prateeky2806's picture
Training in progress, step 1800
eadbb5f
{
"best_metric": 0.834297239780426,
"best_model_checkpoint": "./output_v2/7b_cluster031_Nous-Hermes-llama-2-7b_partitioned_v3_standardized_031/checkpoint-1400",
"epoch": 1.238390092879257,
"global_step": 1800,
"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": 0.9502,
"step": 10
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.9426,
"step": 20
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 0.9682,
"step": 30
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 0.8517,
"step": 40
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 0.8944,
"step": 50
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 0.8608,
"step": 60
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 0.915,
"step": 70
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 0.8349,
"step": 80
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 0.8912,
"step": 90
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 0.876,
"step": 100
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 0.8697,
"step": 110
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 0.8604,
"step": 120
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 0.8654,
"step": 130
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 0.8804,
"step": 140
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 0.8355,
"step": 150
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 0.8704,
"step": 160
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 0.8241,
"step": 170
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 0.8523,
"step": 180
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 0.845,
"step": 190
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 0.8681,
"step": 200
},
{
"epoch": 0.14,
"eval_loss": 0.8539559245109558,
"eval_runtime": 189.7934,
"eval_samples_per_second": 5.269,
"eval_steps_per_second": 2.634,
"step": 200
},
{
"epoch": 0.14,
"mmlu_eval_accuracy": 0.47295530665862845,
"mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
"mmlu_eval_accuracy_anatomy": 0.5,
"mmlu_eval_accuracy_astronomy": 0.375,
"mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
"mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
"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": 0.45454545454545453,
"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": 0.34146341463414637,
"mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
"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": 0.7777777777777778,
"mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
"mmlu_eval_accuracy_high_school_geography": 0.7272727272727273,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.3023255813953488,
"mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
"mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
"mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
"mmlu_eval_accuracy_high_school_psychology": 0.7166666666666667,
"mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
"mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
"mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769,
"mmlu_eval_accuracy_human_aging": 0.6956521739130435,
"mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
"mmlu_eval_accuracy_international_law": 0.6923076923076923,
"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": 0.6744186046511628,
"mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.5757575757575758,
"mmlu_eval_accuracy_philosophy": 0.4117647058823529,
"mmlu_eval_accuracy_prehistory": 0.42857142857142855,
"mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
"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": 0.6666666666666666,
"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": 1.2187759077891358,
"step": 200
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 0.8902,
"step": 210
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 0.9119,
"step": 220
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 0.8849,
"step": 230
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 0.8345,
"step": 240
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 0.8418,
"step": 250
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 0.8305,
"step": 260
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 0.8525,
"step": 270
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 0.8523,
"step": 280
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 0.8511,
"step": 290
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 0.8297,
"step": 300
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 0.8213,
"step": 310
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 0.8225,
"step": 320
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 0.8359,
"step": 330
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 0.8326,
"step": 340
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 0.832,
"step": 350
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 0.8208,
"step": 360
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 0.8427,
"step": 370
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 0.8812,
"step": 380
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 0.9016,
"step": 390
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 0.8694,
"step": 400
},
{
"epoch": 0.28,
"eval_loss": 0.8468822240829468,
"eval_runtime": 189.8705,
"eval_samples_per_second": 5.267,
"eval_steps_per_second": 2.633,
"step": 400
},
{
"epoch": 0.28,
"mmlu_eval_accuracy": 0.45864815481629223,
"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": 0.4375,
"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": 0.38461538461538464,
"mmlu_eval_accuracy_econometrics": 0.16666666666666666,
"mmlu_eval_accuracy_electrical_engineering": 0.375,
"mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
"mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
"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": 0.6111111111111112,
"mmlu_eval_accuracy_high_school_geography": 0.6818181818181818,
"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": 0.34615384615384615,
"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": 0.6818181818181818,
"mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
"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": 0.7272727272727273,
"mmlu_eval_accuracy_marketing": 0.68,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
"mmlu_eval_accuracy_moral_disputes": 0.5,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.5757575757575758,
"mmlu_eval_accuracy_philosophy": 0.4411764705882353,
"mmlu_eval_accuracy_prehistory": 0.42857142857142855,
"mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
"mmlu_eval_accuracy_professional_law": 0.3588235294117647,
"mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
"mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
"mmlu_eval_accuracy_public_relations": 0.5833333333333334,
"mmlu_eval_accuracy_security_studies": 0.5185185185185185,
"mmlu_eval_accuracy_sociology": 0.5454545454545454,
"mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
"mmlu_eval_accuracy_virology": 0.4444444444444444,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 0.9554785659084433,
"step": 400
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 0.8383,
"step": 410
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 0.8315,
"step": 420
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 0.8613,
"step": 430
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 0.865,
"step": 440
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 0.8028,
"step": 450
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 0.8557,
"step": 460
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 0.8834,
"step": 470
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 0.8468,
"step": 480
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 0.8556,
"step": 490
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 0.8591,
"step": 500
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 0.8683,
"step": 510
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 0.8278,
"step": 520
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 0.8208,
"step": 530
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 0.8372,
"step": 540
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 0.8144,
"step": 550
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 0.8783,
"step": 560
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 0.8337,
"step": 570
},
{
"epoch": 0.4,
"learning_rate": 0.0002,
"loss": 0.8572,
"step": 580
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 0.8275,
"step": 590
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 0.8599,
"step": 600
},
{
"epoch": 0.41,
"eval_loss": 0.84308922290802,
"eval_runtime": 189.8586,
"eval_samples_per_second": 5.267,
"eval_steps_per_second": 2.634,
"step": 600
},
{
"epoch": 0.41,
"mmlu_eval_accuracy": 0.45447446268908714,
"mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
"mmlu_eval_accuracy_anatomy": 0.5,
"mmlu_eval_accuracy_astronomy": 0.375,
"mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
"mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
"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": 0.3181818181818182,
"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": 0.375,
"mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
"mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
"mmlu_eval_accuracy_global_facts": 0.4,
"mmlu_eval_accuracy_high_school_biology": 0.3125,
"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": 0.7272727272727273,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.3023255813953488,
"mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
"mmlu_eval_accuracy_high_school_microeconomics": 0.23076923076923078,
"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": 0.5769230769230769,
"mmlu_eval_accuracy_human_aging": 0.6521739130434783,
"mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
"mmlu_eval_accuracy_international_law": 0.6923076923076923,
"mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
"mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
"mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
"mmlu_eval_accuracy_management": 0.45454545454545453,
"mmlu_eval_accuracy_marketing": 0.72,
"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": 0.22,
"mmlu_eval_accuracy_nutrition": 0.6363636363636364,
"mmlu_eval_accuracy_philosophy": 0.5294117647058824,
"mmlu_eval_accuracy_prehistory": 0.4,
"mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
"mmlu_eval_accuracy_professional_law": 0.3235294117647059,
"mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
"mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
"mmlu_eval_accuracy_public_relations": 0.6666666666666666,
"mmlu_eval_accuracy_security_studies": 0.5185185185185185,
"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": 0.9984549878082449,
"step": 600
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 0.8279,
"step": 610
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 0.8363,
"step": 620
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 0.8698,
"step": 630
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 0.8407,
"step": 640
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 0.8555,
"step": 650
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 0.8393,
"step": 660
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 0.8527,
"step": 670
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 0.8666,
"step": 680
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 0.8584,
"step": 690
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 0.8583,
"step": 700
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 0.8524,
"step": 710
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 0.8543,
"step": 720
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 0.8687,
"step": 730
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 0.8593,
"step": 740
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 0.8365,
"step": 750
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 0.8718,
"step": 760
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 0.8362,
"step": 770
},
{
"epoch": 0.54,
"learning_rate": 0.0002,
"loss": 0.8114,
"step": 780
},
{
"epoch": 0.54,
"learning_rate": 0.0002,
"loss": 0.8611,
"step": 790
},
{
"epoch": 0.55,
"learning_rate": 0.0002,
"loss": 0.8521,
"step": 800
},
{
"epoch": 0.55,
"eval_loss": 0.8391241431236267,
"eval_runtime": 189.8022,
"eval_samples_per_second": 5.269,
"eval_steps_per_second": 2.634,
"step": 800
},
{
"epoch": 0.55,
"mmlu_eval_accuracy": 0.4694480562846655,
"mmlu_eval_accuracy_abstract_algebra": 0.45454545454545453,
"mmlu_eval_accuracy_anatomy": 0.5714285714285714,
"mmlu_eval_accuracy_astronomy": 0.3125,
"mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
"mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
"mmlu_eval_accuracy_college_biology": 0.5,
"mmlu_eval_accuracy_college_chemistry": 0.25,
"mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
"mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
"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": 0.38461538461538464,
"mmlu_eval_accuracy_econometrics": 0.25,
"mmlu_eval_accuracy_electrical_engineering": 0.3125,
"mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
"mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
"mmlu_eval_accuracy_global_facts": 0.5,
"mmlu_eval_accuracy_high_school_biology": 0.4375,
"mmlu_eval_accuracy_high_school_chemistry": 0.22727272727272727,
"mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
"mmlu_eval_accuracy_high_school_european_history": 0.4444444444444444,
"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": 0.20689655172413793,
"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": 0.2608695652173913,
"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": 0.3333333333333333,
"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": 0.7272727272727273,
"mmlu_eval_accuracy_marketing": 0.72,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
"mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
"mmlu_eval_accuracy_moral_scenarios": 0.23,
"mmlu_eval_accuracy_nutrition": 0.6666666666666666,
"mmlu_eval_accuracy_philosophy": 0.4411764705882353,
"mmlu_eval_accuracy_prehistory": 0.5428571428571428,
"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": 0.4057971014492754,
"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": 0.6363636363636364,
"mmlu_eval_accuracy_virology": 0.3888888888888889,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.0093606002337006,
"step": 800
},
{
"epoch": 0.56,
"learning_rate": 0.0002,
"loss": 0.838,
"step": 810
},
{
"epoch": 0.56,
"learning_rate": 0.0002,
"loss": 0.8401,
"step": 820
},
{
"epoch": 0.57,
"learning_rate": 0.0002,
"loss": 0.8401,
"step": 830
},
{
"epoch": 0.58,
"learning_rate": 0.0002,
"loss": 0.8148,
"step": 840
},
{
"epoch": 0.58,
"learning_rate": 0.0002,
"loss": 0.8555,
"step": 850
},
{
"epoch": 0.59,
"learning_rate": 0.0002,
"loss": 0.8392,
"step": 860
},
{
"epoch": 0.6,
"learning_rate": 0.0002,
"loss": 0.8361,
"step": 870
},
{
"epoch": 0.61,
"learning_rate": 0.0002,
"loss": 0.8647,
"step": 880
},
{
"epoch": 0.61,
"learning_rate": 0.0002,
"loss": 0.8365,
"step": 890
},
{
"epoch": 0.62,
"learning_rate": 0.0002,
"loss": 0.8224,
"step": 900
},
{
"epoch": 0.63,
"learning_rate": 0.0002,
"loss": 0.8661,
"step": 910
},
{
"epoch": 0.63,
"learning_rate": 0.0002,
"loss": 0.8654,
"step": 920
},
{
"epoch": 0.64,
"learning_rate": 0.0002,
"loss": 0.8116,
"step": 930
},
{
"epoch": 0.65,
"learning_rate": 0.0002,
"loss": 0.8467,
"step": 940
},
{
"epoch": 0.65,
"learning_rate": 0.0002,
"loss": 0.8295,
"step": 950
},
{
"epoch": 0.66,
"learning_rate": 0.0002,
"loss": 0.8249,
"step": 960
},
{
"epoch": 0.67,
"learning_rate": 0.0002,
"loss": 0.8384,
"step": 970
},
{
"epoch": 0.67,
"learning_rate": 0.0002,
"loss": 0.8545,
"step": 980
},
{
"epoch": 0.68,
"learning_rate": 0.0002,
"loss": 0.8742,
"step": 990
},
{
"epoch": 0.69,
"learning_rate": 0.0002,
"loss": 0.8428,
"step": 1000
},
{
"epoch": 0.69,
"eval_loss": 0.8375168442726135,
"eval_runtime": 189.9373,
"eval_samples_per_second": 5.265,
"eval_steps_per_second": 2.632,
"step": 1000
},
{
"epoch": 0.69,
"mmlu_eval_accuracy": 0.4593129155936289,
"mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
"mmlu_eval_accuracy_anatomy": 0.6428571428571429,
"mmlu_eval_accuracy_astronomy": 0.3125,
"mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
"mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
"mmlu_eval_accuracy_college_biology": 0.5,
"mmlu_eval_accuracy_college_chemistry": 0.25,
"mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
"mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
"mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
"mmlu_eval_accuracy_college_physics": 0.45454545454545453,
"mmlu_eval_accuracy_computer_security": 0.45454545454545453,
"mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
"mmlu_eval_accuracy_econometrics": 0.25,
"mmlu_eval_accuracy_electrical_engineering": 0.375,
"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": 0.40625,
"mmlu_eval_accuracy_high_school_chemistry": 0.2727272727272727,
"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": 0.5714285714285714,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
"mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
"mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
"mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
"mmlu_eval_accuracy_high_school_psychology": 0.7166666666666667,
"mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
"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": 0.3333333333333333,
"mmlu_eval_accuracy_international_law": 0.8461538461538461,
"mmlu_eval_accuracy_jurisprudence": 0.18181818181818182,
"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": 0.6627906976744186,
"mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
"mmlu_eval_accuracy_moral_scenarios": 0.23,
"mmlu_eval_accuracy_nutrition": 0.6060606060606061,
"mmlu_eval_accuracy_philosophy": 0.4411764705882353,
"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": 0.45161290322580644,
"mmlu_eval_accuracy_professional_psychology": 0.42028985507246375,
"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": 0.4444444444444444,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 0.9438855402152159,
"step": 1000
},
{
"epoch": 0.69,
"learning_rate": 0.0002,
"loss": 0.8748,
"step": 1010
},
{
"epoch": 0.7,
"learning_rate": 0.0002,
"loss": 0.8454,
"step": 1020
},
{
"epoch": 0.71,
"learning_rate": 0.0002,
"loss": 0.8076,
"step": 1030
},
{
"epoch": 0.72,
"learning_rate": 0.0002,
"loss": 0.8591,
"step": 1040
},
{
"epoch": 0.72,
"learning_rate": 0.0002,
"loss": 0.8583,
"step": 1050
},
{
"epoch": 0.73,
"learning_rate": 0.0002,
"loss": 0.8769,
"step": 1060
},
{
"epoch": 0.74,
"learning_rate": 0.0002,
"loss": 0.8433,
"step": 1070
},
{
"epoch": 0.74,
"learning_rate": 0.0002,
"loss": 0.8534,
"step": 1080
},
{
"epoch": 0.75,
"learning_rate": 0.0002,
"loss": 0.8332,
"step": 1090
},
{
"epoch": 0.76,
"learning_rate": 0.0002,
"loss": 0.8177,
"step": 1100
},
{
"epoch": 0.76,
"learning_rate": 0.0002,
"loss": 0.8254,
"step": 1110
},
{
"epoch": 0.77,
"learning_rate": 0.0002,
"loss": 0.8064,
"step": 1120
},
{
"epoch": 0.78,
"learning_rate": 0.0002,
"loss": 0.8945,
"step": 1130
},
{
"epoch": 0.78,
"learning_rate": 0.0002,
"loss": 0.8316,
"step": 1140
},
{
"epoch": 0.79,
"learning_rate": 0.0002,
"loss": 0.8611,
"step": 1150
},
{
"epoch": 0.8,
"learning_rate": 0.0002,
"loss": 0.8823,
"step": 1160
},
{
"epoch": 0.8,
"learning_rate": 0.0002,
"loss": 0.802,
"step": 1170
},
{
"epoch": 0.81,
"learning_rate": 0.0002,
"loss": 0.8369,
"step": 1180
},
{
"epoch": 0.82,
"learning_rate": 0.0002,
"loss": 0.8221,
"step": 1190
},
{
"epoch": 0.83,
"learning_rate": 0.0002,
"loss": 0.8399,
"step": 1200
},
{
"epoch": 0.83,
"eval_loss": 0.8360938429832458,
"eval_runtime": 189.9481,
"eval_samples_per_second": 5.265,
"eval_steps_per_second": 2.632,
"step": 1200
},
{
"epoch": 0.83,
"mmlu_eval_accuracy": 0.4603456641212963,
"mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
"mmlu_eval_accuracy_anatomy": 0.5714285714285714,
"mmlu_eval_accuracy_astronomy": 0.3125,
"mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
"mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
"mmlu_eval_accuracy_college_biology": 0.5,
"mmlu_eval_accuracy_college_chemistry": 0.25,
"mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
"mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
"mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
"mmlu_eval_accuracy_college_physics": 0.45454545454545453,
"mmlu_eval_accuracy_computer_security": 0.45454545454545453,
"mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
"mmlu_eval_accuracy_econometrics": 0.25,
"mmlu_eval_accuracy_electrical_engineering": 0.3125,
"mmlu_eval_accuracy_elementary_mathematics": 0.2926829268292683,
"mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
"mmlu_eval_accuracy_global_facts": 0.5,
"mmlu_eval_accuracy_high_school_biology": 0.4375,
"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": 0.8181818181818182,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
"mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
"mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
"mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
"mmlu_eval_accuracy_high_school_psychology": 0.7166666666666667,
"mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
"mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
"mmlu_eval_accuracy_high_school_world_history": 0.5,
"mmlu_eval_accuracy_human_aging": 0.6521739130434783,
"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": 0.2727272727272727,
"mmlu_eval_accuracy_management": 0.7272727272727273,
"mmlu_eval_accuracy_marketing": 0.8,
"mmlu_eval_accuracy_medical_genetics": 0.6363636363636364,
"mmlu_eval_accuracy_miscellaneous": 0.686046511627907,
"mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
"mmlu_eval_accuracy_moral_scenarios": 0.26,
"mmlu_eval_accuracy_nutrition": 0.5454545454545454,
"mmlu_eval_accuracy_philosophy": 0.4411764705882353,
"mmlu_eval_accuracy_prehistory": 0.5142857142857142,
"mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
"mmlu_eval_accuracy_professional_law": 0.3058823529411765,
"mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
"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": 0.5454545454545454,
"mmlu_eval_accuracy_virology": 0.4444444444444444,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 0.9296886657240653,
"step": 1200
},
{
"epoch": 0.83,
"learning_rate": 0.0002,
"loss": 0.8106,
"step": 1210
},
{
"epoch": 0.84,
"learning_rate": 0.0002,
"loss": 0.8498,
"step": 1220
},
{
"epoch": 0.85,
"learning_rate": 0.0002,
"loss": 0.8757,
"step": 1230
},
{
"epoch": 0.85,
"learning_rate": 0.0002,
"loss": 0.8307,
"step": 1240
},
{
"epoch": 0.86,
"learning_rate": 0.0002,
"loss": 0.8217,
"step": 1250
},
{
"epoch": 0.87,
"learning_rate": 0.0002,
"loss": 0.8409,
"step": 1260
},
{
"epoch": 0.87,
"learning_rate": 0.0002,
"loss": 0.849,
"step": 1270
},
{
"epoch": 0.88,
"learning_rate": 0.0002,
"loss": 0.8687,
"step": 1280
},
{
"epoch": 0.89,
"learning_rate": 0.0002,
"loss": 0.8664,
"step": 1290
},
{
"epoch": 0.89,
"learning_rate": 0.0002,
"loss": 0.8184,
"step": 1300
},
{
"epoch": 0.9,
"learning_rate": 0.0002,
"loss": 0.8372,
"step": 1310
},
{
"epoch": 0.91,
"learning_rate": 0.0002,
"loss": 0.8708,
"step": 1320
},
{
"epoch": 0.92,
"learning_rate": 0.0002,
"loss": 0.8551,
"step": 1330
},
{
"epoch": 0.92,
"learning_rate": 0.0002,
"loss": 0.8566,
"step": 1340
},
{
"epoch": 0.93,
"learning_rate": 0.0002,
"loss": 0.8325,
"step": 1350
},
{
"epoch": 0.94,
"learning_rate": 0.0002,
"loss": 0.8474,
"step": 1360
},
{
"epoch": 0.94,
"learning_rate": 0.0002,
"loss": 0.802,
"step": 1370
},
{
"epoch": 0.95,
"learning_rate": 0.0002,
"loss": 0.8185,
"step": 1380
},
{
"epoch": 0.96,
"learning_rate": 0.0002,
"loss": 0.8681,
"step": 1390
},
{
"epoch": 0.96,
"learning_rate": 0.0002,
"loss": 0.8103,
"step": 1400
},
{
"epoch": 0.96,
"eval_loss": 0.834297239780426,
"eval_runtime": 189.9889,
"eval_samples_per_second": 5.263,
"eval_steps_per_second": 2.632,
"step": 1400
},
{
"epoch": 0.96,
"mmlu_eval_accuracy": 0.46764732076769827,
"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": 0.4482758620689655,
"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": 0.45454545454545453,
"mmlu_eval_accuracy_college_physics": 0.45454545454545453,
"mmlu_eval_accuracy_computer_security": 0.6363636363636364,
"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": 0.3170731707317073,
"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": 0.3181818181818182,
"mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
"mmlu_eval_accuracy_high_school_european_history": 0.5,
"mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
"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": 0.46153846153846156,
"mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
"mmlu_eval_accuracy_high_school_psychology": 0.7666666666666667,
"mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
"mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
"mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
"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": 0.5555555555555556,
"mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
"mmlu_eval_accuracy_management": 0.7272727272727273,
"mmlu_eval_accuracy_marketing": 0.8,
"mmlu_eval_accuracy_medical_genetics": 0.6363636363636364,
"mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
"mmlu_eval_accuracy_moral_disputes": 0.5,
"mmlu_eval_accuracy_moral_scenarios": 0.23,
"mmlu_eval_accuracy_nutrition": 0.6060606060606061,
"mmlu_eval_accuracy_philosophy": 0.5294117647058824,
"mmlu_eval_accuracy_prehistory": 0.5142857142857142,
"mmlu_eval_accuracy_professional_accounting": 0.1935483870967742,
"mmlu_eval_accuracy_professional_law": 0.3176470588235294,
"mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
"mmlu_eval_accuracy_professional_psychology": 0.42028985507246375,
"mmlu_eval_accuracy_public_relations": 0.5,
"mmlu_eval_accuracy_security_studies": 0.4074074074074074,
"mmlu_eval_accuracy_sociology": 0.6363636363636364,
"mmlu_eval_accuracy_us_foreign_policy": 0.45454545454545453,
"mmlu_eval_accuracy_virology": 0.4444444444444444,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 0.890019636367771,
"step": 1400
},
{
"epoch": 0.97,
"learning_rate": 0.0002,
"loss": 0.8596,
"step": 1410
},
{
"epoch": 0.98,
"learning_rate": 0.0002,
"loss": 0.8457,
"step": 1420
},
{
"epoch": 0.98,
"learning_rate": 0.0002,
"loss": 0.8271,
"step": 1430
},
{
"epoch": 0.99,
"learning_rate": 0.0002,
"loss": 0.8505,
"step": 1440
},
{
"epoch": 1.0,
"learning_rate": 0.0002,
"loss": 0.8166,
"step": 1450
},
{
"epoch": 1.0,
"learning_rate": 0.0002,
"loss": 0.8152,
"step": 1460
},
{
"epoch": 1.01,
"learning_rate": 0.0002,
"loss": 0.7307,
"step": 1470
},
{
"epoch": 1.02,
"learning_rate": 0.0002,
"loss": 0.7916,
"step": 1480
},
{
"epoch": 1.03,
"learning_rate": 0.0002,
"loss": 0.8018,
"step": 1490
},
{
"epoch": 1.03,
"learning_rate": 0.0002,
"loss": 0.7714,
"step": 1500
},
{
"epoch": 1.04,
"learning_rate": 0.0002,
"loss": 0.7902,
"step": 1510
},
{
"epoch": 1.05,
"learning_rate": 0.0002,
"loss": 0.7737,
"step": 1520
},
{
"epoch": 1.05,
"learning_rate": 0.0002,
"loss": 0.7792,
"step": 1530
},
{
"epoch": 1.06,
"learning_rate": 0.0002,
"loss": 0.8008,
"step": 1540
},
{
"epoch": 1.07,
"learning_rate": 0.0002,
"loss": 0.7707,
"step": 1550
},
{
"epoch": 1.07,
"learning_rate": 0.0002,
"loss": 0.7492,
"step": 1560
},
{
"epoch": 1.08,
"learning_rate": 0.0002,
"loss": 0.7548,
"step": 1570
},
{
"epoch": 1.09,
"learning_rate": 0.0002,
"loss": 0.7919,
"step": 1580
},
{
"epoch": 1.09,
"learning_rate": 0.0002,
"loss": 0.7812,
"step": 1590
},
{
"epoch": 1.1,
"learning_rate": 0.0002,
"loss": 0.7882,
"step": 1600
},
{
"epoch": 1.1,
"eval_loss": 0.839587390422821,
"eval_runtime": 190.009,
"eval_samples_per_second": 5.263,
"eval_steps_per_second": 2.631,
"step": 1600
},
{
"epoch": 1.1,
"mmlu_eval_accuracy": 0.46080769644355224,
"mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
"mmlu_eval_accuracy_anatomy": 0.6428571428571429,
"mmlu_eval_accuracy_astronomy": 0.3125,
"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": 0.36363636363636365,
"mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
"mmlu_eval_accuracy_college_physics": 0.5454545454545454,
"mmlu_eval_accuracy_computer_security": 0.36363636363636365,
"mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
"mmlu_eval_accuracy_econometrics": 0.25,
"mmlu_eval_accuracy_electrical_engineering": 0.25,
"mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
"mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
"mmlu_eval_accuracy_global_facts": 0.5,
"mmlu_eval_accuracy_high_school_biology": 0.40625,
"mmlu_eval_accuracy_high_school_chemistry": 0.2727272727272727,
"mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
"mmlu_eval_accuracy_high_school_european_history": 0.4444444444444444,
"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": 0.2413793103448276,
"mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
"mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
"mmlu_eval_accuracy_high_school_psychology": 0.7333333333333333,
"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": 0.6086956521739131,
"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": 0.5555555555555556,
"mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
"mmlu_eval_accuracy_management": 0.7272727272727273,
"mmlu_eval_accuracy_marketing": 0.8,
"mmlu_eval_accuracy_medical_genetics": 0.6363636363636364,
"mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
"mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
"mmlu_eval_accuracy_moral_scenarios": 0.22,
"mmlu_eval_accuracy_nutrition": 0.6363636363636364,
"mmlu_eval_accuracy_philosophy": 0.5,
"mmlu_eval_accuracy_prehistory": 0.5142857142857142,
"mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
"mmlu_eval_accuracy_professional_law": 0.3058823529411765,
"mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
"mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
"mmlu_eval_accuracy_public_relations": 0.5,
"mmlu_eval_accuracy_security_studies": 0.4444444444444444,
"mmlu_eval_accuracy_sociology": 0.6363636363636364,
"mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
"mmlu_eval_accuracy_virology": 0.3888888888888889,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.0650272156566616,
"step": 1600
},
{
"epoch": 1.11,
"learning_rate": 0.0002,
"loss": 0.7811,
"step": 1610
},
{
"epoch": 1.11,
"learning_rate": 0.0002,
"loss": 0.7891,
"step": 1620
},
{
"epoch": 1.12,
"learning_rate": 0.0002,
"loss": 0.7875,
"step": 1630
},
{
"epoch": 1.13,
"learning_rate": 0.0002,
"loss": 0.7795,
"step": 1640
},
{
"epoch": 1.14,
"learning_rate": 0.0002,
"loss": 0.7865,
"step": 1650
},
{
"epoch": 1.14,
"learning_rate": 0.0002,
"loss": 0.8181,
"step": 1660
},
{
"epoch": 1.15,
"learning_rate": 0.0002,
"loss": 0.7624,
"step": 1670
},
{
"epoch": 1.16,
"learning_rate": 0.0002,
"loss": 0.7858,
"step": 1680
},
{
"epoch": 1.16,
"learning_rate": 0.0002,
"loss": 0.8087,
"step": 1690
},
{
"epoch": 1.17,
"learning_rate": 0.0002,
"loss": 0.7708,
"step": 1700
},
{
"epoch": 1.18,
"learning_rate": 0.0002,
"loss": 0.7829,
"step": 1710
},
{
"epoch": 1.18,
"learning_rate": 0.0002,
"loss": 0.8028,
"step": 1720
},
{
"epoch": 1.19,
"learning_rate": 0.0002,
"loss": 0.7593,
"step": 1730
},
{
"epoch": 1.2,
"learning_rate": 0.0002,
"loss": 0.7813,
"step": 1740
},
{
"epoch": 1.2,
"learning_rate": 0.0002,
"loss": 0.7937,
"step": 1750
},
{
"epoch": 1.21,
"learning_rate": 0.0002,
"loss": 0.8233,
"step": 1760
},
{
"epoch": 1.22,
"learning_rate": 0.0002,
"loss": 0.7519,
"step": 1770
},
{
"epoch": 1.22,
"learning_rate": 0.0002,
"loss": 0.7887,
"step": 1780
},
{
"epoch": 1.23,
"learning_rate": 0.0002,
"loss": 0.7741,
"step": 1790
},
{
"epoch": 1.24,
"learning_rate": 0.0002,
"loss": 0.7614,
"step": 1800
},
{
"epoch": 1.24,
"eval_loss": 0.8406212329864502,
"eval_runtime": 189.952,
"eval_samples_per_second": 5.264,
"eval_steps_per_second": 2.632,
"step": 1800
},
{
"epoch": 1.24,
"mmlu_eval_accuracy": 0.46246385378342897,
"mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
"mmlu_eval_accuracy_anatomy": 0.5714285714285714,
"mmlu_eval_accuracy_astronomy": 0.3125,
"mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
"mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
"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": 0.4090909090909091,
"mmlu_eval_accuracy_college_physics": 0.45454545454545453,
"mmlu_eval_accuracy_computer_security": 0.45454545454545453,
"mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
"mmlu_eval_accuracy_econometrics": 0.16666666666666666,
"mmlu_eval_accuracy_electrical_engineering": 0.25,
"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": 0.40625,
"mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
"mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
"mmlu_eval_accuracy_high_school_european_history": 0.4444444444444444,
"mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.3023255813953488,
"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": 0.75,
"mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
"mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
"mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
"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": 0.18181818181818182,
"mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
"mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
"mmlu_eval_accuracy_management": 0.7272727272727273,
"mmlu_eval_accuracy_marketing": 0.76,
"mmlu_eval_accuracy_medical_genetics": 0.6363636363636364,
"mmlu_eval_accuracy_miscellaneous": 0.7093023255813954,
"mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
"mmlu_eval_accuracy_moral_scenarios": 0.23,
"mmlu_eval_accuracy_nutrition": 0.6363636363636364,
"mmlu_eval_accuracy_philosophy": 0.4117647058823529,
"mmlu_eval_accuracy_prehistory": 0.5142857142857142,
"mmlu_eval_accuracy_professional_accounting": 0.1935483870967742,
"mmlu_eval_accuracy_professional_law": 0.31176470588235294,
"mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
"mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
"mmlu_eval_accuracy_public_relations": 0.4166666666666667,
"mmlu_eval_accuracy_security_studies": 0.4444444444444444,
"mmlu_eval_accuracy_sociology": 0.6363636363636364,
"mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
"mmlu_eval_accuracy_virology": 0.5,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.0105097619889611,
"step": 1800
}
],
"max_steps": 5000,
"num_train_epochs": 4,
"total_flos": 3.24962836449067e+17,
"trial_name": null,
"trial_params": null
}