GymValueIterationNetwork

Value Iteration Networks - GYM Environment

This project is extension of the work of Aviv T. et. al. (https://arxiv.org/abs/1602.02867), by applying VIN model to Open AI gym environment - LunarLander-v2 and BipedalWalker-v2.

Dependencies

Python 3.6
Numpy
Matplotlib
Opencv
Pytorch
TensorFlow
Openai GYM
Openai Baseline
Visdom

clone all files with:

$ git clone https://github.com/LiorAl/GymValueIterationNetworks

Train

In order to train LunarLander use:

$ python main.py --env-name "LunarLander-v2" --algo ppo --use-gae --lr 2.5e-4 --clip-param 0.1 --value-loss-coef 0.5 --num-processes 6 --num-steps 128 --num-mini-batch 4 --vis-interval 1 --log-interval 1 --use-linear-lr-decay --use-linear-clip-decay --entropy-coef 0.01

In order to train BipedalWalker use:

$ python main.py --env-name "BipedalWalker-v2" --algo ppo --use-gae --lr 2.5e-4 --clip-param 0.1 --value-loss-coef 0.5 --num-processes 6 --num-steps 128 --num-mini-batch 4 --vis-interval 1 --log-interval 1 --use-linear-lr-decay --use-linear-clip-decay --entropy-coef 0.01

Test

In order to load pretrain model and run LunarLander use:

$ python test_agent.py --load-dir trained_models/ppo --env-name "LunarLander-v2"

BipedalWalker:

python test_agent.py --load-dir trained_models/ppo --env-name "BipedalWalker-v2"

Files Description

ppo -> files for PPO algorithm.
trained_models -> Pretrained models.
BipedalModel.py -> VIN model for BpidealWalker env.
LunarLanderModel.py -> VIN model for LunarLander env.
main.py -> main script for training VIN architecture.