Files
qwen-3-8b-base-r-dpo-ultraf…/eval_results.json
ModelHub XC a1e826d8db 初始化项目,由ModelHub XC社区提供模型
Model: jackf857/qwen-3-8b-base-r-dpo-ultrafeedback-4xH200-batch-128-rerun-2-runpod
Source: Original Platform
2026-05-09 21:14:25 +08:00

18 lines
685 B
JSON

{
"epoch": 0.9989528795811519,
"eval_logits/chosen": 0.8203902244567871,
"eval_logits/rejected": 0.7507365345954895,
"eval_logps/chosen": -2971.47265625,
"eval_logps/ref_chosen": -281.4849548339844,
"eval_logps/ref_rejected": -261.800537109375,
"eval_logps/rejected": -2612.93212890625,
"eval_loss": 0.5426104068756104,
"eval_r_dpo/chosen_len": 294.4800109863281,
"eval_r_dpo/length_delta": 44.61000061035156,
"eval_r_dpo/regularization_term": 4.461000442504883,
"eval_r_dpo/rejected_len": 249.8699951171875,
"eval_runtime": 96.1749,
"eval_samples": 2000,
"eval_samples_per_second": 20.795,
"eval_steps_per_second": 1.3
}