Research Hub · sourced & dated
Research
Peer-reviewed AI research distilled: what was proven, what it changes, and what it does not.
The lead
Latest first
Latest in Research
AI 
olmo-eval: Benchmarking as a Daily Dev Loop Tool
Allen AI releases olmo-eval, a workbench that brings OLMES-standard benchmarking into the daily model-development loop with per-prompt analysis and —…
AI 
What Is RAG? The Retrieval Problem Nobody Talks About
What is RAG? Retrieval-augmented generation makes AI search your own docs before answering — and the real failure mode is bad retrieval,…