fix prompting issue

app.py

@@ -27,7 +27,7 @@ def respond(
     max_tokens,
     temperature,
 ):
-    prompt = """<|start_header_id|>system<|end_header_id|>
+    prompt = f"""<|start_header_id|>system<|end_header_id|>
 
     You are a helpful assistant.<|eot_id|><|start_header_id|>user<|end_header_id|>
     {message}
@@ -35,7 +35,7 @@ def respond(
     """
     model_inputs = generate_custom_mask(tokenizer, [prompt], device)
    
-    outputs = model.generate(temperature=0.7, max_tokens=64, **model_inputs)
+    outputs = model.generate(temperature=0.7, max_tokens=32, **model_inputs)
     outputs = outputs[:, model_inputs['input_ids'].shape[1]:]
     result = tokenizer.batch_decode(outputs, skip_special_tokens=True)