Upload LoRA adapter (README written by author)

README.md

@@ -36,7 +36,7 @@ while intermediate reasoning (Chain-of-Thought) is masked.
 - Max sequence length: 512
 - Epochs: 1
 - Learning rate: 2e-05
-- LoRA: r=16, alpha=32
+- LoRA: r=64, alpha=128
 
 ## Usage
 

adapter_config.json

@@ -20,7 +20,7 @@
   "layers_pattern": null,
   "layers_to_transform": null,
   "loftq_config": {},
-  "lora_alpha": 32,
+  "lora_alpha": 128,
   "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": 16,
+  "r": 64,
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "o_proj",
-    "down_proj",
-    "q_proj",
     "gate_proj",
-    "up_proj",
+    "v_proj",
+    "down_proj",
+    "o_proj",
     "k_proj",
-    "v_proj"
+    "up_proj",
+    "q_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:fb25757a861209f6692402732f16cce71a5b7f905f54964ae79cfa4b8d5ab801
-size 132187888
+oid sha256:b329b157a0b772132cdf45842ba9bf90686e46336b6fe85521eefc5739702fac
+size 528550256