlu_eval_accuracy_astronomy": 0.4375, "mmlu_eval_accuracy_business_ethics": 0.5454545454545454, "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": 0.45454545454545453, "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": 0.38461538461538464, "mmlu_eval_accuracy_econometrics": 0.16666666666666666, "mmlu_eval_accuracy_electrical_engineering": 0.3125, "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536, "mmlu_eval_accuracy_formal_logic": 0.21428571428571427, "mmlu_eval_accuracy_global_facts": 0.6, "mmlu_eval_accuracy_high_school_biology": 0.375, "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182, "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666, "mmlu_eval_accuracy_high_school_european_history": 0.5, "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636, "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666, "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395, "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724, "mmlu_eval_accuracy_high_school_microeconomics": 0.5, "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354, "mmlu_eval_accuracy_high_school_psychology": 0.7166666666666667, "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173, "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364, "mmlu_eval_accuracy_high_school_world_history": 0.5, "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": 0.36363636363636365, "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556, "mmlu_eval_accuracy_machine_learning": 0.2727272727272727, "mmlu_eval_accuracy_management": 0.8181818181818182, "mmlu_eval_accuracy_marketing": 0.8, "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273, "mmlu_eval_accuracy_miscellaneous": 0.686046511627907, "mmlu_eval_accuracy_moral_disputes": 0.5526315789473685, "mmlu_eval_accuracy_moral_scenarios": 0.24, "mmlu_eval_accuracy_nutrition": 0.6060606060606061, "mmlu_eval_accuracy_philosophy": 0.47058823529411764, "mmlu_eval_accuracy_prehistory": 0.5428571428571428, "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225, "mmlu_eval_accuracy_professional_law": 0.34705882352941175, "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": 0.5925925925925926, "mmlu_eval_accuracy_sociology": 0.6818181818181818, "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454, "mmlu_eval_accuracy_virology": 0.4444444444444444, "mmlu_eval_accuracy_world_religions": 0.7368421052631579, "mmlu_loss": 0.9363480592657163, "step": 800 }, { "epoch": 0.73, "learning_rate": 0.0002, "loss": 0.8608, "step": 810 }, { "epoch": 0.74, "learning_rate": 0.0002, "loss": 0.8207, "step": 820 }, { "epoch": 0.75, "learning_rate": 0.0002, "loss": 0.8284, "step": 830 }, { "epoch": 0.76, "learning_rate": 0.0002, "loss": 0.8135, "step": 840 }, { "epoch": 0.77, "learning_rate": 0.0002, "loss": 0.8523, "step": 850 }, { "epoch": 0.78, "learning_rate": 0.0002, "loss": 0.8349, "step": 860 }, { "epoch": 0.79, "learning_rate": 0.0002, "loss": 0.855, "step": 870 }, { "epoch": 0.8, "learning_rate": 0.0002, "loss": 0.8376, "step": 880 }, { "epoch": 0.81, "learning_rate": 0.0002, "loss": 0.8906, "step": 890 }, { "epoch": 0.81, "learning_rate": 0.0002, "loss": 0.8522, "step": 900 }, { "epoch": 0.82, "learning_rate": 0.0002, "loss": 0.8491, "step": 910 }, { "epoch": 0.83, "learning_rate": 0.0002, "loss": 0.8421, "step": 920 }, { "epoch": 0.84, "learning_rate": 0.0002, "loss": 0.8285, "step": 930 }, { "epoch": 0.85, "learning_rate": 0.0002, "loss": 0.8655, "step": 940 }, { "epoch": 0.86, "learning_rate": 0.0002, "loss": 0.8348, "step": 950 }, { "epoch": 0.87, "learning_rate": 0.0002, "loss": 0.8695, "step": 960 }, { "epoch": 0.88, "learning_rate": 0.0002, "loss": 0.8738, "step": 970 }, { "epoch": 0.89, "learning_rate": 0.0002, "loss": 0.8576, "step": 980 }, { "epoch": 0.9, "learning_rate": 0.0002, "loss": 0.7884, "step": 990 }, { "epoch": 0.9, "learning_rate": 0.0002, "loss": 0.8172, "step": 1000 }, { "epoch": 0.9, "eval_loss": 0.8871135711669922, "eval_runtime": 191.669, "eval_samples_per_second": 5.217, "eval_steps_per_second": 2.609, "step": 1000 }, { "epoch": 0.9, "mmlu_eval_accuracy": 0.47325503865005797, "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365, "mmlu_eval_accuracy_anatomy": 0.5714285714285714, "mmlu_eval_accuracy_astronomy": 0.5, "mmlu_eval_accuracy_business_ethics": 0.5454545454545454, "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": 0.45454545454545453, "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": 0.36363636363636365, "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": 0.36585365853658536, "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": 0.3181818181818182, "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": 0.3488372093023256, "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724, "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156, "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": 0.7272727272727273, "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": 0.2727272727272727, "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556, "mmlu_eval_accuracy_machine_learning": 0.2727272727272727, "mmlu_eval_accuracy_management": 0.7272727272727273, "mmlu_eval_accuracy_marketing": 0.8, "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273, "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186, "mmlu_eval_accuracy_moral_disputes": 0.5526315789473685, "mmlu_eval_accuracy_moral_scenarios": 0.24, "mmlu_eval_accuracy_nutrition": 0.5757575757575758, "mmlu_eval_accuracy_philosophy": 0.47058823529411764, "mmlu_eval_accuracy_prehistory": 0.5428571428571428, "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225, "mmlu_eval_accuracy_professional_law": 0.32941176470588235, "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744, "mmlu_eval_accuracy_professional_psychology": 0.36231884057971014, "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": 0.9558495128084723, "step": 1000 } ], "max_steps": 5000, "num_train_epochs": 5, "total_flos": 2.263247661199196e+17, "trial_name": null, "trial_params": null }