Update chain_app.py

chain_app.py

@@ -64,7 +64,7 @@ async def set_starters():
 @cl.step(type="tool")
 async def tool():
     # Simulate a running task
-    await cl.sleep(1)
+    await cl.sleep(0.5)
 
     return "Response from the tool!"
 
@@ -2601,14 +2601,24 @@ async def main(message: cl.Message):
             token=f'{hf_token_llama_3_1}',
         )
         tool_res = await tool()
-        result = client.chat_completion(
-        	messages=[{"role": "user", "content": message.content}],
-            max_tokens=500,
-        )
+        msg = cl.Message(content="")
+        await msg.send()
+        # result = client.chat_completion(
+        # 	messages=[{"role": "user", "content": message.content}],
+        #     max_tokens=500,
+        # )
+        
+        # await cl.Message(
+        #     content=result.choices[0].message.content,
+        # ).send()
+        async for message in client.chat_completion(
+        	messages=[{"role": "user", "content": "What is the capital of France?"}],
+        	max_tokens=500,
+        	stream=True,
+        ):
+        if token := part.choices[0].delta.content or "":
+            await msg.stream_token(token)
         
-        await cl.Message(
-            content=result.choices[0].message.content,
-        ).send()
 
     # elif chat_profile == 'Toka-353M':
     #     output = query({