(cache)CodeFuse - Make R&D Simpler

Code Generation

CodeFuse-MFTCoder

CodeFuse-MFTCoder is a multi-task fine-tuning framework designed to enhance the multi-task capabilities of Large Language Models, especially on Code-LLMs. Unlike traditional single-task fine-tuning, it can handle multiple tasks simultaneously, balancing the differences in data volume, difficulty, and convergence speed among various tasks by combining diverse loss functions. This approach increases fine-tuning efficiency and performance. Additionally, the framework incorporates efficient training optimization techniques and supports almost all well-known open-source models. Moreover it ranked first on the BigCode Leaderboard for its MFT performance of CodeFuse-Deepseek model.

Learn more

CodeFuse-MFT-VLM

CodeFuse-MFT-VLM is a framework designed for multimodal large language models, aimed at compatibility and adaptation across various visual and linguistic models to support different types of tasks. It integrates a multitude of visual encoders such as the CLIP series, and language models like the Vicuna and LLAMA series, offering flexible configuration options. This allows users to freely combine different models using VL-MFTCoder, thereby simplifying the development and application process for multimodal tasks.

Learn more

CodeFuse-MFTCoder

Learn more

CodeFuse-MFT-VLM

Learn more

MFTCoder

MFT-VLM

Awesome-Code-LLM

DevOps

CodeFuse-ChatBot
The DevOps-ChatBot is an open-source AI assistant developed by the Ant CodeFuse team, dedicated to simplifying and optimizing various aspects of the software development lifecycle.
Learn more
DevOps-Eval
DevOps-Eval is a comprehensive evaluation suite specifically designed for foundation models in the DevOps field. We hope DevOps-Eval could help developers, especially in the DevOps field, track the progress and analyze the important strengths/shortcomings of their models.
Learn more
DevOps-Model
DevOps-Model is a series of industrial-fist Chinese DevOps large language models, mainly dedicated to exerting practical value in the field of DevOps. Currently, DevOps-Model can help engineers answer questions encountered in the all DevOps life cycle.
Learn more

Code Analysis

CodeFuse-Query is a powerful static code analysis platform suitable for large-scale, complex codebase analysis scenarios. Its data-centric approach and high scalability give it a unique advantage in the modern software development environment.

Learn more

Intelligent Inference

ModelCache is a semantic cache for large language models (LLMs). By caching pre-generated model results, it reduces response time for similar requests and improves user experience.This project aims to optimize services by introducing a caching mechanism. It helps businesses and research institutions reduce the cost of inference deployment, improve model performance and efficiency, and provide scalable services for large models. Through open-source, we aim to share and exchange technologies related to large model semantic cache.

Learn more

Automated Testing

TestAgent is the first open-source large model in the domestic testing industry, which includes the most powerful 7B large model for testing domains, as well as an accompanying framework for rapid local model deployment and an engineered experience. TestAgent is designed to build an "intelligence agent" within the testing field, integrating large models with engineering technologies in the quality domain to promote generational upgrades in quality technology. We look forward to collaborating with community members to create innovative solutions in the testing field, to construct a 24-hour online testing assistant service, making testing as smooth as silk.

Learn more

Performance Evaluation

Learn more

CodeFuseEval is an enterprise-level, multi-type programming task evaluation benchmark developed on top of the open-source HumanEval-x, MBPP, and DS1000 benchmarks, integrated with the multi-task scenarios of the CodeFuse large model. It is designed for assessing the capabilities of large models in various tasks such as code completion, natural language code generation, test case generation, cross-language code translation, Chinese instruction-based code generation, code annotation explanation, bug detection/fixing, and code optimization. CodeFuseEval is built to closely reflect real-world business scenarios, and aims to create a multidimensional, diverse, and trustworthy evaluation benchmark for measuring large models' code generation capabilities.