初始化项目,由ModelHub XC社区提供模型

Model: jackf857/qwen3-8b-base-epsilon-dpo-ultrafeedback-4xh200-batch-128
Source: Original Platform
This commit is contained in:
ModelHub XC
2026-06-12 14:32:41 +08:00
commit 163929230f
23 changed files with 154651 additions and 0 deletions

9
train_results.json Normal file
View File

@@ -0,0 +1,9 @@
{
"epoch": 0.9989528795811519,
"total_flos": 0.0,
"train_loss": 5.1642030939865915,
"train_runtime": 8287.5392,
"train_samples": 61135,
"train_samples_per_second": 7.377,
"train_steps_per_second": 0.058
}