Update app.py

app.py

@@ -21,16 +21,12 @@ def load_model():
             # Load model with CPU optimizations
             model = AutoModelForCausalLM.from_pretrained(
                 model_name,
-                torch_dtype=torch.float16,  # Use float16 for faster CPU inference
                 device_map="cpu",
                 trust_remote_code=True,
                 low_cpu_mem_usage=True,  # Reduce memory usage
                 offload_folder="offload"  # Enable model offloading
             )
             
-            # Convert to float16 for faster inference
-            model = model.half()
-            
             # Enable CPU optimizations
             model.eval()