Farouk
commit files to HF hub
bfba9b4
{
"best_metric": 0.5720782279968262,
"best_model_checkpoint": "./output_v2/7b_cluster029_Nous-Hermes-llama-2-7b_partitioned_v3_standardized_029/checkpoint-800",
"epoch": 2.8368794326241136,
"global_step": 1400,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 0.6742,
"step": 10
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 0.6589,
"step": 20
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 0.6439,
"step": 30
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 0.611,
"step": 40
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 0.6127,
"step": 50
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 0.6089,
"step": 60
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 0.6102,
"step": 70
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 0.5998,
"step": 80
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 0.6031,
"step": 90
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 0.5942,
"step": 100
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 0.6057,
"step": 110
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 0.5994,
"step": 120
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 0.619,
"step": 130
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 0.6001,
"step": 140
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 0.6004,
"step": 150
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 0.6108,
"step": 160
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 0.5893,
"step": 170
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 0.5896,
"step": 180
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 0.5795,
"step": 190
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 0.587,
"step": 200
},
{
"epoch": 0.41,
"eval_loss": 0.6020743250846863,
"eval_runtime": 220.3104,
"eval_samples_per_second": 4.539,
"eval_steps_per_second": 2.27,
"step": 200
},
{
"epoch": 0.41,
"mmlu_eval_accuracy": 0.4675041957905457,
"mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
"mmlu_eval_accuracy_anatomy": 0.5,
"mmlu_eval_accuracy_astronomy": 0.4375,
"mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
"mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
"mmlu_eval_accuracy_college_biology": 0.4375,
"mmlu_eval_accuracy_college_chemistry": 0.125,
"mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
"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": 0.4375,
"mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
"mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
"mmlu_eval_accuracy_global_facts": 0.6,
"mmlu_eval_accuracy_high_school_biology": 0.34375,
"mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
"mmlu_eval_accuracy_high_school_computer_science": 0.7777777777777778,
"mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
"mmlu_eval_accuracy_high_school_geography": 0.6818181818181818,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.3023255813953488,
"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": 0.7333333333333333,
"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": 0.4166666666666667,
"mmlu_eval_accuracy_international_law": 0.6923076923076923,
"mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
"mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
"mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
"mmlu_eval_accuracy_management": 0.5454545454545454,
"mmlu_eval_accuracy_marketing": 0.72,
"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": 0.24,
"mmlu_eval_accuracy_nutrition": 0.6060606060606061,
"mmlu_eval_accuracy_philosophy": 0.5,
"mmlu_eval_accuracy_prehistory": 0.45714285714285713,
"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": 0.37681159420289856,
"mmlu_eval_accuracy_public_relations": 0.4166666666666667,
"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": 0.5,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.169526866031689,
"step": 200
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 0.6093,
"step": 210
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 0.5798,
"step": 220
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 0.5986,
"step": 230
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 0.5874,
"step": 240
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 0.5871,
"step": 250
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 0.5827,
"step": 260
},
{
"epoch": 0.55,
"learning_rate": 0.0002,
"loss": 0.5883,
"step": 270
},
{
"epoch": 0.57,
"learning_rate": 0.0002,
"loss": 0.593,
"step": 280
},
{
"epoch": 0.59,
"learning_rate": 0.0002,
"loss": 0.5939,
"step": 290
},
{
"epoch": 0.61,
"learning_rate": 0.0002,
"loss": 0.5893,
"step": 300
},
{
"epoch": 0.63,
"learning_rate": 0.0002,
"loss": 0.5809,
"step": 310
},
{
"epoch": 0.65,
"learning_rate": 0.0002,
"loss": 0.5839,
"step": 320
},
{
"epoch": 0.67,
"learning_rate": 0.0002,
"loss": 0.5898,
"step": 330
},
{
"epoch": 0.69,
"learning_rate": 0.0002,
"loss": 0.5787,
"step": 340
},
{
"epoch": 0.71,
"learning_rate": 0.0002,
"loss": 0.588,
"step": 350
},
{
"epoch": 0.73,
"learning_rate": 0.0002,
"loss": 0.5679,
"step": 360
},
{
"epoch": 0.75,
"learning_rate": 0.0002,
"loss": 0.5723,
"step": 370
},
{
"epoch": 0.77,
"learning_rate": 0.0002,
"loss": 0.5878,
"step": 380
},
{
"epoch": 0.79,
"learning_rate": 0.0002,
"loss": 0.582,
"step": 390
},
{
"epoch": 0.81,
"learning_rate": 0.0002,
"loss": 0.5803,
"step": 400
},
{
"epoch": 0.81,
"eval_loss": 0.5868815779685974,
"eval_runtime": 220.3563,
"eval_samples_per_second": 4.538,
"eval_steps_per_second": 2.269,
"step": 400
},
{
"epoch": 0.81,
"mmlu_eval_accuracy": 0.4696096782214941,
"mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
"mmlu_eval_accuracy_anatomy": 0.5,
"mmlu_eval_accuracy_astronomy": 0.4375,
"mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
"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": 0.36363636363636365,
"mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
"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": 0.4230769230769231,
"mmlu_eval_accuracy_econometrics": 0.16666666666666666,
"mmlu_eval_accuracy_electrical_engineering": 0.4375,
"mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
"mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
"mmlu_eval_accuracy_global_facts": 0.5,
"mmlu_eval_accuracy_high_school_biology": 0.34375,
"mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
"mmlu_eval_accuracy_high_school_computer_science": 0.7777777777777778,
"mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
"mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
"mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
"mmlu_eval_accuracy_high_school_microeconomics": 0.5,
"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": 0.5384615384615384,
"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": 0.36363636363636365,
"mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
"mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
"mmlu_eval_accuracy_management": 0.5454545454545454,
"mmlu_eval_accuracy_marketing": 0.72,
"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": 0.5,
"mmlu_eval_accuracy_prehistory": 0.45714285714285713,
"mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
"mmlu_eval_accuracy_professional_law": 0.3588235294117647,
"mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
"mmlu_eval_accuracy_professional_psychology": 0.37681159420289856,
"mmlu_eval_accuracy_public_relations": 0.4166666666666667,
"mmlu_eval_accuracy_security_studies": 0.5555555555555556,
"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": 1.0689004451892394,
"step": 400
},
{
"epoch": 0.83,
"learning_rate": 0.0002,
"loss": 0.5871,
"step": 410
},
{
"epoch": 0.85,
"learning_rate": 0.0002,
"loss": 0.5815,
"step": 420
},
{
"epoch": 0.87,
"learning_rate": 0.0002,
"loss": 0.5658,
"step": 430
},
{
"epoch": 0.89,
"learning_rate": 0.0002,
"loss": 0.5686,
"step": 440
},
{
"epoch": 0.91,
"learning_rate": 0.0002,
"loss": 0.585,
"step": 450
},
{
"epoch": 0.93,
"learning_rate": 0.0002,
"loss": 0.5673,
"step": 460
},
{
"epoch": 0.95,
"learning_rate": 0.0002,
"loss": 0.5691,
"step": 470
},
{
"epoch": 0.97,
"learning_rate": 0.0002,
"loss": 0.5693,
"step": 480
},
{
"epoch": 0.99,
"learning_rate": 0.0002,
"loss": 0.5749,
"step": 490
},
{
"epoch": 1.01,
"learning_rate": 0.0002,
"loss": 0.5457,
"step": 500
},
{
"epoch": 1.03,
"learning_rate": 0.0002,
"loss": 0.5199,
"step": 510
},
{
"epoch": 1.05,
"learning_rate": 0.0002,
"loss": 0.5157,
"step": 520
},
{
"epoch": 1.07,
"learning_rate": 0.0002,
"loss": 0.513,
"step": 530
},
{
"epoch": 1.09,
"learning_rate": 0.0002,
"loss": 0.4971,
"step": 540
},
{
"epoch": 1.11,
"learning_rate": 0.0002,
"loss": 0.504,
"step": 550
},
{
"epoch": 1.13,
"learning_rate": 0.0002,
"loss": 0.5191,
"step": 560
},
{
"epoch": 1.16,
"learning_rate": 0.0002,
"loss": 0.5196,
"step": 570
},
{
"epoch": 1.18,
"learning_rate": 0.0002,
"loss": 0.5236,
"step": 580
},
{
"epoch": 1.2,
"learning_rate": 0.0002,
"loss": 0.4866,
"step": 590
},
{
"epoch": 1.22,
"learning_rate": 0.0002,
"loss": 0.5147,
"step": 600
},
{
"epoch": 1.22,
"eval_loss": 0.5827956199645996,
"eval_runtime": 220.4163,
"eval_samples_per_second": 4.537,
"eval_steps_per_second": 2.268,
"step": 600
},
{
"epoch": 1.22,
"mmlu_eval_accuracy": 0.468505142513405,
"mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
"mmlu_eval_accuracy_anatomy": 0.5714285714285714,
"mmlu_eval_accuracy_astronomy": 0.5625,
"mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
"mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
"mmlu_eval_accuracy_college_biology": 0.4375,
"mmlu_eval_accuracy_college_chemistry": 0.25,
"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": 0.4375,
"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": 0.375,
"mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
"mmlu_eval_accuracy_high_school_computer_science": 0.7777777777777778,
"mmlu_eval_accuracy_high_school_european_history": 0.4444444444444444,
"mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
"mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
"mmlu_eval_accuracy_high_school_microeconomics": 0.5,
"mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
"mmlu_eval_accuracy_high_school_psychology": 0.7333333333333333,
"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": 0.7391304347826086,
"mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
"mmlu_eval_accuracy_international_law": 0.7692307692307693,
"mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
"mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
"mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
"mmlu_eval_accuracy_management": 0.45454545454545453,
"mmlu_eval_accuracy_marketing": 0.72,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
"mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.6363636363636364,
"mmlu_eval_accuracy_philosophy": 0.4411764705882353,
"mmlu_eval_accuracy_prehistory": 0.5142857142857142,
"mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
"mmlu_eval_accuracy_professional_law": 0.32941176470588235,
"mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
"mmlu_eval_accuracy_professional_psychology": 0.37681159420289856,
"mmlu_eval_accuracy_public_relations": 0.4166666666666667,
"mmlu_eval_accuracy_security_studies": 0.6296296296296297,
"mmlu_eval_accuracy_sociology": 0.6818181818181818,
"mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
"mmlu_eval_accuracy_virology": 0.4444444444444444,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.0294917702441426,
"step": 600
},
{
"epoch": 1.24,
"learning_rate": 0.0002,
"loss": 0.5055,
"step": 610
},
{
"epoch": 1.26,
"learning_rate": 0.0002,
"loss": 0.4929,
"step": 620
},
{
"epoch": 1.28,
"learning_rate": 0.0002,
"loss": 0.5117,
"step": 630
},
{
"epoch": 1.3,
"learning_rate": 0.0002,
"loss": 0.5153,
"step": 640
},
{
"epoch": 1.32,
"learning_rate": 0.0002,
"loss": 0.5119,
"step": 650
},
{
"epoch": 1.34,
"learning_rate": 0.0002,
"loss": 0.5098,
"step": 660
},
{
"epoch": 1.36,
"learning_rate": 0.0002,
"loss": 0.4965,
"step": 670
},
{
"epoch": 1.38,
"learning_rate": 0.0002,
"loss": 0.5085,
"step": 680
},
{
"epoch": 1.4,
"learning_rate": 0.0002,
"loss": 0.5002,
"step": 690
},
{
"epoch": 1.42,
"learning_rate": 0.0002,
"loss": 0.4984,
"step": 700
},
{
"epoch": 1.44,
"learning_rate": 0.0002,
"loss": 0.5161,
"step": 710
},
{
"epoch": 1.46,
"learning_rate": 0.0002,
"loss": 0.524,
"step": 720
},
{
"epoch": 1.48,
"learning_rate": 0.0002,
"loss": 0.5245,
"step": 730
},
{
"epoch": 1.5,
"learning_rate": 0.0002,
"loss": 0.5011,
"step": 740
},
{
"epoch": 1.52,
"learning_rate": 0.0002,
"loss": 0.5004,
"step": 750
},
{
"epoch": 1.54,
"learning_rate": 0.0002,
"loss": 0.4794,
"step": 760
},
{
"epoch": 1.56,
"learning_rate": 0.0002,
"loss": 0.499,
"step": 770
},
{
"epoch": 1.58,
"learning_rate": 0.0002,
"loss": 0.5247,
"step": 780
},
{
"epoch": 1.6,
"learning_rate": 0.0002,
"loss": 0.5012,
"step": 790
},
{
"epoch": 1.62,
"learning_rate": 0.0002,
"loss": 0.4885,
"step": 800
},
{
"epoch": 1.62,
"eval_loss": 0.5720782279968262,
"eval_runtime": 220.4387,
"eval_samples_per_second": 4.536,
"eval_steps_per_second": 2.268,
"step": 800
},
{
"epoch": 1.62,
"mmlu_eval_accuracy": 0.4778576165582804,
"mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
"mmlu_eval_accuracy_anatomy": 0.5714285714285714,
"mmlu_eval_accuracy_astronomy": 0.5625,
"mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
"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": 0.36363636363636365,
"mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
"mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
"mmlu_eval_accuracy_college_physics": 0.45454545454545453,
"mmlu_eval_accuracy_computer_security": 0.36363636363636365,
"mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
"mmlu_eval_accuracy_econometrics": 0.16666666666666666,
"mmlu_eval_accuracy_electrical_engineering": 0.4375,
"mmlu_eval_accuracy_elementary_mathematics": 0.2926829268292683,
"mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
"mmlu_eval_accuracy_global_facts": 0.6,
"mmlu_eval_accuracy_high_school_biology": 0.40625,
"mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
"mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_european_history": 0.5,
"mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
"mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
"mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
"mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
"mmlu_eval_accuracy_high_school_psychology": 0.7,
"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": 0.7391304347826086,
"mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
"mmlu_eval_accuracy_international_law": 0.7692307692307693,
"mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
"mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
"mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
"mmlu_eval_accuracy_management": 0.5454545454545454,
"mmlu_eval_accuracy_marketing": 0.72,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
"mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.6060606060606061,
"mmlu_eval_accuracy_philosophy": 0.4411764705882353,
"mmlu_eval_accuracy_prehistory": 0.5428571428571428,
"mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
"mmlu_eval_accuracy_professional_law": 0.3352941176470588,
"mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
"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": 0.6363636363636364,
"mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
"mmlu_eval_accuracy_virology": 0.5,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.0124665262474248,
"step": 800
},
{
"epoch": 1.64,
"learning_rate": 0.0002,
"loss": 0.4962,
"step": 810
},
{
"epoch": 1.66,
"learning_rate": 0.0002,
"loss": 0.5166,
"step": 820
},
{
"epoch": 1.68,
"learning_rate": 0.0002,
"loss": 0.4949,
"step": 830
},
{
"epoch": 1.7,
"learning_rate": 0.0002,
"loss": 0.5175,
"step": 840
},
{
"epoch": 1.72,
"learning_rate": 0.0002,
"loss": 0.5084,
"step": 850
},
{
"epoch": 1.74,
"learning_rate": 0.0002,
"loss": 0.4876,
"step": 860
},
{
"epoch": 1.76,
"learning_rate": 0.0002,
"loss": 0.5081,
"step": 870
},
{
"epoch": 1.78,
"learning_rate": 0.0002,
"loss": 0.5069,
"step": 880
},
{
"epoch": 1.8,
"learning_rate": 0.0002,
"loss": 0.4948,
"step": 890
},
{
"epoch": 1.82,
"learning_rate": 0.0002,
"loss": 0.5043,
"step": 900
},
{
"epoch": 1.84,
"learning_rate": 0.0002,
"loss": 0.4945,
"step": 910
},
{
"epoch": 1.86,
"learning_rate": 0.0002,
"loss": 0.4856,
"step": 920
},
{
"epoch": 1.88,
"learning_rate": 0.0002,
"loss": 0.5026,
"step": 930
},
{
"epoch": 1.9,
"learning_rate": 0.0002,
"loss": 0.5059,
"step": 940
},
{
"epoch": 1.93,
"learning_rate": 0.0002,
"loss": 0.4825,
"step": 950
},
{
"epoch": 1.95,
"learning_rate": 0.0002,
"loss": 0.5058,
"step": 960
},
{
"epoch": 1.97,
"learning_rate": 0.0002,
"loss": 0.4945,
"step": 970
},
{
"epoch": 1.99,
"learning_rate": 0.0002,
"loss": 0.5196,
"step": 980
},
{
"epoch": 2.01,
"learning_rate": 0.0002,
"loss": 0.4916,
"step": 990
},
{
"epoch": 2.03,
"learning_rate": 0.0002,
"loss": 0.408,
"step": 1000
},
{
"epoch": 2.03,
"eval_loss": 0.5841771364212036,
"eval_runtime": 220.5232,
"eval_samples_per_second": 4.535,
"eval_steps_per_second": 2.267,
"step": 1000
},
{
"epoch": 2.03,
"mmlu_eval_accuracy": 0.47790692788177713,
"mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
"mmlu_eval_accuracy_anatomy": 0.5,
"mmlu_eval_accuracy_astronomy": 0.5625,
"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": 0.25,
"mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
"mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
"mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
"mmlu_eval_accuracy_college_physics": 0.45454545454545453,
"mmlu_eval_accuracy_computer_security": 0.36363636363636365,
"mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
"mmlu_eval_accuracy_econometrics": 0.16666666666666666,
"mmlu_eval_accuracy_electrical_engineering": 0.4375,
"mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
"mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
"mmlu_eval_accuracy_global_facts": 0.5,
"mmlu_eval_accuracy_high_school_biology": 0.34375,
"mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
"mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
"mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
"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": 0.5384615384615384,
"mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
"mmlu_eval_accuracy_high_school_psychology": 0.7166666666666667,
"mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
"mmlu_eval_accuracy_high_school_us_history": 0.5454545454545454,
"mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
"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": 0.18181818181818182,
"mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
"mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
"mmlu_eval_accuracy_management": 0.45454545454545453,
"mmlu_eval_accuracy_marketing": 0.72,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
"mmlu_eval_accuracy_moral_disputes": 0.5,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.6363636363636364,
"mmlu_eval_accuracy_philosophy": 0.4411764705882353,
"mmlu_eval_accuracy_prehistory": 0.5428571428571428,
"mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
"mmlu_eval_accuracy_professional_law": 0.36470588235294116,
"mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
"mmlu_eval_accuracy_professional_psychology": 0.391304347826087,
"mmlu_eval_accuracy_public_relations": 0.5,
"mmlu_eval_accuracy_security_studies": 0.6296296296296297,
"mmlu_eval_accuracy_sociology": 0.7272727272727273,
"mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
"mmlu_eval_accuracy_virology": 0.5555555555555556,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.051499547013726,
"step": 1000
},
{
"epoch": 2.05,
"learning_rate": 0.0002,
"loss": 0.383,
"step": 1010
},
{
"epoch": 2.07,
"learning_rate": 0.0002,
"loss": 0.4062,
"step": 1020
},
{
"epoch": 2.09,
"learning_rate": 0.0002,
"loss": 0.4133,
"step": 1030
},
{
"epoch": 2.11,
"learning_rate": 0.0002,
"loss": 0.4063,
"step": 1040
},
{
"epoch": 2.13,
"learning_rate": 0.0002,
"loss": 0.4027,
"step": 1050
},
{
"epoch": 2.15,
"learning_rate": 0.0002,
"loss": 0.3987,
"step": 1060
},
{
"epoch": 2.17,
"learning_rate": 0.0002,
"loss": 0.4103,
"step": 1070
},
{
"epoch": 2.19,
"learning_rate": 0.0002,
"loss": 0.417,
"step": 1080
},
{
"epoch": 2.21,
"learning_rate": 0.0002,
"loss": 0.4004,
"step": 1090
},
{
"epoch": 2.23,
"learning_rate": 0.0002,
"loss": 0.4134,
"step": 1100
},
{
"epoch": 2.25,
"learning_rate": 0.0002,
"loss": 0.4072,
"step": 1110
},
{
"epoch": 2.27,
"learning_rate": 0.0002,
"loss": 0.4016,
"step": 1120
},
{
"epoch": 2.29,
"learning_rate": 0.0002,
"loss": 0.4238,
"step": 1130
},
{
"epoch": 2.31,
"learning_rate": 0.0002,
"loss": 0.3953,
"step": 1140
},
{
"epoch": 2.33,
"learning_rate": 0.0002,
"loss": 0.3951,
"step": 1150
},
{
"epoch": 2.35,
"learning_rate": 0.0002,
"loss": 0.3971,
"step": 1160
},
{
"epoch": 2.37,
"learning_rate": 0.0002,
"loss": 0.3963,
"step": 1170
},
{
"epoch": 2.39,
"learning_rate": 0.0002,
"loss": 0.4151,
"step": 1180
},
{
"epoch": 2.41,
"learning_rate": 0.0002,
"loss": 0.4158,
"step": 1190
},
{
"epoch": 2.43,
"learning_rate": 0.0002,
"loss": 0.3982,
"step": 1200
},
{
"epoch": 2.43,
"eval_loss": 0.5915235877037048,
"eval_runtime": 220.412,
"eval_samples_per_second": 4.537,
"eval_steps_per_second": 2.268,
"step": 1200
},
{
"epoch": 2.43,
"mmlu_eval_accuracy": 0.4767521778303616,
"mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
"mmlu_eval_accuracy_anatomy": 0.5714285714285714,
"mmlu_eval_accuracy_astronomy": 0.5625,
"mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
"mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
"mmlu_eval_accuracy_college_biology": 0.4375,
"mmlu_eval_accuracy_college_chemistry": 0.0,
"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": 0.34615384615384615,
"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": 0.2857142857142857,
"mmlu_eval_accuracy_global_facts": 0.5,
"mmlu_eval_accuracy_high_school_biology": 0.375,
"mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
"mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
"mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
"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": 0.5,
"mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
"mmlu_eval_accuracy_high_school_psychology": 0.75,
"mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
"mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
"mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
"mmlu_eval_accuracy_human_aging": 0.7391304347826086,
"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": 0.6111111111111112,
"mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
"mmlu_eval_accuracy_management": 0.5454545454545454,
"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": 0.24,
"mmlu_eval_accuracy_nutrition": 0.6666666666666666,
"mmlu_eval_accuracy_philosophy": 0.5,
"mmlu_eval_accuracy_prehistory": 0.5428571428571428,
"mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
"mmlu_eval_accuracy_professional_law": 0.35294117647058826,
"mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
"mmlu_eval_accuracy_professional_psychology": 0.42028985507246375,
"mmlu_eval_accuracy_public_relations": 0.5833333333333334,
"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": 0.5,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.0756092985639685,
"step": 1200
},
{
"epoch": 2.45,
"learning_rate": 0.0002,
"loss": 0.3996,
"step": 1210
},
{
"epoch": 2.47,
"learning_rate": 0.0002,
"loss": 0.4137,
"step": 1220
},
{
"epoch": 2.49,
"learning_rate": 0.0002,
"loss": 0.4108,
"step": 1230
},
{
"epoch": 2.51,
"learning_rate": 0.0002,
"loss": 0.4003,
"step": 1240
},
{
"epoch": 2.53,
"learning_rate": 0.0002,
"loss": 0.3988,
"step": 1250
},
{
"epoch": 2.55,
"learning_rate": 0.0002,
"loss": 0.4088,
"step": 1260
},
{
"epoch": 2.57,
"learning_rate": 0.0002,
"loss": 0.4043,
"step": 1270
},
{
"epoch": 2.59,
"learning_rate": 0.0002,
"loss": 0.4166,
"step": 1280
},
{
"epoch": 2.61,
"learning_rate": 0.0002,
"loss": 0.4169,
"step": 1290
},
{
"epoch": 2.63,
"learning_rate": 0.0002,
"loss": 0.4004,
"step": 1300
},
{
"epoch": 2.65,
"learning_rate": 0.0002,
"loss": 0.3969,
"step": 1310
},
{
"epoch": 2.67,
"learning_rate": 0.0002,
"loss": 0.3988,
"step": 1320
},
{
"epoch": 2.7,
"learning_rate": 0.0002,
"loss": 0.4141,
"step": 1330
},
{
"epoch": 2.72,
"learning_rate": 0.0002,
"loss": 0.4163,
"step": 1340
},
{
"epoch": 2.74,
"learning_rate": 0.0002,
"loss": 0.404,
"step": 1350
},
{
"epoch": 2.76,
"learning_rate": 0.0002,
"loss": 0.4017,
"step": 1360
},
{
"epoch": 2.78,
"learning_rate": 0.0002,
"loss": 0.405,
"step": 1370
},
{
"epoch": 2.8,
"learning_rate": 0.0002,
"loss": 0.4144,
"step": 1380
},
{
"epoch": 2.82,
"learning_rate": 0.0002,
"loss": 0.4043,
"step": 1390
},
{
"epoch": 2.84,
"learning_rate": 0.0002,
"loss": 0.4095,
"step": 1400
},
{
"epoch": 2.84,
"eval_loss": 0.5842967629432678,
"eval_runtime": 220.3559,
"eval_samples_per_second": 4.538,
"eval_steps_per_second": 2.269,
"step": 1400
},
{
"epoch": 2.84,
"mmlu_eval_accuracy": 0.47256392807151826,
"mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
"mmlu_eval_accuracy_anatomy": 0.5,
"mmlu_eval_accuracy_astronomy": 0.5625,
"mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
"mmlu_eval_accuracy_clinical_knowledge": 0.3793103448275862,
"mmlu_eval_accuracy_college_biology": 0.25,
"mmlu_eval_accuracy_college_chemistry": 0.0,
"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": 0.46153846153846156,
"mmlu_eval_accuracy_econometrics": 0.25,
"mmlu_eval_accuracy_electrical_engineering": 0.5,
"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": 0.36363636363636365,
"mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_european_history": 0.5,
"mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
"mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
"mmlu_eval_accuracy_high_school_microeconomics": 0.5,
"mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
"mmlu_eval_accuracy_high_school_psychology": 0.7,
"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": 0.5,
"mmlu_eval_accuracy_international_law": 0.6923076923076923,
"mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
"mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
"mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
"mmlu_eval_accuracy_management": 0.45454545454545453,
"mmlu_eval_accuracy_marketing": 0.76,
"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": 0.6363636363636364,
"mmlu_eval_accuracy_philosophy": 0.47058823529411764,
"mmlu_eval_accuracy_prehistory": 0.5142857142857142,
"mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
"mmlu_eval_accuracy_professional_law": 0.35294117647058826,
"mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
"mmlu_eval_accuracy_professional_psychology": 0.391304347826087,
"mmlu_eval_accuracy_public_relations": 0.5,
"mmlu_eval_accuracy_security_studies": 0.5925925925925926,
"mmlu_eval_accuracy_sociology": 0.7272727272727273,
"mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
"mmlu_eval_accuracy_virology": 0.5555555555555556,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.0570991446820936,
"step": 1400
}
],
"max_steps": 5000,
"num_train_epochs": 11,
"total_flos": 3.427554805864612e+17,
"trial_name": null,
"trial_params": null
}