prateeky2806's picture
Training in progress, step 1600
ee02f7a
{
"best_metric": 0.6245253086090088,
"best_model_checkpoint": "./output_v2/7b_cluster012_Nous-Hermes-llama-2-7b_partitioned_v3_standardized_012/checkpoint-1600",
"epoch": 1.8529241459177765,
"global_step": 1600,
"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": 0.7418,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 0.7269,
"step": 20
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 0.7061,
"step": 30
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 0.6809,
"step": 40
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 0.6742,
"step": 50
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 0.6789,
"step": 60
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 0.6628,
"step": 70
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 0.6995,
"step": 80
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 0.6774,
"step": 90
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 0.6716,
"step": 100
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 0.6628,
"step": 110
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 0.6696,
"step": 120
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 0.648,
"step": 130
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 0.6823,
"step": 140
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 0.6532,
"step": 150
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 0.669,
"step": 160
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 0.6639,
"step": 170
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 0.6582,
"step": 180
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 0.6381,
"step": 190
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 0.6342,
"step": 200
},
{
"epoch": 0.23,
"eval_loss": 0.6798371076583862,
"eval_runtime": 220.8775,
"eval_samples_per_second": 4.527,
"eval_steps_per_second": 2.264,
"step": 200
},
{
"epoch": 0.23,
"mmlu_eval_accuracy": 0.4661323913730994,
"mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
"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": 0.36363636363636365,
"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": 0.36363636363636365,
"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": 0.3902439024390244,
"mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
"mmlu_eval_accuracy_global_facts": 0.5,
"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": 0.6111111111111112,
"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": 0.4230769230769231,
"mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
"mmlu_eval_accuracy_high_school_psychology": 0.7166666666666667,
"mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
"mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
"mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769,
"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": 0.2727272727272727,
"mmlu_eval_accuracy_management": 0.6363636363636364,
"mmlu_eval_accuracy_marketing": 0.76,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
"mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
"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": 0.3225806451612903,
"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": 0.5454545454545454,
"mmlu_eval_accuracy_virology": 0.3888888888888889,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.1315238676556718,
"step": 200
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 0.6574,
"step": 210
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 0.6601,
"step": 220
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 0.6499,
"step": 230
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 0.6469,
"step": 240
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 0.6546,
"step": 250
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 0.6681,
"step": 260
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 0.6403,
"step": 270
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 0.676,
"step": 280
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 0.6548,
"step": 290
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 0.6394,
"step": 300
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 0.6173,
"step": 310
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 0.6425,
"step": 320
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 0.6379,
"step": 330
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 0.6191,
"step": 340
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 0.6645,
"step": 350
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 0.5998,
"step": 360
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 0.6402,
"step": 370
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 0.6291,
"step": 380
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 0.6202,
"step": 390
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 0.6532,
"step": 400
},
{
"epoch": 0.46,
"eval_loss": 0.6604505777359009,
"eval_runtime": 220.7584,
"eval_samples_per_second": 4.53,
"eval_steps_per_second": 2.265,
"step": 400
},
{
"epoch": 0.46,
"mmlu_eval_accuracy": 0.466397920486829,
"mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
"mmlu_eval_accuracy_anatomy": 0.5714285714285714,
"mmlu_eval_accuracy_astronomy": 0.375,
"mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
"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": 0.36363636363636365,
"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": 0.36363636363636365,
"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": 0.3902439024390244,
"mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
"mmlu_eval_accuracy_global_facts": 0.5,
"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": 0.5,
"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": 0.20689655172413793,
"mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
"mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
"mmlu_eval_accuracy_high_school_psychology": 0.75,
"mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
"mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
"mmlu_eval_accuracy_high_school_world_history": 0.5,
"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": 0.36363636363636365,
"mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
"mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
"mmlu_eval_accuracy_management": 0.5454545454545454,
"mmlu_eval_accuracy_marketing": 0.76,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.686046511627907,
"mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
"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": 0.3764705882352941,
"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": 0.5925925925925926,
"mmlu_eval_accuracy_sociology": 0.6818181818181818,
"mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
"mmlu_eval_accuracy_virology": 0.3888888888888889,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 0.9161743937308265,
"step": 400
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 0.6487,
"step": 410
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 0.6248,
"step": 420
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 0.6421,
"step": 430
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 0.6679,
"step": 440
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 0.6202,
"step": 450
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 0.7009,
"step": 460
},
{
"epoch": 0.54,
"learning_rate": 0.0002,
"loss": 0.627,
"step": 470
},
{
"epoch": 0.56,
"learning_rate": 0.0002,
"loss": 0.6294,
"step": 480
},
{
"epoch": 0.57,
"learning_rate": 0.0002,
"loss": 0.6254,
"step": 490
},
{
"epoch": 0.58,
"learning_rate": 0.0002,
"loss": 0.6174,
"step": 500
},
{
"epoch": 0.59,
"learning_rate": 0.0002,
"loss": 0.6225,
"step": 510
},
{
"epoch": 0.6,
"learning_rate": 0.0002,
"loss": 0.6703,
"step": 520
},
{
"epoch": 0.61,
"learning_rate": 0.0002,
"loss": 0.6232,
"step": 530
},
{
"epoch": 0.63,
"learning_rate": 0.0002,
"loss": 0.6606,
"step": 540
},
{
"epoch": 0.64,
"learning_rate": 0.0002,
"loss": 0.5992,
"step": 550
},
{
"epoch": 0.65,
"learning_rate": 0.0002,
"loss": 0.6118,
"step": 560
},
{
"epoch": 0.66,
"learning_rate": 0.0002,
"loss": 0.642,
"step": 570
},
{
"epoch": 0.67,
"learning_rate": 0.0002,
"loss": 0.6149,
"step": 580
},
{
"epoch": 0.68,
"learning_rate": 0.0002,
"loss": 0.6148,
"step": 590
},
{
"epoch": 0.69,
"learning_rate": 0.0002,
"loss": 0.6258,
"step": 600
},
{
"epoch": 0.69,
"eval_loss": 0.6469126343727112,
"eval_runtime": 220.793,
"eval_samples_per_second": 4.529,
"eval_steps_per_second": 2.265,
"step": 600
},
{
"epoch": 0.69,
"mmlu_eval_accuracy": 0.47179539026686096,
"mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
"mmlu_eval_accuracy_anatomy": 0.5714285714285714,
"mmlu_eval_accuracy_astronomy": 0.5,
"mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
"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": 0.36363636363636365,
"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": 0.375,
"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": 0.34375,
"mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
"mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_european_history": 0.5,
"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": 0.20689655172413793,
"mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
"mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
"mmlu_eval_accuracy_high_school_psychology": 0.7166666666666667,
"mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
"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": 0.5,
"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": 0.2727272727272727,
"mmlu_eval_accuracy_management": 0.5454545454545454,
"mmlu_eval_accuracy_marketing": 0.8,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
"mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.6363636363636364,
"mmlu_eval_accuracy_philosophy": 0.5294117647058824,
"mmlu_eval_accuracy_prehistory": 0.45714285714285713,
"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": 0.37681159420289856,
"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": 0.3888888888888889,
"mmlu_eval_accuracy_world_religions": 0.7894736842105263,
"mmlu_loss": 0.9272451737102889,
"step": 600
},
{
"epoch": 0.71,
"learning_rate": 0.0002,
"loss": 0.5855,
"step": 610
},
{
"epoch": 0.72,
"learning_rate": 0.0002,
"loss": 0.6213,
"step": 620
},
{
"epoch": 0.73,
"learning_rate": 0.0002,
"loss": 0.6392,
"step": 630
},
{
"epoch": 0.74,
"learning_rate": 0.0002,
"loss": 0.6488,
"step": 640
},
{
"epoch": 0.75,
"learning_rate": 0.0002,
"loss": 0.6377,
"step": 650
},
{
"epoch": 0.76,
"learning_rate": 0.0002,
"loss": 0.6119,
"step": 660
},
{
"epoch": 0.78,
"learning_rate": 0.0002,
"loss": 0.609,
"step": 670
},
{
"epoch": 0.79,
"learning_rate": 0.0002,
"loss": 0.6189,
"step": 680
},
{
"epoch": 0.8,
"learning_rate": 0.0002,
"loss": 0.6444,
"step": 690
},
{
"epoch": 0.81,
"learning_rate": 0.0002,
"loss": 0.6365,
"step": 700
},
{
"epoch": 0.82,
"learning_rate": 0.0002,
"loss": 0.6219,
"step": 710
},
{
"epoch": 0.83,
"learning_rate": 0.0002,
"loss": 0.626,
"step": 720
},
{
"epoch": 0.85,
"learning_rate": 0.0002,
"loss": 0.583,
"step": 730
},
{
"epoch": 0.86,
"learning_rate": 0.0002,
"loss": 0.6017,
"step": 740
},
{
"epoch": 0.87,
"learning_rate": 0.0002,
"loss": 0.635,
"step": 750
},
{
"epoch": 0.88,
"learning_rate": 0.0002,
"loss": 0.6216,
"step": 760
},
{
"epoch": 0.89,
"learning_rate": 0.0002,
"loss": 0.5759,
"step": 770
},
{
"epoch": 0.9,
"learning_rate": 0.0002,
"loss": 0.5969,
"step": 780
},
{
"epoch": 0.91,
"learning_rate": 0.0002,
"loss": 0.6033,
"step": 790
},
{
"epoch": 0.93,
"learning_rate": 0.0002,
"loss": 0.6057,
"step": 800
},
{
"epoch": 0.93,
"eval_loss": 0.6383256912231445,
"eval_runtime": 220.6008,
"eval_samples_per_second": 4.533,
"eval_steps_per_second": 2.267,
"step": 800
},
{
"epoch": 0.93,
"mmlu_eval_accuracy": 0.46887372107587305,
"mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
"mmlu_eval_accuracy_anatomy": 0.5,
"mmlu_eval_accuracy_astronomy": 0.3125,
"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": 0.36363636363636365,
"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": 0.375,
"mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
"mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
"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": 0.5555555555555556,
"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": 0.2413793103448276,
"mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
"mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
"mmlu_eval_accuracy_high_school_psychology": 0.75,
"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": 0.6956521739130435,
"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": 0.2727272727272727,
"mmlu_eval_accuracy_management": 0.6363636363636364,
"mmlu_eval_accuracy_marketing": 0.8,
"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": 0.5,
"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": 0.4166666666666667,
"mmlu_eval_accuracy_security_studies": 0.5555555555555556,
"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": 0.7894736842105263,
"mmlu_loss": 1.0044854765577975,
"step": 800
},
{
"epoch": 0.94,
"learning_rate": 0.0002,
"loss": 0.6071,
"step": 810
},
{
"epoch": 0.95,
"learning_rate": 0.0002,
"loss": 0.5743,
"step": 820
},
{
"epoch": 0.96,
"learning_rate": 0.0002,
"loss": 0.6023,
"step": 830
},
{
"epoch": 0.97,
"learning_rate": 0.0002,
"loss": 0.6004,
"step": 840
},
{
"epoch": 0.98,
"learning_rate": 0.0002,
"loss": 0.5766,
"step": 850
},
{
"epoch": 1.0,
"learning_rate": 0.0002,
"loss": 0.63,
"step": 860
},
{
"epoch": 1.01,
"learning_rate": 0.0002,
"loss": 0.5598,
"step": 870
},
{
"epoch": 1.02,
"learning_rate": 0.0002,
"loss": 0.5241,
"step": 880
},
{
"epoch": 1.03,
"learning_rate": 0.0002,
"loss": 0.5345,
"step": 890
},
{
"epoch": 1.04,
"learning_rate": 0.0002,
"loss": 0.5461,
"step": 900
},
{
"epoch": 1.05,
"learning_rate": 0.0002,
"loss": 0.5186,
"step": 910
},
{
"epoch": 1.07,
"learning_rate": 0.0002,
"loss": 0.5461,
"step": 920
},
{
"epoch": 1.08,
"learning_rate": 0.0002,
"loss": 0.5642,
"step": 930
},
{
"epoch": 1.09,
"learning_rate": 0.0002,
"loss": 0.5216,
"step": 940
},
{
"epoch": 1.1,
"learning_rate": 0.0002,
"loss": 0.5416,
"step": 950
},
{
"epoch": 1.11,
"learning_rate": 0.0002,
"loss": 0.5295,
"step": 960
},
{
"epoch": 1.12,
"learning_rate": 0.0002,
"loss": 0.5407,
"step": 970
},
{
"epoch": 1.13,
"learning_rate": 0.0002,
"loss": 0.547,
"step": 980
},
{
"epoch": 1.15,
"learning_rate": 0.0002,
"loss": 0.5566,
"step": 990
},
{
"epoch": 1.16,
"learning_rate": 0.0002,
"loss": 0.5895,
"step": 1000
},
{
"epoch": 1.16,
"eval_loss": 0.6423189043998718,
"eval_runtime": 221.1599,
"eval_samples_per_second": 4.522,
"eval_steps_per_second": 2.261,
"step": 1000
},
{
"epoch": 1.16,
"mmlu_eval_accuracy": 0.4701589950226067,
"mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
"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": 0.5,
"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": 0.38461538461538464,
"mmlu_eval_accuracy_econometrics": 0.16666666666666666,
"mmlu_eval_accuracy_electrical_engineering": 0.375,
"mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
"mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
"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": 0.5555555555555556,
"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": 0.2413793103448276,
"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": 0.6818181818181818,
"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": 0.7692307692307693,
"mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
"mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
"mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
"mmlu_eval_accuracy_management": 0.6363636363636364,
"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": 0.26,
"mmlu_eval_accuracy_nutrition": 0.6060606060606061,
"mmlu_eval_accuracy_philosophy": 0.4411764705882353,
"mmlu_eval_accuracy_prehistory": 0.45714285714285713,
"mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
"mmlu_eval_accuracy_professional_law": 0.34705882352941175,
"mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
"mmlu_eval_accuracy_professional_psychology": 0.391304347826087,
"mmlu_eval_accuracy_public_relations": 0.4166666666666667,
"mmlu_eval_accuracy_security_studies": 0.5185185185185185,
"mmlu_eval_accuracy_sociology": 0.7727272727272727,
"mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
"mmlu_eval_accuracy_virology": 0.3888888888888889,
"mmlu_eval_accuracy_world_religions": 0.7894736842105263,
"mmlu_loss": 0.9071105094748129,
"step": 1000
},
{
"epoch": 1.17,
"learning_rate": 0.0002,
"loss": 0.5462,
"step": 1010
},
{
"epoch": 1.18,
"learning_rate": 0.0002,
"loss": 0.5465,
"step": 1020
},
{
"epoch": 1.19,
"learning_rate": 0.0002,
"loss": 0.5454,
"step": 1030
},
{
"epoch": 1.2,
"learning_rate": 0.0002,
"loss": 0.5208,
"step": 1040
},
{
"epoch": 1.22,
"learning_rate": 0.0002,
"loss": 0.5409,
"step": 1050
},
{
"epoch": 1.23,
"learning_rate": 0.0002,
"loss": 0.5412,
"step": 1060
},
{
"epoch": 1.24,
"learning_rate": 0.0002,
"loss": 0.5251,
"step": 1070
},
{
"epoch": 1.25,
"learning_rate": 0.0002,
"loss": 0.5528,
"step": 1080
},
{
"epoch": 1.26,
"learning_rate": 0.0002,
"loss": 0.5449,
"step": 1090
},
{
"epoch": 1.27,
"learning_rate": 0.0002,
"loss": 0.5383,
"step": 1100
},
{
"epoch": 1.29,
"learning_rate": 0.0002,
"loss": 0.5544,
"step": 1110
},
{
"epoch": 1.3,
"learning_rate": 0.0002,
"loss": 0.5287,
"step": 1120
},
{
"epoch": 1.31,
"learning_rate": 0.0002,
"loss": 0.5309,
"step": 1130
},
{
"epoch": 1.32,
"learning_rate": 0.0002,
"loss": 0.5159,
"step": 1140
},
{
"epoch": 1.33,
"learning_rate": 0.0002,
"loss": 0.5343,
"step": 1150
},
{
"epoch": 1.34,
"learning_rate": 0.0002,
"loss": 0.5282,
"step": 1160
},
{
"epoch": 1.35,
"learning_rate": 0.0002,
"loss": 0.5315,
"step": 1170
},
{
"epoch": 1.37,
"learning_rate": 0.0002,
"loss": 0.5492,
"step": 1180
},
{
"epoch": 1.38,
"learning_rate": 0.0002,
"loss": 0.5429,
"step": 1190
},
{
"epoch": 1.39,
"learning_rate": 0.0002,
"loss": 0.5208,
"step": 1200
},
{
"epoch": 1.39,
"eval_loss": 0.6369008421897888,
"eval_runtime": 221.011,
"eval_samples_per_second": 4.525,
"eval_steps_per_second": 2.262,
"step": 1200
},
{
"epoch": 1.39,
"mmlu_eval_accuracy": 0.47643388223673894,
"mmlu_eval_accuracy_abstract_algebra": 0.45454545454545453,
"mmlu_eval_accuracy_anatomy": 0.6428571428571429,
"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": 0.5,
"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": 0.5454545454545454,
"mmlu_eval_accuracy_computer_security": 0.36363636363636365,
"mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
"mmlu_eval_accuracy_econometrics": 0.16666666666666666,
"mmlu_eval_accuracy_electrical_engineering": 0.375,
"mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
"mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
"mmlu_eval_accuracy_global_facts": 0.4,
"mmlu_eval_accuracy_high_school_biology": 0.40625,
"mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
"mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
"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": 0.5714285714285714,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
"mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
"mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
"mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
"mmlu_eval_accuracy_high_school_psychology": 0.75,
"mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
"mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
"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": 0.6923076923076923,
"mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
"mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
"mmlu_eval_accuracy_machine_learning": 0.09090909090909091,
"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": 0.4473684210526316,
"mmlu_eval_accuracy_moral_scenarios": 0.26,
"mmlu_eval_accuracy_nutrition": 0.6363636363636364,
"mmlu_eval_accuracy_philosophy": 0.4411764705882353,
"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": 0.45161290322580644,
"mmlu_eval_accuracy_professional_psychology": 0.37681159420289856,
"mmlu_eval_accuracy_public_relations": 0.4166666666666667,
"mmlu_eval_accuracy_security_studies": 0.5185185185185185,
"mmlu_eval_accuracy_sociology": 0.7727272727272727,
"mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
"mmlu_eval_accuracy_virology": 0.3888888888888889,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 0.9347293432364264,
"step": 1200
},
{
"epoch": 1.4,
"learning_rate": 0.0002,
"loss": 0.5522,
"step": 1210
},
{
"epoch": 1.41,
"learning_rate": 0.0002,
"loss": 0.5528,
"step": 1220
},
{
"epoch": 1.42,
"learning_rate": 0.0002,
"loss": 0.5721,
"step": 1230
},
{
"epoch": 1.44,
"learning_rate": 0.0002,
"loss": 0.5326,
"step": 1240
},
{
"epoch": 1.45,
"learning_rate": 0.0002,
"loss": 0.5497,
"step": 1250
},
{
"epoch": 1.46,
"learning_rate": 0.0002,
"loss": 0.5476,
"step": 1260
},
{
"epoch": 1.47,
"learning_rate": 0.0002,
"loss": 0.5285,
"step": 1270
},
{
"epoch": 1.48,
"learning_rate": 0.0002,
"loss": 0.5827,
"step": 1280
},
{
"epoch": 1.49,
"learning_rate": 0.0002,
"loss": 0.5332,
"step": 1290
},
{
"epoch": 1.51,
"learning_rate": 0.0002,
"loss": 0.5504,
"step": 1300
},
{
"epoch": 1.52,
"learning_rate": 0.0002,
"loss": 0.5909,
"step": 1310
},
{
"epoch": 1.53,
"learning_rate": 0.0002,
"loss": 0.5266,
"step": 1320
},
{
"epoch": 1.54,
"learning_rate": 0.0002,
"loss": 0.5614,
"step": 1330
},
{
"epoch": 1.55,
"learning_rate": 0.0002,
"loss": 0.517,
"step": 1340
},
{
"epoch": 1.56,
"learning_rate": 0.0002,
"loss": 0.5402,
"step": 1350
},
{
"epoch": 1.57,
"learning_rate": 0.0002,
"loss": 0.5511,
"step": 1360
},
{
"epoch": 1.59,
"learning_rate": 0.0002,
"loss": 0.5385,
"step": 1370
},
{
"epoch": 1.6,
"learning_rate": 0.0002,
"loss": 0.5431,
"step": 1380
},
{
"epoch": 1.61,
"learning_rate": 0.0002,
"loss": 0.5528,
"step": 1390
},
{
"epoch": 1.62,
"learning_rate": 0.0002,
"loss": 0.5535,
"step": 1400
},
{
"epoch": 1.62,
"eval_loss": 0.6297795176506042,
"eval_runtime": 221.766,
"eval_samples_per_second": 4.509,
"eval_steps_per_second": 2.255,
"step": 1400
},
{
"epoch": 1.62,
"mmlu_eval_accuracy": 0.46949284116120593,
"mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
"mmlu_eval_accuracy_anatomy": 0.7142857142857143,
"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": 0.5,
"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": 0.5454545454545454,
"mmlu_eval_accuracy_computer_security": 0.36363636363636365,
"mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
"mmlu_eval_accuracy_econometrics": 0.16666666666666666,
"mmlu_eval_accuracy_electrical_engineering": 0.375,
"mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
"mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
"mmlu_eval_accuracy_global_facts": 0.4,
"mmlu_eval_accuracy_high_school_biology": 0.40625,
"mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
"mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
"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": 0.20689655172413793,
"mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
"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": 0.7272727272727273,
"mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
"mmlu_eval_accuracy_human_aging": 0.6086956521739131,
"mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
"mmlu_eval_accuracy_international_law": 0.6923076923076923,
"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": 0.686046511627907,
"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": 0.2903225806451613,
"mmlu_eval_accuracy_professional_law": 0.32941176470588235,
"mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
"mmlu_eval_accuracy_professional_psychology": 0.36231884057971014,
"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": 0.5454545454545454,
"mmlu_eval_accuracy_virology": 0.3888888888888889,
"mmlu_eval_accuracy_world_religions": 0.6842105263157895,
"mmlu_loss": 0.917322517929516,
"step": 1400
},
{
"epoch": 1.63,
"learning_rate": 0.0002,
"loss": 0.5332,
"step": 1410
},
{
"epoch": 1.64,
"learning_rate": 0.0002,
"loss": 0.5563,
"step": 1420
},
{
"epoch": 1.66,
"learning_rate": 0.0002,
"loss": 0.5173,
"step": 1430
},
{
"epoch": 1.67,
"learning_rate": 0.0002,
"loss": 0.5432,
"step": 1440
},
{
"epoch": 1.68,
"learning_rate": 0.0002,
"loss": 0.5492,
"step": 1450
},
{
"epoch": 1.69,
"learning_rate": 0.0002,
"loss": 0.5482,
"step": 1460
},
{
"epoch": 1.7,
"learning_rate": 0.0002,
"loss": 0.5191,
"step": 1470
},
{
"epoch": 1.71,
"learning_rate": 0.0002,
"loss": 0.5578,
"step": 1480
},
{
"epoch": 1.73,
"learning_rate": 0.0002,
"loss": 0.5405,
"step": 1490
},
{
"epoch": 1.74,
"learning_rate": 0.0002,
"loss": 0.5499,
"step": 1500
},
{
"epoch": 1.75,
"learning_rate": 0.0002,
"loss": 0.5204,
"step": 1510
},
{
"epoch": 1.76,
"learning_rate": 0.0002,
"loss": 0.5327,
"step": 1520
},
{
"epoch": 1.77,
"learning_rate": 0.0002,
"loss": 0.5495,
"step": 1530
},
{
"epoch": 1.78,
"learning_rate": 0.0002,
"loss": 0.5527,
"step": 1540
},
{
"epoch": 1.8,
"learning_rate": 0.0002,
"loss": 0.5569,
"step": 1550
},
{
"epoch": 1.81,
"learning_rate": 0.0002,
"loss": 0.5626,
"step": 1560
},
{
"epoch": 1.82,
"learning_rate": 0.0002,
"loss": 0.5432,
"step": 1570
},
{
"epoch": 1.83,
"learning_rate": 0.0002,
"loss": 0.5432,
"step": 1580
},
{
"epoch": 1.84,
"learning_rate": 0.0002,
"loss": 0.5325,
"step": 1590
},
{
"epoch": 1.85,
"learning_rate": 0.0002,
"loss": 0.5136,
"step": 1600
},
{
"epoch": 1.85,
"eval_loss": 0.6245253086090088,
"eval_runtime": 222.3896,
"eval_samples_per_second": 4.497,
"eval_steps_per_second": 2.248,
"step": 1600
},
{
"epoch": 1.85,
"mmlu_eval_accuracy": 0.47687615586569587,
"mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
"mmlu_eval_accuracy_anatomy": 0.7142857142857143,
"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": 0.5625,
"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": 0.45454545454545453,
"mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
"mmlu_eval_accuracy_econometrics": 0.16666666666666666,
"mmlu_eval_accuracy_electrical_engineering": 0.3125,
"mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
"mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
"mmlu_eval_accuracy_global_facts": 0.4,
"mmlu_eval_accuracy_high_school_biology": 0.34375,
"mmlu_eval_accuracy_high_school_chemistry": 0.5,
"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": 0.5714285714285714,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
"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": 0.7833333333333333,
"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": 0.6923076923076923,
"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": 0.8,
"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": 0.6363636363636364,
"mmlu_eval_accuracy_philosophy": 0.5,
"mmlu_eval_accuracy_prehistory": 0.4,
"mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
"mmlu_eval_accuracy_professional_law": 0.34705882352941175,
"mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
"mmlu_eval_accuracy_professional_psychology": 0.34782608695652173,
"mmlu_eval_accuracy_public_relations": 0.5,
"mmlu_eval_accuracy_security_studies": 0.5555555555555556,
"mmlu_eval_accuracy_sociology": 0.7272727272727273,
"mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
"mmlu_eval_accuracy_virology": 0.3888888888888889,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.0008745580600384,
"step": 1600
}
],
"max_steps": 5000,
"num_train_epochs": 6,
"total_flos": 4.014370057017508e+17,
"trial_name": null,
"trial_params": null
}