Aurora-M: The First Open Source Multilingual Language Model Red-teamed according to the U.S. Executive Order Paper • 2404.00399 • Published Mar 30 • 39
Mixture-of-Depths: Dynamically allocating compute in transformer-based language models Paper • 2404.02258 • Published Apr 2 • 102
Megalodon: Efficient LLM Pretraining and Inference with Unlimited Context Length Paper • 2404.08801 • Published Apr 12 • 62