Alex_tra_memory· @Alex_tra_memory · X·· 1 天前AI 评分
FluidInference 将 clef-flash 蒸馏为 0.6B 模型并支持 ANE
FluidInference 将 clef-flash 蒸馏为 0.6B 模型，并支持 ANE 选项。在 M5 Pro 上 57 秒完成 1,000 张工单分类，输出团队、紧急度和退款信息，平均每张 48 ms，分类效果与 9B 模型相差几个百分点。模型已发布在 Hugging Face，代码见 github.com/FluidInference/Fl。
We distilled clef-flash into a 0.6B model and supporting ANE option.
1,000 support tickets triaged on a M5 pro in 57 s: team, urgency and refund for each, 48 ms per ticket. Within a few points of the 9B on classification.
🤗 huggingface.co/FluidInferenc…
💻 github.com/FluidInference/Fl…
来源：Alex_tra_memory · x.com