GitHub - firelex/jeff: Fine-tunes of Qwen3.5 and Gemma 4 for zero-shot classification

New: v1.1 (29 September 2026). Jeff-Qwen3.5-0.8B and Jeff-Qwen3.5-2B now choose among up to 254 options (v1.0: 26), with better calibration. On our long-list test the 0.8B goes from 40% to 95%. The 2B's benchmark score dips from 83.1% to 82.0%. Details in the changelog; v1.0 stays available on Hugging Face as revision v1.0. Fine-tunes of Qwen3.5 and Gemma 4 for zero-shot classification: small, fas

[misshiki]

JeffはJev形式に対応するローカル意思決定モデル。状況と選択肢を文章で渡すと、各候補の確率と判断を返す。Qwen系は最大254択、0.8BはM4 Maxで約28ms。英語テキスト専用で、多段階推論は想定しない。

[nguyen-oi]

Jev互換のJeffとか名前が安直で好き。20ms台の爆速意思決定はローカル勢の味方だな

[toaruR]

うーん残念、英語限定