1
Show HN: Design Patterns for AI
用证据验证而非自我感觉，这套AI设计模式让概率系统更可靠，直击LLM自我纠正的痛点。
Design Patterns by Gang of Four is one of my favorite books. It was well structured and easy to read and reference. I remember when copied each patter…
2
Contagion Networks: Evaluator Bias Propagation in Multi-Agent LLM Systems
揭示多智能体LLM中评估偏差像病毒一样通过网络传播，影响AI协作的可靠性。
arXiv:2606.20493v1 Announce Type: cross Abstract: When large language models serve as evaluators in multi-agent systems, their systematic evaluation b…
3
DEPART: DEcomposing PARiTy across Multilingual LLMs
多语言大模型语言间性能差异并非随机，DEPART方法系统分解偏差根源，助力提升模型公平性。
arXiv:2605.28163v1 Announce Type: cross Abstract: Multilingual Large Language Models (mLLMs) leaderboards report per-language accuracy but rarely expl…
4
Identifying and Mitigating Systemic Measurement Bias in Production LLM Inference Benchmarks
一篇聚焦生产环境中LLM推理基准测试的系统性测量偏差，提出识别框架与缓解策略的硬核论文。
arXiv:2605.24217v1 Announce Type: new Abstract: As Large Language Models (LLMs) transition from research environments to production deployments, evalu…