mnist1d
mnist1d copied to clipboard
A 1D analogue of the MNIST dataset for measuring spatial biases and answering Science of Deep Learning questions.
As discussed in #8 here is a small restructuring of the repo to become a full python package. I'd like to add one more thing before this is ready for...
Hi, thanks for contributing this dataset as open source. Motivated by "Understanding Deep Learning", I'd love to start playing with it. I saw that the repo is not ready for...
Hey thank you for this, but pickled objects in Python can be used to invoke remote code execution. It means I have to trust you in order to use this...
Thanks for providing this nice dataset! In the README file (and also in the [paper at arXiv](https://arxiv.org/abs/2011.14439)), it is stated that one can obtain a 68% (test(?)) accuracy using an...
When trying to use the repo in a Colab notebook, in the import section: ``` # Setup import torch import torch.nn as nn import torch.nn.functional as F import torch.optim as...
Hi, for a course, I wrote a `pytorch` compatible Dataset class for MNIST1D. The examples in this repo don't require one as such as most do not rely on a...
Hi, Any chance to make the package available on `conda-forge`?
Hi :wave: I'm new here and trying to learn more about SSL techniques. This dataset seemed like a great place to start! **First try at implementing a masked autoencoder** The...