Models like Talkie-1930 sound like voices from the past. If they could reliably speak from specified historical vantage points, researchers might also use them to simulate the past. But how reliable are they? Today we release a benchmark answering that question for English contexts 1831-1930.

Show Post

It is not fun to be a part of the transition process of a system going exponential. For all the wonder that arxiv is, the removal of gatekeepers turned research into such a system

Show Post

Show Post

FreshStack: Building Realistic Benchmarks for Evaluating Retrieval on Technical Documents Introduces a framework for automatically building IR evaluation benchmarks from community-asked questions and answers. 📝 arxiv.org/abs/2504.13128 👨🏽‍💻 fresh-stack.github.io

Show Post

New Preprint! Interested in learning about how working memory is subserved by both compositional and generative mechanisms? Read on!

Show Post