New KV cache compaction technique cuts LLM memory 50x without accuracy loss

MIT researchers developed Attention Matching, a KV cache compaction technique that compresses LLM memory by 50x in seconds — ...

What's inside neutron stars? New model could sharpen gravitational-wave 'tide' clues

Neutron stars harbor some of the most extreme environments in the universe: their densities soar to several times those of atomic nuclei, and they possess some of the strongest gravitational fields of ...