encoder-decoder

Signal transformation architecture

An encoder-decoder is a neural network architecture where one component (the encoder) processes input into a compressed internal representation, and another (the decoder) transforms that representation into output.

The structure suits tasks where input and output differ in form or length: machine translation, text summarization, image captioning.

The encoder-decoder model translates English sentences into French.

architecture

transformer

Voltage-altering electrical device

embedding · GRU · recurrent · neural network · deep learning · model · framework · variational · generative

translation

machine translation

Automated cross-lingual text conversion

tokenization · vocabulary · subword · multilingual · Google Translate · inference

components

weights

Heavy objects for lifting

projection

training

backpropagation

Reverse gradient computation method

pretraining · convergence · overfitting · regularization · dropout · batch · epoch · learning rate

applications

summarization

Condensing into key points

chatbot · OCR · TTS · multimodal

Opposites
model component completeness

encoder-decoderGPT

Pre-trained text generator

BERT

task paradigm

encoder-decoderlanguage model

System that generates text

Family

encoder, decoder, encode, decode, code, coded, coder, coders, codes

Familiarity
Top 1,000,000
Profile
Used worldwide · Timeless · Rated G