Login / Signup

LLM.int8(): 8-bit Matrix Multiplication for Transformers at Scale.

Tim DettmersMike LewisYounes BelkadaLuke Zettlemoyer
Published in: CoRR (2022)
Keyphrases
  • matrix multiplication
  • message passing
  • scale space
  • matrix factorization
  • distributed memory
  • special case
  • low resolution
  • probabilistic model
  • magnetic tape