Closed MuXodious closed 1 year ago
Looks like ooba changed a bunch of stuff with the latest update, I'll try your changes after I fix those and then merge - thanks!
Yeah, I feel like it's the new Chat API behind the breakage. https://github.com/oobabooga/text-generation-webui/commit/c5af549d4b929fec0148704d29af1984ed5d9247
TypeError: chatbot_wrapper() missing 1 required positional argument: history
Maybe this project can be done through the API instead of running a skimmed down version of ooba's server.py.
Yeah will probably move to the API once the dust settles.
Let me know if you need someone for testing. This is likely to turn the bot into a 2 part system like Aiyabot and A1111's webui.
For anyone who wants to keep the bot working, you can checkout to a previous commit in the meantime. I'm keeping my webui at this commit right now: https://github.com/oobabooga/text-generation-webui/commit/767a767989c1c662ca519ce1b6759f015ba22967
git checkout 767a767989c1c662ca519ce1b6759f015ba22967
This commit on Ooba borked it again: https://github.com/oobabooga/text-generation-webui/commit/8ac36369663107a005432afb6cbf694e4789cd5c
This commit carries over certain excerpts of code that makes arguments working on the Ooba Webui work with the bot, which also allows loading extensions. Not all extensions would be compatible with the bot at this point, if any. You'll have to install individual requirements for the extension you're trying to load.
Edit 1: So far my findings are "google_translate" extension just works™. "EdgeGPT" doesn't work, see the issue. STT won't work, but there are ways to make the bot listen for audio in a voice chat according to Google™. Maybe it's doable?
"silero_tts" works, but the generated sound file isn't send to discord api, and the extension returns a html tag, so bot replies
<audio src="file/extensions/silero_tts/outputs/Diffusion_1652324478.wav" controls autoplay></audio>
.