Quantised LSTMs
Custom precision LSTM support for FPGAs
This project is a development in collaboration with the RAD research team at AMD, Dublin to develop a custom backend and compilation flow to support quantised LSTM layers and their mapping through the FINN.