prateeky2806's picture
Training in progress, step 1800
c7808df
raw
history blame
55.5 kB
{
"best_metric": 0.48671162128448486,
"best_model_checkpoint": "./output_v2/7b_cluster09_Nous-Hermes-llama-2-7b_partitioned_v3_standardized_09/checkpoint-1600",
"epoch": 2.219482120838471,
"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": 0.597,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 0.5778,
"step": 20
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 0.5677,
"step": 30
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 0.5528,
"step": 40
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 0.5558,
"step": 50
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 0.5571,
"step": 60
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 0.5499,
"step": 70
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 0.5491,
"step": 80
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 0.5407,
"step": 90
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 0.5492,
"step": 100
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 0.5258,
"step": 110
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 0.5217,
"step": 120
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 0.538,
"step": 130
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 0.5265,
"step": 140
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 0.5344,
"step": 150
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 0.5361,
"step": 160
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 0.5186,
"step": 170
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 0.5312,
"step": 180
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 0.5395,
"step": 190
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 0.5399,
"step": 200
},
{
"epoch": 0.25,
"eval_loss": 0.533891499042511,
"eval_runtime": 249.6236,
"eval_samples_per_second": 4.006,
"eval_steps_per_second": 2.003,
"step": 200
},
{
"epoch": 0.25,
"mmlu_eval_accuracy": 0.46207163729626294,
"mmlu_eval_accuracy_abstract_algebra": 0.09090909090909091,
"mmlu_eval_accuracy_anatomy": 0.6428571428571429,
"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": 0.36363636363636365,
"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": 0.375,
"mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
"mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
"mmlu_eval_accuracy_global_facts": 0.5,
"mmlu_eval_accuracy_high_school_biology": 0.34375,
"mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
"mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
"mmlu_eval_accuracy_high_school_geography": 0.6363636363636364,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
"mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
"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": 0.2608695652173913,
"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": 0.45454545454545453,
"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": 0.6744186046511628,
"mmlu_eval_accuracy_moral_disputes": 0.5,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.6060606060606061,
"mmlu_eval_accuracy_philosophy": 0.47058823529411764,
"mmlu_eval_accuracy_prehistory": 0.4857142857142857,
"mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
"mmlu_eval_accuracy_professional_law": 0.34705882352941175,
"mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
"mmlu_eval_accuracy_professional_psychology": 0.37681159420289856,
"mmlu_eval_accuracy_public_relations": 0.4166666666666667,
"mmlu_eval_accuracy_security_studies": 0.48148148148148145,
"mmlu_eval_accuracy_sociology": 0.6363636363636364,
"mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
"mmlu_eval_accuracy_virology": 0.3888888888888889,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.267449478023046,
"step": 200
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 0.524,
"step": 210
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 0.5484,
"step": 220
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 0.5247,
"step": 230
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 0.5305,
"step": 240
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 0.5179,
"step": 250
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 0.5408,
"step": 260
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 0.5472,
"step": 270
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 0.5136,
"step": 280
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 0.5262,
"step": 290
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 0.5361,
"step": 300
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 0.5007,
"step": 310
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 0.5211,
"step": 320
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 0.5217,
"step": 330
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 0.5337,
"step": 340
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 0.5113,
"step": 350
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 0.518,
"step": 360
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 0.5151,
"step": 370
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 0.5133,
"step": 380
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 0.5083,
"step": 390
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 0.5235,
"step": 400
},
{
"epoch": 0.49,
"eval_loss": 0.5213926434516907,
"eval_runtime": 249.5749,
"eval_samples_per_second": 4.007,
"eval_steps_per_second": 2.003,
"step": 400
},
{
"epoch": 0.49,
"mmlu_eval_accuracy": 0.45812855653406065,
"mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
"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": 0.375,
"mmlu_eval_accuracy_college_chemistry": 0.125,
"mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
"mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
"mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
"mmlu_eval_accuracy_college_physics": 0.45454545454545453,
"mmlu_eval_accuracy_computer_security": 0.36363636363636365,
"mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
"mmlu_eval_accuracy_econometrics": 0.16666666666666666,
"mmlu_eval_accuracy_electrical_engineering": 0.375,
"mmlu_eval_accuracy_elementary_mathematics": 0.2682926829268293,
"mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
"mmlu_eval_accuracy_global_facts": 0.6,
"mmlu_eval_accuracy_high_school_biology": 0.375,
"mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
"mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
"mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
"mmlu_eval_accuracy_high_school_geography": 0.6818181818181818,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
"mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
"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": 0.34782608695652173,
"mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
"mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
"mmlu_eval_accuracy_human_aging": 0.7391304347826086,
"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": 0.45454545454545453,
"mmlu_eval_accuracy_marketing": 0.8,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
"mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.6060606060606061,
"mmlu_eval_accuracy_philosophy": 0.4411764705882353,
"mmlu_eval_accuracy_prehistory": 0.5428571428571428,
"mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
"mmlu_eval_accuracy_professional_law": 0.3352941176470588,
"mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
"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": 0.6363636363636364,
"mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
"mmlu_eval_accuracy_virology": 0.3333333333333333,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.1127305320092031,
"step": 400
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 0.5194,
"step": 410
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 0.5279,
"step": 420
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 0.5105,
"step": 430
},
{
"epoch": 0.54,
"learning_rate": 0.0002,
"loss": 0.5427,
"step": 440
},
{
"epoch": 0.55,
"learning_rate": 0.0002,
"loss": 0.5276,
"step": 450
},
{
"epoch": 0.57,
"learning_rate": 0.0002,
"loss": 0.4865,
"step": 460
},
{
"epoch": 0.58,
"learning_rate": 0.0002,
"loss": 0.5161,
"step": 470
},
{
"epoch": 0.59,
"learning_rate": 0.0002,
"loss": 0.513,
"step": 480
},
{
"epoch": 0.6,
"learning_rate": 0.0002,
"loss": 0.5284,
"step": 490
},
{
"epoch": 0.62,
"learning_rate": 0.0002,
"loss": 0.5101,
"step": 500
},
{
"epoch": 0.63,
"learning_rate": 0.0002,
"loss": 0.5218,
"step": 510
},
{
"epoch": 0.64,
"learning_rate": 0.0002,
"loss": 0.5087,
"step": 520
},
{
"epoch": 0.65,
"learning_rate": 0.0002,
"loss": 0.5157,
"step": 530
},
{
"epoch": 0.67,
"learning_rate": 0.0002,
"loss": 0.501,
"step": 540
},
{
"epoch": 0.68,
"learning_rate": 0.0002,
"loss": 0.508,
"step": 550
},
{
"epoch": 0.69,
"learning_rate": 0.0002,
"loss": 0.5199,
"step": 560
},
{
"epoch": 0.7,
"learning_rate": 0.0002,
"loss": 0.5043,
"step": 570
},
{
"epoch": 0.72,
"learning_rate": 0.0002,
"loss": 0.5069,
"step": 580
},
{
"epoch": 0.73,
"learning_rate": 0.0002,
"loss": 0.5258,
"step": 590
},
{
"epoch": 0.74,
"learning_rate": 0.0002,
"loss": 0.5189,
"step": 600
},
{
"epoch": 0.74,
"eval_loss": 0.5119001865386963,
"eval_runtime": 249.8867,
"eval_samples_per_second": 4.002,
"eval_steps_per_second": 2.001,
"step": 600
},
{
"epoch": 0.74,
"mmlu_eval_accuracy": 0.45806114323766334,
"mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
"mmlu_eval_accuracy_anatomy": 0.6428571428571429,
"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": 0.375,
"mmlu_eval_accuracy_college_chemistry": 0.125,
"mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
"mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
"mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
"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": 0.3170731707317073,
"mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
"mmlu_eval_accuracy_global_facts": 0.6,
"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": 0.5555555555555556,
"mmlu_eval_accuracy_high_school_geography": 0.6818181818181818,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
"mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
"mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
"mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
"mmlu_eval_accuracy_high_school_psychology": 0.7666666666666667,
"mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
"mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
"mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769,
"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": 0.45454545454545453,
"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": 0.6627906976744186,
"mmlu_eval_accuracy_moral_disputes": 0.5,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.5757575757575758,
"mmlu_eval_accuracy_philosophy": 0.4117647058823529,
"mmlu_eval_accuracy_prehistory": 0.4857142857142857,
"mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
"mmlu_eval_accuracy_professional_law": 0.3352941176470588,
"mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
"mmlu_eval_accuracy_professional_psychology": 0.36231884057971014,
"mmlu_eval_accuracy_public_relations": 0.4166666666666667,
"mmlu_eval_accuracy_security_studies": 0.5185185185185185,
"mmlu_eval_accuracy_sociology": 0.6363636363636364,
"mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
"mmlu_eval_accuracy_virology": 0.3333333333333333,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.1718710024425318,
"step": 600
},
{
"epoch": 0.75,
"learning_rate": 0.0002,
"loss": 0.5234,
"step": 610
},
{
"epoch": 0.76,
"learning_rate": 0.0002,
"loss": 0.5205,
"step": 620
},
{
"epoch": 0.78,
"learning_rate": 0.0002,
"loss": 0.5146,
"step": 630
},
{
"epoch": 0.79,
"learning_rate": 0.0002,
"loss": 0.5094,
"step": 640
},
{
"epoch": 0.8,
"learning_rate": 0.0002,
"loss": 0.4959,
"step": 650
},
{
"epoch": 0.81,
"learning_rate": 0.0002,
"loss": 0.5001,
"step": 660
},
{
"epoch": 0.83,
"learning_rate": 0.0002,
"loss": 0.5007,
"step": 670
},
{
"epoch": 0.84,
"learning_rate": 0.0002,
"loss": 0.5029,
"step": 680
},
{
"epoch": 0.85,
"learning_rate": 0.0002,
"loss": 0.5143,
"step": 690
},
{
"epoch": 0.86,
"learning_rate": 0.0002,
"loss": 0.4983,
"step": 700
},
{
"epoch": 0.88,
"learning_rate": 0.0002,
"loss": 0.4995,
"step": 710
},
{
"epoch": 0.89,
"learning_rate": 0.0002,
"loss": 0.5072,
"step": 720
},
{
"epoch": 0.9,
"learning_rate": 0.0002,
"loss": 0.499,
"step": 730
},
{
"epoch": 0.91,
"learning_rate": 0.0002,
"loss": 0.505,
"step": 740
},
{
"epoch": 0.92,
"learning_rate": 0.0002,
"loss": 0.4917,
"step": 750
},
{
"epoch": 0.94,
"learning_rate": 0.0002,
"loss": 0.4983,
"step": 760
},
{
"epoch": 0.95,
"learning_rate": 0.0002,
"loss": 0.4946,
"step": 770
},
{
"epoch": 0.96,
"learning_rate": 0.0002,
"loss": 0.4931,
"step": 780
},
{
"epoch": 0.97,
"learning_rate": 0.0002,
"loss": 0.4836,
"step": 790
},
{
"epoch": 0.99,
"learning_rate": 0.0002,
"loss": 0.5001,
"step": 800
},
{
"epoch": 0.99,
"eval_loss": 0.5022817254066467,
"eval_runtime": 249.7465,
"eval_samples_per_second": 4.004,
"eval_steps_per_second": 2.002,
"step": 800
},
{
"epoch": 0.99,
"mmlu_eval_accuracy": 0.45223086902107806,
"mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
"mmlu_eval_accuracy_anatomy": 0.6428571428571429,
"mmlu_eval_accuracy_astronomy": 0.4375,
"mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
"mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
"mmlu_eval_accuracy_college_biology": 0.4375,
"mmlu_eval_accuracy_college_chemistry": 0.125,
"mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
"mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
"mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
"mmlu_eval_accuracy_college_physics": 0.45454545454545453,
"mmlu_eval_accuracy_computer_security": 0.36363636363636365,
"mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
"mmlu_eval_accuracy_econometrics": 0.16666666666666666,
"mmlu_eval_accuracy_electrical_engineering": 0.375,
"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": 0.34375,
"mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
"mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
"mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
"mmlu_eval_accuracy_high_school_geography": 0.6363636363636364,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
"mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
"mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
"mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
"mmlu_eval_accuracy_high_school_psychology": 0.75,
"mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
"mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
"mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769,
"mmlu_eval_accuracy_human_aging": 0.7391304347826086,
"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": 0.18181818181818182,
"mmlu_eval_accuracy_management": 0.45454545454545453,
"mmlu_eval_accuracy_marketing": 0.8,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
"mmlu_eval_accuracy_moral_disputes": 0.5,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.6060606060606061,
"mmlu_eval_accuracy_philosophy": 0.4117647058823529,
"mmlu_eval_accuracy_prehistory": 0.4857142857142857,
"mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
"mmlu_eval_accuracy_professional_law": 0.35294117647058826,
"mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
"mmlu_eval_accuracy_professional_psychology": 0.36231884057971014,
"mmlu_eval_accuracy_public_relations": 0.4166666666666667,
"mmlu_eval_accuracy_security_studies": 0.5185185185185185,
"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": 0.6842105263157895,
"mmlu_loss": 1.064671132450938,
"step": 800
},
{
"epoch": 1.0,
"learning_rate": 0.0002,
"loss": 0.5135,
"step": 810
},
{
"epoch": 1.01,
"learning_rate": 0.0002,
"loss": 0.4532,
"step": 820
},
{
"epoch": 1.02,
"learning_rate": 0.0002,
"loss": 0.4483,
"step": 830
},
{
"epoch": 1.04,
"learning_rate": 0.0002,
"loss": 0.4507,
"step": 840
},
{
"epoch": 1.05,
"learning_rate": 0.0002,
"loss": 0.4572,
"step": 850
},
{
"epoch": 1.06,
"learning_rate": 0.0002,
"loss": 0.4346,
"step": 860
},
{
"epoch": 1.07,
"learning_rate": 0.0002,
"loss": 0.4306,
"step": 870
},
{
"epoch": 1.09,
"learning_rate": 0.0002,
"loss": 0.439,
"step": 880
},
{
"epoch": 1.1,
"learning_rate": 0.0002,
"loss": 0.4215,
"step": 890
},
{
"epoch": 1.11,
"learning_rate": 0.0002,
"loss": 0.4608,
"step": 900
},
{
"epoch": 1.12,
"learning_rate": 0.0002,
"loss": 0.4345,
"step": 910
},
{
"epoch": 1.13,
"learning_rate": 0.0002,
"loss": 0.422,
"step": 920
},
{
"epoch": 1.15,
"learning_rate": 0.0002,
"loss": 0.4444,
"step": 930
},
{
"epoch": 1.16,
"learning_rate": 0.0002,
"loss": 0.4649,
"step": 940
},
{
"epoch": 1.17,
"learning_rate": 0.0002,
"loss": 0.4508,
"step": 950
},
{
"epoch": 1.18,
"learning_rate": 0.0002,
"loss": 0.439,
"step": 960
},
{
"epoch": 1.2,
"learning_rate": 0.0002,
"loss": 0.4347,
"step": 970
},
{
"epoch": 1.21,
"learning_rate": 0.0002,
"loss": 0.4413,
"step": 980
},
{
"epoch": 1.22,
"learning_rate": 0.0002,
"loss": 0.4337,
"step": 990
},
{
"epoch": 1.23,
"learning_rate": 0.0002,
"loss": 0.4358,
"step": 1000
},
{
"epoch": 1.23,
"eval_loss": 0.5019292235374451,
"eval_runtime": 249.7097,
"eval_samples_per_second": 4.005,
"eval_steps_per_second": 2.002,
"step": 1000
},
{
"epoch": 1.23,
"mmlu_eval_accuracy": 0.46197732544268794,
"mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
"mmlu_eval_accuracy_anatomy": 0.6428571428571429,
"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": 0.5,
"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": 0.36363636363636365,
"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": 0.3170731707317073,
"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": 0.4444444444444444,
"mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
"mmlu_eval_accuracy_high_school_geography": 0.6363636363636364,
"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": 0.7333333333333333,
"mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
"mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
"mmlu_eval_accuracy_high_school_world_history": 0.6153846153846154,
"mmlu_eval_accuracy_human_aging": 0.6956521739130435,
"mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
"mmlu_eval_accuracy_international_law": 0.7692307692307693,
"mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
"mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
"mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
"mmlu_eval_accuracy_management": 0.7272727272727273,
"mmlu_eval_accuracy_marketing": 0.8,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
"mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.5757575757575758,
"mmlu_eval_accuracy_philosophy": 0.4117647058823529,
"mmlu_eval_accuracy_prehistory": 0.5428571428571428,
"mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
"mmlu_eval_accuracy_professional_law": 0.32941176470588235,
"mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
"mmlu_eval_accuracy_professional_psychology": 0.3188405797101449,
"mmlu_eval_accuracy_public_relations": 0.4166666666666667,
"mmlu_eval_accuracy_security_studies": 0.5185185185185185,
"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": 1.103624303117589,
"step": 1000
},
{
"epoch": 1.25,
"learning_rate": 0.0002,
"loss": 0.4428,
"step": 1010
},
{
"epoch": 1.26,
"learning_rate": 0.0002,
"loss": 0.4306,
"step": 1020
},
{
"epoch": 1.27,
"learning_rate": 0.0002,
"loss": 0.4585,
"step": 1030
},
{
"epoch": 1.28,
"learning_rate": 0.0002,
"loss": 0.4323,
"step": 1040
},
{
"epoch": 1.29,
"learning_rate": 0.0002,
"loss": 0.4333,
"step": 1050
},
{
"epoch": 1.31,
"learning_rate": 0.0002,
"loss": 0.4364,
"step": 1060
},
{
"epoch": 1.32,
"learning_rate": 0.0002,
"loss": 0.4256,
"step": 1070
},
{
"epoch": 1.33,
"learning_rate": 0.0002,
"loss": 0.4197,
"step": 1080
},
{
"epoch": 1.34,
"learning_rate": 0.0002,
"loss": 0.4382,
"step": 1090
},
{
"epoch": 1.36,
"learning_rate": 0.0002,
"loss": 0.4489,
"step": 1100
},
{
"epoch": 1.37,
"learning_rate": 0.0002,
"loss": 0.4152,
"step": 1110
},
{
"epoch": 1.38,
"learning_rate": 0.0002,
"loss": 0.425,
"step": 1120
},
{
"epoch": 1.39,
"learning_rate": 0.0002,
"loss": 0.4537,
"step": 1130
},
{
"epoch": 1.41,
"learning_rate": 0.0002,
"loss": 0.4496,
"step": 1140
},
{
"epoch": 1.42,
"learning_rate": 0.0002,
"loss": 0.4266,
"step": 1150
},
{
"epoch": 1.43,
"learning_rate": 0.0002,
"loss": 0.4449,
"step": 1160
},
{
"epoch": 1.44,
"learning_rate": 0.0002,
"loss": 0.4381,
"step": 1170
},
{
"epoch": 1.45,
"learning_rate": 0.0002,
"loss": 0.4272,
"step": 1180
},
{
"epoch": 1.47,
"learning_rate": 0.0002,
"loss": 0.4366,
"step": 1190
},
{
"epoch": 1.48,
"learning_rate": 0.0002,
"loss": 0.428,
"step": 1200
},
{
"epoch": 1.48,
"eval_loss": 0.4976211488246918,
"eval_runtime": 249.5918,
"eval_samples_per_second": 4.007,
"eval_steps_per_second": 2.003,
"step": 1200
},
{
"epoch": 1.48,
"mmlu_eval_accuracy": 0.4642275199494228,
"mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
"mmlu_eval_accuracy_anatomy": 0.6428571428571429,
"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": 0.5,
"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": 0.5454545454545454,
"mmlu_eval_accuracy_computer_security": 0.36363636363636365,
"mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
"mmlu_eval_accuracy_econometrics": 0.16666666666666666,
"mmlu_eval_accuracy_electrical_engineering": 0.4375,
"mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
"mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
"mmlu_eval_accuracy_global_facts": 0.6,
"mmlu_eval_accuracy_high_school_biology": 0.375,
"mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
"mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
"mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_geography": 0.6818181818181818,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
"mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
"mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
"mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
"mmlu_eval_accuracy_high_school_psychology": 0.7833333333333333,
"mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
"mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
"mmlu_eval_accuracy_high_school_world_history": 0.6153846153846154,
"mmlu_eval_accuracy_human_aging": 0.7391304347826086,
"mmlu_eval_accuracy_human_sexuality": 0.5,
"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": 0.18181818181818182,
"mmlu_eval_accuracy_management": 0.6363636363636364,
"mmlu_eval_accuracy_marketing": 0.76,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
"mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.48484848484848486,
"mmlu_eval_accuracy_philosophy": 0.38235294117647056,
"mmlu_eval_accuracy_prehistory": 0.5428571428571428,
"mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
"mmlu_eval_accuracy_professional_law": 0.34705882352941175,
"mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
"mmlu_eval_accuracy_professional_psychology": 0.34782608695652173,
"mmlu_eval_accuracy_public_relations": 0.4166666666666667,
"mmlu_eval_accuracy_security_studies": 0.5185185185185185,
"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": 1.0431902918079503,
"step": 1200
},
{
"epoch": 1.49,
"learning_rate": 0.0002,
"loss": 0.4388,
"step": 1210
},
{
"epoch": 1.5,
"learning_rate": 0.0002,
"loss": 0.4396,
"step": 1220
},
{
"epoch": 1.52,
"learning_rate": 0.0002,
"loss": 0.448,
"step": 1230
},
{
"epoch": 1.53,
"learning_rate": 0.0002,
"loss": 0.4294,
"step": 1240
},
{
"epoch": 1.54,
"learning_rate": 0.0002,
"loss": 0.427,
"step": 1250
},
{
"epoch": 1.55,
"learning_rate": 0.0002,
"loss": 0.4464,
"step": 1260
},
{
"epoch": 1.57,
"learning_rate": 0.0002,
"loss": 0.4319,
"step": 1270
},
{
"epoch": 1.58,
"learning_rate": 0.0002,
"loss": 0.4297,
"step": 1280
},
{
"epoch": 1.59,
"learning_rate": 0.0002,
"loss": 0.4266,
"step": 1290
},
{
"epoch": 1.6,
"learning_rate": 0.0002,
"loss": 0.4567,
"step": 1300
},
{
"epoch": 1.62,
"learning_rate": 0.0002,
"loss": 0.4496,
"step": 1310
},
{
"epoch": 1.63,
"learning_rate": 0.0002,
"loss": 0.4408,
"step": 1320
},
{
"epoch": 1.64,
"learning_rate": 0.0002,
"loss": 0.4487,
"step": 1330
},
{
"epoch": 1.65,
"learning_rate": 0.0002,
"loss": 0.4168,
"step": 1340
},
{
"epoch": 1.66,
"learning_rate": 0.0002,
"loss": 0.4418,
"step": 1350
},
{
"epoch": 1.68,
"learning_rate": 0.0002,
"loss": 0.4288,
"step": 1360
},
{
"epoch": 1.69,
"learning_rate": 0.0002,
"loss": 0.4305,
"step": 1370
},
{
"epoch": 1.7,
"learning_rate": 0.0002,
"loss": 0.425,
"step": 1380
},
{
"epoch": 1.71,
"learning_rate": 0.0002,
"loss": 0.4247,
"step": 1390
},
{
"epoch": 1.73,
"learning_rate": 0.0002,
"loss": 0.4276,
"step": 1400
},
{
"epoch": 1.73,
"eval_loss": 0.49078691005706787,
"eval_runtime": 249.9521,
"eval_samples_per_second": 4.001,
"eval_steps_per_second": 2.0,
"step": 1400
},
{
"epoch": 1.73,
"mmlu_eval_accuracy": 0.46579887678101295,
"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": 0.41379310344827586,
"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": 0.45454545454545453,
"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": 0.6,
"mmlu_eval_accuracy_high_school_biology": 0.375,
"mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
"mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
"mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
"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": 0.5,
"mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
"mmlu_eval_accuracy_high_school_psychology": 0.7833333333333333,
"mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
"mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
"mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769,
"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": 0.5454545454545454,
"mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
"mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
"mmlu_eval_accuracy_management": 0.45454545454545453,
"mmlu_eval_accuracy_marketing": 0.72,
"mmlu_eval_accuracy_medical_genetics": 0.6363636363636364,
"mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
"mmlu_eval_accuracy_moral_disputes": 0.5,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.5454545454545454,
"mmlu_eval_accuracy_philosophy": 0.38235294117647056,
"mmlu_eval_accuracy_prehistory": 0.5714285714285714,
"mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
"mmlu_eval_accuracy_professional_law": 0.35294117647058826,
"mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
"mmlu_eval_accuracy_professional_psychology": 0.3188405797101449,
"mmlu_eval_accuracy_public_relations": 0.4166666666666667,
"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": 1.1886581055839442,
"step": 1400
},
{
"epoch": 1.74,
"learning_rate": 0.0002,
"loss": 0.4448,
"step": 1410
},
{
"epoch": 1.75,
"learning_rate": 0.0002,
"loss": 0.4305,
"step": 1420
},
{
"epoch": 1.76,
"learning_rate": 0.0002,
"loss": 0.4262,
"step": 1430
},
{
"epoch": 1.78,
"learning_rate": 0.0002,
"loss": 0.4274,
"step": 1440
},
{
"epoch": 1.79,
"learning_rate": 0.0002,
"loss": 0.4375,
"step": 1450
},
{
"epoch": 1.8,
"learning_rate": 0.0002,
"loss": 0.4295,
"step": 1460
},
{
"epoch": 1.81,
"learning_rate": 0.0002,
"loss": 0.439,
"step": 1470
},
{
"epoch": 1.82,
"learning_rate": 0.0002,
"loss": 0.4182,
"step": 1480
},
{
"epoch": 1.84,
"learning_rate": 0.0002,
"loss": 0.4162,
"step": 1490
},
{
"epoch": 1.85,
"learning_rate": 0.0002,
"loss": 0.4348,
"step": 1500
},
{
"epoch": 1.86,
"learning_rate": 0.0002,
"loss": 0.4407,
"step": 1510
},
{
"epoch": 1.87,
"learning_rate": 0.0002,
"loss": 0.4213,
"step": 1520
},
{
"epoch": 1.89,
"learning_rate": 0.0002,
"loss": 0.4188,
"step": 1530
},
{
"epoch": 1.9,
"learning_rate": 0.0002,
"loss": 0.4591,
"step": 1540
},
{
"epoch": 1.91,
"learning_rate": 0.0002,
"loss": 0.4098,
"step": 1550
},
{
"epoch": 1.92,
"learning_rate": 0.0002,
"loss": 0.4331,
"step": 1560
},
{
"epoch": 1.94,
"learning_rate": 0.0002,
"loss": 0.4383,
"step": 1570
},
{
"epoch": 1.95,
"learning_rate": 0.0002,
"loss": 0.4334,
"step": 1580
},
{
"epoch": 1.96,
"learning_rate": 0.0002,
"loss": 0.4363,
"step": 1590
},
{
"epoch": 1.97,
"learning_rate": 0.0002,
"loss": 0.4227,
"step": 1600
},
{
"epoch": 1.97,
"eval_loss": 0.48671162128448486,
"eval_runtime": 249.6163,
"eval_samples_per_second": 4.006,
"eval_steps_per_second": 2.003,
"step": 1600
},
{
"epoch": 1.97,
"mmlu_eval_accuracy": 0.4663407571795711,
"mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
"mmlu_eval_accuracy_anatomy": 0.5714285714285714,
"mmlu_eval_accuracy_astronomy": 0.4375,
"mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
"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": 0.2727272727272727,
"mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
"mmlu_eval_accuracy_college_physics": 0.5454545454545454,
"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": 0.36585365853658536,
"mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
"mmlu_eval_accuracy_global_facts": 0.5,
"mmlu_eval_accuracy_high_school_biology": 0.375,
"mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
"mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
"mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
"mmlu_eval_accuracy_high_school_geography": 0.6818181818181818,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
"mmlu_eval_accuracy_high_school_mathematics": 0.1724137931034483,
"mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
"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": 0.7272727272727273,
"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": 0.45454545454545453,
"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": 0.6363636363636364,
"mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
"mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
"mmlu_eval_accuracy_moral_scenarios": 0.23,
"mmlu_eval_accuracy_nutrition": 0.48484848484848486,
"mmlu_eval_accuracy_philosophy": 0.3235294117647059,
"mmlu_eval_accuracy_prehistory": 0.5428571428571428,
"mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
"mmlu_eval_accuracy_professional_law": 0.32941176470588235,
"mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
"mmlu_eval_accuracy_professional_psychology": 0.30434782608695654,
"mmlu_eval_accuracy_public_relations": 0.4166666666666667,
"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": 0.3888888888888889,
"mmlu_eval_accuracy_world_religions": 0.6842105263157895,
"mmlu_loss": 1.050787725118682,
"step": 1600
},
{
"epoch": 1.99,
"learning_rate": 0.0002,
"loss": 0.4297,
"step": 1610
},
{
"epoch": 2.0,
"learning_rate": 0.0002,
"loss": 0.4376,
"step": 1620
},
{
"epoch": 2.01,
"learning_rate": 0.0002,
"loss": 0.3567,
"step": 1630
},
{
"epoch": 2.02,
"learning_rate": 0.0002,
"loss": 0.332,
"step": 1640
},
{
"epoch": 2.03,
"learning_rate": 0.0002,
"loss": 0.3284,
"step": 1650
},
{
"epoch": 2.05,
"learning_rate": 0.0002,
"loss": 0.3347,
"step": 1660
},
{
"epoch": 2.06,
"learning_rate": 0.0002,
"loss": 0.3498,
"step": 1670
},
{
"epoch": 2.07,
"learning_rate": 0.0002,
"loss": 0.3541,
"step": 1680
},
{
"epoch": 2.08,
"learning_rate": 0.0002,
"loss": 0.34,
"step": 1690
},
{
"epoch": 2.1,
"learning_rate": 0.0002,
"loss": 0.3507,
"step": 1700
},
{
"epoch": 2.11,
"learning_rate": 0.0002,
"loss": 0.3518,
"step": 1710
},
{
"epoch": 2.12,
"learning_rate": 0.0002,
"loss": 0.3515,
"step": 1720
},
{
"epoch": 2.13,
"learning_rate": 0.0002,
"loss": 0.3432,
"step": 1730
},
{
"epoch": 2.15,
"learning_rate": 0.0002,
"loss": 0.355,
"step": 1740
},
{
"epoch": 2.16,
"learning_rate": 0.0002,
"loss": 0.3463,
"step": 1750
},
{
"epoch": 2.17,
"learning_rate": 0.0002,
"loss": 0.337,
"step": 1760
},
{
"epoch": 2.18,
"learning_rate": 0.0002,
"loss": 0.3524,
"step": 1770
},
{
"epoch": 2.19,
"learning_rate": 0.0002,
"loss": 0.3414,
"step": 1780
},
{
"epoch": 2.21,
"learning_rate": 0.0002,
"loss": 0.3505,
"step": 1790
},
{
"epoch": 2.22,
"learning_rate": 0.0002,
"loss": 0.3371,
"step": 1800
},
{
"epoch": 2.22,
"eval_loss": 0.5045116543769836,
"eval_runtime": 249.7171,
"eval_samples_per_second": 4.005,
"eval_steps_per_second": 2.002,
"step": 1800
},
{
"epoch": 2.22,
"mmlu_eval_accuracy": 0.4648096685434819,
"mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
"mmlu_eval_accuracy_anatomy": 0.5714285714285714,
"mmlu_eval_accuracy_astronomy": 0.375,
"mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
"mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
"mmlu_eval_accuracy_college_biology": 0.4375,
"mmlu_eval_accuracy_college_chemistry": 0.125,
"mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
"mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
"mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
"mmlu_eval_accuracy_college_physics": 0.5454545454545454,
"mmlu_eval_accuracy_computer_security": 0.5454545454545454,
"mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
"mmlu_eval_accuracy_econometrics": 0.16666666666666666,
"mmlu_eval_accuracy_electrical_engineering": 0.4375,
"mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
"mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
"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": 0.6111111111111112,
"mmlu_eval_accuracy_high_school_geography": 0.7272727272727273,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
"mmlu_eval_accuracy_high_school_mathematics": 0.1724137931034483,
"mmlu_eval_accuracy_high_school_microeconomics": 0.5,
"mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
"mmlu_eval_accuracy_high_school_psychology": 0.75,
"mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
"mmlu_eval_accuracy_high_school_us_history": 0.8181818181818182,
"mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769,
"mmlu_eval_accuracy_human_aging": 0.7391304347826086,
"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": 0.2727272727272727,
"mmlu_eval_accuracy_management": 0.5454545454545454,
"mmlu_eval_accuracy_marketing": 0.68,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
"mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
"mmlu_eval_accuracy_moral_scenarios": 0.23,
"mmlu_eval_accuracy_nutrition": 0.5151515151515151,
"mmlu_eval_accuracy_philosophy": 0.4117647058823529,
"mmlu_eval_accuracy_prehistory": 0.5142857142857142,
"mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
"mmlu_eval_accuracy_professional_law": 0.3588235294117647,
"mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
"mmlu_eval_accuracy_professional_psychology": 0.3333333333333333,
"mmlu_eval_accuracy_public_relations": 0.3333333333333333,
"mmlu_eval_accuracy_security_studies": 0.5185185185185185,
"mmlu_eval_accuracy_sociology": 0.6363636363636364,
"mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
"mmlu_eval_accuracy_virology": 0.4444444444444444,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.1527424410141167,
"step": 1800
}
],
"max_steps": 5000,
"num_train_epochs": 7,
"total_flos": 4.892810169024184e+17,
"trial_name": null,
"trial_params": null
}