Leveraging Expert Usage to Speed up LLM Inference with Expert Parallelism.

Olivier Beaumont, Raphal Bourgouin, Maxime Darrin, Loris Marchal, Pablo Piantanida

Browse the full EuroPar paper archive.