pharaouk's picture
a
cf47524
raw
history blame
24.8 kB
{
"best_metric": 0.32424649596214294,
"best_model_checkpoint": "./output_v2/7b_cluster03_Nous-Hermes-llama-2-7b_partitioned_v3_standardized_03/checkpoint-200",
"epoch": 2.7467811158798283,
"global_step": 800,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 0.8627,
"step": 10
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 0.4107,
"step": 20
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 0.3691,
"step": 30
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 0.3979,
"step": 40
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 0.3495,
"step": 50
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 0.3693,
"step": 60
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 0.357,
"step": 70
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 0.3456,
"step": 80
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 0.3122,
"step": 90
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 0.3131,
"step": 100
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 0.3678,
"step": 110
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 0.3364,
"step": 120
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 0.324,
"step": 130
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 0.353,
"step": 140
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 0.3473,
"step": 150
},
{
"epoch": 0.55,
"learning_rate": 0.0002,
"loss": 0.363,
"step": 160
},
{
"epoch": 0.58,
"learning_rate": 0.0002,
"loss": 0.3192,
"step": 170
},
{
"epoch": 0.62,
"learning_rate": 0.0002,
"loss": 0.3125,
"step": 180
},
{
"epoch": 0.65,
"learning_rate": 0.0002,
"loss": 0.3001,
"step": 190
},
{
"epoch": 0.69,
"learning_rate": 0.0002,
"loss": 0.3164,
"step": 200
},
{
"epoch": 0.69,
"eval_loss": 0.32424649596214294,
"eval_runtime": 166.2846,
"eval_samples_per_second": 6.014,
"eval_steps_per_second": 3.007,
"step": 200
},
{
"epoch": 0.69,
"mmlu_eval_accuracy": 0.46206132856742516,
"mmlu_eval_accuracy_abstract_algebra": 0.09090909090909091,
"mmlu_eval_accuracy_anatomy": 0.6428571428571429,
"mmlu_eval_accuracy_astronomy": 0.4375,
"mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
"mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
"mmlu_eval_accuracy_college_biology": 0.375,
"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": 0.45454545454545453,
"mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
"mmlu_eval_accuracy_econometrics": 0.25,
"mmlu_eval_accuracy_electrical_engineering": 0.5625,
"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": 0.3181818181818182,
"mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
"mmlu_eval_accuracy_high_school_european_history": 0.5,
"mmlu_eval_accuracy_high_school_geography": 0.7272727272727273,
"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": 0.4230769230769231,
"mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
"mmlu_eval_accuracy_high_school_psychology": 0.7666666666666667,
"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": 0.6521739130434783,
"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": 0.36363636363636365,
"mmlu_eval_accuracy_marketing": 0.72,
"mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
"mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
"mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.6363636363636364,
"mmlu_eval_accuracy_philosophy": 0.47058823529411764,
"mmlu_eval_accuracy_prehistory": 0.4857142857142857,
"mmlu_eval_accuracy_professional_accounting": 0.1935483870967742,
"mmlu_eval_accuracy_professional_law": 0.3,
"mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
"mmlu_eval_accuracy_professional_psychology": 0.391304347826087,
"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": 0.3888888888888889,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 0.954094658943754,
"step": 200
},
{
"epoch": 0.72,
"learning_rate": 0.0002,
"loss": 0.3516,
"step": 210
},
{
"epoch": 0.76,
"learning_rate": 0.0002,
"loss": 0.2846,
"step": 220
},
{
"epoch": 0.79,
"learning_rate": 0.0002,
"loss": 0.3048,
"step": 230
},
{
"epoch": 0.82,
"learning_rate": 0.0002,
"loss": 0.315,
"step": 240
},
{
"epoch": 0.86,
"learning_rate": 0.0002,
"loss": 0.3229,
"step": 250
},
{
"epoch": 0.89,
"learning_rate": 0.0002,
"loss": 0.3218,
"step": 260
},
{
"epoch": 0.93,
"learning_rate": 0.0002,
"loss": 0.3188,
"step": 270
},
{
"epoch": 0.96,
"learning_rate": 0.0002,
"loss": 0.3246,
"step": 280
},
{
"epoch": 1.0,
"learning_rate": 0.0002,
"loss": 0.2986,
"step": 290
},
{
"epoch": 1.03,
"learning_rate": 0.0002,
"loss": 0.2905,
"step": 300
},
{
"epoch": 1.06,
"learning_rate": 0.0002,
"loss": 0.2847,
"step": 310
},
{
"epoch": 1.1,
"learning_rate": 0.0002,
"loss": 0.2384,
"step": 320
},
{
"epoch": 1.13,
"learning_rate": 0.0002,
"loss": 0.2815,
"step": 330
},
{
"epoch": 1.17,
"learning_rate": 0.0002,
"loss": 0.2737,
"step": 340
},
{
"epoch": 1.2,
"learning_rate": 0.0002,
"loss": 0.2985,
"step": 350
},
{
"epoch": 1.24,
"learning_rate": 0.0002,
"loss": 0.2747,
"step": 360
},
{
"epoch": 1.27,
"learning_rate": 0.0002,
"loss": 0.2691,
"step": 370
},
{
"epoch": 1.3,
"learning_rate": 0.0002,
"loss": 0.2823,
"step": 380
},
{
"epoch": 1.34,
"learning_rate": 0.0002,
"loss": 0.2389,
"step": 390
},
{
"epoch": 1.37,
"learning_rate": 0.0002,
"loss": 0.2552,
"step": 400
},
{
"epoch": 1.37,
"eval_loss": 0.3267834484577179,
"eval_runtime": 166.4022,
"eval_samples_per_second": 6.01,
"eval_steps_per_second": 3.005,
"step": 400
},
{
"epoch": 1.37,
"mmlu_eval_accuracy": 0.4687168428980155,
"mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
"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": 0.375,
"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": 0.45454545454545453,
"mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
"mmlu_eval_accuracy_econometrics": 0.25,
"mmlu_eval_accuracy_electrical_engineering": 0.5625,
"mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
"mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
"mmlu_eval_accuracy_global_facts": 0.5,
"mmlu_eval_accuracy_high_school_biology": 0.40625,
"mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
"mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
"mmlu_eval_accuracy_high_school_european_history": 0.5,
"mmlu_eval_accuracy_high_school_geography": 0.7272727272727273,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
"mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
"mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
"mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
"mmlu_eval_accuracy_high_school_psychology": 0.8,
"mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
"mmlu_eval_accuracy_high_school_us_history": 0.7727272727272727,
"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": 0.8461538461538461,
"mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
"mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
"mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
"mmlu_eval_accuracy_management": 0.45454545454545453,
"mmlu_eval_accuracy_marketing": 0.72,
"mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
"mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
"mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.696969696969697,
"mmlu_eval_accuracy_philosophy": 0.4117647058823529,
"mmlu_eval_accuracy_prehistory": 0.4857142857142857,
"mmlu_eval_accuracy_professional_accounting": 0.1935483870967742,
"mmlu_eval_accuracy_professional_law": 0.3235294117647059,
"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": 0.4444444444444444,
"mmlu_eval_accuracy_sociology": 0.7272727272727273,
"mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
"mmlu_eval_accuracy_virology": 0.4444444444444444,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 0.9237010620787311,
"step": 400
},
{
"epoch": 1.41,
"learning_rate": 0.0002,
"loss": 0.2629,
"step": 410
},
{
"epoch": 1.44,
"learning_rate": 0.0002,
"loss": 0.2455,
"step": 420
},
{
"epoch": 1.48,
"learning_rate": 0.0002,
"loss": 0.2857,
"step": 430
},
{
"epoch": 1.51,
"learning_rate": 0.0002,
"loss": 0.2728,
"step": 440
},
{
"epoch": 1.55,
"learning_rate": 0.0002,
"loss": 0.2814,
"step": 450
},
{
"epoch": 1.58,
"learning_rate": 0.0002,
"loss": 0.2993,
"step": 460
},
{
"epoch": 1.61,
"learning_rate": 0.0002,
"loss": 0.282,
"step": 470
},
{
"epoch": 1.65,
"learning_rate": 0.0002,
"loss": 0.269,
"step": 480
},
{
"epoch": 1.68,
"learning_rate": 0.0002,
"loss": 0.2534,
"step": 490
},
{
"epoch": 1.72,
"learning_rate": 0.0002,
"loss": 0.2965,
"step": 500
},
{
"epoch": 1.75,
"learning_rate": 0.0002,
"loss": 0.2872,
"step": 510
},
{
"epoch": 1.79,
"learning_rate": 0.0002,
"loss": 0.2742,
"step": 520
},
{
"epoch": 1.82,
"learning_rate": 0.0002,
"loss": 0.2568,
"step": 530
},
{
"epoch": 1.85,
"learning_rate": 0.0002,
"loss": 0.2722,
"step": 540
},
{
"epoch": 1.89,
"learning_rate": 0.0002,
"loss": 0.2502,
"step": 550
},
{
"epoch": 1.92,
"learning_rate": 0.0002,
"loss": 0.2755,
"step": 560
},
{
"epoch": 1.96,
"learning_rate": 0.0002,
"loss": 0.2695,
"step": 570
},
{
"epoch": 1.99,
"learning_rate": 0.0002,
"loss": 0.281,
"step": 580
},
{
"epoch": 2.03,
"learning_rate": 0.0002,
"loss": 0.2127,
"step": 590
},
{
"epoch": 2.06,
"learning_rate": 0.0002,
"loss": 0.1943,
"step": 600
},
{
"epoch": 2.06,
"eval_loss": 0.3318677246570587,
"eval_runtime": 166.3325,
"eval_samples_per_second": 6.012,
"eval_steps_per_second": 3.006,
"step": 600
},
{
"epoch": 2.06,
"mmlu_eval_accuracy": 0.463905720895551,
"mmlu_eval_accuracy_abstract_algebra": 0.09090909090909091,
"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": 0.25,
"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": 0.45454545454545453,
"mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
"mmlu_eval_accuracy_econometrics": 0.25,
"mmlu_eval_accuracy_electrical_engineering": 0.5625,
"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": 0.34375,
"mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
"mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
"mmlu_eval_accuracy_high_school_european_history": 0.4444444444444444,
"mmlu_eval_accuracy_high_school_geography": 0.7272727272727273,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
"mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
"mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
"mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
"mmlu_eval_accuracy_high_school_psychology": 0.75,
"mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
"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": 0.4166666666666667,
"mmlu_eval_accuracy_international_law": 0.8461538461538461,
"mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
"mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
"mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
"mmlu_eval_accuracy_management": 0.7272727272727273,
"mmlu_eval_accuracy_marketing": 0.72,
"mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
"mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
"mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
"mmlu_eval_accuracy_moral_scenarios": 0.23,
"mmlu_eval_accuracy_nutrition": 0.6363636363636364,
"mmlu_eval_accuracy_philosophy": 0.4411764705882353,
"mmlu_eval_accuracy_prehistory": 0.4857142857142857,
"mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
"mmlu_eval_accuracy_professional_law": 0.3352941176470588,
"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": 0.4444444444444444,
"mmlu_eval_accuracy_sociology": 0.7272727272727273,
"mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
"mmlu_eval_accuracy_virology": 0.4444444444444444,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 0.8365475967156825,
"step": 600
},
{
"epoch": 2.09,
"learning_rate": 0.0002,
"loss": 0.1929,
"step": 610
},
{
"epoch": 2.13,
"learning_rate": 0.0002,
"loss": 0.2081,
"step": 620
},
{
"epoch": 2.16,
"learning_rate": 0.0002,
"loss": 0.2088,
"step": 630
},
{
"epoch": 2.2,
"learning_rate": 0.0002,
"loss": 0.2116,
"step": 640
},
{
"epoch": 2.23,
"learning_rate": 0.0002,
"loss": 0.2112,
"step": 650
},
{
"epoch": 2.27,
"learning_rate": 0.0002,
"loss": 0.2223,
"step": 660
},
{
"epoch": 2.3,
"learning_rate": 0.0002,
"loss": 0.2396,
"step": 670
},
{
"epoch": 2.33,
"learning_rate": 0.0002,
"loss": 0.2324,
"step": 680
},
{
"epoch": 2.37,
"learning_rate": 0.0002,
"loss": 0.2156,
"step": 690
},
{
"epoch": 2.4,
"learning_rate": 0.0002,
"loss": 0.2321,
"step": 700
},
{
"epoch": 2.44,
"learning_rate": 0.0002,
"loss": 0.2162,
"step": 710
},
{
"epoch": 2.47,
"learning_rate": 0.0002,
"loss": 0.232,
"step": 720
},
{
"epoch": 2.51,
"learning_rate": 0.0002,
"loss": 0.2447,
"step": 730
},
{
"epoch": 2.54,
"learning_rate": 0.0002,
"loss": 0.2169,
"step": 740
},
{
"epoch": 2.58,
"learning_rate": 0.0002,
"loss": 0.2489,
"step": 750
},
{
"epoch": 2.61,
"learning_rate": 0.0002,
"loss": 0.2107,
"step": 760
},
{
"epoch": 2.64,
"learning_rate": 0.0002,
"loss": 0.2313,
"step": 770
},
{
"epoch": 2.68,
"learning_rate": 0.0002,
"loss": 0.213,
"step": 780
},
{
"epoch": 2.71,
"learning_rate": 0.0002,
"loss": 0.2313,
"step": 790
},
{
"epoch": 2.75,
"learning_rate": 0.0002,
"loss": 0.2378,
"step": 800
},
{
"epoch": 2.75,
"eval_loss": 0.33069494366645813,
"eval_runtime": 166.8414,
"eval_samples_per_second": 5.994,
"eval_steps_per_second": 2.997,
"step": 800
},
{
"epoch": 2.75,
"mmlu_eval_accuracy": 0.46892147950888885,
"mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
"mmlu_eval_accuracy_anatomy": 0.5,
"mmlu_eval_accuracy_astronomy": 0.4375,
"mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
"mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
"mmlu_eval_accuracy_college_biology": 0.375,
"mmlu_eval_accuracy_college_chemistry": 0.375,
"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": 0.45454545454545453,
"mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
"mmlu_eval_accuracy_econometrics": 0.25,
"mmlu_eval_accuracy_electrical_engineering": 0.5625,
"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": 0.34375,
"mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
"mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
"mmlu_eval_accuracy_high_school_european_history": 0.4444444444444444,
"mmlu_eval_accuracy_high_school_geography": 0.7272727272727273,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
"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": 0.7833333333333333,
"mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
"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": 0.4166666666666667,
"mmlu_eval_accuracy_international_law": 0.8461538461538461,
"mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
"mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
"mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
"mmlu_eval_accuracy_management": 0.5454545454545454,
"mmlu_eval_accuracy_marketing": 0.68,
"mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
"mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
"mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
"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": 0.0967741935483871,
"mmlu_eval_accuracy_professional_law": 0.31176470588235294,
"mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
"mmlu_eval_accuracy_professional_psychology": 0.34782608695652173,
"mmlu_eval_accuracy_public_relations": 0.5,
"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": 0.3888888888888889,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 0.8771737180163904,
"step": 800
}
],
"max_steps": 5000,
"num_train_epochs": 18,
"total_flos": 3.838045329904435e+16,
"trial_name": null,
"trial_params": null
}