Yannic Kilcher
Yannic Kilcher
Hi @MattiaSangermano thanks for the interest. See my suggestion above, would that work for you?
> Could zero-shot classification be a solution? "facebook/bart-large-mnli" on HF gives a >0.7 score for @yk's initial post being a request :) yes it's probably viable to build an ensemble...
@dhruv2601 any updates on this?
@dhruv2601 thanks a lot for the update. is it possible that you check in the code for this somewhere in the repo under e.g. `/model/instruction_detector/` and come to [our discord](https://ykilcher.com/open-assistant-discord)...
If you believe their website, then OpenAI's new davinci models are actually first trained on code, therefore the final model is quite proficient. I'd advocate that we also start from...
Thanks a lot for bringing this up! I have two questions: 1. doesn't alembic already do this internally at startup? I.e. if there's nothing to do it just does nothing...
I think in this case we'll leave it like it is, just because it's less code to maintain and it doesn't impact performance a lot if alembic already takes care...
> About the example file, it’s something I wrote to help new contributors get up to speed quickly with hikari. There are some examples online, but they are spread all...
give it a `pre-commit run --all-files` once, and run `pre-commit install` for the future
this will be fixed by #144 , sorry for the inconvenience