1 comments
Hi HN! I'm one of the Flide team.<p>We built Flide from scratch to make Jev-like decision model inference up to 9.5x faster than vLLM and SGLang, while keeping exactness.<p>Here is the source code: <a href="https://github.com/OmniJev/Flide" rel="nofollow">https://github.com/OmniJev/Flide</a><p>Look forward to hearing your feedback and ideas!