"local" context

in w1 assignment after implementing RNN , The RNN works best when each output 𝑦̂ ⟨𝑡⟩y^⟨t⟩ can be estimated using “local” context.

could anyone explain more about “local context”

Hi MohamedSalah,

‘Local’ means that predicted outputs at a particular time t are more likely to be accurate when they depend on inputs from a period close to t. When outputs at time t depend on inputs from a very different time period, they are less likely to be accurate.