shekkizh / neuralnetworks.thought-experiments

Observations and notes to understand the workings of neural network models and other thought experiments using Tensorflow

neuralnetworks.thought-experiments

A repo of observations and notes to understand the workings of neural network models and other simple thought experiments using Tensorflow.

• Activations - Notes on activation functions with experimental results to compare models on MNIST and CIFAR10
• Generative Models - Observations on generative machine learning and results using Auto Encoders, VAEs and GANs
• Model Pruning - Results on doing pruning in neural networks. (mainly OBD as of now.)

neuralnetworks.thought-experiments's Issues

Optimal Brain Damage

Hello! I am very interested in the work of "Optimal Brain Damage" in "Model Pruning". But I didn't understand the code. How do I use the OBD algorithm to complete model pruning?

Disappeared images in GAN notes

Hi there!

Thanks for your great notes about the difficulty of training GANs. I see that some of the images in your notes are missing. I think the images are from the examples of changing a dimension in the latent z space. Can you please fix it when you have time?

Thanks