ENVS = [
#'flow-ring-random-v0',
#'flow-ring-controller-v0',
'flow-merge-random-v0',
'flow-merge-controller-v0',
]
mounts = []
mounts.append(
doodad.MountLocal(local_dir='~/code/d4rl',
mount_point='/code/d4rl',
pythonpath=True))
sweeper = launcher.DoodadSweeper(mounts=mounts,
docker_img='justinfu/brac_flow:0.3',
docker_output_dir='/root/tmp/offlinerl',
gcp_bucket_name='justin-doodad',
gcp_image='ubuntu-1804-docker-gpu',
gcp_project='qlearning000')
for env_name in ENVS:
env = gym.make(env_name)
_, dataset = os.path.split(env.dataset_filepath)
dirname, _ = os.path.splitext(dataset)
params = {
'env_name': [env_name],
'seed': range(3),
'agent_name': ['brac_primal'],
'total_train_steps': [100000],
'sub_dir': ['auto'],
'model_arch': [1],
'flow-merge-controller-v0',
]
ENVS.extend(FLOW_ENVS)
mounts = []
mounts.append(doodad.MountLocal(local_dir='~/code/awr',
mount_point='/code/awr', pythonpath=True))
mounts.append(doodad.MountLocal(local_dir='~/code/d4rl',
mount_point='/code/d4rl', pythonpath=True))
sweeper = launcher.DoodadSweeper(
mounts=mounts,
docker_img='justinfu/awr:0.1',
gcp_bucket_name='justin-doodad',
gcp_image='ubuntu-1804-docker-gpu',
gcp_project='qlearning000'
)
flow_sweeper = launcher.DoodadSweeper(
mounts=mounts,
docker_img='justinfu/awr_flow:0.1',
gcp_bucket_name='justin-doodad',
gcp_image='ubuntu-1804-docker-gpu',
gcp_project='qlearning000'
)
for env_name in ENVS:
env = gym.make(env_name)
#'flow-ring-random-v0',
#'flow-ring-controller-v0',
"flow-merge-random-v0",
"flow-merge-controller-v0",
]
mounts = []
mounts.append(
doodad.MountLocal(local_dir="~/code/d4rl",
mount_point="/code/d4rl",
pythonpath=True))
sweeper = launcher.DoodadSweeper(
mounts=mounts,
docker_img="justinfu/brac_flow:0.3",
docker_output_dir="/root/tmp/offlinerl",
gcp_bucket_name="justin-doodad",
gcp_image="ubuntu-1804-docker-gpu",
gcp_project="qlearning000",
)
for env_name in ENVS:
env = gym.make(env_name)
_, dataset = os.path.split(env.dataset_filepath)
dirname, _ = os.path.splitext(dataset)
params = {
"env_name": [env_name],
"seed": range(3),
"agent_name": ["brac_primal"],
"total_train_steps": [100000],
"sub_dir": ["auto"],
def setUp(self):
self.sweeper = launcher.DoodadSweeper()