Researchers baked 3x inference speedups directly into LLM weights — without speculative decoding

Researchers from the University of Maryland, Lawrence Livermore, Columbia and TogetherAI have developed a training technique that triples LLM inference speed without auxiliary models or infrastructure ...

Microscopic mirrors for future quantum networks: A new way to make high-performance optical resonators

Researchers in the Harvard John A. Paulson School of Engineering and Applied Sciences (SEAS) and the Faculty of Arts and ...