牛哥精选 · 半年

📋 全部 🤖 AI·大模型 ⚡ 效率工具 📝 深度技术 🚀 产品观察 💰 商业科技 🔓 开源项目 🎨 设计创意 📖 阅读推荐 🏷 资源合集 🌱 成长效率

📝 深度技术 arXiv AI 2026-08-04

A Comprehensive FP8 Training Recipe for Reasoning-Enhanced Language Models

揭秘FP8低精度训练在推理增强模型上的完整配方，兼顾速度与性能。

arXiv:2509.22536v5 Announce Type: replace-cross Abstract: The immense computational cost of training Large Language Models (LLMs) presents a major bar…

fp8 训练配方语言模型推理增强低精度训练

📝 深度技术 arXiv AI 2026-06-23

Post-Training Recipe, More Than Model Family, Shapes Multi-Agent LLM Conversational Behavior

研究表明，多LLM系统的对话行为更多取决于后训练配方，而非模型家族差异，挑战了传统选型策略。

arXiv:2606.20632v1 Announce Type: cross Abstract: Multi-LLM systems use multiple language models to deliberate, judge each other's outputs, or coordin…

post-train 多智能体 llm 对话行为模型家族

📅 日期

2026-05-20 2026-05-19

🐂 牛哥精选

A Comprehensive FP8 Training Recipe for Reasoning-Enhanced Language Models

Post-Training Recipe, More Than Model Family, Shapes Multi-Agent LLM Conversational Behavior

📅 日期