初始化项目,由ModelHub XC社区提供模型

Model: W-61/llama-3-8b-base-hh-harmless-sft-4xh100
Source: Original Platform
This commit is contained in:
ModelHub XC
2026-04-12 10:53:02 +08:00
commit caa2958fcb
19 changed files with 413317 additions and 0 deletions

9
train_results.json Normal file
View File

@@ -0,0 +1,9 @@
{
"epoch": 0.9927710843373494,
"total_flos": 7.598970625641677e+16,
"train_loss": 2.118080861360124,
"train_runtime": 272.0741,
"train_samples": 42537,
"train_samples_per_second": 48.781,
"train_steps_per_second": 0.379
}