prateeky2806's picture
Training in progress, step 200
9f7c774
raw
history blame
6.56 kB
{
"best_metric": 0.45071524381637573,
"best_model_checkpoint": "./output_v2/7b_cluster014_Nous-Hermes-llama-2-7b_partitioned_v3_standardized_014/checkpoint-200",
"epoch": 0.35382574082264484,
"global_step": 200,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 0.6451,
"step": 10
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 0.5699,
"step": 20
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 0.5073,
"step": 30
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 0.4662,
"step": 40
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 0.4545,
"step": 50
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 0.4675,
"step": 60
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 0.4524,
"step": 70
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 0.4799,
"step": 80
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 0.5122,
"step": 90
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 0.461,
"step": 100
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 0.4393,
"step": 110
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 0.4981,
"step": 120
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 0.4686,
"step": 130
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 0.469,
"step": 140
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 0.4926,
"step": 150
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 0.4213,
"step": 160
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 0.4412,
"step": 170
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 0.4607,
"step": 180
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 0.4537,
"step": 190
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 0.4358,
"step": 200
},
{
"epoch": 0.35,
"eval_loss": 0.45071524381637573,
"eval_runtime": 191.6209,
"eval_samples_per_second": 5.219,
"eval_steps_per_second": 2.609,
"step": 200
},
{
"epoch": 0.35,
"mmlu_eval_accuracy": 0.4662069900433653,
"mmlu_eval_accuracy_abstract_algebra": 0.09090909090909091,
"mmlu_eval_accuracy_anatomy": 0.5714285714285714,
"mmlu_eval_accuracy_astronomy": 0.4375,
"mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
"mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
"mmlu_eval_accuracy_college_biology": 0.4375,
"mmlu_eval_accuracy_college_chemistry": 0.125,
"mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
"mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
"mmlu_eval_accuracy_college_medicine": 0.4090909090909091,
"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": 0.4375,
"mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
"mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
"mmlu_eval_accuracy_global_facts": 0.6,
"mmlu_eval_accuracy_high_school_biology": 0.375,
"mmlu_eval_accuracy_high_school_chemistry": 0.2727272727272727,
"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": 0.3103448275862069,
"mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
"mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
"mmlu_eval_accuracy_high_school_psychology": 0.7333333333333333,
"mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
"mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
"mmlu_eval_accuracy_high_school_world_history": 0.5,
"mmlu_eval_accuracy_human_aging": 0.7391304347826086,
"mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
"mmlu_eval_accuracy_international_law": 0.7692307692307693,
"mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
"mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
"mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
"mmlu_eval_accuracy_management": 0.5454545454545454,
"mmlu_eval_accuracy_marketing": 0.68,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
"mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.6060606060606061,
"mmlu_eval_accuracy_philosophy": 0.4411764705882353,
"mmlu_eval_accuracy_prehistory": 0.5428571428571428,
"mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
"mmlu_eval_accuracy_professional_law": 0.3411764705882353,
"mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
"mmlu_eval_accuracy_professional_psychology": 0.3333333333333333,
"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": 1.2129778240663887,
"step": 200
}
],
"max_steps": 5000,
"num_train_epochs": 9,
"total_flos": 2.902219844685005e+16,
"trial_name": null,
"trial_params": null
}