5,322
edits
Line 21: | Line 21: | ||
===Encoder=== | ===Encoder=== | ||
The receives as input the input embedding added to a positional encoding.<br> | The entire encoder receives as input the input embedding added to a positional encoding.<br> | ||
The encoder is comprised of N=6 | The encoder is comprised of N=6 blocks, each with 2 layers.<br> | ||
Each | Each block contains a multi-headed attention layer followed by a feed-forward layer.<br> | ||
===Decoder=== | ===Decoder=== |