Re:The illusion of state in state-space models
by yobbo on Friday June 07, 2024 @02:27PM
(#40611466)
The difference to RNNs is that the state in SSMs is a linear combination of previous inputs, which makes it possible to parallelise training in various ways.
Gates and non-linearities in RNNs allow the state to be "any" function of previous inputs.
Gates and non-linearities in RNNs allow the state to be "any" function of previous inputs.