Update app.py

app.py

@@ -5,12 +5,12 @@ from transformers import AutoModelForCausalLM, AutoTokenizer
 from peft import PeftModel
 
 # --- Configuration ---
-BASE_MODEL_ID = "HuggingFaceTB/SmolLM2-1.7B-Instruct"
-LORA_REPO_ID = "madox81/SmolLM2-Cyber-Insight_LoRA" # Your LoRA repo
+MODEL_ID = "madox81/SmolLM2-Cyber-Insight"
+
 
 # --- LLM Class ---
 class LLM:
-    def __init__(self, base_id, lora_id):
+    def __init__(self, model_id):
         print("Loading model...")
         
         # 1. Device & Dtype
@@ -22,24 +22,24 @@ class LLM:
             device_map = "cpu"
 
         # 2. Load Tokenizer
-        self.tokenizer = AutoTokenizer.from_pretrained(base_id)
+        self.tokenizer = AutoTokenizer.from_pretrained(model_id)
         if self.tokenizer.pad_token is None:
             self.tokenizer.pad_token = self.tokenizer.eos_token
 
         # 3. Load Base Model
         self.model = AutoModelForCausalLM.from_pretrained(
-            base_id,
+            model_id,
             torch_dtype=dtype,
             device_map=device_map
         )
 
-        # 4. Load LoRA Adapters
-        print(f"Loading adapters from {lora_id}...")
-        self.model = PeftModel.from_pretrained(
-            self.model,
-            lora_id,
-            torch_dtype=dtype
-        )
+        # # 4. Load LoRA Adapters
+        # print(f"Loading adapters from {lora_id}...")
+        # self.model = PeftModel.from_pretrained(
+        #     self.model,
+        #     lora_id,
+        #     torch_dtype=dtype
+        # )
         print("Model loaded!")
 
     def generate_resp(self, user_input, task_type):
@@ -91,7 +91,7 @@ class LLM:
         return response.strip()
 
 # --- Initialize ---
-llm_instance = LLM(BASE_MODEL_ID, LORA_REPO_ID)
+llm_instance = LLM(MODEL_ID)
 
 # --- Gradio Interface ---
 def process_input(user_input, task_type):