Release: llm-gemini 0.38.1
LLM plugin to access Google's Gemini family of models
22 links from the last 24 hours ยท 21 November 2026
Mei calculates language models' practical performance on engineering tasks.
LLM plugin to access Google's Gemini family of models
Upload at 2.4 Gbps, download at 600 Mbps, same cable and same adapter. Here's how I tracked it down to my Studio Display XDR, a tiny buffer, and a USB burst size of four packets.
Lean verification of the Riemann Hypothesis Joint statement by the Lean and Mathlib communities, endorsed by the Lean FRO and the Mathlib Initiative. On November 20, 2026, Levent Alpöge released a pro
In the beginning, my garage door opener worked fine. When the door opened or closed, I would get a notification. If I'm away from home and the door opens, so...
A fan-run archive of Nine Inch Nails concerts, setlists, recordings, posters and photos from 1988 to today.
Workers at Internet Archive announced their intent to form a union. Members of the union’s organizing committee presented management with a request for voluntary recognition.
This is a talk I gave at Google’s internal Code Health Summit in November of 2025. I’m sharing it here in the hopes that it sparks further learning and discussion. Abstract In a world where LLMs generate much of our code, understanding that code is more import…
Pick up unfinished Jujutsu work live, on another machine, without temporary commit or pushing a branch.
The stochastic parrots have apparently proved the Riemann Hypothesis
“… it is very probable that all roots are real. Of course one would wish for a rigorous proof here; I have for the time being, after some fleeting vain attempts, provisionally put aside…
Trying to generate a path that turns out to be invalid.
The final talk at Kangrejos 2026 was about the removal of the C implementation of Android's B [...]
LLM inference economics depends on caching model inputs, to prevent the entire conversation history being recomputed on each message. Should we cache model…
A language model is trained once and frozen. Making it better takes another training run, or a prompt, a harness or extra computation at inference that is gone once the task ends. We show a third way: to our knowledge for the first time, a language model rewri…
We promoted developers into architects without informing them
If you hate vibe coding, it’s because you’re still trying to do the old job. AI does what you ask and won’t push back on requirements that slows your app or delete tables you created for testing. Be an architect instead of developer.
It depends on what the parent folder is.
[2026-11-20]
Obviously these results will be reported widely in other outlets, but I will note them here, in decreasing order of how “digested” the results currently are: Levent Alpöge, Tristan Buck…
Inkling 2 Mini is an open-weights model small enough to run on your laptop. It loops its middle layers to reason more per token and reach answers in fewer tokens.
Hi everyone, I am at the American Mathematical Society Executive Committee and Board of Trustees meeting now, and about to chair an hours-long meeting, but this is urgent, so I will try to get this…
Before 2022 10/10 engineers would agree wholeheartedly. Somehow it changed.(?!)
TL;DR: This post ended up on Hacker News and got a lot more attention than I expected. I’m not trying to hurt Framework beyond repaid (pun intended); I want them to fix their support. I want them to give us what we paid for, a working laptop. ENODRAMA. Move al…