prateeky2806's picture
Training in progress, step 2200
e2caa5e
raw
history blame
67.3 kB
{
"best_metric": 0.4432196617126465,
"best_model_checkpoint": "./output_v2/7b_cluster017_Nous-Hermes-llama-2-7b_partitioned_v3_standardized_017/checkpoint-1600",
"epoch": 2.6284348864994027,
"global_step": 2200,
"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": 0.5801,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 0.6179,
"step": 20
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 0.5163,
"step": 30
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 0.5249,
"step": 40
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 0.5421,
"step": 50
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 0.4993,
"step": 60
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 0.5421,
"step": 70
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 0.4769,
"step": 80
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 0.5084,
"step": 90
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 0.4731,
"step": 100
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 0.5069,
"step": 110
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 0.4659,
"step": 120
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 0.4863,
"step": 130
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 0.5124,
"step": 140
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 0.5311,
"step": 150
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 0.5032,
"step": 160
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 0.5065,
"step": 170
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 0.4613,
"step": 180
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 0.517,
"step": 190
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 0.4761,
"step": 200
},
{
"epoch": 0.24,
"eval_loss": 0.4977516829967499,
"eval_runtime": 178.665,
"eval_samples_per_second": 5.597,
"eval_steps_per_second": 2.799,
"step": 200
},
{
"epoch": 0.24,
"mmlu_eval_accuracy": 0.4731690276039549,
"mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
"mmlu_eval_accuracy_anatomy": 0.6428571428571429,
"mmlu_eval_accuracy_astronomy": 0.4375,
"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": 0.125,
"mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
"mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
"mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
"mmlu_eval_accuracy_college_physics": 0.45454545454545453,
"mmlu_eval_accuracy_computer_security": 0.45454545454545453,
"mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
"mmlu_eval_accuracy_econometrics": 0.16666666666666666,
"mmlu_eval_accuracy_electrical_engineering": 0.5625,
"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": 0.5555555555555556,
"mmlu_eval_accuracy_high_school_european_history": 0.5,
"mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
"mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
"mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
"mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
"mmlu_eval_accuracy_high_school_psychology": 0.7333333333333333,
"mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
"mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
"mmlu_eval_accuracy_high_school_world_history": 0.5,
"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": 0.6363636363636364,
"mmlu_eval_accuracy_marketing": 0.68,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
"mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
"mmlu_eval_accuracy_moral_scenarios": 0.23,
"mmlu_eval_accuracy_nutrition": 0.6363636363636364,
"mmlu_eval_accuracy_philosophy": 0.47058823529411764,
"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": 0.41935483870967744,
"mmlu_eval_accuracy_professional_psychology": 0.34782608695652173,
"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": 0.4444444444444444,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.201889930774431,
"step": 200
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 0.447,
"step": 210
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 0.5419,
"step": 220
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 0.46,
"step": 230
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 0.481,
"step": 240
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 0.4279,
"step": 250
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 0.462,
"step": 260
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 0.4866,
"step": 270
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 0.4565,
"step": 280
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 0.4579,
"step": 290
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 0.4585,
"step": 300
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 0.466,
"step": 310
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 0.4766,
"step": 320
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 0.4682,
"step": 330
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 0.4467,
"step": 340
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 0.4675,
"step": 350
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 0.4816,
"step": 360
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 0.4439,
"step": 370
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 0.4553,
"step": 380
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 0.4707,
"step": 390
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 0.4389,
"step": 400
},
{
"epoch": 0.48,
"eval_loss": 0.4804040491580963,
"eval_runtime": 178.9419,
"eval_samples_per_second": 5.588,
"eval_steps_per_second": 2.794,
"step": 400
},
{
"epoch": 0.48,
"mmlu_eval_accuracy": 0.4686810757119835,
"mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
"mmlu_eval_accuracy_anatomy": 0.5,
"mmlu_eval_accuracy_astronomy": 0.4375,
"mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
"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": 0.2727272727272727,
"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": 0.4230769230769231,
"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": 0.4090909090909091,
"mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
"mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.3023255813953488,
"mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
"mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
"mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
"mmlu_eval_accuracy_high_school_psychology": 0.7666666666666667,
"mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
"mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
"mmlu_eval_accuracy_high_school_world_history": 0.5,
"mmlu_eval_accuracy_human_aging": 0.6956521739130435,
"mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
"mmlu_eval_accuracy_international_law": 0.6923076923076923,
"mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
"mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
"mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
"mmlu_eval_accuracy_management": 0.5454545454545454,
"mmlu_eval_accuracy_marketing": 0.76,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
"mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
"mmlu_eval_accuracy_moral_scenarios": 0.23,
"mmlu_eval_accuracy_nutrition": 0.6060606060606061,
"mmlu_eval_accuracy_philosophy": 0.47058823529411764,
"mmlu_eval_accuracy_prehistory": 0.5142857142857142,
"mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
"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": 0.5833333333333334,
"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": 1.168120609523422,
"step": 400
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 0.49,
"step": 410
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 0.4614,
"step": 420
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 0.4711,
"step": 430
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 0.4557,
"step": 440
},
{
"epoch": 0.54,
"learning_rate": 0.0002,
"loss": 0.4454,
"step": 450
},
{
"epoch": 0.55,
"learning_rate": 0.0002,
"loss": 0.4819,
"step": 460
},
{
"epoch": 0.56,
"learning_rate": 0.0002,
"loss": 0.4694,
"step": 470
},
{
"epoch": 0.57,
"learning_rate": 0.0002,
"loss": 0.4602,
"step": 480
},
{
"epoch": 0.59,
"learning_rate": 0.0002,
"loss": 0.4528,
"step": 490
},
{
"epoch": 0.6,
"learning_rate": 0.0002,
"loss": 0.4415,
"step": 500
},
{
"epoch": 0.61,
"learning_rate": 0.0002,
"loss": 0.4597,
"step": 510
},
{
"epoch": 0.62,
"learning_rate": 0.0002,
"loss": 0.437,
"step": 520
},
{
"epoch": 0.63,
"learning_rate": 0.0002,
"loss": 0.4649,
"step": 530
},
{
"epoch": 0.65,
"learning_rate": 0.0002,
"loss": 0.4552,
"step": 540
},
{
"epoch": 0.66,
"learning_rate": 0.0002,
"loss": 0.4517,
"step": 550
},
{
"epoch": 0.67,
"learning_rate": 0.0002,
"loss": 0.4324,
"step": 560
},
{
"epoch": 0.68,
"learning_rate": 0.0002,
"loss": 0.4473,
"step": 570
},
{
"epoch": 0.69,
"learning_rate": 0.0002,
"loss": 0.4611,
"step": 580
},
{
"epoch": 0.7,
"learning_rate": 0.0002,
"loss": 0.4378,
"step": 590
},
{
"epoch": 0.72,
"learning_rate": 0.0002,
"loss": 0.4337,
"step": 600
},
{
"epoch": 0.72,
"eval_loss": 0.4673095643520355,
"eval_runtime": 178.8409,
"eval_samples_per_second": 5.592,
"eval_steps_per_second": 2.796,
"step": 600
},
{
"epoch": 0.72,
"mmlu_eval_accuracy": 0.4657957744296099,
"mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
"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": 0.36363636363636365,
"mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
"mmlu_eval_accuracy_college_physics": 0.45454545454545453,
"mmlu_eval_accuracy_computer_security": 0.45454545454545453,
"mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
"mmlu_eval_accuracy_econometrics": 0.25,
"mmlu_eval_accuracy_electrical_engineering": 0.3125,
"mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
"mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
"mmlu_eval_accuracy_global_facts": 0.5,
"mmlu_eval_accuracy_high_school_biology": 0.40625,
"mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
"mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
"mmlu_eval_accuracy_high_school_european_history": 0.5,
"mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
"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": 0.75,
"mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
"mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
"mmlu_eval_accuracy_high_school_world_history": 0.5,
"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": 0.68,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.627906976744186,
"mmlu_eval_accuracy_moral_disputes": 0.5,
"mmlu_eval_accuracy_moral_scenarios": 0.23,
"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": 0.3176470588235294,
"mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
"mmlu_eval_accuracy_professional_psychology": 0.34782608695652173,
"mmlu_eval_accuracy_public_relations": 0.5,
"mmlu_eval_accuracy_security_studies": 0.48148148148148145,
"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": 1.1695979737893096,
"step": 600
},
{
"epoch": 0.73,
"learning_rate": 0.0002,
"loss": 0.4255,
"step": 610
},
{
"epoch": 0.74,
"learning_rate": 0.0002,
"loss": 0.4492,
"step": 620
},
{
"epoch": 0.75,
"learning_rate": 0.0002,
"loss": 0.4353,
"step": 630
},
{
"epoch": 0.76,
"learning_rate": 0.0002,
"loss": 0.4388,
"step": 640
},
{
"epoch": 0.78,
"learning_rate": 0.0002,
"loss": 0.4402,
"step": 650
},
{
"epoch": 0.79,
"learning_rate": 0.0002,
"loss": 0.4568,
"step": 660
},
{
"epoch": 0.8,
"learning_rate": 0.0002,
"loss": 0.4703,
"step": 670
},
{
"epoch": 0.81,
"learning_rate": 0.0002,
"loss": 0.4561,
"step": 680
},
{
"epoch": 0.82,
"learning_rate": 0.0002,
"loss": 0.4745,
"step": 690
},
{
"epoch": 0.84,
"learning_rate": 0.0002,
"loss": 0.4384,
"step": 700
},
{
"epoch": 0.85,
"learning_rate": 0.0002,
"loss": 0.4472,
"step": 710
},
{
"epoch": 0.86,
"learning_rate": 0.0002,
"loss": 0.4607,
"step": 720
},
{
"epoch": 0.87,
"learning_rate": 0.0002,
"loss": 0.4876,
"step": 730
},
{
"epoch": 0.88,
"learning_rate": 0.0002,
"loss": 0.4575,
"step": 740
},
{
"epoch": 0.9,
"learning_rate": 0.0002,
"loss": 0.4578,
"step": 750
},
{
"epoch": 0.91,
"learning_rate": 0.0002,
"loss": 0.4417,
"step": 760
},
{
"epoch": 0.92,
"learning_rate": 0.0002,
"loss": 0.4305,
"step": 770
},
{
"epoch": 0.93,
"learning_rate": 0.0002,
"loss": 0.4478,
"step": 780
},
{
"epoch": 0.94,
"learning_rate": 0.0002,
"loss": 0.4395,
"step": 790
},
{
"epoch": 0.96,
"learning_rate": 0.0002,
"loss": 0.4188,
"step": 800
},
{
"epoch": 0.96,
"eval_loss": 0.4581112265586853,
"eval_runtime": 178.9111,
"eval_samples_per_second": 5.589,
"eval_steps_per_second": 2.795,
"step": 800
},
{
"epoch": 0.96,
"mmlu_eval_accuracy": 0.4708997812184524,
"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": 0.36363636363636365,
"mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
"mmlu_eval_accuracy_college_physics": 0.5454545454545454,
"mmlu_eval_accuracy_computer_security": 0.36363636363636365,
"mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
"mmlu_eval_accuracy_econometrics": 0.25,
"mmlu_eval_accuracy_electrical_engineering": 0.375,
"mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
"mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
"mmlu_eval_accuracy_global_facts": 0.5,
"mmlu_eval_accuracy_high_school_biology": 0.40625,
"mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
"mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
"mmlu_eval_accuracy_high_school_european_history": 0.5,
"mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.3953488372093023,
"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": 0.43478260869565216,
"mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
"mmlu_eval_accuracy_high_school_world_history": 0.5,
"mmlu_eval_accuracy_human_aging": 0.6956521739130435,
"mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
"mmlu_eval_accuracy_international_law": 0.7692307692307693,
"mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
"mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
"mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
"mmlu_eval_accuracy_management": 0.6363636363636364,
"mmlu_eval_accuracy_marketing": 0.76,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
"mmlu_eval_accuracy_moral_disputes": 0.5,
"mmlu_eval_accuracy_moral_scenarios": 0.23,
"mmlu_eval_accuracy_nutrition": 0.5757575757575758,
"mmlu_eval_accuracy_philosophy": 0.47058823529411764,
"mmlu_eval_accuracy_prehistory": 0.5428571428571428,
"mmlu_eval_accuracy_professional_accounting": 0.1935483870967742,
"mmlu_eval_accuracy_professional_law": 0.36470588235294116,
"mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
"mmlu_eval_accuracy_professional_psychology": 0.34782608695652173,
"mmlu_eval_accuracy_public_relations": 0.5,
"mmlu_eval_accuracy_security_studies": 0.48148148148148145,
"mmlu_eval_accuracy_sociology": 0.6363636363636364,
"mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
"mmlu_eval_accuracy_virology": 0.3888888888888889,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.1651555353060714,
"step": 800
},
{
"epoch": 0.97,
"learning_rate": 0.0002,
"loss": 0.4485,
"step": 810
},
{
"epoch": 0.98,
"learning_rate": 0.0002,
"loss": 0.4215,
"step": 820
},
{
"epoch": 0.99,
"learning_rate": 0.0002,
"loss": 0.4323,
"step": 830
},
{
"epoch": 1.0,
"learning_rate": 0.0002,
"loss": 0.4166,
"step": 840
},
{
"epoch": 1.02,
"learning_rate": 0.0002,
"loss": 0.363,
"step": 850
},
{
"epoch": 1.03,
"learning_rate": 0.0002,
"loss": 0.425,
"step": 860
},
{
"epoch": 1.04,
"learning_rate": 0.0002,
"loss": 0.3954,
"step": 870
},
{
"epoch": 1.05,
"learning_rate": 0.0002,
"loss": 0.4146,
"step": 880
},
{
"epoch": 1.06,
"learning_rate": 0.0002,
"loss": 0.3617,
"step": 890
},
{
"epoch": 1.08,
"learning_rate": 0.0002,
"loss": 0.3758,
"step": 900
},
{
"epoch": 1.09,
"learning_rate": 0.0002,
"loss": 0.4184,
"step": 910
},
{
"epoch": 1.1,
"learning_rate": 0.0002,
"loss": 0.386,
"step": 920
},
{
"epoch": 1.11,
"learning_rate": 0.0002,
"loss": 0.3808,
"step": 930
},
{
"epoch": 1.12,
"learning_rate": 0.0002,
"loss": 0.3842,
"step": 940
},
{
"epoch": 1.14,
"learning_rate": 0.0002,
"loss": 0.3884,
"step": 950
},
{
"epoch": 1.15,
"learning_rate": 0.0002,
"loss": 0.3743,
"step": 960
},
{
"epoch": 1.16,
"learning_rate": 0.0002,
"loss": 0.3834,
"step": 970
},
{
"epoch": 1.17,
"learning_rate": 0.0002,
"loss": 0.3883,
"step": 980
},
{
"epoch": 1.18,
"learning_rate": 0.0002,
"loss": 0.3831,
"step": 990
},
{
"epoch": 1.19,
"learning_rate": 0.0002,
"loss": 0.3961,
"step": 1000
},
{
"epoch": 1.19,
"eval_loss": 0.4571826756000519,
"eval_runtime": 178.8234,
"eval_samples_per_second": 5.592,
"eval_steps_per_second": 2.796,
"step": 1000
},
{
"epoch": 1.19,
"mmlu_eval_accuracy": 0.46518318736306025,
"mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
"mmlu_eval_accuracy_anatomy": 0.5714285714285714,
"mmlu_eval_accuracy_astronomy": 0.3125,
"mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
"mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
"mmlu_eval_accuracy_college_biology": 0.4375,
"mmlu_eval_accuracy_college_chemistry": 0.125,
"mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
"mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
"mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
"mmlu_eval_accuracy_college_physics": 0.5454545454545454,
"mmlu_eval_accuracy_computer_security": 0.36363636363636365,
"mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
"mmlu_eval_accuracy_econometrics": 0.25,
"mmlu_eval_accuracy_electrical_engineering": 0.3125,
"mmlu_eval_accuracy_elementary_mathematics": 0.2682926829268293,
"mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
"mmlu_eval_accuracy_global_facts": 0.5,
"mmlu_eval_accuracy_high_school_biology": 0.40625,
"mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
"mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
"mmlu_eval_accuracy_high_school_european_history": 0.4444444444444444,
"mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
"mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
"mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
"mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
"mmlu_eval_accuracy_high_school_psychology": 0.7666666666666667,
"mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
"mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
"mmlu_eval_accuracy_high_school_world_history": 0.5,
"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": 0.45454545454545453,
"mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
"mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
"mmlu_eval_accuracy_management": 0.6363636363636364,
"mmlu_eval_accuracy_marketing": 0.68,
"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": 0.24,
"mmlu_eval_accuracy_nutrition": 0.6666666666666666,
"mmlu_eval_accuracy_philosophy": 0.4411764705882353,
"mmlu_eval_accuracy_prehistory": 0.5428571428571428,
"mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
"mmlu_eval_accuracy_professional_law": 0.3352941176470588,
"mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
"mmlu_eval_accuracy_professional_psychology": 0.37681159420289856,
"mmlu_eval_accuracy_public_relations": 0.5,
"mmlu_eval_accuracy_security_studies": 0.48148148148148145,
"mmlu_eval_accuracy_sociology": 0.6818181818181818,
"mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
"mmlu_eval_accuracy_virology": 0.3888888888888889,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.1887711029567856,
"step": 1000
},
{
"epoch": 1.21,
"learning_rate": 0.0002,
"loss": 0.4023,
"step": 1010
},
{
"epoch": 1.22,
"learning_rate": 0.0002,
"loss": 0.3878,
"step": 1020
},
{
"epoch": 1.23,
"learning_rate": 0.0002,
"loss": 0.3734,
"step": 1030
},
{
"epoch": 1.24,
"learning_rate": 0.0002,
"loss": 0.3773,
"step": 1040
},
{
"epoch": 1.25,
"learning_rate": 0.0002,
"loss": 0.4206,
"step": 1050
},
{
"epoch": 1.27,
"learning_rate": 0.0002,
"loss": 0.3705,
"step": 1060
},
{
"epoch": 1.28,
"learning_rate": 0.0002,
"loss": 0.3942,
"step": 1070
},
{
"epoch": 1.29,
"learning_rate": 0.0002,
"loss": 0.4041,
"step": 1080
},
{
"epoch": 1.3,
"learning_rate": 0.0002,
"loss": 0.422,
"step": 1090
},
{
"epoch": 1.31,
"learning_rate": 0.0002,
"loss": 0.3907,
"step": 1100
},
{
"epoch": 1.33,
"learning_rate": 0.0002,
"loss": 0.3764,
"step": 1110
},
{
"epoch": 1.34,
"learning_rate": 0.0002,
"loss": 0.4011,
"step": 1120
},
{
"epoch": 1.35,
"learning_rate": 0.0002,
"loss": 0.3779,
"step": 1130
},
{
"epoch": 1.36,
"learning_rate": 0.0002,
"loss": 0.3858,
"step": 1140
},
{
"epoch": 1.37,
"learning_rate": 0.0002,
"loss": 0.4028,
"step": 1150
},
{
"epoch": 1.39,
"learning_rate": 0.0002,
"loss": 0.3845,
"step": 1160
},
{
"epoch": 1.4,
"learning_rate": 0.0002,
"loss": 0.3939,
"step": 1170
},
{
"epoch": 1.41,
"learning_rate": 0.0002,
"loss": 0.3591,
"step": 1180
},
{
"epoch": 1.42,
"learning_rate": 0.0002,
"loss": 0.384,
"step": 1190
},
{
"epoch": 1.43,
"learning_rate": 0.0002,
"loss": 0.4026,
"step": 1200
},
{
"epoch": 1.43,
"eval_loss": 0.4532225430011749,
"eval_runtime": 178.7925,
"eval_samples_per_second": 5.593,
"eval_steps_per_second": 2.797,
"step": 1200
},
{
"epoch": 1.43,
"mmlu_eval_accuracy": 0.45916720892795637,
"mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
"mmlu_eval_accuracy_anatomy": 0.5,
"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": 0.3125,
"mmlu_eval_accuracy_college_chemistry": 0.125,
"mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
"mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
"mmlu_eval_accuracy_college_medicine": 0.45454545454545453,
"mmlu_eval_accuracy_college_physics": 0.5454545454545454,
"mmlu_eval_accuracy_computer_security": 0.36363636363636365,
"mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
"mmlu_eval_accuracy_econometrics": 0.16666666666666666,
"mmlu_eval_accuracy_electrical_engineering": 0.375,
"mmlu_eval_accuracy_elementary_mathematics": 0.24390243902439024,
"mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
"mmlu_eval_accuracy_global_facts": 0.4,
"mmlu_eval_accuracy_high_school_biology": 0.40625,
"mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
"mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
"mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
"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": 0.3103448275862069,
"mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
"mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
"mmlu_eval_accuracy_high_school_psychology": 0.75,
"mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
"mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
"mmlu_eval_accuracy_high_school_world_history": 0.5,
"mmlu_eval_accuracy_human_aging": 0.6956521739130435,
"mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
"mmlu_eval_accuracy_international_law": 0.7692307692307693,
"mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
"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": 0.6744186046511628,
"mmlu_eval_accuracy_moral_disputes": 0.5526315789473685,
"mmlu_eval_accuracy_moral_scenarios": 0.26,
"mmlu_eval_accuracy_nutrition": 0.5757575757575758,
"mmlu_eval_accuracy_philosophy": 0.4411764705882353,
"mmlu_eval_accuracy_prehistory": 0.5428571428571428,
"mmlu_eval_accuracy_professional_accounting": 0.16129032258064516,
"mmlu_eval_accuracy_professional_law": 0.3352941176470588,
"mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
"mmlu_eval_accuracy_professional_psychology": 0.42028985507246375,
"mmlu_eval_accuracy_public_relations": 0.5,
"mmlu_eval_accuracy_security_studies": 0.48148148148148145,
"mmlu_eval_accuracy_sociology": 0.6363636363636364,
"mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
"mmlu_eval_accuracy_virology": 0.3888888888888889,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.0201030533478714,
"step": 1200
},
{
"epoch": 1.45,
"learning_rate": 0.0002,
"loss": 0.3933,
"step": 1210
},
{
"epoch": 1.46,
"learning_rate": 0.0002,
"loss": 0.3587,
"step": 1220
},
{
"epoch": 1.47,
"learning_rate": 0.0002,
"loss": 0.3688,
"step": 1230
},
{
"epoch": 1.48,
"learning_rate": 0.0002,
"loss": 0.3514,
"step": 1240
},
{
"epoch": 1.49,
"learning_rate": 0.0002,
"loss": 0.3732,
"step": 1250
},
{
"epoch": 1.51,
"learning_rate": 0.0002,
"loss": 0.3918,
"step": 1260
},
{
"epoch": 1.52,
"learning_rate": 0.0002,
"loss": 0.387,
"step": 1270
},
{
"epoch": 1.53,
"learning_rate": 0.0002,
"loss": 0.3821,
"step": 1280
},
{
"epoch": 1.54,
"learning_rate": 0.0002,
"loss": 0.3814,
"step": 1290
},
{
"epoch": 1.55,
"learning_rate": 0.0002,
"loss": 0.3925,
"step": 1300
},
{
"epoch": 1.57,
"learning_rate": 0.0002,
"loss": 0.3949,
"step": 1310
},
{
"epoch": 1.58,
"learning_rate": 0.0002,
"loss": 0.3587,
"step": 1320
},
{
"epoch": 1.59,
"learning_rate": 0.0002,
"loss": 0.3992,
"step": 1330
},
{
"epoch": 1.6,
"learning_rate": 0.0002,
"loss": 0.3987,
"step": 1340
},
{
"epoch": 1.61,
"learning_rate": 0.0002,
"loss": 0.3744,
"step": 1350
},
{
"epoch": 1.62,
"learning_rate": 0.0002,
"loss": 0.3741,
"step": 1360
},
{
"epoch": 1.64,
"learning_rate": 0.0002,
"loss": 0.3756,
"step": 1370
},
{
"epoch": 1.65,
"learning_rate": 0.0002,
"loss": 0.3862,
"step": 1380
},
{
"epoch": 1.66,
"learning_rate": 0.0002,
"loss": 0.361,
"step": 1390
},
{
"epoch": 1.67,
"learning_rate": 0.0002,
"loss": 0.3712,
"step": 1400
},
{
"epoch": 1.67,
"eval_loss": 0.4475863575935364,
"eval_runtime": 179.1443,
"eval_samples_per_second": 5.582,
"eval_steps_per_second": 2.791,
"step": 1400
},
{
"epoch": 1.67,
"mmlu_eval_accuracy": 0.4631075705147942,
"mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
"mmlu_eval_accuracy_anatomy": 0.5714285714285714,
"mmlu_eval_accuracy_astronomy": 0.375,
"mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
"mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
"mmlu_eval_accuracy_college_biology": 0.375,
"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": 0.36363636363636365,
"mmlu_eval_accuracy_college_physics": 0.6363636363636364,
"mmlu_eval_accuracy_computer_security": 0.2727272727272727,
"mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
"mmlu_eval_accuracy_econometrics": 0.25,
"mmlu_eval_accuracy_electrical_engineering": 0.4375,
"mmlu_eval_accuracy_elementary_mathematics": 0.2682926829268293,
"mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
"mmlu_eval_accuracy_global_facts": 0.5,
"mmlu_eval_accuracy_high_school_biology": 0.4375,
"mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
"mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
"mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
"mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.3023255813953488,
"mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
"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": 0.4782608695652174,
"mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
"mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
"mmlu_eval_accuracy_human_aging": 0.6086956521739131,
"mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
"mmlu_eval_accuracy_international_law": 0.6923076923076923,
"mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
"mmlu_eval_accuracy_logical_fallacies": 0.5,
"mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
"mmlu_eval_accuracy_management": 0.6363636363636364,
"mmlu_eval_accuracy_marketing": 0.64,
"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": 0.29,
"mmlu_eval_accuracy_nutrition": 0.6060606060606061,
"mmlu_eval_accuracy_philosophy": 0.5294117647058824,
"mmlu_eval_accuracy_prehistory": 0.45714285714285713,
"mmlu_eval_accuracy_professional_accounting": 0.16129032258064516,
"mmlu_eval_accuracy_professional_law": 0.35294117647058826,
"mmlu_eval_accuracy_professional_medicine": 0.3548387096774194,
"mmlu_eval_accuracy_professional_psychology": 0.3188405797101449,
"mmlu_eval_accuracy_public_relations": 0.5,
"mmlu_eval_accuracy_security_studies": 0.48148148148148145,
"mmlu_eval_accuracy_sociology": 0.6363636363636364,
"mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
"mmlu_eval_accuracy_virology": 0.3888888888888889,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.180907432194167,
"step": 1400
},
{
"epoch": 1.68,
"learning_rate": 0.0002,
"loss": 0.3929,
"step": 1410
},
{
"epoch": 1.7,
"learning_rate": 0.0002,
"loss": 0.3882,
"step": 1420
},
{
"epoch": 1.71,
"learning_rate": 0.0002,
"loss": 0.3797,
"step": 1430
},
{
"epoch": 1.72,
"learning_rate": 0.0002,
"loss": 0.3992,
"step": 1440
},
{
"epoch": 1.73,
"learning_rate": 0.0002,
"loss": 0.3774,
"step": 1450
},
{
"epoch": 1.74,
"learning_rate": 0.0002,
"loss": 0.3677,
"step": 1460
},
{
"epoch": 1.76,
"learning_rate": 0.0002,
"loss": 0.3697,
"step": 1470
},
{
"epoch": 1.77,
"learning_rate": 0.0002,
"loss": 0.3683,
"step": 1480
},
{
"epoch": 1.78,
"learning_rate": 0.0002,
"loss": 0.3703,
"step": 1490
},
{
"epoch": 1.79,
"learning_rate": 0.0002,
"loss": 0.3953,
"step": 1500
},
{
"epoch": 1.8,
"learning_rate": 0.0002,
"loss": 0.4016,
"step": 1510
},
{
"epoch": 1.82,
"learning_rate": 0.0002,
"loss": 0.374,
"step": 1520
},
{
"epoch": 1.83,
"learning_rate": 0.0002,
"loss": 0.3753,
"step": 1530
},
{
"epoch": 1.84,
"learning_rate": 0.0002,
"loss": 0.3884,
"step": 1540
},
{
"epoch": 1.85,
"learning_rate": 0.0002,
"loss": 0.3588,
"step": 1550
},
{
"epoch": 1.86,
"learning_rate": 0.0002,
"loss": 0.3988,
"step": 1560
},
{
"epoch": 1.88,
"learning_rate": 0.0002,
"loss": 0.3697,
"step": 1570
},
{
"epoch": 1.89,
"learning_rate": 0.0002,
"loss": 0.3937,
"step": 1580
},
{
"epoch": 1.9,
"learning_rate": 0.0002,
"loss": 0.3856,
"step": 1590
},
{
"epoch": 1.91,
"learning_rate": 0.0002,
"loss": 0.363,
"step": 1600
},
{
"epoch": 1.91,
"eval_loss": 0.4432196617126465,
"eval_runtime": 178.8367,
"eval_samples_per_second": 5.592,
"eval_steps_per_second": 2.796,
"step": 1600
},
{
"epoch": 1.91,
"mmlu_eval_accuracy": 0.46789885257340896,
"mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
"mmlu_eval_accuracy_anatomy": 0.5,
"mmlu_eval_accuracy_astronomy": 0.5,
"mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
"mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
"mmlu_eval_accuracy_college_biology": 0.375,
"mmlu_eval_accuracy_college_chemistry": 0.25,
"mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
"mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
"mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
"mmlu_eval_accuracy_college_physics": 0.45454545454545453,
"mmlu_eval_accuracy_computer_security": 0.18181818181818182,
"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": 0.2682926829268293,
"mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
"mmlu_eval_accuracy_global_facts": 0.5,
"mmlu_eval_accuracy_high_school_biology": 0.40625,
"mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
"mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
"mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
"mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
"mmlu_eval_accuracy_high_school_mathematics": 0.3448275862068966,
"mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
"mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
"mmlu_eval_accuracy_high_school_psychology": 0.7833333333333333,
"mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
"mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
"mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769,
"mmlu_eval_accuracy_human_aging": 0.6086956521739131,
"mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
"mmlu_eval_accuracy_international_law": 0.6923076923076923,
"mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
"mmlu_eval_accuracy_logical_fallacies": 0.5,
"mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
"mmlu_eval_accuracy_management": 0.6363636363636364,
"mmlu_eval_accuracy_marketing": 0.68,
"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": 0.25,
"mmlu_eval_accuracy_nutrition": 0.6060606060606061,
"mmlu_eval_accuracy_philosophy": 0.5294117647058824,
"mmlu_eval_accuracy_prehistory": 0.5714285714285714,
"mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
"mmlu_eval_accuracy_professional_law": 0.3411764705882353,
"mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
"mmlu_eval_accuracy_professional_psychology": 0.37681159420289856,
"mmlu_eval_accuracy_public_relations": 0.5,
"mmlu_eval_accuracy_security_studies": 0.48148148148148145,
"mmlu_eval_accuracy_sociology": 0.6363636363636364,
"mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
"mmlu_eval_accuracy_virology": 0.4444444444444444,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.0490072815345877,
"step": 1600
},
{
"epoch": 1.92,
"learning_rate": 0.0002,
"loss": 0.3555,
"step": 1610
},
{
"epoch": 1.94,
"learning_rate": 0.0002,
"loss": 0.4285,
"step": 1620
},
{
"epoch": 1.95,
"learning_rate": 0.0002,
"loss": 0.3708,
"step": 1630
},
{
"epoch": 1.96,
"learning_rate": 0.0002,
"loss": 0.3781,
"step": 1640
},
{
"epoch": 1.97,
"learning_rate": 0.0002,
"loss": 0.368,
"step": 1650
},
{
"epoch": 1.98,
"learning_rate": 0.0002,
"loss": 0.3784,
"step": 1660
},
{
"epoch": 2.0,
"learning_rate": 0.0002,
"loss": 0.3709,
"step": 1670
},
{
"epoch": 2.01,
"learning_rate": 0.0002,
"loss": 0.3228,
"step": 1680
},
{
"epoch": 2.02,
"learning_rate": 0.0002,
"loss": 0.3244,
"step": 1690
},
{
"epoch": 2.03,
"learning_rate": 0.0002,
"loss": 0.304,
"step": 1700
},
{
"epoch": 2.04,
"learning_rate": 0.0002,
"loss": 0.2811,
"step": 1710
},
{
"epoch": 2.05,
"learning_rate": 0.0002,
"loss": 0.3046,
"step": 1720
},
{
"epoch": 2.07,
"learning_rate": 0.0002,
"loss": 0.3062,
"step": 1730
},
{
"epoch": 2.08,
"learning_rate": 0.0002,
"loss": 0.2967,
"step": 1740
},
{
"epoch": 2.09,
"learning_rate": 0.0002,
"loss": 0.2936,
"step": 1750
},
{
"epoch": 2.1,
"learning_rate": 0.0002,
"loss": 0.2899,
"step": 1760
},
{
"epoch": 2.11,
"learning_rate": 0.0002,
"loss": 0.2915,
"step": 1770
},
{
"epoch": 2.13,
"learning_rate": 0.0002,
"loss": 0.3226,
"step": 1780
},
{
"epoch": 2.14,
"learning_rate": 0.0002,
"loss": 0.3043,
"step": 1790
},
{
"epoch": 2.15,
"learning_rate": 0.0002,
"loss": 0.3054,
"step": 1800
},
{
"epoch": 2.15,
"eval_loss": 0.45656338334083557,
"eval_runtime": 179.3489,
"eval_samples_per_second": 5.576,
"eval_steps_per_second": 2.788,
"step": 1800
},
{
"epoch": 2.15,
"mmlu_eval_accuracy": 0.45495968338399634,
"mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
"mmlu_eval_accuracy_anatomy": 0.5,
"mmlu_eval_accuracy_astronomy": 0.375,
"mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
"mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
"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": 0.36363636363636365,
"mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
"mmlu_eval_accuracy_college_physics": 0.45454545454545453,
"mmlu_eval_accuracy_computer_security": 0.18181818181818182,
"mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
"mmlu_eval_accuracy_econometrics": 0.16666666666666666,
"mmlu_eval_accuracy_electrical_engineering": 0.25,
"mmlu_eval_accuracy_elementary_mathematics": 0.2682926829268293,
"mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
"mmlu_eval_accuracy_global_facts": 0.4,
"mmlu_eval_accuracy_high_school_biology": 0.40625,
"mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
"mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
"mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
"mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
"mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
"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": 0.6818181818181818,
"mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
"mmlu_eval_accuracy_human_aging": 0.5652173913043478,
"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": 0.5,
"mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
"mmlu_eval_accuracy_management": 0.7272727272727273,
"mmlu_eval_accuracy_marketing": 0.64,
"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": 0.27,
"mmlu_eval_accuracy_nutrition": 0.6060606060606061,
"mmlu_eval_accuracy_philosophy": 0.5,
"mmlu_eval_accuracy_prehistory": 0.5428571428571428,
"mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
"mmlu_eval_accuracy_professional_law": 0.3411764705882353,
"mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
"mmlu_eval_accuracy_professional_psychology": 0.34782608695652173,
"mmlu_eval_accuracy_public_relations": 0.5,
"mmlu_eval_accuracy_security_studies": 0.5185185185185185,
"mmlu_eval_accuracy_sociology": 0.5454545454545454,
"mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
"mmlu_eval_accuracy_virology": 0.3888888888888889,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.0489876337213242,
"step": 1800
},
{
"epoch": 2.16,
"learning_rate": 0.0002,
"loss": 0.2987,
"step": 1810
},
{
"epoch": 2.17,
"learning_rate": 0.0002,
"loss": 0.2989,
"step": 1820
},
{
"epoch": 2.19,
"learning_rate": 0.0002,
"loss": 0.3031,
"step": 1830
},
{
"epoch": 2.2,
"learning_rate": 0.0002,
"loss": 0.3111,
"step": 1840
},
{
"epoch": 2.21,
"learning_rate": 0.0002,
"loss": 0.2976,
"step": 1850
},
{
"epoch": 2.22,
"learning_rate": 0.0002,
"loss": 0.2884,
"step": 1860
},
{
"epoch": 2.23,
"learning_rate": 0.0002,
"loss": 0.2923,
"step": 1870
},
{
"epoch": 2.25,
"learning_rate": 0.0002,
"loss": 0.3239,
"step": 1880
},
{
"epoch": 2.26,
"learning_rate": 0.0002,
"loss": 0.3251,
"step": 1890
},
{
"epoch": 2.27,
"learning_rate": 0.0002,
"loss": 0.3147,
"step": 1900
},
{
"epoch": 2.28,
"learning_rate": 0.0002,
"loss": 0.3403,
"step": 1910
},
{
"epoch": 2.29,
"learning_rate": 0.0002,
"loss": 0.3004,
"step": 1920
},
{
"epoch": 2.31,
"learning_rate": 0.0002,
"loss": 0.3127,
"step": 1930
},
{
"epoch": 2.32,
"learning_rate": 0.0002,
"loss": 0.3,
"step": 1940
},
{
"epoch": 2.33,
"learning_rate": 0.0002,
"loss": 0.3137,
"step": 1950
},
{
"epoch": 2.34,
"learning_rate": 0.0002,
"loss": 0.3002,
"step": 1960
},
{
"epoch": 2.35,
"learning_rate": 0.0002,
"loss": 0.3065,
"step": 1970
},
{
"epoch": 2.37,
"learning_rate": 0.0002,
"loss": 0.2968,
"step": 1980
},
{
"epoch": 2.38,
"learning_rate": 0.0002,
"loss": 0.3102,
"step": 1990
},
{
"epoch": 2.39,
"learning_rate": 0.0002,
"loss": 0.3331,
"step": 2000
},
{
"epoch": 2.39,
"eval_loss": 0.453141987323761,
"eval_runtime": 179.2289,
"eval_samples_per_second": 5.579,
"eval_steps_per_second": 2.79,
"step": 2000
},
{
"epoch": 2.39,
"mmlu_eval_accuracy": 0.45313748079668487,
"mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
"mmlu_eval_accuracy_anatomy": 0.5,
"mmlu_eval_accuracy_astronomy": 0.375,
"mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
"mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
"mmlu_eval_accuracy_college_biology": 0.4375,
"mmlu_eval_accuracy_college_chemistry": 0.25,
"mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
"mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
"mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
"mmlu_eval_accuracy_college_physics": 0.45454545454545453,
"mmlu_eval_accuracy_computer_security": 0.18181818181818182,
"mmlu_eval_accuracy_conceptual_physics": 0.3076923076923077,
"mmlu_eval_accuracy_econometrics": 0.16666666666666666,
"mmlu_eval_accuracy_electrical_engineering": 0.3125,
"mmlu_eval_accuracy_elementary_mathematics": 0.2682926829268293,
"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": 0.36363636363636365,
"mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
"mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
"mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
"mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
"mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
"mmlu_eval_accuracy_high_school_psychology": 0.7666666666666667,
"mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
"mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
"mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
"mmlu_eval_accuracy_human_aging": 0.6521739130434783,
"mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
"mmlu_eval_accuracy_international_law": 0.6923076923076923,
"mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
"mmlu_eval_accuracy_logical_fallacies": 0.5,
"mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
"mmlu_eval_accuracy_management": 0.6363636363636364,
"mmlu_eval_accuracy_marketing": 0.64,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
"mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
"mmlu_eval_accuracy_moral_scenarios": 0.29,
"mmlu_eval_accuracy_nutrition": 0.6060606060606061,
"mmlu_eval_accuracy_philosophy": 0.5588235294117647,
"mmlu_eval_accuracy_prehistory": 0.5142857142857142,
"mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
"mmlu_eval_accuracy_professional_law": 0.34705882352941175,
"mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
"mmlu_eval_accuracy_professional_psychology": 0.37681159420289856,
"mmlu_eval_accuracy_public_relations": 0.5,
"mmlu_eval_accuracy_security_studies": 0.48148148148148145,
"mmlu_eval_accuracy_sociology": 0.5909090909090909,
"mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
"mmlu_eval_accuracy_virology": 0.4444444444444444,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.14181383113637,
"step": 2000
},
{
"epoch": 2.4,
"learning_rate": 0.0002,
"loss": 0.2785,
"step": 2010
},
{
"epoch": 2.41,
"learning_rate": 0.0002,
"loss": 0.3038,
"step": 2020
},
{
"epoch": 2.43,
"learning_rate": 0.0002,
"loss": 0.3259,
"step": 2030
},
{
"epoch": 2.44,
"learning_rate": 0.0002,
"loss": 0.3308,
"step": 2040
},
{
"epoch": 2.45,
"learning_rate": 0.0002,
"loss": 0.325,
"step": 2050
},
{
"epoch": 2.46,
"learning_rate": 0.0002,
"loss": 0.3271,
"step": 2060
},
{
"epoch": 2.47,
"learning_rate": 0.0002,
"loss": 0.3187,
"step": 2070
},
{
"epoch": 2.49,
"learning_rate": 0.0002,
"loss": 0.3192,
"step": 2080
},
{
"epoch": 2.5,
"learning_rate": 0.0002,
"loss": 0.3227,
"step": 2090
},
{
"epoch": 2.51,
"learning_rate": 0.0002,
"loss": 0.2969,
"step": 2100
},
{
"epoch": 2.52,
"learning_rate": 0.0002,
"loss": 0.334,
"step": 2110
},
{
"epoch": 2.53,
"learning_rate": 0.0002,
"loss": 0.3266,
"step": 2120
},
{
"epoch": 2.54,
"learning_rate": 0.0002,
"loss": 0.3085,
"step": 2130
},
{
"epoch": 2.56,
"learning_rate": 0.0002,
"loss": 0.2917,
"step": 2140
},
{
"epoch": 2.57,
"learning_rate": 0.0002,
"loss": 0.3142,
"step": 2150
},
{
"epoch": 2.58,
"learning_rate": 0.0002,
"loss": 0.2954,
"step": 2160
},
{
"epoch": 2.59,
"learning_rate": 0.0002,
"loss": 0.3341,
"step": 2170
},
{
"epoch": 2.6,
"learning_rate": 0.0002,
"loss": 0.3129,
"step": 2180
},
{
"epoch": 2.62,
"learning_rate": 0.0002,
"loss": 0.2964,
"step": 2190
},
{
"epoch": 2.63,
"learning_rate": 0.0002,
"loss": 0.3069,
"step": 2200
},
{
"epoch": 2.63,
"eval_loss": 0.45312267541885376,
"eval_runtime": 179.1132,
"eval_samples_per_second": 5.583,
"eval_steps_per_second": 2.792,
"step": 2200
},
{
"epoch": 2.63,
"mmlu_eval_accuracy": 0.45786090120959316,
"mmlu_eval_accuracy_abstract_algebra": 0.45454545454545453,
"mmlu_eval_accuracy_anatomy": 0.5,
"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": 0.25,
"mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
"mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
"mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
"mmlu_eval_accuracy_college_physics": 0.45454545454545453,
"mmlu_eval_accuracy_computer_security": 0.2727272727272727,
"mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
"mmlu_eval_accuracy_econometrics": 0.16666666666666666,
"mmlu_eval_accuracy_electrical_engineering": 0.25,
"mmlu_eval_accuracy_elementary_mathematics": 0.2926829268292683,
"mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
"mmlu_eval_accuracy_global_facts": 0.4,
"mmlu_eval_accuracy_high_school_biology": 0.4375,
"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": 0.5238095238095238,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
"mmlu_eval_accuracy_high_school_mathematics": 0.3448275862068966,
"mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
"mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
"mmlu_eval_accuracy_high_school_psychology": 0.7166666666666667,
"mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
"mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
"mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
"mmlu_eval_accuracy_human_aging": 0.6086956521739131,
"mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
"mmlu_eval_accuracy_international_law": 0.6923076923076923,
"mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
"mmlu_eval_accuracy_logical_fallacies": 0.5,
"mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
"mmlu_eval_accuracy_management": 0.6363636363636364,
"mmlu_eval_accuracy_marketing": 0.64,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.6976744186046512,
"mmlu_eval_accuracy_moral_disputes": 0.5,
"mmlu_eval_accuracy_moral_scenarios": 0.27,
"mmlu_eval_accuracy_nutrition": 0.6060606060606061,
"mmlu_eval_accuracy_philosophy": 0.5588235294117647,
"mmlu_eval_accuracy_prehistory": 0.5428571428571428,
"mmlu_eval_accuracy_professional_accounting": 0.1935483870967742,
"mmlu_eval_accuracy_professional_law": 0.3588235294117647,
"mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
"mmlu_eval_accuracy_professional_psychology": 0.391304347826087,
"mmlu_eval_accuracy_public_relations": 0.5,
"mmlu_eval_accuracy_security_studies": 0.5185185185185185,
"mmlu_eval_accuracy_sociology": 0.5454545454545454,
"mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
"mmlu_eval_accuracy_virology": 0.4444444444444444,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.1481178610855542,
"step": 2200
}
],
"max_steps": 5000,
"num_train_epochs": 6,
"total_flos": 4.497885537845576e+17,
"trial_name": null,
"trial_params": null
}