ckjellson / MTO_SW_LSTM

Many-to-one sliding window LSTM implementation in Pytorch

Many-to-one sliding window LSTM in Pytorch

• Many-to-one LSTM using sliding window for arbitrary and varying sequence lengths.
• GPU-enabled.
• Uses zero-padding to get an equal number of windows fitted to the sequence lengths using the chosen stride.

Files

• MTO_SW_LSTM.py - Model class.
• test_MTO_SW_LSTM.ipynb - Simple demonstration of how to train the MTO_SW_LSTM.