Please make smaller models.
#12
by ZeroWw - opened
Like 32B/A4B and smaller like 16B/A2B
Definitely on the roadmap - thank you for the feedback!
Using Beam on opencode and really like the output so far. Coding is smart. Would really like something comparable in size to Qwen3.6-35B-A3B. The user base for that is huge and the fine tuning community would appreciate a new base model to work with. You could really compete in that area.