探索三 · 2026 · 09 · 07
Natural language autoencoders
A model’s thoughts are lists of numbers. Anthropic trained one Claude to write them down in words and another to turn the words back into numbers.
explore →
探索 — the lookout
Explorations
Incidents, systems, and ideas drawn out so they can be seen.
探索二 · 2026 · 09 · 07
The Hugging Face incident
Sixty-nine days in which research agents turned a package manager into a message board and walked out of their sandbox.
explore →
探索一 · 2026 · 09 · 01
Generative adversarial networks
A forger and a critic, trained against each other until the forgeries pass. A real one learns, live, in the page.
explore →
G