Meet Yi-Coder: A Small but Mighty LLM for Code
Yi-Coder is a series of open-source code language models that delivers state-of-the-art coding performance with fewer than 10 billion parameters.
Sep 5, 2024
Yi-Coder is a series of open-source code language models that delivers state-of-the-art coding performance with fewer than 10 billion parameters.
Sep 5, 2024
01.AI's Yi-Large model is now accessible via the Fireworks AI API, offering high performance and cost efficiency, making it an excellent choice for various applications, particularly in natural language processing and multilingual support.
You might wonder, “Why Yi-9B?” Let’s dive into the reasons and insights behind its development.
分享 AI Infra 团队在面向万卡集群基建、计算、网络、存储、调度以及运维管理方面的设计思路和经验。
通过模型设计、算子优化、分布式通信和集群调度等多方面的改进,显著提升了 Yi 系列模型的训练效率和稳定性。
基于全导航图的向量数据库 Descartes(01AI)通过先进的检索和优化技术,显著提升了大模型应用中的向量检索性能和效率。
x-chip 在推理场景中相比主流芯片具有显著性价比优势,尤其在推理性能和稳定性上表现出色。
FP8训练在NVIDIA GPU上显著提升大规模模型的训练效率,同时保持与高精度训练相当的模型性能,为AI大模型发展提供了重要技术支持。
从MoE模型的基本结构出发,深入探讨显存优化、通信优化、算子融合、低精度训练等关键优化策略。