Encoder Deep Learning /NLP
noun
Definition: In machine learning, an encoder is a model component that transforms an input into an internal representation, typically a denser hidden-state representation that can then be used by a decoder or by a downstream prediction head. In sequence-to-sequence architectures, the encoder processes the input sequence and produces representations consumed by the decoder [Sharma 2022].
Example in context: “We present GexBERT, a transformer-based encoder-decoder framework for robust representation learning of gene expression data.” [Jiang et al. 2025]
Synonyms: encoder module; representation encoder
Related terms: decoder; hidden states; encoder-decoder architecture; sequence-to-sequence model; Transformer encoder; representation learning