prateeky2806's picture
Training in progress, step 1000
aebf6f9
raw
history blame
30.8 kB
{
"best_metric": 0.5078858137130737,
"best_model_checkpoint": "./output_v2/7b_cluster05_Nous-Hermes-llama-2-7b_partitioned_v3_standardized_05/checkpoint-600",
"epoch": 2.973977695167286,
"global_step": 1000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 0.6832,
"step": 10
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 0.5389,
"step": 20
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 0.5072,
"step": 30
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 0.5211,
"step": 40
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 0.5888,
"step": 50
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 0.534,
"step": 60
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 0.5684,
"step": 70
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 0.514,
"step": 80
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 0.5518,
"step": 90
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 0.5034,
"step": 100
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 0.542,
"step": 110
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 0.4847,
"step": 120
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 0.4772,
"step": 130
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 0.5196,
"step": 140
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 0.4672,
"step": 150
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 0.4913,
"step": 160
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 0.5498,
"step": 170
},
{
"epoch": 0.54,
"learning_rate": 0.0002,
"loss": 0.5328,
"step": 180
},
{
"epoch": 0.57,
"learning_rate": 0.0002,
"loss": 0.5313,
"step": 190
},
{
"epoch": 0.59,
"learning_rate": 0.0002,
"loss": 0.515,
"step": 200
},
{
"epoch": 0.59,
"eval_loss": 0.531296968460083,
"eval_runtime": 174.4771,
"eval_samples_per_second": 5.731,
"eval_steps_per_second": 2.866,
"step": 200
},
{
"epoch": 0.59,
"mmlu_eval_accuracy": 0.46292469330066577,
"mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
"mmlu_eval_accuracy_anatomy": 0.5,
"mmlu_eval_accuracy_astronomy": 0.5,
"mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
"mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
"mmlu_eval_accuracy_college_biology": 0.3125,
"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": 0.18181818181818182,
"mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
"mmlu_eval_accuracy_econometrics": 0.16666666666666666,
"mmlu_eval_accuracy_electrical_engineering": 0.5,
"mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
"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": 0.45454545454545453,
"mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
"mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_geography": 0.7272727272727273,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
"mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
"mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
"mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
"mmlu_eval_accuracy_high_school_psychology": 0.65,
"mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
"mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
"mmlu_eval_accuracy_high_school_world_history": 0.46153846153846156,
"mmlu_eval_accuracy_human_aging": 0.6521739130434783,
"mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
"mmlu_eval_accuracy_international_law": 0.8461538461538461,
"mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
"mmlu_eval_accuracy_logical_fallacies": 0.5,
"mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
"mmlu_eval_accuracy_management": 0.6363636363636364,
"mmlu_eval_accuracy_marketing": 0.72,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
"mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.5454545454545454,
"mmlu_eval_accuracy_philosophy": 0.5,
"mmlu_eval_accuracy_prehistory": 0.4,
"mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
"mmlu_eval_accuracy_professional_law": 0.3352941176470588,
"mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
"mmlu_eval_accuracy_professional_psychology": 0.42028985507246375,
"mmlu_eval_accuracy_public_relations": 0.5833333333333334,
"mmlu_eval_accuracy_security_studies": 0.5185185185185185,
"mmlu_eval_accuracy_sociology": 0.5909090909090909,
"mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
"mmlu_eval_accuracy_virology": 0.3888888888888889,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.3857738420406147,
"step": 200
},
{
"epoch": 0.62,
"learning_rate": 0.0002,
"loss": 0.4582,
"step": 210
},
{
"epoch": 0.65,
"learning_rate": 0.0002,
"loss": 0.4796,
"step": 220
},
{
"epoch": 0.68,
"learning_rate": 0.0002,
"loss": 0.5081,
"step": 230
},
{
"epoch": 0.71,
"learning_rate": 0.0002,
"loss": 0.5206,
"step": 240
},
{
"epoch": 0.74,
"learning_rate": 0.0002,
"loss": 0.4961,
"step": 250
},
{
"epoch": 0.77,
"learning_rate": 0.0002,
"loss": 0.5219,
"step": 260
},
{
"epoch": 0.8,
"learning_rate": 0.0002,
"loss": 0.5311,
"step": 270
},
{
"epoch": 0.83,
"learning_rate": 0.0002,
"loss": 0.5039,
"step": 280
},
{
"epoch": 0.86,
"learning_rate": 0.0002,
"loss": 0.4994,
"step": 290
},
{
"epoch": 0.89,
"learning_rate": 0.0002,
"loss": 0.4804,
"step": 300
},
{
"epoch": 0.92,
"learning_rate": 0.0002,
"loss": 0.4791,
"step": 310
},
{
"epoch": 0.95,
"learning_rate": 0.0002,
"loss": 0.4822,
"step": 320
},
{
"epoch": 0.98,
"learning_rate": 0.0002,
"loss": 0.5157,
"step": 330
},
{
"epoch": 1.01,
"learning_rate": 0.0002,
"loss": 0.5184,
"step": 340
},
{
"epoch": 1.04,
"learning_rate": 0.0002,
"loss": 0.4562,
"step": 350
},
{
"epoch": 1.07,
"learning_rate": 0.0002,
"loss": 0.4117,
"step": 360
},
{
"epoch": 1.1,
"learning_rate": 0.0002,
"loss": 0.451,
"step": 370
},
{
"epoch": 1.13,
"learning_rate": 0.0002,
"loss": 0.4237,
"step": 380
},
{
"epoch": 1.16,
"learning_rate": 0.0002,
"loss": 0.4243,
"step": 390
},
{
"epoch": 1.19,
"learning_rate": 0.0002,
"loss": 0.4409,
"step": 400
},
{
"epoch": 1.19,
"eval_loss": 0.5141976475715637,
"eval_runtime": 174.8657,
"eval_samples_per_second": 5.719,
"eval_steps_per_second": 2.859,
"step": 400
},
{
"epoch": 1.19,
"mmlu_eval_accuracy": 0.45097298030174665,
"mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
"mmlu_eval_accuracy_anatomy": 0.5,
"mmlu_eval_accuracy_astronomy": 0.5,
"mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
"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": 0.2727272727272727,
"mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
"mmlu_eval_accuracy_college_physics": 0.5454545454545454,
"mmlu_eval_accuracy_computer_security": 0.2727272727272727,
"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": 0.3902439024390244,
"mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
"mmlu_eval_accuracy_global_facts": 0.6,
"mmlu_eval_accuracy_high_school_biology": 0.3125,
"mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
"mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
"mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
"mmlu_eval_accuracy_high_school_geography": 0.7272727272727273,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
"mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
"mmlu_eval_accuracy_high_school_microeconomics": 0.5,
"mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
"mmlu_eval_accuracy_high_school_psychology": 0.6833333333333333,
"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": 0.7692307692307693,
"mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
"mmlu_eval_accuracy_logical_fallacies": 0.5,
"mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
"mmlu_eval_accuracy_management": 0.5454545454545454,
"mmlu_eval_accuracy_marketing": 0.68,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
"mmlu_eval_accuracy_moral_disputes": 0.5,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.5151515151515151,
"mmlu_eval_accuracy_philosophy": 0.47058823529411764,
"mmlu_eval_accuracy_prehistory": 0.45714285714285713,
"mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
"mmlu_eval_accuracy_professional_law": 0.3411764705882353,
"mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
"mmlu_eval_accuracy_professional_psychology": 0.391304347826087,
"mmlu_eval_accuracy_public_relations": 0.5833333333333334,
"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": 1.0380505950970687,
"step": 400
},
{
"epoch": 1.22,
"learning_rate": 0.0002,
"loss": 0.4176,
"step": 410
},
{
"epoch": 1.25,
"learning_rate": 0.0002,
"loss": 0.445,
"step": 420
},
{
"epoch": 1.28,
"learning_rate": 0.0002,
"loss": 0.4968,
"step": 430
},
{
"epoch": 1.31,
"learning_rate": 0.0002,
"loss": 0.4573,
"step": 440
},
{
"epoch": 1.34,
"learning_rate": 0.0002,
"loss": 0.4097,
"step": 450
},
{
"epoch": 1.37,
"learning_rate": 0.0002,
"loss": 0.4215,
"step": 460
},
{
"epoch": 1.4,
"learning_rate": 0.0002,
"loss": 0.4754,
"step": 470
},
{
"epoch": 1.43,
"learning_rate": 0.0002,
"loss": 0.4463,
"step": 480
},
{
"epoch": 1.46,
"learning_rate": 0.0002,
"loss": 0.4027,
"step": 490
},
{
"epoch": 1.49,
"learning_rate": 0.0002,
"loss": 0.4361,
"step": 500
},
{
"epoch": 1.52,
"learning_rate": 0.0002,
"loss": 0.4458,
"step": 510
},
{
"epoch": 1.55,
"learning_rate": 0.0002,
"loss": 0.4445,
"step": 520
},
{
"epoch": 1.58,
"learning_rate": 0.0002,
"loss": 0.4117,
"step": 530
},
{
"epoch": 1.61,
"learning_rate": 0.0002,
"loss": 0.4609,
"step": 540
},
{
"epoch": 1.64,
"learning_rate": 0.0002,
"loss": 0.4511,
"step": 550
},
{
"epoch": 1.67,
"learning_rate": 0.0002,
"loss": 0.4385,
"step": 560
},
{
"epoch": 1.7,
"learning_rate": 0.0002,
"loss": 0.4451,
"step": 570
},
{
"epoch": 1.72,
"learning_rate": 0.0002,
"loss": 0.4414,
"step": 580
},
{
"epoch": 1.75,
"learning_rate": 0.0002,
"loss": 0.4235,
"step": 590
},
{
"epoch": 1.78,
"learning_rate": 0.0002,
"loss": 0.4954,
"step": 600
},
{
"epoch": 1.78,
"eval_loss": 0.5078858137130737,
"eval_runtime": 174.9277,
"eval_samples_per_second": 5.717,
"eval_steps_per_second": 2.858,
"step": 600
},
{
"epoch": 1.78,
"mmlu_eval_accuracy": 0.4596214580927039,
"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": 0.4827586206896552,
"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": 0.2727272727272727,
"mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
"mmlu_eval_accuracy_college_physics": 0.5454545454545454,
"mmlu_eval_accuracy_computer_security": 0.2727272727272727,
"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": 0.43902439024390244,
"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": 0.4444444444444444,
"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": 0.6190476190476191,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
"mmlu_eval_accuracy_high_school_mathematics": 0.3103448275862069,
"mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
"mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
"mmlu_eval_accuracy_high_school_psychology": 0.6833333333333333,
"mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
"mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
"mmlu_eval_accuracy_high_school_world_history": 0.46153846153846156,
"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": 0.2727272727272727,
"mmlu_eval_accuracy_logical_fallacies": 0.5,
"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": 0.47368421052631576,
"mmlu_eval_accuracy_moral_scenarios": 0.23,
"mmlu_eval_accuracy_nutrition": 0.5757575757575758,
"mmlu_eval_accuracy_philosophy": 0.47058823529411764,
"mmlu_eval_accuracy_prehistory": 0.4857142857142857,
"mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
"mmlu_eval_accuracy_professional_law": 0.3588235294117647,
"mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
"mmlu_eval_accuracy_professional_psychology": 0.42028985507246375,
"mmlu_eval_accuracy_public_relations": 0.6666666666666666,
"mmlu_eval_accuracy_security_studies": 0.5555555555555556,
"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": 1.0377603609980552,
"step": 600
},
{
"epoch": 1.81,
"learning_rate": 0.0002,
"loss": 0.4865,
"step": 610
},
{
"epoch": 1.84,
"learning_rate": 0.0002,
"loss": 0.4451,
"step": 620
},
{
"epoch": 1.87,
"learning_rate": 0.0002,
"loss": 0.424,
"step": 630
},
{
"epoch": 1.9,
"learning_rate": 0.0002,
"loss": 0.4473,
"step": 640
},
{
"epoch": 1.93,
"learning_rate": 0.0002,
"loss": 0.4627,
"step": 650
},
{
"epoch": 1.96,
"learning_rate": 0.0002,
"loss": 0.4298,
"step": 660
},
{
"epoch": 1.99,
"learning_rate": 0.0002,
"loss": 0.4561,
"step": 670
},
{
"epoch": 2.02,
"learning_rate": 0.0002,
"loss": 0.3726,
"step": 680
},
{
"epoch": 2.05,
"learning_rate": 0.0002,
"loss": 0.3548,
"step": 690
},
{
"epoch": 2.08,
"learning_rate": 0.0002,
"loss": 0.3565,
"step": 700
},
{
"epoch": 2.11,
"learning_rate": 0.0002,
"loss": 0.3133,
"step": 710
},
{
"epoch": 2.14,
"learning_rate": 0.0002,
"loss": 0.3475,
"step": 720
},
{
"epoch": 2.17,
"learning_rate": 0.0002,
"loss": 0.3761,
"step": 730
},
{
"epoch": 2.2,
"learning_rate": 0.0002,
"loss": 0.3336,
"step": 740
},
{
"epoch": 2.23,
"learning_rate": 0.0002,
"loss": 0.392,
"step": 750
},
{
"epoch": 2.26,
"learning_rate": 0.0002,
"loss": 0.3556,
"step": 760
},
{
"epoch": 2.29,
"learning_rate": 0.0002,
"loss": 0.3706,
"step": 770
},
{
"epoch": 2.32,
"learning_rate": 0.0002,
"loss": 0.3426,
"step": 780
},
{
"epoch": 2.35,
"learning_rate": 0.0002,
"loss": 0.3273,
"step": 790
},
{
"epoch": 2.38,
"learning_rate": 0.0002,
"loss": 0.3772,
"step": 800
},
{
"epoch": 2.38,
"eval_loss": 0.5177344083786011,
"eval_runtime": 175.3396,
"eval_samples_per_second": 5.703,
"eval_steps_per_second": 2.852,
"step": 800
},
{
"epoch": 2.38,
"mmlu_eval_accuracy": 0.4452625151934351,
"mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
"mmlu_eval_accuracy_anatomy": 0.5714285714285714,
"mmlu_eval_accuracy_astronomy": 0.5,
"mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
"mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
"mmlu_eval_accuracy_college_biology": 0.375,
"mmlu_eval_accuracy_college_chemistry": 0.0,
"mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
"mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
"mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
"mmlu_eval_accuracy_college_physics": 0.45454545454545453,
"mmlu_eval_accuracy_computer_security": 0.2727272727272727,
"mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
"mmlu_eval_accuracy_econometrics": 0.16666666666666666,
"mmlu_eval_accuracy_electrical_engineering": 0.3125,
"mmlu_eval_accuracy_elementary_mathematics": 0.4634146341463415,
"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": 0.4444444444444444,
"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": 0.6190476190476191,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
"mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
"mmlu_eval_accuracy_high_school_microeconomics": 0.2692307692307692,
"mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
"mmlu_eval_accuracy_high_school_psychology": 0.7,
"mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
"mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
"mmlu_eval_accuracy_high_school_world_history": 0.46153846153846156,
"mmlu_eval_accuracy_human_aging": 0.6521739130434783,
"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": 0.5,
"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": 0.47368421052631576,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.5151515151515151,
"mmlu_eval_accuracy_philosophy": 0.4411764705882353,
"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": 0.42028985507246375,
"mmlu_eval_accuracy_public_relations": 0.6666666666666666,
"mmlu_eval_accuracy_security_studies": 0.48148148148148145,
"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": 0.9420785120841573,
"step": 800
},
{
"epoch": 2.41,
"learning_rate": 0.0002,
"loss": 0.3678,
"step": 810
},
{
"epoch": 2.44,
"learning_rate": 0.0002,
"loss": 0.316,
"step": 820
},
{
"epoch": 2.47,
"learning_rate": 0.0002,
"loss": 0.3669,
"step": 830
},
{
"epoch": 2.5,
"learning_rate": 0.0002,
"loss": 0.3955,
"step": 840
},
{
"epoch": 2.53,
"learning_rate": 0.0002,
"loss": 0.3854,
"step": 850
},
{
"epoch": 2.56,
"learning_rate": 0.0002,
"loss": 0.3514,
"step": 860
},
{
"epoch": 2.59,
"learning_rate": 0.0002,
"loss": 0.3491,
"step": 870
},
{
"epoch": 2.62,
"learning_rate": 0.0002,
"loss": 0.3567,
"step": 880
},
{
"epoch": 2.65,
"learning_rate": 0.0002,
"loss": 0.3839,
"step": 890
},
{
"epoch": 2.68,
"learning_rate": 0.0002,
"loss": 0.3291,
"step": 900
},
{
"epoch": 2.71,
"learning_rate": 0.0002,
"loss": 0.3917,
"step": 910
},
{
"epoch": 2.74,
"learning_rate": 0.0002,
"loss": 0.3812,
"step": 920
},
{
"epoch": 2.77,
"learning_rate": 0.0002,
"loss": 0.3496,
"step": 930
},
{
"epoch": 2.8,
"learning_rate": 0.0002,
"loss": 0.3339,
"step": 940
},
{
"epoch": 2.83,
"learning_rate": 0.0002,
"loss": 0.3565,
"step": 950
},
{
"epoch": 2.86,
"learning_rate": 0.0002,
"loss": 0.3825,
"step": 960
},
{
"epoch": 2.88,
"learning_rate": 0.0002,
"loss": 0.4028,
"step": 970
},
{
"epoch": 2.91,
"learning_rate": 0.0002,
"loss": 0.3621,
"step": 980
},
{
"epoch": 2.94,
"learning_rate": 0.0002,
"loss": 0.3345,
"step": 990
},
{
"epoch": 2.97,
"learning_rate": 0.0002,
"loss": 0.4121,
"step": 1000
},
{
"epoch": 2.97,
"eval_loss": 0.5176346898078918,
"eval_runtime": 175.3431,
"eval_samples_per_second": 5.703,
"eval_steps_per_second": 2.852,
"step": 1000
},
{
"epoch": 2.97,
"mmlu_eval_accuracy": 0.43483776787791517,
"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": 0.4827586206896552,
"mmlu_eval_accuracy_college_biology": 0.3125,
"mmlu_eval_accuracy_college_chemistry": 0.0,
"mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
"mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
"mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
"mmlu_eval_accuracy_college_physics": 0.45454545454545453,
"mmlu_eval_accuracy_computer_security": 0.18181818181818182,
"mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
"mmlu_eval_accuracy_econometrics": 0.25,
"mmlu_eval_accuracy_electrical_engineering": 0.375,
"mmlu_eval_accuracy_elementary_mathematics": 0.43902439024390244,
"mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
"mmlu_eval_accuracy_global_facts": 0.3,
"mmlu_eval_accuracy_high_school_biology": 0.25,
"mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
"mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
"mmlu_eval_accuracy_high_school_european_history": 0.4444444444444444,
"mmlu_eval_accuracy_high_school_geography": 0.7272727272727273,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.4186046511627907,
"mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
"mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
"mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
"mmlu_eval_accuracy_high_school_psychology": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
"mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
"mmlu_eval_accuracy_high_school_world_history": 0.46153846153846156,
"mmlu_eval_accuracy_human_aging": 0.6521739130434783,
"mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
"mmlu_eval_accuracy_international_law": 0.7692307692307693,
"mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
"mmlu_eval_accuracy_logical_fallacies": 0.5,
"mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
"mmlu_eval_accuracy_management": 0.36363636363636365,
"mmlu_eval_accuracy_marketing": 0.68,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
"mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.45454545454545453,
"mmlu_eval_accuracy_philosophy": 0.4117647058823529,
"mmlu_eval_accuracy_prehistory": 0.42857142857142855,
"mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
"mmlu_eval_accuracy_professional_law": 0.32941176470588235,
"mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
"mmlu_eval_accuracy_professional_psychology": 0.42028985507246375,
"mmlu_eval_accuracy_public_relations": 0.6666666666666666,
"mmlu_eval_accuracy_security_studies": 0.4444444444444444,
"mmlu_eval_accuracy_sociology": 0.5909090909090909,
"mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
"mmlu_eval_accuracy_virology": 0.5,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 0.9825846157073197,
"step": 1000
}
],
"max_steps": 5000,
"num_train_epochs": 15,
"total_flos": 2.0310520321046938e+17,
"trial_name": null,
"trial_params": null
}