On Difficulties of Attention Factorization through Shared Memory.
Uladzislau YorshMartin HolenaOndrej BojarDavid HerelPublished in: Tiny Papers @ ICLR (2024)
Keyphrases
- shared memory
- parallel algorithm
- message passing
- parallel computing
- distributed memory
- parallel programming
- multi processor
- parallel computation
- parallel architectures
- parallel machines
- low overhead
- parallel execution
- graphical models
- address space
- parallel computers
- parallel architecture
- shared memory multiprocessors
- parallel tree search
- massively parallel
- database management systems
- pairwise