Training tasks with long temporal dependencies
Work done in collaboration with Wayne Soo
Applies annealed skip-connections through time to support training of biological RNNs on tasks with long (across trial) temporal dependencies.
Currently implemented Reversal Learning and classic WCST (set task variable in main_workspace.py)
To run, call main_workspace.py with command line arguments where is the number of time steps to skip over