Abstract

In the last years, neural networks have evolved from laboratory environments to the state-of-the-art for many real-world problems. Our hypothesis is that neural network models (i.e., their weights and biases) evolve on unique, smooth trajectories in weight space during training. Following, a population of such neural network models (refereed to as “model zoo”) would form topological structures in weight space. We think that the geometry, curvature and smoothness of these structures contain information about the state of training and can be reveal latent properties of individual models. With such zoos, one could investigate novel approaches for (i) model analysis, (ii) discover unknown learning dynamics, (iii) learn rich representations of such populations, or (iv) exploit the model zoos for generative modelling of neural network weights and biases. Unfortunately, the lack of standardized model zoos and available benchmarks significantly increases the friction for further research about populations of neural networks. With this work, we publish a novel dataset of model zoos containing systematically generated and diverse populations of neural network models for further research. In total the proposed model zoo dataset is based on six image datasets, consist of 24 model zoos with varying hyperparameter combinations are generated and includes 47’360 unique neural network models resulting in over 2’415’360 collected model states. Additionally, to the model zoo data we provide an in-depth analysis of the zoos and provide benchmarks for multiple downstream tasks as mentioned before.

Dataset

This dataset is part of a larger collection of model zoos and contains the zoos trained on the labelled samples from Fashion-MNIST. All zoos with extensive information and code can be found at www.modelzoos.cc.

This repository contains two types of files: the raw model zoos as collections of models (file names beginning with "fmnist_"), as well as preprocessed model zoos wrapped in a custom pytorch dataset class (filenames beginning with "dataset"). Zoos are trained in three configurations varying the seed only (seed), varying hyperparameters with fixed seeds (hyp_fix) or varying hyperparameters with random seeds (hyp_rand). The index_dict.json files contain information on how to read the vectorized models.

For more information on the zoos and code to access and use the zoos, please see www.modelzoos.cc.

OSCAR, the Occluded Stereo dataset for Convolutional Architectures with Recurrence. Version: 2.0 (dataset as presented in our JOV 2021 journal publication "Recurrent Processing Improves Occluded Object Recognition and Gives Rise to Perceptual Hysteresis")

If you make use of the dataset, please cite as follows:

Ernst, M. R., Burwick, T., & Triesch, J. (2021). Recurrent Processing Improves Occluded Object Recognition and Gives Rise to Perceptual Hysteresis. In Journal of Vision

Contents

readme.md - detailed description and sample pictures

img.zip - folder that contains images for the readme file

licence.md - licence agreement for using the datasets

os-fmnist2c.zip - compressed archive of the occluded stereo FashionMNIST dataset (centered, ~1.1GB)

os-fmnist2r.zip - compressed archive of the occluded stereo FashionMNIST dataset (random, ~1.2GB)

os-mnist2c.zip - compressed archive of the occluded stereo MNIST dataset (centered, ~865MB)

os-mnist2r.zip - compressed archive of the occluded stereo MNIST dataset (random, ~851MB)

os-ycb2.zip - compressed archive of the occluded stereo ycb-object dataset (~1.1GB)

os-ycb2_highres.zip - compressed archive of the occluded stereo ycb-object dataset (high resolution, ~9.8GB)

OSCARv2_dataset.py - python script to directly load image data from folder, pytorch dataset