Brain-inspired sparse training enables Transformers and LLMs to perform as fully connected

Open in new window