A arxiv.org Megalodon: Efficient LLM Pretraining and Inference with Unlimited Context Length 168 points amichail 2 years ago 28 comments
