prateeky2806's picture
Training in progress, step 600
5069e06
raw
history blame
18.7 kB
{
"best_metric": 0.6469126343727112,
"best_model_checkpoint": "./output_v2/7b_cluster012_Nous-Hermes-llama-2-7b_partitioned_v3_standardized_012/checkpoint-600",
"epoch": 0.6948465547191662,
"global_step": 600,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.7418,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 0.7269,
"step": 20
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 0.7061,
"step": 30
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 0.6809,
"step": 40
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 0.6742,
"step": 50
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 0.6789,
"step": 60
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 0.6628,
"step": 70
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 0.6995,
"step": 80
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 0.6774,
"step": 90
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 0.6716,
"step": 100
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 0.6628,
"step": 110
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 0.6696,
"step": 120
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 0.648,
"step": 130
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 0.6823,
"step": 140
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 0.6532,
"step": 150
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 0.669,
"step": 160
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 0.6639,
"step": 170
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 0.6582,
"step": 180
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 0.6381,
"step": 190
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 0.6342,
"step": 200
},
{
"epoch": 0.23,
"eval_loss": 0.6798371076583862,
"eval_runtime": 220.8775,
"eval_samples_per_second": 4.527,
"eval_steps_per_second": 2.264,
"step": 200
},
{
"epoch": 0.23,
"mmlu_eval_accuracy": 0.4661323913730994,
"mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
"mmlu_eval_accuracy_anatomy": 0.5714285714285714,
"mmlu_eval_accuracy_astronomy": 0.5,
"mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
"mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
"mmlu_eval_accuracy_college_biology": 0.4375,
"mmlu_eval_accuracy_college_chemistry": 0.125,
"mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
"mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
"mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
"mmlu_eval_accuracy_college_physics": 0.45454545454545453,
"mmlu_eval_accuracy_computer_security": 0.36363636363636365,
"mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
"mmlu_eval_accuracy_econometrics": 0.16666666666666666,
"mmlu_eval_accuracy_electrical_engineering": 0.3125,
"mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
"mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
"mmlu_eval_accuracy_global_facts": 0.5,
"mmlu_eval_accuracy_high_school_biology": 0.375,
"mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
"mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
"mmlu_eval_accuracy_high_school_geography": 0.6818181818181818,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
"mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
"mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
"mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
"mmlu_eval_accuracy_high_school_psychology": 0.7166666666666667,
"mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
"mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
"mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769,
"mmlu_eval_accuracy_human_aging": 0.6956521739130435,
"mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
"mmlu_eval_accuracy_international_law": 0.7692307692307693,
"mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
"mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
"mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
"mmlu_eval_accuracy_management": 0.6363636363636364,
"mmlu_eval_accuracy_marketing": 0.76,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
"mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.5757575757575758,
"mmlu_eval_accuracy_philosophy": 0.5,
"mmlu_eval_accuracy_prehistory": 0.45714285714285713,
"mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
"mmlu_eval_accuracy_professional_law": 0.35294117647058826,
"mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
"mmlu_eval_accuracy_professional_psychology": 0.391304347826087,
"mmlu_eval_accuracy_public_relations": 0.5,
"mmlu_eval_accuracy_security_studies": 0.5185185185185185,
"mmlu_eval_accuracy_sociology": 0.6818181818181818,
"mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
"mmlu_eval_accuracy_virology": 0.3888888888888889,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.1315238676556718,
"step": 200
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 0.6574,
"step": 210
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 0.6601,
"step": 220
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 0.6499,
"step": 230
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 0.6469,
"step": 240
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 0.6546,
"step": 250
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 0.6681,
"step": 260
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 0.6403,
"step": 270
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 0.676,
"step": 280
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 0.6548,
"step": 290
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 0.6394,
"step": 300
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 0.6173,
"step": 310
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 0.6425,
"step": 320
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 0.6379,
"step": 330
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 0.6191,
"step": 340
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 0.6645,
"step": 350
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 0.5998,
"step": 360
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 0.6402,
"step": 370
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 0.6291,
"step": 380
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 0.6202,
"step": 390
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 0.6532,
"step": 400
},
{
"epoch": 0.46,
"eval_loss": 0.6604505777359009,
"eval_runtime": 220.7584,
"eval_samples_per_second": 4.53,
"eval_steps_per_second": 2.265,
"step": 400
},
{
"epoch": 0.46,
"mmlu_eval_accuracy": 0.466397920486829,
"mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
"mmlu_eval_accuracy_anatomy": 0.5714285714285714,
"mmlu_eval_accuracy_astronomy": 0.375,
"mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
"mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
"mmlu_eval_accuracy_college_biology": 0.4375,
"mmlu_eval_accuracy_college_chemistry": 0.125,
"mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
"mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
"mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
"mmlu_eval_accuracy_college_physics": 0.45454545454545453,
"mmlu_eval_accuracy_computer_security": 0.36363636363636365,
"mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
"mmlu_eval_accuracy_econometrics": 0.16666666666666666,
"mmlu_eval_accuracy_electrical_engineering": 0.3125,
"mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
"mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
"mmlu_eval_accuracy_global_facts": 0.5,
"mmlu_eval_accuracy_high_school_biology": 0.375,
"mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
"mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_european_history": 0.5,
"mmlu_eval_accuracy_high_school_geography": 0.6818181818181818,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
"mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
"mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
"mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
"mmlu_eval_accuracy_high_school_psychology": 0.75,
"mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
"mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
"mmlu_eval_accuracy_high_school_world_history": 0.5,
"mmlu_eval_accuracy_human_aging": 0.6956521739130435,
"mmlu_eval_accuracy_human_sexuality": 0.5,
"mmlu_eval_accuracy_international_law": 0.7692307692307693,
"mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
"mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
"mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
"mmlu_eval_accuracy_management": 0.5454545454545454,
"mmlu_eval_accuracy_marketing": 0.76,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.686046511627907,
"mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.6060606060606061,
"mmlu_eval_accuracy_philosophy": 0.4411764705882353,
"mmlu_eval_accuracy_prehistory": 0.42857142857142855,
"mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
"mmlu_eval_accuracy_professional_law": 0.3764705882352941,
"mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
"mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
"mmlu_eval_accuracy_public_relations": 0.5,
"mmlu_eval_accuracy_security_studies": 0.5925925925925926,
"mmlu_eval_accuracy_sociology": 0.6818181818181818,
"mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
"mmlu_eval_accuracy_virology": 0.3888888888888889,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 0.9161743937308265,
"step": 400
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 0.6487,
"step": 410
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 0.6248,
"step": 420
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 0.6421,
"step": 430
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 0.6679,
"step": 440
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 0.6202,
"step": 450
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 0.7009,
"step": 460
},
{
"epoch": 0.54,
"learning_rate": 0.0002,
"loss": 0.627,
"step": 470
},
{
"epoch": 0.56,
"learning_rate": 0.0002,
"loss": 0.6294,
"step": 480
},
{
"epoch": 0.57,
"learning_rate": 0.0002,
"loss": 0.6254,
"step": 490
},
{
"epoch": 0.58,
"learning_rate": 0.0002,
"loss": 0.6174,
"step": 500
},
{
"epoch": 0.59,
"learning_rate": 0.0002,
"loss": 0.6225,
"step": 510
},
{
"epoch": 0.6,
"learning_rate": 0.0002,
"loss": 0.6703,
"step": 520
},
{
"epoch": 0.61,
"learning_rate": 0.0002,
"loss": 0.6232,
"step": 530
},
{
"epoch": 0.63,
"learning_rate": 0.0002,
"loss": 0.6606,
"step": 540
},
{
"epoch": 0.64,
"learning_rate": 0.0002,
"loss": 0.5992,
"step": 550
},
{
"epoch": 0.65,
"learning_rate": 0.0002,
"loss": 0.6118,
"step": 560
},
{
"epoch": 0.66,
"learning_rate": 0.0002,
"loss": 0.642,
"step": 570
},
{
"epoch": 0.67,
"learning_rate": 0.0002,
"loss": 0.6149,
"step": 580
},
{
"epoch": 0.68,
"learning_rate": 0.0002,
"loss": 0.6148,
"step": 590
},
{
"epoch": 0.69,
"learning_rate": 0.0002,
"loss": 0.6258,
"step": 600
},
{
"epoch": 0.69,
"eval_loss": 0.6469126343727112,
"eval_runtime": 220.793,
"eval_samples_per_second": 4.529,
"eval_steps_per_second": 2.265,
"step": 600
},
{
"epoch": 0.69,
"mmlu_eval_accuracy": 0.47179539026686096,
"mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
"mmlu_eval_accuracy_anatomy": 0.5714285714285714,
"mmlu_eval_accuracy_astronomy": 0.5,
"mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
"mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
"mmlu_eval_accuracy_college_biology": 0.4375,
"mmlu_eval_accuracy_college_chemistry": 0.125,
"mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
"mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
"mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
"mmlu_eval_accuracy_college_physics": 0.45454545454545453,
"mmlu_eval_accuracy_computer_security": 0.45454545454545453,
"mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
"mmlu_eval_accuracy_econometrics": 0.16666666666666666,
"mmlu_eval_accuracy_electrical_engineering": 0.375,
"mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
"mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
"mmlu_eval_accuracy_global_facts": 0.5,
"mmlu_eval_accuracy_high_school_biology": 0.34375,
"mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
"mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_european_history": 0.5,
"mmlu_eval_accuracy_high_school_geography": 0.7272727272727273,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
"mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
"mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
"mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
"mmlu_eval_accuracy_high_school_psychology": 0.7166666666666667,
"mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
"mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
"mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
"mmlu_eval_accuracy_human_aging": 0.6956521739130435,
"mmlu_eval_accuracy_human_sexuality": 0.5,
"mmlu_eval_accuracy_international_law": 0.6923076923076923,
"mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
"mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
"mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
"mmlu_eval_accuracy_management": 0.5454545454545454,
"mmlu_eval_accuracy_marketing": 0.8,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
"mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.6363636363636364,
"mmlu_eval_accuracy_philosophy": 0.5294117647058824,
"mmlu_eval_accuracy_prehistory": 0.45714285714285713,
"mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
"mmlu_eval_accuracy_professional_law": 0.3352941176470588,
"mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
"mmlu_eval_accuracy_professional_psychology": 0.37681159420289856,
"mmlu_eval_accuracy_public_relations": 0.5,
"mmlu_eval_accuracy_security_studies": 0.5555555555555556,
"mmlu_eval_accuracy_sociology": 0.6818181818181818,
"mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
"mmlu_eval_accuracy_virology": 0.3888888888888889,
"mmlu_eval_accuracy_world_religions": 0.7894736842105263,
"mmlu_loss": 0.9272451737102889,
"step": 600
}
],
"max_steps": 5000,
"num_train_epochs": 6,
"total_flos": 1.5041596134437683e+17,
"trial_name": null,
"trial_params": null
}