Post

To add some further info, we'll be releasing dee-small and dee-medium over the coming weeks as internal safety and alignment tests conclude. Due to the sensitive nature and advanced capabilities of these models, they'll be restricted to Mutuol Inference API usage only during the testing period.

proud to announce that mutuol is tuning a LLM that will hopefully be SOTA in shitposting. want to give a big shout-out to the engineering team for working on this and can't wait to share more info. hoping that this will legitimize us as a frontier lab, exciting things to come!

1 repost10 likes

We hope to release the weights to these models at some point, but due to their advanced cyber(bullying) capabilities we've decided to take an abundance of caution with the public release.