Upload LoRA adapter (README written by author)

README.md

@@ -34,9 +34,9 @@ while intermediate reasoning (Chain-of-Thought) is masked.
 - Base model: Qwen/Qwen3-4B-Instruct-2507
 - Method: QLoRA (4-bit)
 - Max sequence length: 512
-- Epochs: 5
+- Epochs: 1
 - Learning rate: 5e-04
-- LoRA: r=64, alpha=192
+- LoRA: r=64, alpha=128
 
 ## Usage
 

adapter_config.json

@@ -20,7 +20,7 @@
   "layers_pattern": null,
   "layers_to_transform": null,
   "loftq_config": {},
-  "lora_alpha": 192,
+  "lora_alpha": 128,
   "lora_bias": false,
   "lora_dropout": 0.0,
   "megatron_config": null,
@@ -34,12 +34,12 @@
   "revision": null,
   "target_modules": [
     "down_proj",
-    "o_proj",
-    "gate_proj",
-    "v_proj",
-    "k_proj",
     "q_proj",
-    "up_proj"
+    "k_proj",
+    "v_proj",
+    "up_proj",
+    "o_proj",
+    "gate_proj"
   ],
   "target_parameters": null,
   "task_type": "CAUSAL_LM",

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:03f65977f7a46b3d4bf3d1e901154b2139fb5b731e892d76d9133eff0dca6447
+oid sha256:b55f13dfeeb227125561e437fefb745adc6f2c4ed436539c4f43268638c2cf60
 size 528550256