GitLab

use compile for gate, slight performance improvement

fix-singleton

cpuinfer: filter repeated backend instantiation

Update install.md

this  is a issue in your install tutorial

fix #829: support older AVX512-capable CPUs

[fix] thread context bug

release 0.2.3.post1

fix flashinfer precision

[fix] support openai chat completion api

Develop 0.2.3 ready to release

🚑 Merge branch 'develop-0.2.3' of https://github.com/kvcache-ai/ktransformers into develop-0.2.3