Advertisement
Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- Hyperparameter,value
- Number of Episodes, 2000
- Number of Timesteps, 1000
- Print Checkpoint step every, 4
- Training Batch Size, 64
- Discount Rate / Gamma, 0.99
- Learning Rate / alpha, 5e-4
- Number of Hidden Layers, 2
- Fully Connected Layer 1 Units, 64
- Fully Connected Layer 2 Units, 64
- TAU, 1e-3
- Epsilon, 0.1
- Epsilon-Min, 0.01
- Epsilon-Decay, 0.995
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement