- Try to be self-contained as possible within a single Jupyter notebook.
- Try to use
tf.layer
for more unified implementation.
- Linear regression
- MLP: Multi Layer Perceptron for MNIST Classification
- CNN: Convolutional Neural Networks for MNIST Classification
- MDN: Mixture Density Network for Toy Regression Task
- PPO: Proximal Policy Optimization for PyBullet Ant
- SAC: Soft Actor-Critic for PyBullet Ant
- VAE: Variational AutoEncoder
- WAE: Wasserstein AutoEncoder
- GAN: Generative Adversarial Networks
- TF Model save and restore
- Parallel Processing with Ray
- Parallel Rollouts with TF Model
- Distributed PPO with Ray
- Using Custom Forward and Backward Paths in TF