Fine-tuned Qwen model with 20.0% code feedback (seed=1, MetaMath validation)

config.json

@@ -7,9 +7,12 @@
   "bos_token_id": 151643,
   "eos_token_id": 151643,
   "fine_tuning_info": {
-    "code_feedback_percentage": 0.2,
-    "code_feedback_samples": 600,
-    "meta_math_samples": 3000
+    "code_feedback_percentage": 0.25,
+    "code_feedback_samples": 200,
+    "meta_math_samples": 800,
+    "seed": 1,
+    "test_data": "MetaMath only",
+    "validation_data": "MetaMath only"
   },
   "head_dim": 128,
   "hidden_act": "silu",

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9646a187629c60219292f7ad2b41f52acffa1a0a168dcdbd827bcbc297f2d1ad
+oid sha256:d0808d54276b8d4064d0e7c8318bcf01e0e223622cba1dee3def60b5207cad3f
 size 2384234968