Corey Lammie

Fix logic error in `code_gen.py`

1

This fixes the logic error as discussed in #20.

Add Support for `torch.nn.LSTM` and `torch.nn.LSTMCell` Modules

2

Support should be added for `torch.nn.LSTM` and `torch.nn.LSTMCell` modules.

Add Support for `torch.nn.RNN` and `torch.nn.RNNCell` Modules

1

Support should be added for `torch.nn.RNN` and `torch.nn.RNNCell` modules.

- Common plotting functions/utilities could be developed under a new submodule entitled `memtorch.plot`. - Heatmaps of the conductance states of devices should be easily generatable.

Peripheral circuitry temperate variation could be modeled to enable wider device-to-system testing simulations.

Parallelize/Optimize Device Simulation Logic

5

Currently, when performing inference, or programming devices in passive arrays (0T1R arrangements), devices are simulated in a sequential manor. CUDA kernels and other optimization methods can be used to drastically...

To avoid compounding non-idealities, two seperate conductance matrices can be stored for each layer/module: (1) with all non-idealities applied, and (2) with no non-idealities applied. A list can be added...

Currently, all convolutional layer types support `groups`=1 and `groups`=`in_channels` operation modes. Support should be added for `groups`=2, where the operation becomes equivalent to having two conv layers side by side,...

Support should be added for `torch.nn.GRU` and `torch.nn.GRUCell` modules.

In extension to #87, currently, `torch.nn.Sequential` containers within `torch.nn.ModuleDict` and `torch.nn.ModuleList` containers are not able to be patched. It is unclear to what extent containers to be nested, so extensive...