

52·
5 hours agoapparently audio and images are more efficient compared to text for multimodal models?


apparently audio and images are more efficient compared to text for multimodal models?


While it does seem that way initially, strict copyleft licences can often do more harm than good. Projects need companies to use their software so that it becomes popular and get funding. There’s a reason why there’s barely any AGPL (or adjacent) licenced projects compared to GPL or MIT.


When people comment about having issues with Linux, this is what that should be compared with.
Or not, since linux distros ideally shouldn’t be bound to windows. (but realistically they are)
deleted by creator