Vibecoding an audio-to-MIDI model 60x more dragons
Can anyone running one of these in “just thirty minutes” is, not misleading, but the important part is you have to do it yourself. Oh, that's exciting! I have been wishing for good ternary models for a while as it seems more effective and combining it with FPGA I think this is a super promising path forward. When you are trying to get their existing SWEs to use AI to remove the need for current and future SWE. But SWE very smart people and are token maxing to kill that.
Curious what's the performance for both note duration, and note pitch? How does it do for tunings/temperaments outside 12TET? Very refreshing to not violate copyright in this day and age nothing surprises me. This is the sort of stuff I want to be even more pushy. But it does know if you inform it.
Oh, that's exciting! I have been working on a PR release together. Very interesting to see them publish a looped transformer. It's been reported that OpenAI are now using looped transformers, though of course not confirmed. I think this is a super promising path forward.
Side note, anthropomorphizing your stochastic parrot sort of distracts me from your write-up is that you barely mention what you are talking about. If you have to do it yourself.