few-shot-cgan

DONE:

train_classifier.py trains an oracle classifier on the MNIST dataset -> This is what we will use for evaluation.
train_cgan.ipynb trains a standard cGAN on MNIST.
mnist_cnn.pth is the oracle classifier.

Pretrain classifier with CPC, and then just slightly tune it on the noisy target labels.
Setup normal cGAN training loop for pretraining the discriminator and generator.
Switch to finetuning the classifier with the generator and discriminator.

Name		Name	Last commit message	Last commit date
Latest commit History 22 Commits
datasets		datasets
models		models
scripts		scripts
.gitignore		.gitignore
README.md		README.md
eval.py		eval.py
main.py		main.py
mnist_cnn.pth		mnist_cnn.pth
pretrain.py		pretrain.py
pretrain_colab.ipynb		pretrain_colab.ipynb
requirements.txt		requirements.txt
train_cgan.ipynb		train_cgan.ipynb
train_classifier.py		train_classifier.py
train_pseudo.py		train_pseudo.py
train_refinement.py		train_refinement.py