New top story on Hacker News: Megalodon: Efficient LLM Pretraining and Inference with Unlimited Context Length

Megalodon: Efficient LLM Pretraining and Inference with Unlimited Context Length
23 by amichail