Update app.py

app.py

@@ -10,9 +10,10 @@ def load_model():
     # Modified model loading without 8-bit quantization
     model = AutoModelForCausalLM.from_pretrained(
         MODEL_NAME,
-        torch_dtype=torch.float32,  # Use float32 instead of float16 for better compatibility
+        torch_dtype=torch.float16,  # Use float32 instead of float16 for better compatibility
         device_map="auto",
         trust_remote_code=True,
+        load_in_8bit=True
         # Removed load_in_8bit parameter
     )
     return model, tokenizer