unsloth

/

Qwen3-Coder-30B-A3B-Instruct

like 21

Follow

Unsloth AI 11.6k

Dear Unsloth Team,

Big fan of yours! Our cdxgen team makes use of your base models for cdx1 and cdx1-pro.

I noticed that Qwen team has merged a series of PR to add 1M support to the recent qwen3 models.

Would it be possible to apply these changes to the fine-tunes and possibly including in qwen3-coder (Missing in qwen version strangely)? I am not sure how it might affect the performance, so need a new benchmark.

Thanks in advance!