adis98/dl_reprod

Code repo for the reproducibility project

DL_reprod

Code repository for the reproducibility project of course CS4240 Deep Learning, EWI faculty, TU Delft Group 10: Aditya Shankar, Mathieu Jung-Muller, Pietro Piccini, Sreeparna Deb Supervisor: Casper van Engelenburg This project consists of a reproducibility project based on the paper: A Simple Framework for Contrastive Learning of Visual Representations, https://arxiv.org/abs/2002.05709. Our work in this repository is complementary to our blog post: https://medium.com/@m.jungmuller/visualizing-feature-representations-in-simclr-a742d87c8d3.

PyTorch checkpoints: https://drive.google.com/drive/folders/17b_VDIqJNmqkZi35AgkKYCt9Qwk4krUN?usp=sharing

Model code is available in the SimCLR directory.

The converter for the PyTorch checkpoints is helpers/convert.py.

The training steps and accuracy results are available in training_steps directory.

PCA and TSNE images are in pca_tsne_views directory.