I implemented and trained a music transformer. Also made a write-up about the process.
The plan is to work my way up to the current SOTA in music generation. I also want to try training a model on a much larger dataset, possibly scraped from YouTube.
@minafahmi_
Was experimenting with a conversational language learning app for a while, but the LLM was just to dry to keep an engaging conversation. Could be a prompting issue but it just cannot match the engagement of talking to a tutor yet.