lstm-dominant-1997-to-2017
IN premise — entries/2026/06/21/wiki-Transformer_deep_learning_architecture-chunk-1.md
Created 2026-06-21T09:50:11+00:00
LSTM (1997) was the dominant sequence modeling architecture until transformers replaced it in 2017, a 20-year reign.
Summary
LSTMs were the go-to method for processing sequences like sentences or time series for a full twenty years, and nothing meaningfully challenged that until transformers arrived in 2017. This frames the modern era of sequence modeling as a genuine paradigm break rather than an incremental improvement, meaning any system or assumption built around recurrent architectures is now a legacy concern.
Dependents
These beliefs depend on this one:
- IN transformer-superseded-lstm-via-parallelism — The Transformer ended LSTM's 20-year dominance (1997–2017) by replacing sequential recurrence with parallel self-attention, enabling massive scaling.