rlworkgroup / garage

A toolkit for reproducible reinforcement learning research.
MIT License
1.88k stars 310 forks source link

Docs page "Implement a custom Encoder/Decoder" #2102

Open ryanjulian opened 4 years ago

ryanjulian commented 4 years ago

Should cover TensorFlow, PyTorch, and NumPy