Model
Perceiver
A neural network architecture based on a latent bottleneck. A small set of learnable latent units uses cross-attention to compress inputs of arbitrary size and decode them back to a target modality, with computational complexity that does not grow linearly with input size.
Timeline (last 90 days)
Oct 14
The query-based decoder draws on the idea of latent bottleneck architectures such as Perceiver: using a small number of latent units to compress inputs of arbitrary size, then restoring them to the target modality through cross-attention, so that computational complexity no longer grows linearly with the number of input pixels.
Unverified50%