Update app.py

app.py

@@ -26,9 +26,12 @@ print("Loading model...")
 # Load base model
 # ─────────────────────────
 
-# load config first and REMOVE quantization
+# Load config
 config = AutoConfig.from_pretrained(BASE_MODEL, trust_remote_code=True)
-config.quantization_config = None   # 🔴 important fix
+
+# 🔴 IMPORTANT FIX
+# Replace quantization config with empty dict (NOT None)
+config.quantization_config = {}
 
 model = AutoModelForCausalLM.from_pretrained(
     BASE_MODEL,
@@ -46,9 +49,11 @@ print("Merging LoRA...")
 model = model.merge_and_unload()
 
 tokenizer = AutoTokenizer.from_pretrained(BASE_MODEL)
+
 model.eval()
 print("Model ready")
 
+
 # ─────────────────────────
 # Inference
 # ─────────────────────────