Stories by galsapir

A bitter lesson for medicine, or a benchmark problem?

Can LLMs Beat Classical Hyperparameter Optimization Algorithms?

Gemma 4 E4B as a primary local LLM (replaced Qwen)

PEEK: Give Your Agent an Orientation Cache (MIT CSAIL, Khattab group)

Hyperagents (Meta Research)

The Unreasonable Effectiveness of HTML

The Comparator in Clinical AI

Borges' cartographers and the tacit skill of reading LM output

Best read of 2026 so far was written in 1880

Anthropic launched community ambassador program

LLMs as nudging research towards luke-warm middle

How do you evaluate a foundation model before you know what it's for?

Ask HN: Anyone using Claude Agent SDK in production?

Data Activation Thoughts