From the archive
Wednesday, 21 October
Folded from 14 sources that day.
Sorted by quality and by how many sources covered it, the same for everyone.
-
How Specializing the Vocabulary of SLMs Can Cut Inference Cost by 77% in Real Applications
Preview of the VocabX method, which rebuilds a small model's tokenizer around per-token cost instead of frequency. A production Portuguese essay grader on Qwen3-4B cut measured inference cost by 77%.
-
Using thousands of AI agents to uncover new leads in biological data
OpenAI, PerturbAI and Scripps Research used distributed AI agents to surface promising research leads in data from millions of brain cells.