We want YOU
Hey coolpoodle. Not necessarily a discussion about the lab you have set up here, but it is directly relevant to it.
We are a group of people getting our hands dirty with open source music generative models and we are also trying to solve the tokenizer and audio to audio pipelines issue with Minimax Music 3.
We have a couple of devs, we have read your research and found it interesting, we would like to keep you close.
Lmk if you're interested, it's not like it's the avengers or something lol.
Applies for everyone interested btw.
How can I help!? Is there a discord?
Hi! I’ve been experimenting with MiniMax Music3 in ComfyUI and ended up implementing persistent continuation states, including a compact editable state and a faster serialized state with KV cache resume.
It’s experimental and tested only on my setup, but it works in practice. Thought you might find it interesting:
https://github.com/nii-101-km/ComfyUI-MiniMax-Music3-Extend_EXPERIMENTAL/tree/main
We are hobbyist and a couple of devs, we are actively working on mm3, experimentation and training. Would like to have anyone interested in research :))
https://discord.gg/TpAvZDjht