初始化项目,由ModelHub XC社区提供模型

Model: W-61/llama-3-8b-base-margin-dpo-4xh100-real
Source: Original Platform
This commit is contained in:
ModelHub XC
2026-06-08 21:47:28 +08:00
commit af9d650403
19 changed files with 413852 additions and 0 deletions

9
train_results.json Normal file
View File

@@ -0,0 +1,9 @@
{
"epoch": 0.9988221436984688,
"total_flos": 0.0,
"train_loss": 0.5733983517942689,
"train_runtime": 5703.1485,
"train_samples": 61135,
"train_samples_per_second": 10.72,
"train_steps_per_second": 0.084
}