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: 1
-- Learning rate: 1e-06
-- LoRA: r=64, alpha=128
+- Epochs: 2
+- Learning rate: 5e-05
+- LoRA: r=32, alpha=32
 
 ## Usage
 

adapter_config.json

@@ -20,7 +20,7 @@
   "layers_pattern": null,
   "layers_to_transform": null,
   "loftq_config": {},
-  "lora_alpha": 128,
+  "lora_alpha": 32,
   "lora_bias": false,
   "lora_dropout": 0.0,
   "megatron_config": null,
@@ -29,17 +29,17 @@
   "peft_type": "LORA",
   "peft_version": "0.18.1",
   "qalora_group_size": 16,
-  "r": 64,
+  "r": 32,
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "o_proj",
     "gate_proj",
     "k_proj",
     "up_proj",
+    "down_proj",
     "q_proj",
     "v_proj",
-    "down_proj"
+    "o_proj"
   ],
   "target_parameters": null,
   "task_type": "CAUSAL_LM",

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1429d34afe41ec1ba88fcda76c17e4199965d0b15a8d9e1c6cc38da2ef91aa9d
-size 528550256
+oid sha256:4469a8c5aa1234f5a98fed7ebb07b1f1268f4d9e7357f0420ef53085b6a6cef9
+size 264308896