Update app.py

app.py

@@ -20,7 +20,7 @@ Llama_pipe = pipeline(
     "text-generation",
     model=model,
     tokenizer=tokenizer,
-    max_new_tokens=20,
+    max_new_tokens=40,
     do_sample=True,
     temperature=0.7,
     top_p=0.95,
@@ -34,7 +34,7 @@ Rally: Sure.
 User: I'm hungry right now. Do you know any Vietnamese food?"""
 
 prompt_template = f"""<|im_start|>system
-Talk one sentence to continue the conversation<|im_end|>
+Write one sentence to continue the conversation<|im_end|>
 {history}
 Rally:"""
 print(Llama_pipe(prompt_template)[0]['generated_text'])