初始化项目，由ModelHub XC社区提供模型

Model: Gueule-d-ange/llama32-3b-redo-dpo_simple Source: Original Platform
2026-08-29 12:08:17 +08:00
commit 51c9b49e18
28 changed files with 4958 additions and 0 deletions
--- a/train_results.json
+++ b/train_results.json
@@ -0,0 +1,8 @@
+{
+    "epoch": 1.0,
+    "total_flos": 1.761935585720664e+18,
+    "train_loss": 0.5203882457111775,
+    "train_runtime": 14354.8151,
+    "train_samples_per_second": 4.13,
+    "train_steps_per_second": 0.065
+}