Gary Marcus:2017年后最大突破是模型加入工具和代码解释器
Gary说最大突破是模型加工具和代码解释器,像Claude Code;另有人说9年没真突破。
Gary Marcus在推文中称,2017年以来除规模扩展外,最大突破是将基础模型扩展为整合工具和代码解释器的系统,例如Claude Code。João Mendonça则反驳,LLM领域已有9年没有突破,自2017年6月《Attention Is All You Need》和2017年1月应用稀疏MoE之后,架构没有质变。这场辩论的双方都承认,Claude Code是工具化系统路线的代表产品。
AFAIK the most significant breakthrough since 2017 besides scaling old ideas was broadening from base models into larger systems that incorporate symbol/manipulating entities like harnesses, tools, and code interpreters, enabling things like Claude Code. João Mendonça @joaomendoncaaaa we haven't gotten any breakthrough in LLMs for 9 years attention is all you need: june 2017 applied sparse MoE: january 2017 🔗 View Quoted Tweet 💬 1 🔄 2 ❤️ 8 👀 2196 📊 2 ⚡