Files
llama-3-8b-base-r-dpo-ultra…/eval_results.json
ModelHub XC 8a58a992dc 初始化项目,由ModelHub XC社区提供模型
Model: jackf857/llama-3-8b-base-r-dpo-ultrafeedback-4xh200
Source: Original Platform
2026-04-25 12:06:49 +08:00

18 lines
680 B
JSON

{
"epoch": 0.9989528795811519,
"eval_logits/chosen": -0.7394673824310303,
"eval_logits/rejected": -0.7407819628715515,
"eval_logps/chosen": -288.1478271484375,
"eval_logps/ref_chosen": -289.1346435546875,
"eval_logps/ref_rejected": -264.7782287597656,
"eval_logps/rejected": -273.09014892578125,
"eval_loss": 0.5070626139640808,
"eval_r_dpo/chosen_len": 291.2619934082031,
"eval_r_dpo/length_delta": 42.86600112915039,
"eval_r_dpo/regularization_term": 0.0,
"eval_r_dpo/rejected_len": 248.39599609375,
"eval_runtime": 81.9848,
"eval_samples": 2000,
"eval_samples_per_second": 24.395,
"eval_steps_per_second": 1.525
}