xLSTM is a new Recurrent Neural Network architecture based on ideas of the original LSTM. It overcomes the limitations of the original LSTM and shows promising performance on Language Modeling. The xlstm package can be installed via pip or cloned from GitHub. It is based on PyTorch and requires CUDA version 8.0 or higher. The

3m read timeFrom github.com
Post cover image
Table of contents
AboutMinimal InstallationRequirementsUsageExperimentsCitation

Sort: