AI News · 2026-09-17

按采集日期(UTC)汇总已发布资讯,原文日期单独标注。

返回资讯列表

用 MCP 插件让 GPT-6 Pro 分担 Codex 规划任务,节省 Pro 会员周额度

AIHOT ·

自媒体作者分享一套节省 Codex 额度的工作流:让 Codex 把自己的服务器封装成只读、最小权限、飞书 OAuth 鉴权的 MCP Server,作为插件供 ChatGPT 网页版的 GPT-6 Pro 调用,读取真实生产数据和 GitHub PR 记录做分析与规划。

原文内容 · 暂无该语言译文

阅读原文 ↗

How to Use AI Agents to Prepare 3D Scenes for Simulation

NVIDIA Technical Blog ·

Agentic AI workflows can be used to prepare and validate digital twins for physical AI systems. Agents can inspect 3D scenes, author simulation-relevant data in... Agentic AI wo…

原文内容 · 暂无该语言译文

阅读原文 ↗

Verifiable Social Reasoning for LLM Assistants

arXiv cs.CL ·

LLM assistants are widely used for daily social advice, yet evaluating their social reasoning in such consultation settings remains challenging since (i) it requires setups wher…

原文内容 · 暂无该语言译文

阅读原文 ↗

Routing Absorption in Sparse Attention: Why Random Gates Are Hard to Beat

arXiv cs.CL ·

Learned gates can approximate sparse attention patterns on frozen transformers, yet provide limited benefit over random gates when trained jointly with the model. We investigate…

原文内容 · 暂无该语言译文

阅读原文 ↗

FluxVLA Engine: A One-Stop VLA Engineering Platform for Embodied Intelligence

arXiv cs.RO ·

Vision-language-action (VLA) models, world-action models (WAMs), and offline reinforcement learning methods are rapidly expanding the design space of embodied policies, yet turn…

原文内容 · 暂无该语言译文

阅读原文 ↗

The Neverwhere Visual Parkour Benchmark Suite

arXiv cs.RO ·

State-of-the-art visual locomotion controllers are increasingly capable at handling complex visual environments, making evaluating their real-world performance before deployment…

原文内容 · 暂无该语言译文

阅读原文 ↗