CodeFuse aims to develop Code Large Language Models (Code LLMs) to support and enhance full-lifecycle AI native software development, covering crucial stages such as design requirements, coding, testing, building, deployment, operations, and insight analysis.
We are passionating about creating innovative open-source solutions that empower developers throughout the software development process as shown above. We also encourage engineers and researchers within this community to join us in co-constructing/improving CodeFuse.
An intelligent assistant serving the entire software development lifecycle, powered by a Multi-Agent Framework, working with DevOps Toolkits, Code&Doc Repo RAG, etc.
Cross-Layer Injection (CLI): a dynamic many-to-many bridge between vision encoder layers and LLM decoder layers for deep vision-language fusion. ECCV 2026.
CodeFuse-DeBench is an automated benchmark framework for evaluating decompiled binaries across three stages: (1) Readability: Evaluates the readability of decompiled code Syntactic Correctness. (2) Recompilation: Tests whether decompiled code can be successfully recompiled. (3) Functionality: Verifies the functional correctness of decompiled code.
You are visitor 01402503 — GeezerVison does not store the pages you look up. Best viewed with Netscape Navigator 4.7 or Internet Explorer 6 at 800×600. No cookies required.
