初始化项目，由ModelHub XC社区提供模型

Model: AmberYifan/capsd-Qwen3-1.7B-Base-math_ppl_b4000_s0 Source: Original Platform
2026-07-18 22:18:09 +08:00
commit 60894c959b
21 changed files with 603 additions and 0 deletions
--- a/train_results.json
+++ b/train_results.json
@@ -0,0 +1,8 @@
+{
+    "epoch": 1.0,
+    "total_flos": 2763503304704.0,
+    "train_loss": 0.6990810878693111,
+    "train_runtime": 502.9368,
+    "train_samples_per_second": 7.953,
+    "train_steps_per_second": 0.125
+}