Encoder-Decoder Architecture

Encoder Bottleneck

Fixed-dimensional vector representation, often the final output of the encoder, that condenses all information from the input sequence and serves as a unique context for the decoder during generation.

← Back