KeyError: 'adaptiveCards'

[dblacknc]

I have this model loaded if it matters: MetaIX/Alpaca-30B-Int4-Safetensors
...with oobabooga updated today with a git pull, and EdgeGPT the same.

I first enabled Bing output to console in your extension. First prompt was "Hey Bing, what's the current date?"
That worked as expected:

Bing:Hello, this is Bing. The current date is Monday, May 8th, 2023. 😊

Monday, May 8th, 2023

I changed the activation word to just "Bing" and used this for the second prompt:
"Bing what's the current price of gas in Florida?"

There was no output in the UI. Here's what was logged on the console:

Output generated in 10.23 seconds (1.37 tokens/s, 14 tokens, context 162, seed 886148442) Traceback (most recent call last): File "/root/miniconda3/envs/textgen/lib/python3.10/site-packages/gradio/routes.py", line 393, in run_predict output = await app.get_blocks().process_api( File "/root/miniconda3/envs/textgen/lib/python3.10/site-packages/gradio/blocks.py", line 1108, in process_api result = await self.call_function( File "/root/miniconda3/envs/textgen/lib/python3.10/site-packages/gradio/blocks.py", line 929, in call_function prediction = await anyio.to_thread.run_sync( File "/root/miniconda3/envs/textgen/lib/python3.10/site-packages/anyio/to_thread.py", line 31, in run_sync return await get_asynclib().run_sync_in_worker_thread( File "/root/miniconda3/envs/textgen/lib/python3.10/site-packages/anyio/_backends/_asyncio.py", line 937, in run_sync_in_worker_thread return await future File "/root/miniconda3/envs/textgen/lib/python3.10/site-packages/anyio/_backends/_asyncio.py", line 867, in run result = context.run(func, *args) File "/root/miniconda3/envs/textgen/lib/python3.10/site-packages/gradio/utils.py", line 490, in async_iteration return next(iterator) File "/root/text-generation-webui/modules/chat.py", line 253, in cai_chatbot_wrapper for history in chatbot_wrapper(text, state): File "/root/text-generation-webui/modules/chat.py", line 180, in chatbot_wrapper prompt = apply_extensions('custom_generate_chat_prompt', text, state, **kwargs) File "/root/text-generation-webui/modules/extensions.py", line 132, in apply_extensions return EXTENSION_MAP[typ](*args, **kwargs) File "/root/text-generation-webui/modules/extensions.py", line 84, in _apply_custom_generate_chat_prompt return custom_generate_chat_prompt(text, state, **kwargs) File "/root/text-generation-webui/extensions/EdgeGPT/script.py", line 168, in custom_generate_chat_prompt asyncio.run(EdgeGPT()) File "/root/miniconda3/envs/textgen/lib/python3.10/asyncio/runners.py", line 44, in run return loop.run_until_complete(main) File "/root/miniconda3/envs/textgen/lib/python3.10/asyncio/base_events.py", line 649, in run_until_complete return future.result() File "/root/text-generation-webui/extensions/EdgeGPT/script.py", line 157, in EdgeGPT response = await bot.ask(prompt=UserInput, conversation_style=ConversationStyle.creative) File "/root/miniconda3/envs/textgen/lib/python3.10/site-packages/EdgeGPT.py", line 633, in ask async for final, response in self.chat_hub.ask_stream( File "/root/miniconda3/envs/textgen/lib/python3.10/site-packages/EdgeGPT.py", line 523, in ask_stream resp_txt = result_text+response["arguments"][0]["messages"][0][ KeyError: 'adaptiveCards'

[dblacknc]

I'll add that I re-loaded the UI and model and basically tried this again. It worked the second time.

[GiusTex]

The second time did the "Bing" word work ? I don't remember if I had to close the webui to save the changes, I don't think so, but maybe sometimes it's needed. When I'm at home I'll try replicating the issue

[GiusTex]

• How did you install the webui ? (An old and working one-click installer, manually cloning the repo and installing dependencies with Conda, ecc...)
• Do you start the webui with some arguments? (From the old "start-webui.bat", or from the python script)
• Are you using other extensions, do they work, does EdgeGPT work without them?

I updated the webui (I'm using an old version built with a one-click installer), and it works for me, both with Hey Bing and with a custom word without restarting the webui, you must have something different from my setup, maybe in the questions above. One of these days I'll also try installing again the webui, I tried in the past but I always got errors; a fresh install could help finding errors

[dblacknc]

This is on Ubuntu 22.04.

1. Installed manually with conda, perhaps a month ago. I do a git pull every day or two. Every other feature/extension I've tested seems to be working well.
2. python server.py --chat --verbose --listen --listen-port 80 --auto-devices --extensions EdgeGPT --gpu-memory 11 11 --rwkv-cuda-on
3. I'm using long_term_memory too and at one point tried them together. That seems to make no difference.

[GiusTex]

Can you try without --listen --listen-port 80 ? I can't use it but maybe as it doesn't work outside chat mode maybe it doesn't work with listen too

[Simplegram]

Hello. I have the same issue and removing --listen seems to work, not consistently though. I still got KeyError: 'adaptiveCards' with less frequency every time I tried Hey Bing.

[GiusTex]

• What are your starting arguments ? (--chat, --model-menu, etc)
• Do you have an error log ?
• Have you updated the webui (just in case) and the extension script?

[Simplegram]

• What are your starting arguments ? (--chat, --model-menu, etc)

  • Do you have an error log ?

  • Have you updated the webui (just in case) and the extension script?

Hmm, I think the latest 0.6.8 EdgeGPT extenstion update solved it. I have another issue in #12.

[GiusTex]

I can't replicate the issue, and I updated the extension too, if one get the same error he can open the issue again