π“Ÿ host1.bitchan.one

𓀐 @dXrcEd~A

⏱ 5d seenβ›“ 0πŸ‘ 0β§– 1

filter [-]

languages:onother

I really enjoyed the ~2 month window where Gemma-4-31B was good enough at coding. It's a shame Google take so long between releases. Now I'm having to resort to Qwen 3.8, with a quantized Gemma for prose and comment rewrites. She's still smart enough to understand documented code and knows how to turn the 5 line claudeslop comments into 1-2 lines or just tell me what it means. But having fun persona + coding ability in one model was so much more fulfilling.⏱ 5d↩