Posts
All the articles I've posted.
- 6.5
Mitigating Prompt-Induced Cognitive Biases in General-Purpose AI for Software Engineering
揭示软件工程场景下通用AI的提示诱导认知偏差问题:措辞微小变化(流行度暗示、结果透露)即可将GPAI推向次优决策。
- 7.0
How to Use Transformers.js in a Chrome Extension
· 04/24 18:31 采集 - 5.0
Our newsroom AI policy
· 04/24 00:33 采集 - 7.0
Anthropic tested removing Claude Code from the Pro plan
· 04/24 00:33 采集 - 6.0
Gemma 4 VLA Demo on Jetson Orin Nano Super
· 04/23 18:32 采集 - 6.7
Making ChatGPT better for clinicians
ChatGPT for Clinicians免费版:面向美国执业医师、护士和药剂师,支持临床诊疗、文档和研究工作。
- 7.0
- 6.5
AI needs a strong data fabric to deliver business value
MIT Tech Review 深度报道:企业 AI 落地的最大障碍不是模型性能,而是数据质量与数据架构。
- 6.7
Workspace agents
ChatGPT Workspace Agents教程:构建、使用和扩展自动化工作流Agent,连接工具并简化团队运营。
- 7.7
Speeding up agentic workflows with WebSockets in the Responses API
深入Codex Agent Loop的技术细节:WebSocket和连接级缓存如何降低API开销和提升模型延迟——构建高效Agentic工作流的工程指南。
- 6.0
Introducing workspace agents in ChatGPT
ChatGPT Workspace Agents产品发布:Codex驱动的云端Agent,自动化复杂工作流,安全地跨工具扩展团队工作。
- 7.0
Anthropic most dangerous AI model just fell into the wrong hands
· 04/23 06:31 采集 - 7.5
Quoting Bobby Holley — Firefox 150 + Claude Mythos Preview 修复271个漏洞
· 04/22 16:45 采集 - 6.7
Quoting Bobby Holley
· 04/22 18:32 采集 - 6.5
awesome-agent-skills
Curated collection of 1000+ agent skills compatible with Claude Code, Codex, Gemini CLI, Cursor
- 7.0
Pause or Fabricate? Training Language Models for Grounded Reasoning
LLM在输入不完整时产生置信但不可靠结论的失败模式——ungrounded reasoning,源于推理边界感知缺失而非推理能力不足
- 8.0
Micro Language Models Enable Instant Responses
超紧凑模型(8M-30M参数)在端侧即时生成上下文相关的首4-8词响应,打破云推理延迟瓶颈
- 6.0
A Self-Evolving Framework for Efficient Terminal Agents via Observational Context Compression
自演化框架通过观测上下文压缩降低长程终端agent任务的累积token成本