freewilly2-juniper-lora / trainer_state.json
frank098's picture
First model version
fd8cca8
raw
history blame
18.4 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.2972972972972974,
"global_step": 600,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.02,
"learning_rate": 0.0001,
"loss": 0.7874,
"step": 10
},
{
"epoch": 0.04,
"learning_rate": 0.0001,
"loss": 0.7241,
"step": 20
},
{
"epoch": 0.06,
"learning_rate": 0.0001,
"loss": 0.6801,
"step": 30
},
{
"epoch": 0.09,
"learning_rate": 0.0001,
"loss": 0.7074,
"step": 40
},
{
"epoch": 0.11,
"learning_rate": 0.0001,
"loss": 1.2215,
"step": 50
},
{
"epoch": 0.13,
"learning_rate": 0.0001,
"loss": 0.6756,
"step": 60
},
{
"epoch": 0.15,
"learning_rate": 0.0001,
"loss": 0.6333,
"step": 70
},
{
"epoch": 0.17,
"learning_rate": 0.0001,
"loss": 0.6226,
"step": 80
},
{
"epoch": 0.19,
"learning_rate": 0.0001,
"loss": 0.6498,
"step": 90
},
{
"epoch": 0.22,
"learning_rate": 0.0001,
"loss": 1.0335,
"step": 100
},
{
"epoch": 0.24,
"learning_rate": 0.0001,
"loss": 0.6274,
"step": 110
},
{
"epoch": 0.26,
"learning_rate": 0.0001,
"loss": 0.5948,
"step": 120
},
{
"epoch": 0.28,
"learning_rate": 0.0001,
"loss": 0.617,
"step": 130
},
{
"epoch": 0.3,
"learning_rate": 0.0001,
"loss": 0.6596,
"step": 140
},
{
"epoch": 0.32,
"learning_rate": 0.0001,
"loss": 1.1614,
"step": 150
},
{
"epoch": 0.35,
"learning_rate": 0.0001,
"loss": 0.6195,
"step": 160
},
{
"epoch": 0.37,
"learning_rate": 0.0001,
"loss": 0.5798,
"step": 170
},
{
"epoch": 0.39,
"learning_rate": 0.0001,
"loss": 0.5892,
"step": 180
},
{
"epoch": 0.4,
"eval_loss": 0.6906093955039978,
"eval_runtime": 1140.1455,
"eval_samples_per_second": 0.877,
"eval_steps_per_second": 0.877,
"step": 187
},
{
"epoch": 0.4,
"mmlu_eval_accuracy": 0.6468424709524904,
"mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
"mmlu_eval_accuracy_anatomy": 0.5,
"mmlu_eval_accuracy_astronomy": 0.8125,
"mmlu_eval_accuracy_business_ethics": 0.8181818181818182,
"mmlu_eval_accuracy_clinical_knowledge": 0.6551724137931034,
"mmlu_eval_accuracy_college_biology": 0.6875,
"mmlu_eval_accuracy_college_chemistry": 0.25,
"mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
"mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
"mmlu_eval_accuracy_college_medicine": 0.6818181818181818,
"mmlu_eval_accuracy_college_physics": 0.36363636363636365,
"mmlu_eval_accuracy_computer_security": 0.6363636363636364,
"mmlu_eval_accuracy_conceptual_physics": 0.6153846153846154,
"mmlu_eval_accuracy_econometrics": 0.5,
"mmlu_eval_accuracy_electrical_engineering": 0.6875,
"mmlu_eval_accuracy_elementary_mathematics": 0.5121951219512195,
"mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
"mmlu_eval_accuracy_global_facts": 0.5,
"mmlu_eval_accuracy_high_school_biology": 0.8125,
"mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
"mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_european_history": 0.8333333333333334,
"mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.8571428571428571,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.6976744186046512,
"mmlu_eval_accuracy_high_school_mathematics": 0.3793103448275862,
"mmlu_eval_accuracy_high_school_microeconomics": 0.8076923076923077,
"mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
"mmlu_eval_accuracy_high_school_psychology": 0.9166666666666666,
"mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
"mmlu_eval_accuracy_high_school_us_history": 0.9090909090909091,
"mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
"mmlu_eval_accuracy_human_aging": 0.782608695652174,
"mmlu_eval_accuracy_human_sexuality": 0.5833333333333334,
"mmlu_eval_accuracy_international_law": 0.9230769230769231,
"mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
"mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
"mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
"mmlu_eval_accuracy_management": 0.9090909090909091,
"mmlu_eval_accuracy_marketing": 0.88,
"mmlu_eval_accuracy_medical_genetics": 1.0,
"mmlu_eval_accuracy_miscellaneous": 0.8023255813953488,
"mmlu_eval_accuracy_moral_disputes": 0.6842105263157895,
"mmlu_eval_accuracy_moral_scenarios": 0.47,
"mmlu_eval_accuracy_nutrition": 0.7272727272727273,
"mmlu_eval_accuracy_philosophy": 0.7647058823529411,
"mmlu_eval_accuracy_prehistory": 0.6285714285714286,
"mmlu_eval_accuracy_professional_accounting": 0.6129032258064516,
"mmlu_eval_accuracy_professional_law": 0.5470588235294118,
"mmlu_eval_accuracy_professional_medicine": 0.6451612903225806,
"mmlu_eval_accuracy_professional_psychology": 0.7101449275362319,
"mmlu_eval_accuracy_public_relations": 0.5,
"mmlu_eval_accuracy_security_studies": 0.7037037037037037,
"mmlu_eval_accuracy_sociology": 0.9545454545454546,
"mmlu_eval_accuracy_us_foreign_policy": 1.0,
"mmlu_eval_accuracy_virology": 0.5555555555555556,
"mmlu_eval_accuracy_world_religions": 0.8947368421052632,
"mmlu_loss": 1.1459644647311495,
"step": 187
},
{
"epoch": 0.41,
"learning_rate": 0.0001,
"loss": 0.63,
"step": 190
},
{
"epoch": 0.43,
"learning_rate": 0.0001,
"loss": 1.0713,
"step": 200
},
{
"epoch": 0.45,
"learning_rate": 0.0001,
"loss": 0.5899,
"step": 210
},
{
"epoch": 0.48,
"learning_rate": 0.0001,
"loss": 0.5853,
"step": 220
},
{
"epoch": 0.5,
"learning_rate": 0.0001,
"loss": 0.5808,
"step": 230
},
{
"epoch": 0.52,
"learning_rate": 0.0001,
"loss": 0.6217,
"step": 240
},
{
"epoch": 0.54,
"learning_rate": 0.0001,
"loss": 0.9323,
"step": 250
},
{
"epoch": 0.56,
"learning_rate": 0.0001,
"loss": 0.5872,
"step": 260
},
{
"epoch": 0.58,
"learning_rate": 0.0001,
"loss": 0.547,
"step": 270
},
{
"epoch": 0.61,
"learning_rate": 0.0001,
"loss": 0.5627,
"step": 280
},
{
"epoch": 0.63,
"learning_rate": 0.0001,
"loss": 0.6126,
"step": 290
},
{
"epoch": 0.65,
"learning_rate": 0.0001,
"loss": 0.9871,
"step": 300
},
{
"epoch": 0.67,
"learning_rate": 0.0001,
"loss": 0.568,
"step": 310
},
{
"epoch": 0.69,
"learning_rate": 0.0001,
"loss": 0.561,
"step": 320
},
{
"epoch": 0.71,
"learning_rate": 0.0001,
"loss": 0.5675,
"step": 330
},
{
"epoch": 0.74,
"learning_rate": 0.0001,
"loss": 0.5864,
"step": 340
},
{
"epoch": 0.76,
"learning_rate": 0.0001,
"loss": 1.0407,
"step": 350
},
{
"epoch": 0.78,
"learning_rate": 0.0001,
"loss": 0.5774,
"step": 360
},
{
"epoch": 0.8,
"learning_rate": 0.0001,
"loss": 0.5696,
"step": 370
},
{
"epoch": 0.81,
"eval_loss": 0.6527115702629089,
"eval_runtime": 1139.7605,
"eval_samples_per_second": 0.877,
"eval_steps_per_second": 0.877,
"step": 374
},
{
"epoch": 0.81,
"mmlu_eval_accuracy": 0.651057156408483,
"mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
"mmlu_eval_accuracy_anatomy": 0.5,
"mmlu_eval_accuracy_astronomy": 0.8125,
"mmlu_eval_accuracy_business_ethics": 0.8181818181818182,
"mmlu_eval_accuracy_clinical_knowledge": 0.6551724137931034,
"mmlu_eval_accuracy_college_biology": 0.625,
"mmlu_eval_accuracy_college_chemistry": 0.25,
"mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
"mmlu_eval_accuracy_college_mathematics": 0.45454545454545453,
"mmlu_eval_accuracy_college_medicine": 0.6818181818181818,
"mmlu_eval_accuracy_college_physics": 0.45454545454545453,
"mmlu_eval_accuracy_computer_security": 0.6363636363636364,
"mmlu_eval_accuracy_conceptual_physics": 0.6153846153846154,
"mmlu_eval_accuracy_econometrics": 0.5,
"mmlu_eval_accuracy_electrical_engineering": 0.6875,
"mmlu_eval_accuracy_elementary_mathematics": 0.5121951219512195,
"mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
"mmlu_eval_accuracy_global_facts": 0.5,
"mmlu_eval_accuracy_high_school_biology": 0.78125,
"mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
"mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
"mmlu_eval_accuracy_high_school_european_history": 0.8333333333333334,
"mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.8571428571428571,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.6976744186046512,
"mmlu_eval_accuracy_high_school_mathematics": 0.4482758620689655,
"mmlu_eval_accuracy_high_school_microeconomics": 0.8076923076923077,
"mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
"mmlu_eval_accuracy_high_school_psychology": 0.95,
"mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
"mmlu_eval_accuracy_high_school_us_history": 0.9090909090909091,
"mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
"mmlu_eval_accuracy_human_aging": 0.782608695652174,
"mmlu_eval_accuracy_human_sexuality": 0.5833333333333334,
"mmlu_eval_accuracy_international_law": 0.9230769230769231,
"mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
"mmlu_eval_accuracy_logical_fallacies": 0.7222222222222222,
"mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
"mmlu_eval_accuracy_management": 0.9090909090909091,
"mmlu_eval_accuracy_marketing": 0.88,
"mmlu_eval_accuracy_medical_genetics": 1.0,
"mmlu_eval_accuracy_miscellaneous": 0.8023255813953488,
"mmlu_eval_accuracy_moral_disputes": 0.6842105263157895,
"mmlu_eval_accuracy_moral_scenarios": 0.49,
"mmlu_eval_accuracy_nutrition": 0.696969696969697,
"mmlu_eval_accuracy_philosophy": 0.7647058823529411,
"mmlu_eval_accuracy_prehistory": 0.6,
"mmlu_eval_accuracy_professional_accounting": 0.6451612903225806,
"mmlu_eval_accuracy_professional_law": 0.5529411764705883,
"mmlu_eval_accuracy_professional_medicine": 0.6451612903225806,
"mmlu_eval_accuracy_professional_psychology": 0.7101449275362319,
"mmlu_eval_accuracy_public_relations": 0.5,
"mmlu_eval_accuracy_security_studies": 0.6666666666666666,
"mmlu_eval_accuracy_sociology": 0.9545454545454546,
"mmlu_eval_accuracy_us_foreign_policy": 1.0,
"mmlu_eval_accuracy_virology": 0.5555555555555556,
"mmlu_eval_accuracy_world_religions": 0.8947368421052632,
"mmlu_loss": 1.1366865825197574,
"step": 374
},
{
"epoch": 0.82,
"learning_rate": 0.0001,
"loss": 0.5626,
"step": 380
},
{
"epoch": 0.84,
"learning_rate": 0.0001,
"loss": 0.597,
"step": 390
},
{
"epoch": 0.86,
"learning_rate": 0.0001,
"loss": 0.9613,
"step": 400
},
{
"epoch": 0.89,
"learning_rate": 0.0001,
"loss": 0.5486,
"step": 410
},
{
"epoch": 0.91,
"learning_rate": 0.0001,
"loss": 0.5496,
"step": 420
},
{
"epoch": 0.93,
"learning_rate": 0.0001,
"loss": 0.5468,
"step": 430
},
{
"epoch": 0.95,
"learning_rate": 0.0001,
"loss": 0.5914,
"step": 440
},
{
"epoch": 0.97,
"learning_rate": 0.0001,
"loss": 0.9408,
"step": 450
},
{
"epoch": 0.99,
"learning_rate": 0.0001,
"loss": 0.5851,
"step": 460
},
{
"epoch": 1.02,
"learning_rate": 0.0001,
"loss": 0.6053,
"step": 470
},
{
"epoch": 1.04,
"learning_rate": 0.0001,
"loss": 0.4955,
"step": 480
},
{
"epoch": 1.06,
"learning_rate": 0.0001,
"loss": 0.5032,
"step": 490
},
{
"epoch": 1.08,
"learning_rate": 0.0001,
"loss": 0.5175,
"step": 500
},
{
"epoch": 1.1,
"learning_rate": 0.0001,
"loss": 0.6447,
"step": 510
},
{
"epoch": 1.12,
"learning_rate": 0.0001,
"loss": 0.6054,
"step": 520
},
{
"epoch": 1.15,
"learning_rate": 0.0001,
"loss": 0.5034,
"step": 530
},
{
"epoch": 1.17,
"learning_rate": 0.0001,
"loss": 0.5105,
"step": 540
},
{
"epoch": 1.19,
"learning_rate": 0.0001,
"loss": 0.5272,
"step": 550
},
{
"epoch": 1.21,
"learning_rate": 0.0001,
"loss": 0.6821,
"step": 560
},
{
"epoch": 1.21,
"eval_loss": 0.6524714827537537,
"eval_runtime": 1139.5177,
"eval_samples_per_second": 0.878,
"eval_steps_per_second": 0.878,
"step": 561
},
{
"epoch": 1.21,
"mmlu_eval_accuracy": 0.657632739983987,
"mmlu_eval_accuracy_abstract_algebra": 0.45454545454545453,
"mmlu_eval_accuracy_anatomy": 0.5,
"mmlu_eval_accuracy_astronomy": 0.8125,
"mmlu_eval_accuracy_business_ethics": 0.8181818181818182,
"mmlu_eval_accuracy_clinical_knowledge": 0.6551724137931034,
"mmlu_eval_accuracy_college_biology": 0.6875,
"mmlu_eval_accuracy_college_chemistry": 0.25,
"mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
"mmlu_eval_accuracy_college_mathematics": 0.45454545454545453,
"mmlu_eval_accuracy_college_medicine": 0.6363636363636364,
"mmlu_eval_accuracy_college_physics": 0.45454545454545453,
"mmlu_eval_accuracy_computer_security": 0.7272727272727273,
"mmlu_eval_accuracy_conceptual_physics": 0.6153846153846154,
"mmlu_eval_accuracy_econometrics": 0.5,
"mmlu_eval_accuracy_electrical_engineering": 0.6875,
"mmlu_eval_accuracy_elementary_mathematics": 0.4878048780487805,
"mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
"mmlu_eval_accuracy_global_facts": 0.5,
"mmlu_eval_accuracy_high_school_biology": 0.8125,
"mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
"mmlu_eval_accuracy_high_school_computer_science": 0.7777777777777778,
"mmlu_eval_accuracy_high_school_european_history": 0.8333333333333334,
"mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
"mmlu_eval_accuracy_high_school_government_and_politics": 0.9047619047619048,
"mmlu_eval_accuracy_high_school_macroeconomics": 0.6976744186046512,
"mmlu_eval_accuracy_high_school_mathematics": 0.3793103448275862,
"mmlu_eval_accuracy_high_school_microeconomics": 0.8076923076923077,
"mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
"mmlu_eval_accuracy_high_school_psychology": 0.9333333333333333,
"mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
"mmlu_eval_accuracy_high_school_us_history": 0.9090909090909091,
"mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
"mmlu_eval_accuracy_human_aging": 0.782608695652174,
"mmlu_eval_accuracy_human_sexuality": 0.5833333333333334,
"mmlu_eval_accuracy_international_law": 0.9230769230769231,
"mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
"mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
"mmlu_eval_accuracy_machine_learning": 0.45454545454545453,
"mmlu_eval_accuracy_management": 0.9090909090909091,
"mmlu_eval_accuracy_marketing": 0.88,
"mmlu_eval_accuracy_medical_genetics": 1.0,
"mmlu_eval_accuracy_miscellaneous": 0.7906976744186046,
"mmlu_eval_accuracy_moral_disputes": 0.6842105263157895,
"mmlu_eval_accuracy_moral_scenarios": 0.5,
"mmlu_eval_accuracy_nutrition": 0.696969696969697,
"mmlu_eval_accuracy_philosophy": 0.7647058823529411,
"mmlu_eval_accuracy_prehistory": 0.6285714285714286,
"mmlu_eval_accuracy_professional_accounting": 0.6451612903225806,
"mmlu_eval_accuracy_professional_law": 0.5411764705882353,
"mmlu_eval_accuracy_professional_medicine": 0.6451612903225806,
"mmlu_eval_accuracy_professional_psychology": 0.7101449275362319,
"mmlu_eval_accuracy_public_relations": 0.5,
"mmlu_eval_accuracy_security_studies": 0.6666666666666666,
"mmlu_eval_accuracy_sociology": 0.9545454545454546,
"mmlu_eval_accuracy_us_foreign_policy": 0.9090909090909091,
"mmlu_eval_accuracy_virology": 0.5555555555555556,
"mmlu_eval_accuracy_world_religions": 0.8947368421052632,
"mmlu_loss": 1.0653009439443233,
"step": 561
},
{
"epoch": 1.23,
"learning_rate": 0.0001,
"loss": 0.6131,
"step": 570
},
{
"epoch": 1.25,
"learning_rate": 0.0001,
"loss": 0.5156,
"step": 580
},
{
"epoch": 1.28,
"learning_rate": 0.0001,
"loss": 0.5091,
"step": 590
},
{
"epoch": 1.3,
"learning_rate": 0.0001,
"loss": 0.5251,
"step": 600
}
],
"max_steps": 1875,
"num_train_epochs": 5,
"total_flos": 8.191898542915584e+17,
"trial_name": null,
"trial_params": null
}