mRNNTorch

Effectively design, train, and analyze multi-regional recurrent neural networks in pytorch.

Highlights

• Flexible: design models with json files or manually in a custom model

• Specialized Analysis: custom analysis module allows for targeting particular sets of regions

• Built with Pytorch: easily extends to common pytorch functionality

Quickstart

Download the repository and pip install in root:

pip install -e .

import and use in custom rnn model:

import mrnntorch
from mrnntorch.mrnn import mRNN

class RNN(nn.Module):
   def __init__(self, config):
     super().__init__()
     self.mrnn = mRNN(config)

   def forward(self, input, x, h):
     xn, hn = self.mrnn(input, x, h)
     return xn, hn

Documentation

Note

This project is under active development. If something is unclear, open an issue.

• mrnntorch