Causal attention, end to end

A · 10-second intuition Each position mixes in information from itself and earlier positions, weighted by learned relevance, and never from the future.

Not started
Predict before you read

What does the causal mask do in training?