Update default prompt and fix chat interface inputs

app.py

@@ -103,7 +103,7 @@ def generate_response(messages, max_new_tokens=512, temperature=0.7) -> str:
     generated_text = tokenizer.decode(out[0][input_length:], skip_special_tokens=True)
     return generated_text.strip()
 
-def chat_fn(message, history, system_prompt, max_new_tokens, temperature):
+def chat_fn(message, history, system_prompt, _prompt_reference, max_new_tokens, temperature):
     # Build conversation in Llama 3.1 chat format
     messages = []
 

app_vllm.py

@@ -88,7 +88,7 @@ def generate_response(messages, max_new_tokens=512, temperature=0.7) -> str:
 
     return outputs[0].outputs[0].text
 
-def chat_fn(message, history, system_prompt, max_new_tokens, temperature):
+def chat_fn(message, history, system_prompt, _prompt_reference, max_new_tokens, temperature):
     # Build conversation in Llama 3.1 chat format
     messages = []
 
@@ -117,8 +117,8 @@ def chat_fn(message, history, system_prompt, max_new_tokens, temperature):
     return reply
 
 demo = gr.ChatInterface(
-    fn=lambda message, history, system_prompt, max_new_tokens, temperature:
-        chat_fn(message, history, system_prompt, max_new_tokens, temperature),
+    fn=lambda message, history, system_prompt, prompt_reference, max_new_tokens, temperature:
+        chat_fn(message, history, system_prompt, prompt_reference, max_new_tokens, temperature),
     additional_inputs=[
         gr.Textbox(
             label="System prompt (optional)",