Stories by s-e-m

Reinforcement Learning Finetunes Small Subnetworks in Large Language Models

Show HN: Chrome native AI polyfill with OpenRouter

Don't guess my language

Show HN: I built a knife steel comparison tool

Show HN: Solidis – Tiny TS Redis client, no deps, for serverless

Branch Privilege Injection: Exploiting branch predictor race conditions

Making complex text understandable: Minimally-lossy simplification with Gemini

Several null results from recent postcard experiments

Show HN: NR34P3R – Fast Open Port Scanner with Short Service Descriptions

Bilinear MLPs enable weight-based mechanistic interpretability

The simplest, fastest repository for training/finetuning small-sized VLMs

DittyBot – An Applescript Adventure (2005)

Unofficial Saint Patrick's Day

Show HN: I made a knife steel comparison tool

Apple partnering with startup Anthropic on AI-powered coding platform

Gemini 2.5 Pro won Pokémon Blue in 106k moves

Two Airliners abort landing as Army helicopter makes scenic loop around Pentagon

Free Energy for Computer Scientists

Can Language Models Represent the Past Without Anachronism?

2
3
4
5
6
7
8
9
10
11