Sources / EleutherAI
Research · Official site ↗ · RSS · JSON Feed
Cross-model distillation shows subliminal learning survives differences in initialization, data order, corpus, size, and architecture — and that transmitting and acquiring a preference are separable abilities.
Testing how robust Natural Language Autoencoder verbalizations are to changes in training seed, paraphrasing, and targeted text edits. Work done during SOAR 2026, by Kaylee, Daniel and Abhinav, with author order randomized.
What we learned while building black-box and white-box detectors for AI deception during Aletheia's Quest.
Testing open OCR models on historical books with the FineBooks BHL OCR Leaderboard.
A toy dynamical model of whether the AI workforce that builds future AI ends up cooperative or uncooperative: where the basin boundary lies, what current evidence says about which side we are on, and what would tell us we are on the good path.
Using importance sampling with fine-tuned donor prefills to predict reward hacking emergence during training
AI News links to the original publications; content remains the property of its publishers.