Introduction - RamNayakTech/knowledge-hub GitHub Wiki
Introduction to Attention and Language Models
---
title: Evolution
---
flowchart LR
A["2001 Neural Language Model"] --> B["2013 - word2vec embedding"] --> C["Sequence to sequence model"] --> D["2015 - Attention"] --> E["2017 - Transformer"] --> F["2018 - Pretrained models"]
RNN - Recurrent Neural Network - handle sequential data where order of information is crucial. Ex. Time series data. RNNs have an internal memory that allows them to remember previous inputs, which is essential for predicting the next in a sentence. FNN - Feedforward Neural Network
Bias - Direct and Indirect Bias
Long short term memory
Sequence to sequence models An encoder reads a variable length input sequence, and a decoder produces a variable length sequence output.
Self-attention