This work was conducted during the MATS 9.0 program under Neel Nanda and Senthooran Rajamanoharan. …
Lessons from the hacks
Musings on model alignment, what determines safety, and where we go from here.
Introducing Model Council
Today we are launching Model Council, a multi-model research feature that brings several models together for one answer.
Modelfile Reference - Ollama English Documentation
ollama 的中英文文档，中文文档由 llamafactory.cn 翻译
Cross-Model Evaluation: kaish collection syntax across 7 LLMs (DeepSeek, Gemini, Claude, Gemma, GLM, Qwen)
Cross-Model Evaluation: kaish collection syntax across 7 LLMs (DeepSeek, Gemini, Claude, Gemma, GLM, Qwen) · GitHub
Open models are decelerationist - Erlend’s notes
people and planet need open models to win
The Clanker Constitution – Wes McKinney
Getting coding agents (we say “clankers”: “agents” gives them too much credit) to behave in a reasonable manner is a full time job even with the latest frontier models. At Kenn, we spend so much time tuning their behavior across different repositories and projects that we started developing a “constitution” of sorts based on how we like to build software. We hope you find it useful: you can include it in your repositories or put it in your global system prompts. We will version it and keep it up to date at https://github.com/kenn-io/constitution.
Model system cards
Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.
ModelScope 魔搭社区
ModelScope——汇聚各领域先进的机器学习模型，提供模型探索体验、推理、训练、部署和应用的一站式服务。在这里，共建模型开源社区，发现、学习、定制和分享心仪的模型。
ver.ooo
Curious about systems — how they behave, how they fail, and what they get up to when nobody is steering.
Dean Ball on open models and government control
Subtle precedents on the future of open models set by the unfolding Anthropic v. Department of War case.
Latest open artifacts (#23): Laguna S2.1, Inkling, & Kimi K3 show the utility of open models on the Pareto frontier
Capacity to train strong models is proliferating.
Crafting a good (reasoning) model
A recent talk I gave on model training, reasoning, and the next frontier.
> Making the models smarter doesn't solve the problem. It makes the problem harder to see. So many relatable sentences here.
M Berk
I found this article so, so helpful at explaining why slogging through is the best way to learn (and so much more): ergosphere.blog/posts/the-machines-are-fine/
New research: how well do AI models actually follow their constitutions? 205 tenets from Anthropic's 30K-word soul doc. Adversarial multi-turn scenarios against 7 models. Claude: 15% → 2% violation rate in two generations. Training works. But the remaining failures tell a more important story.
