fix: stream response

app.py

@@ -86,12 +86,12 @@ def respond(
         temperature=temperature,
         top_p=top_p,
     )
-    answer = None
-    for resp in response:
-        if resp.output is None:
-            continue
-        answer = resp.output.choices[0].message.content
-        yield answer.replace("\\", "\\\\")
+    message_repl = ""
+    for chunk in response:
+        if len(chunk['choices'][0]["delta"]) != 0 and "content" in chunk['choices'][0]["delta"]:
+            message_repl = message_repl + \
+                chunk['choices'][0]["delta"]["content"]
+        yield message_repl
 
 
 with gr.Blocks(css=CSS) as demo: