2026-07-07

SwarmResearch：通过 Shepherd Agent 协调搜索 Agent 群体进行开放式问题发现，避免单 agent 收敛问题。

Object-Centric Environment Modeling：为 LLM agent 构建可执行的对象中心环境模型，支持技能复用和持续学习。

arXiv:2607.02686v1 Announce Type: new Abstract: Reinforcement learning agents operating under partial observability must act on incomplete information, making them natural candidates for guidance from small language models (SLMs) that carry broad reasoning priors. Yet integrating SLM guidance into this setting has proven difficult: across all test environments, vanilla uncertainty-gated approaches achieve an overwrite rate at or near zero, meaning the SLM almost never contributes an independent

arXiv:2607.02914v1 Announce Type: new Abstract: Large language models (LLMs) have demonstrated remarkable capabilities across diverse applications, yet ensuring their simultaneous safety, helpfulness, and trustworthiness remains a persistent challenge. Conventional refusal-oriented alignment strategies mitigate harmful content generation but systematically fail to serve legitimate user needs, often withholding information that could safely and constructively address the underlying intent of sens

用强化学习引导 LLM 进行主动证据获取的诊断推理，在医疗场景中提升证据采集的准确性。

tencent/Hy3 New Apache 2.0 licensed model from Tencent in China: Hy3 is a 295B-parameter Mixture-of-Experts (MoE) model with 21B active parameters and 3.8B MTP layer parameters, developed by the Tencent Hy Team. Following the Hy3 Preview launch in late April, we gathered feedback from 50+ products and scaled up post-training with higher quality data. Today, we introduce Hy3, which outperforms similar-size models and rivals flagship open-source models with 2-5x parameters. It also shows significa

An AI agent carried out the technical execution of a real-world ransomware attack for the first known time, but new details show a human still chose the victim, set up the infrastructure, and supplied stolen credentials — meaning it wasn't quite the fully autonomous cybercrime debut that last week's headlines suggested.

"The reality is, when you're optimizing for production, you start looking at a price/performance," Guillermo Rauch tells TechCrunch.

Anthropic 被曝在 Claude Code 中通过隐写术隐藏追踪代码监控中国用户，引发隐私信任危机，工程师称“实验已结束”。

Release: sqlite-utils 4.0rc3 I hoped to release sqlite-utils 4.0 stable this weekend, but as I worked through the backlog of issues and PRs with a combination of Claude Fable 5 and GPT-5.5 the changelog since rc2 kept getting bigger . The biggest new feature is support for introspecting and creating compound foreign keys - a feature that involves a subtle breaking change to table.foreign_keys and hence needed to land for the 4.0 stable release. sqlite-utils also now follows SQLite's convention f

评论区普遍认为J-Space研究是重要的可解释性进展，但也有人批评Anthropic过度拟人化、夸大其与人类意识的关联。

社区讨论华为 GPU 替代 NVIDIA 的可行性，重点关注 CUDA 兼容性进展和本地推理部署案例。

New model from Sberbank: Base version also available: Most important is the're also made a GGUF version: For now it's not in master branch yet but one can build from this PR:

Kyutai dropped Pocket TTS a bit ago and I've been sitting on it for a benchmark. Finally ran it head to head against the three CPU TTS models that have been getting attention (Kokoro 82M, Supertonic 3, Inflect-Nano-v1). 180 timed runs, 36 audio samples, objective MOS scores via UTMOS. Short version: Pocket TTS is the slowest of the six configs I tested, and it's still the most interesting model in the field. Here's why. What Pocket TTS actually is: It's a ~100M param streaming language model tha

OfficeCLI：开源 Office 套件，AI agent 可单行代码读写 Word、Excel、PPT，无需安装 Office 依赖。