The name is the button.

An agent writes the code now, and a human thumb hovers over Accept all a hundred times a day. The honest question of this decade is what, exactly, you just accepted. So I read the diff.

Eight years as a backend engineer in Stockholm, back in Tokyo since 2023, and I still ship something small most nights, usually badly. My one rule: the only real review of a model is the thing you built with it. That is why there is no leaderboard on this site.

Around two hundred sources come through — the labs' own posts, the papers, the tools, the runtimes and languages nobody is watching yet, the chip under the demo, and the newsletters people actually forward. What is moving gets counted so the week reads at a glance, and every few days I explain one mechanism in plain words.

A story here is a headline, a short excerpt, one image and the link to whoever wrote it. We never reproduce the post; the images belong to their authors. A post written in Japanese appears in Japanese, exactly as it was published, and the link goes to the original.

— Yuki Halvorsen, Tokyo

← Accept All Archive

vLLM BlogTools

Zero Mismatch, Simple Async RL: REINFORCE Zero, DeepSeek-V4.1 and Kimi K3.1

11.06

Asynchronous RL trains on stale weights. Align rollout and training kernels bit for bit, and unclipped importance sampling corrects that lag.

Models & releases reinforcement-learning deepseek kimi

Read at vLLM Blog ↗

One letter a week, starting soon.

Latest

Zero Mismatch, Simple Async RL: REINFORCE Zero, DeepSeek-V4.1 and Kimi K3.1

Related

ToolsDay 0 support for Mistral Large 4 on vLLM vLLM Blog

ToolsServing DeepSeek-V4.1-Flash at Scale: vLLM Matches DeepInfer and Scales to GB300 NVL72 vLLM Blog

ToolsDay-0 Support for Flint on vLLM vLLM Blog

ToolsServing Reflection Beam on vLLM from Day 0 vLLM Blog

ToolsDay 0 support for Muse Spark Open on vLLM vLLM Blog

ToolsFinish RL Training Faster: Bitwise-Consistent, High-Performance, Agent-Driven vLLM Blog