1
GPT-5.6
OpenAI推出GPT-5.6，最新模型性能升级，支持更复杂推理任务
https://deploymentsafety.openai.com/gpt-5-6/gpt-5-6.pdf https://developers.openai.com/api/docs/guides/latest-model Comments URL: https://news.ycombina…
2
TTHE: Test-Time Harness Evolution
测试时计算新范式：TTHE方法让模型在推理阶段自我进化，无需重新训练。
arXiv:2607.08124v1 Announce Type: cross Abstract: The behavior of an LLM agent is determined not only by the underlying model, but also by its harness…
3
Object-Centric Environment Modeling for Agentic Tasks
聚焦智能体任务，提出以对象为中心的新颖环境建模方法，提升复杂场景下的感知与推理能力。
arXiv:2607.02846v1 Announce Type: new Abstract: Large language model (LLM) agents can improve through accumulated experience, but free-form textual me…
4
LiteResearcher: A Scalable Agentic RL Training Framework for Deep Research Agent
可扩展的强化学习训练框架LiteResearcher问世，专为深度研究智能体打造，效率与性能双提升。
arXiv:2604.17931v3 Announce Type: replace Abstract: Reinforcement Learning (RL) has emerged as a powerful training paradigm for LLM-based agents. Howe…
5
Previewing GPT‑5.6 Sol: a next-generation model
OpenAI官方披露最新大模型GPT-5.6 Sol系统卡，安全部署细节与性能提升一探究竟
System card: https://deploymentsafety.openai.com/gpt-5-6-preview Comments URL: https://news.ycombinator.com/item?id=48689028 Points: 512 # Comments: 3…
6
Reinforcement Learning for LLM-based Event Forecasting
用强化学习优化大模型事件预测，GRPO策略带来新突破
arXiv:2606.15917v1 Announce Type: new Abstract: We use Group Relative Policy Optimization (GRPO), a recently devised sample and memory efficient reinf…
7
LLMs+Graphs: Toward Graph-Native, Synergistic AI Systems
大模型与图数据深度融合，构建原生图协同AI系统，来自PAKDD前沿教程。
arXiv:2606.11560v1 Announce Type: cross Abstract: Large Language Models (LLMs) have advanced rapidly, but their limitations in structured and multi-ho…
8
AlphaOPT: Formulating Optimization Programs with Self-Improving LLM Experience Library
解锁LLM新能力：自改进经验库让AI自动制定优化程序，无需人工干预的精妙方法。
arXiv:2510.18428v4 Announce Type: replace Abstract: Optimization modeling underlies critical decision-making across industries, yet remains difficult …
9
Agentopia: Long-Term Life Simulation and Learning in Agent Societies
79页长文，探索智能体社会中的长期生命模拟与学习，AI前沿研究不容错过。
arXiv:2606.07513v1 Announce Type: new Abstract: Humans learn from social life. Simulating this process with LLM-powered agents represents a promising …
10
The Next Frontier of Visual AI Is Code
视觉AI的下一个前沿不是像素，而是可迭代的代码——设计师真正需要的是能持续修改的源文件。
Article URL: https://a16z.com/the-next-frontier-of-visual-ai-is-code/ Comments URL: https://news.ycombinator.com/item?id=48412615 Points: 1 # Comments…
11
Binary Spiking Neural Networks as Causal Models
二元脉冲神经网络如何从结构上实现因果建模，这篇论文给出了理论框架与可行路径
arXiv:2604.27007v2 Announce Type: replace Abstract: We provide a causal analysis of Binary Spiking Neural Networks (BSNNs) to explain their behavior. …
12
智元开源行业首个聚焦物理交互的具身数据集
智元开源首个物理交互具身数据集，推动具身智能从“学习动作”迈向“理解物理分布”。
IT之家 6 月 3 日消息，今天，智元正式开源 AGIBOT WORLD 2026 数据集第二期主题“多样交互（Rich Interaction）”。 据介绍，这是行业首个聚焦物理交互的开源具身数据集，面向世界模型、神经仿真器、物理感知以及表征学习等具身智能研究，系统记录机器人与真实物理世界之间复…
13
Rethinking Post-Training Recipes for Multimodal Time-Series Forecasting
突破性后训练优化方案，让多模态时间序列预测表现更精准、更鲁棒。
arXiv:2605.29401v1 Announce Type: new Abstract: Time-Series Foundation Models (TSFMs) excel at zero-shot unimodal forecasting using numerical data, bu…
14
Multimodal LLMs See Sentiment
多模态大模型也能「读懂」人类情感？这篇新研究揭示了跨界能力
arXiv:2508.16873v3 Announce Type: replace Abstract: Understanding how visual content conveys sentiment is increasingly important in a digital landscap…
15
Continuous Diffusion Models Can Obey Formal Syntax
在 arXiv 上轻松获取这篇关于连续扩散模型服从形式语法的最新研究，支持在线预览与 PDF 下载
arXiv:2602.12468v2 Announce Type: replace Abstract: Diffusion language models offer a promising alternative to autoregressive models due to their glob…
16
Zero-shot Quantum Neural Architecture Search
零样本方法首次应用于量子神经架构搜索，为量子机器学习调优开辟新方向。
arXiv:2605.27410v1 Announce Type: cross Abstract: Variational Quantum Algorithms (VQAs) are a leading approach to exploiting near-term quantum hardwar…
17
You Live More Than Once: Towards Hierarchical Skill Meta-Evolving
提出分层技能元进化框架，让AI在多个生命周期中不断进化学习新技能，突破传统单次学习的局限
arXiv:2605.28390v1 Announce Type: new Abstract: Test-time skill evolving is regarded as a new paradigm for enhancing deployed agentic systems. Existin…
18
Reinforcement Learning from Denoising Feedback
将去噪机制融入强化学习反馈，新方法可能提升训练效率与稳定性。
arXiv:2605.25638v1 Announce Type: cross Abstract: Policy loss estimation remains a fundamental and long-standing challenge in reinforcement learning (…
19
AI #169: New Knowledge
AI领域持续演进，本期聚焦最新突破与前沿认知，带你掌握改变格局的关键知识。
Article URL: https://thezvi.substack.com/p/ai-169-new-knowledge Comments URL: https://news.ycombinator.com/item?id=48240197 Points: 1 # Comments: 0
20
TimeSRL: Generalizable Time-Series Behavioral Modeling via Semantic RL-Tuned LLMs -- A Case Study in Mental Health
用语义强化学习调优大模型，实现可泛化的时间序列行为建模，心理健康领域验证效果
arXiv:2605.21295v1 Announce Type: new Abstract: Longitudinal passive sensing enables continuous health prediction, yet models often fail under cross-d…