prateeky2806's picture
Training in progress, step 200
3fcfc71
{
"best_metric": 0.7327473163604736,
"best_model_checkpoint": "./output_v2/7b_cluster011_Nous-Hermes-llama-2-7b_partitioned_v3_standardized_011/checkpoint-200",
"epoch": 0.4626951995373048,
"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": 0.7957,
"step": 10
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 0.718,
"step": 20
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 0.73,
"step": 30
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 0.7826,
"step": 40
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 0.7013,
"step": 50
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 0.7353,
"step": 60
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 0.6666,
"step": 70
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 0.7649,
"step": 80
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 0.7018,
"step": 90
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 0.7173,
"step": 100
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 0.7857,
"step": 110
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 0.7154,
"step": 120
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 0.7485,
"step": 130
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 0.7114,
"step": 140
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 0.7333,
"step": 150
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 0.6549,
"step": 160
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 0.6765,
"step": 170
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 0.677,
"step": 180
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 0.6763,
"step": 190
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 0.6638,
"step": 200
},
{
"epoch": 0.46,
"eval_loss": 0.7327473163604736,
"eval_runtime": 246.3779,
"eval_samples_per_second": 4.059,
"eval_steps_per_second": 2.029,
"step": 200
},
{
"epoch": 0.46,
"mmlu_eval_accuracy": 0.4592376175825003,
"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": 0.4375,
"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": 0.2727272727272727,
"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": 0.2926829268292683,
"mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
"mmlu_eval_accuracy_global_facts": 0.4,
"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": 0.5555555555555556,
"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": 0.38461538461538464,
"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": 0.6818181818181818,
"mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
"mmlu_eval_accuracy_human_aging": 0.6956521739130435,
"mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
"mmlu_eval_accuracy_international_law": 0.6923076923076923,
"mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
"mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
"mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
"mmlu_eval_accuracy_management": 0.6363636363636364,
"mmlu_eval_accuracy_marketing": 0.72,
"mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
"mmlu_eval_accuracy_miscellaneous": 0.7093023255813954,
"mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
"mmlu_eval_accuracy_moral_scenarios": 0.24,
"mmlu_eval_accuracy_nutrition": 0.5757575757575758,
"mmlu_eval_accuracy_philosophy": 0.38235294117647056,
"mmlu_eval_accuracy_prehistory": 0.45714285714285713,
"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": 0.391304347826087,
"mmlu_eval_accuracy_public_relations": 0.4166666666666667,
"mmlu_eval_accuracy_security_studies": 0.5185185185185185,
"mmlu_eval_accuracy_sociology": 0.7272727272727273,
"mmlu_eval_accuracy_us_foreign_policy": 0.45454545454545453,
"mmlu_eval_accuracy_virology": 0.4444444444444444,
"mmlu_eval_accuracy_world_religions": 0.7368421052631579,
"mmlu_loss": 1.173365458184683,
"step": 200
}
],
"max_steps": 5000,
"num_train_epochs": 12,
"total_flos": 5.784949765973606e+16,
"trial_name": null,
"trial_params": null
}