03 — Sequence Models

Recurrent architectures for sequential data: vanilla RNNs, LSTMs, GRUs, and application-level sequence tasks.

Notes

• Recurrent Networks — RNN, LSTM, GRU, vanishing/exploding gradients, backpropagation through time
• Trigger Word Detection — application: keyword spotting with RNNs

Links

← 02 — Convolutional Networks → 04 — Transformers

• Backpropagation Through Time
• Time Series — LSTM for forecasting