data_for_skillblender / create_conda_env_rlgpu.sh
e-zorzi's picture
Upload folder using huggingface_hub
83e0ecd verified
Raw History Blame Contribute Delete
1.14 kB
#!/bin/bash
ROOT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
# should match env name from YAML
ENV_NAME=rlgpu
pushd "${ROOT_DIR}/python"
# setup conda
CONDA_DIR="$(conda info --base)"
source "${CONDA_DIR}/etc/profile.d/conda.sh"
# deactivate the env, if it is active
ACTIVE_ENV_NAME="$(basename ${CONDA_PREFIX})"
if [ "${ENV_NAME}" = "${ACTIVE_ENV_NAME}" ]; then
conda deactivate
fi
# !!! this removes existing version of the env
conda remove -y -n "${ENV_NAME}" --all
# create the env from YAML
conda env create -f ./rlgpu_conda_env.yml
if [ $? -ne 0 ]; then
echo "*** Failed to create env"
exit 1
fi
# activate env
conda activate "${ENV_NAME}"
if [ $? -ne 0 ]; then
echo "*** Failed to activate env"
exit 1
fi
# double check that the correct env is active
ACTIVE_ENV_NAME="$(basename ${CONDA_PREFIX})"
if [ "${ENV_NAME}" != "${ACTIVE_ENV_NAME}" ]; then
echo "*** Env is not active, aborting"
exit 1
fi
# install isaacgym package
pip install -e .
popd
echo "SUCCESS"