pugavkomm / cgtasknet

cgtasknet

Table of Contents

About
Installation Requirements
Tasks TODO

About

A library aimed at studying the dynamics of spiking neural networks while solving various cognitive tasks:

1. Romo task;
2. Decision-making task;
3. Context decision-making task;
4. Antisaccade task;
5. Pro-saccade task;
6. Go tasks family.

---

Installation

Installation from source code

---

1. git clone https://github.com/Pugavkomm/cgtasknet.git
2. You shold install pytorch (stable version)
3. python setup.py install

You can create your environment and install all the libraries you need to create networks

1. git clone https://github.com/Pugavkomm/cgtasknet.git
2. Create env: python3 -m venv env
3. Activate env: linux: source env/bin/activate | windows: .\env\Scripts\activate
4. You shold install pytorch (stable version)
5. pip3 install -r requirements.txt
6. python setup.py install

Installation from pypi

TODO

Docker

TODO

Requirements

Main dependencies:

1. torch, norse -- Model and learning;
2. numpy -- Prepare datasets;
3. matplotlib -- Data visualization (Currently not in use!);

Tasks

Several classes of cognitive tasks are considered:

1. Romo task
2. Context decision making task
3. decision making task
4. Working memory tasks: DM with delay (just romo task), context DM with delay.
5. Go/No-Go

TODO

• Add my own wrapper over the loading model to load parameters (and save) as well;
• Add some tests for instrument_subgroups;
• Add somet tests for instrument_pca;
• Add some tests for all models.