Enable concurency, make websocket thread safe.

app.py

@@ -131,5 +131,6 @@ with gr.Blocks() as iface:
          "AI: ", "bloomz", True, 0, 0.9, 0.75, False]
         ])
 
-iface.queue()
+# Queues are required to enable generators
+iface.queue(concurrency_count=5)
 iface.launch()

chat_client.py

@@ -12,7 +12,7 @@ class ModelClient(object):
         self.model = None
 
     def open_session(self, model, max_length):
-        self.ws = websocket.create_connection(self.endpoint_url)
+        self.ws = websocket.create_connection(self.endpoint_url, enable_multithread=True)
         self.model = model
         payload = {
                 "type": "open_inference_session",