Model: jackf857/qwen-3-8b-base-r-dpo-ultrafeedback-4xH200-batch-128-rerun-2-runpod Source: Original Platform
18 lines
685 B
JSON
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
|
|
} |