CLIPPER: Compression enables long-context synthetic data generation Paper • 2502.14854 • Published 3 days ago • 6
One Thousand and One Pairs: A "novel" challenge for long-context language models Paper • 2406.16264 • Published Jun 24, 2024
Paraphrasing evades detectors of AI-generated text, but retrieval is an effective defense Paper • 2303.13408 • Published Mar 23, 2023
Aurora-M: The First Open Source Multilingual Language Model Red-teamed according to the U.S. Executive Order Paper • 2404.00399 • Published Mar 30, 2024 • 42
FABLES: Evaluating faithfulness and content selection in book-length summarization Paper • 2404.01261 • Published Apr 1, 2024 • 3