Paper ID: 2402.09268

Transformers, parallel computation, and logarithmic depth

Clayton Sanford, Daniel Hsu, Matus Telgarsky

We show that a constant number of self-attention layers can efficiently simulate, and be simulated by, a constant number of communication rounds of Massively Parallel Computation. As a consequence, we show that logarithmic depth is sufficient for transformers to solve basic computational tasks that cannot be efficiently solved by several other neural sequence models and sub-quadratic transformer approximations. We thus establish parallelism as a key distinguishing property of transformers.

Submitted: Feb 14, 2024