requirements_latest.txt

# TODO: separate requirements for evo/rl (while ensuring they do not conflict).
# under the gcc verison of: gcc (Ubuntu 9.4.0-1ubuntu1~20.04.1) 9.4.0 (Tirona) ; python 3.10

chardet       # something is missing this to import name 'COMMON_SAFE_ASCII_CHARACTERS'
deap
dm-tree        # rllib dependency
einops
graphviz
gym
gymnasium     # gymnasium[all] will cause some box2d gcc beef
# gym==0.19.0     # minerl will require 0.19.0
hydra-core
hydra-submitit-launcher
imageio
lz4                 # for hydra
neat-python
numba
numpy
# numpy==1.23.4
# numpy==1.21  # numba requires this version or lower
opencv-python      # M1 mac users may need to install this separately (or try pip install opencv-python separately; requires numpy 1.23+)
pandas
pingouin
psutil
pygame
pyglet
PyOpenGL
pyscreenshot
ribs
seaborn
scikit-image
# submitit
tensorboard
tensorboardX      # for having the tensorboard logging (rllib use this to write the logging!)
torch
torchinfo
tqdm
wandb       # after installing wandb, run `wandb login` to login to wandb

# git submodule update --init qdpy/       # add submodule first before python -m pip install -r requirements.txt
# ./submodules/qdpy

# NOTE: only need this for diverse generators, not RL.
# git submodule update --init pytorch-neat/       # add submodule first before python -m pip install -r requirements.txt
# python -m pip install neat-python
./submodules/pytorch-neat

# ray[rllib]
ray==2.3.0  



# Would be great to try using dqd-rl inside evolve.py... until then, we don't need dqd.
#./dqd


# For fancy curriculum of metric targets for controllable RL
# ./teachDeepRL

#PyGObject

# Leftovers from gym-pcgrl with stable-baselines 2
# stable-baselines==2.9.0
# tensorflow==1.15
# torch==1.9.0
# gym==0.21.0    # or 0.16.0
# cloudpickle==2.0.0   # or 1.2.2