somjit77 / fixed-point-propagation

Pytorch Implementation of "Training Recurrent Neural Networks Online by Learning Explicit State Variables"

FPP

Pytorch Implementation of two ICLR papers

1. Training Recurrent Neural Networks Online by Learning Explicit State Variables. Somjit Nath, Vincent Liu, Alan Chan, Xin Li, Adam White and Martha White.
2. Unbiased Online Recurrent Optimization. Corentin Tallec, Yann Ollivier.

Running

python main_torch.py --dataset cw --data_size 10000 --model fpp --state_update True --overlap True --T 10 --num_update 1 --batch_size 1 --lr 0.001 --buffer_size 100 --num_run 10

You can see the optional arguments by running:

python main_torch.py --help