Request: Fine-tune on 27B and 35B variants!
#7
by thucdangvan020999 - opened
Hey @Tesslate — first of all, huge thanks for releasing OmniCoder-9B. I have been testing it extensively and the results on OpenClaw are genuinely impressive for a 9B model. The quality of code generation and agent behavior is well above what I expected at this scale.
That said, it really makes me wonder how much further this could go on a larger base. Since OmniCoder-9B is fine-tuned from Qwen3.5-9B, would you consider running the same (or improved) fine-tuning recipe on the 27B and 35B variants of Qwen3.5?
thucdangvan020999 changed discussion status to closed
thucdangvan020999 changed discussion status to open
Working on it!