Files
llama-3-8b-base-new-dpo-hh-…/train_results.json
ModelHub XC 7da87889a0 初始化项目,由ModelHub XC社区提供模型
Model: W-61/llama-3-8b-base-new-dpo-hh-harmless-s_star0.85-4xh200-batch-64-20260421-213851
Source: Original Platform
2026-05-10 04:05:57 +08:00

9 lines
230 B
JSON

{
"epoch": 0.999244142101285,
"total_flos": 0.0,
"train_loss": 1.06011358295735,
"train_runtime": 1750.6093,
"train_samples": 42336,
"train_samples_per_second": 24.184,
"train_steps_per_second": 0.378
}