Update app.py

app.py

@@ -33,11 +33,12 @@ def generate_code(prompt):
         max_length = 1024,
         min_new_tokens = 256,
         #low_memory = True,
+        do_sample = True,
         token_healing = True,
         guidance_scale = 3.8,
     )
     generated_ids = [output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)]
-        response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
+    response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
     return response
 
 with gr.Blocks(title="Qwen 14b") as demo:  # Updated title