Osculari

https://www.repostatus.org/badges/latest/active.svg https://github.com/ArashAkbarinia/osculari/actions/workflows/python-package.yml/badge.svg https://img.shields.io/pypi/v/osculari.svg https://img.shields.io/pypi/pyversions/osculari.svg https://static.pepy.tech/badge/osculari https://codecov.io/gh/ArashAkbarinia/osculari/branch/main/graph/badge.svg https://img.shields.io/badge/PyTorch_1.9.1+-ee4c2c?logo=pytorch&logoColor=white https://img.shields.io/pypi/l/osculari.svg https://zenodo.org/badge/717052640.svg

Osculari (ōsculārī; Latin; to embrace, kiss) is a Python package providing an easy interface for different psychophysical techniques to explore and interpret the internal presentation of artificial neural networks.

  • Supporting the following pretrained models:
  • Managing convolution and transformer architectures.

  • Allowing to readout the network at any given depth.

  • Training a linear classifier on top of the extract features from any network/layer.

  • Experimenting with 2AFC and 4AFC paradigms.