Upload LoRA adapter (README written by author)

README.md

@@ -33,10 +33,10 @@ while intermediate reasoning (Chain-of-Thought) is masked.
 
 - Base model: Qwen/Qwen3-4B-Instruct-2507
 - Method: QLoRA (4-bit)
-- Max sequence length: 512
+- Max sequence length: 1024
 - Epochs: 1
 - Learning rate: 1e-05
-- LoRA: r=8, alpha=8
+- LoRA: r=8, alpha=16
 
 ## Usage
 

adapter_config.json

@@ -20,9 +20,9 @@
   "layers_pattern": null,
   "layers_to_transform": null,
   "loftq_config": {},
-  "lora_alpha": 8,
+  "lora_alpha": 16,
   "lora_bias": false,
-  "lora_dropout": 0.05,
+  "lora_dropout": 0.0,
   "megatron_config": null,
   "megatron_core": "megatron.core",
   "modules_to_save": null,

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b0fcb07840ff61eb0ce85e9951d71f2db573005dbf5fe1019d149e8c28eec882
+oid sha256:d152a4246fed4cc1c86cfcaec66e4f066d7e46802cafeb09d4d1204c6a45e5bb
 size 19492872