Source code for torchrl.collectors.distributed.sync
# Copyright (c) Meta Platforms, Inc. and affiliates.
#
# This source code is licensed under the MIT license found in the
# LICENSE file in the root directory of this source tree.
r"""Generic distributed data-collector using torch.distributed backend."""
from __future__ import annotations
import os
import socket
from copy import copy, deepcopy
from datetime import timedelta
from typing import Callable, List, OrderedDict
import torch.cuda
from tensordict import TensorDict
from torch import nn
from torchrl._utils import _ProcessNoWarn, logger as torchrl_logger, VERBOSE
from torchrl.collectors import MultiaSyncDataCollector
from torchrl.collectors.collectors import (
DataCollectorBase,
DEFAULT_EXPLORATION_TYPE,
MultiSyncDataCollector,
SyncDataCollector,
)
from torchrl.collectors.distributed.default_configs import (
DEFAULT_SLURM_CONF,
MAX_TIME_TO_CONNECT,
)
from torchrl.collectors.utils import split_trajectories
from torchrl.data.utils import CloudpickleWrapper
from torchrl.envs.common import EnvBase
from torchrl.envs.env_creator import EnvCreator
from torchrl.envs.utils import _convert_exploration_type
SUBMITIT_ERR = None
try:
import submitit
_has_submitit = True
except ModuleNotFoundError as err:
_has_submitit = False
SUBMITIT_ERR = err
def _distributed_init_collection_node(
rank,
rank0_ip,
tcpport,
world_size,
backend,
collector_class,
num_workers,
env_make,
policy,
frames_per_batch,
collector_kwargs,
update_interval,
total_frames,
verbose=VERBOSE,
):
os.environ["MASTER_ADDR"] = str(rank0_ip)
os.environ["MASTER_PORT"] = str(tcpport)
if verbose:
torchrl_logger.info(
f"node with rank {rank} -- creating collector of type {collector_class}"
)
if not issubclass(collector_class, SyncDataCollector):
env_make = [env_make] * num_workers
else:
collector_kwargs["return_same_td"] = True
if num_workers != 1:
raise RuntimeError(
"SyncDataCollector and subclasses can only support a single environment."
)
if isinstance(policy, nn.Module):
policy_weights = TensorDict(dict(policy.named_parameters()), [])
# TODO: Do we want this?
# updates the policy weights to avoid them to be shared
if all(
param.device == torch.device("cpu") for param in policy_weights.values()
):
policy = deepcopy(policy)
policy_weights = TensorDict(dict(policy.named_parameters()), [])
policy_weights = policy_weights.apply(lambda x: x.data)
else:
policy_weights = TensorDict({}, [])
collector = collector_class(
env_make,
policy,
frames_per_batch=frames_per_batch,
split_trajs=False,
total_frames=total_frames,
**collector_kwargs,
)
torchrl_logger.info(f"IP address: {rank0_ip} \ttcp port: {tcpport}")
if verbose:
torchrl_logger.info(f"node with rank {rank} -- launching distributed")
torch.distributed.init_process_group(
backend,
rank=rank,
world_size=world_size,
timeout=timedelta(MAX_TIME_TO_CONNECT),
# init_method=f"tcp://{rank0_ip}:{tcpport}",
)
if verbose:
torchrl_logger.info(f"node with rank {rank} -- creating store")
if verbose:
torchrl_logger.info(f"node with rank {rank} -- loop")
policy_weights.irecv(0)
frames = 0
for i, data in enumerate(collector):
data.isend(dst=0)
frames += data.numel()
if (
frames < total_frames
and (i + 1) % update_interval == 0
and not policy_weights.is_empty()
):
policy_weights.irecv(0)
if not collector.closed:
collector.shutdown()
del collector
return
[docs]class DistributedSyncDataCollector(DataCollectorBase):
"""A distributed synchronous data collector with torch.distributed backend.
Args:
create_env_fn (Callable or List[Callabled]): list of Callables, each returning an
instance of :class:`~torchrl.envs.EnvBase`.
policy (Callable): Policy to be executed in the environment.
Must accept :class:`tensordict.tensordict.TensorDictBase` object as input.
If ``None`` is provided, the policy used will be a
:class:`~torchrl.collectors.RandomPolicy` instance with the environment
``action_spec``.
Accepted policies are usually subclasses of :class:`~tensordict.nn.TensorDictModuleBase`.
This is the recommended usage of the collector.
Other callables are accepted too:
If the policy is not a ``TensorDictModuleBase`` (e.g., a regular :class:`~torch.nn.Module`
instances) it will be wrapped in a `nn.Module` first.
Then, the collector will try to assess if these
modules require wrapping in a :class:`~tensordict.nn.TensorDictModule` or not.
- If the policy forward signature matches any of ``forward(self, tensordict)``,
``forward(self, td)`` or ``forward(self, <anything>: TensorDictBase)`` (or
any typing with a single argument typed as a subclass of ``TensorDictBase``)
then the policy won't be wrapped in a :class:`~tensordict.nn.TensorDictModule`.
- In all other cases an attempt to wrap it will be undergone as such: ``TensorDictModule(policy, in_keys=env_obs_key, out_keys=env.action_keys)``.
Keyword Args:
frames_per_batch (int): A keyword-only argument representing the total
number of elements in a batch.
total_frames (int): A keyword-only argument representing the total
number of frames returned by the collector
during its lifespan. If the ``total_frames`` is not divisible by
``frames_per_batch``, an exception is raised.
Endless collectors can be created by passing ``total_frames=-1``.
Defaults to ``-1`` (endless collector).
device (int, str or torch.device, optional): The generic device of the
collector. The ``device`` args fills any non-specified device: if
``device`` is not ``None`` and any of ``storing_device``, ``policy_device`` or
``env_device`` is not specified, its value will be set to ``device``.
Defaults to ``None`` (No default device).
Lists of devices are supported.
storing_device (int, str or torch.device, optional): The *remote* device on which
the output :class:`~tensordict.TensorDict` will be stored.
If ``device`` is passed and ``storing_device`` is ``None``, it will
default to the value indicated by ``device``.
For long trajectories, it may be necessary to store the data on a different
device than the one where the policy and env are executed.
Defaults to ``None`` (the output tensordict isn't on a specific device,
leaf tensors sit on the device where they were created).
Lists of devices are supported.
env_device (int, str or torch.device, optional): The *remote* device on which
the environment should be cast (or executed if that functionality is
supported). If not specified and the env has a non-``None`` device,
``env_device`` will default to that value. If ``device`` is passed
and ``env_device=None``, it will default to ``device``. If the value
as such specified of ``env_device`` differs from ``policy_device``
and one of them is not ``None``, the data will be cast to ``env_device``
before being passed to the env (i.e., passing different devices to
policy and env is supported). Defaults to ``None``.
Lists of devices are supported.
policy_device (int, str or torch.device, optional): The *remote* device on which
the policy should be cast.
If ``device`` is passed and ``policy_device=None``, it will default
to ``device``. If the value as such specified of ``policy_device``
differs from ``env_device`` and one of them is not ``None``,
the data will be cast to ``policy_device`` before being passed to
the policy (i.e., passing different devices to policy and env is
supported). Defaults to ``None``.
Lists of devices are supported.
max_frames_per_traj (int, optional): Maximum steps per trajectory.
Note that a trajectory can span across multiple batches (unless
``reset_at_each_iter`` is set to ``True``, see below).
Once a trajectory reaches ``n_steps``, the environment is reset.
If the environment wraps multiple environments together, the number
of steps is tracked for each environment independently. Negative
values are allowed, in which case this argument is ignored.
Defaults to ``None`` (i.e., no maximum number of steps).
init_random_frames (int, optional): Number of frames for which the
policy is ignored before it is called. This feature is mainly
intended to be used in offline/model-based settings, where a
batch of random trajectories can be used to initialize training.
If provided, it will be rounded up to the closest multiple of frames_per_batch.
Defaults to ``None`` (i.e. no random frames).
reset_at_each_iter (bool, optional): Whether environments should be reset
at the beginning of a batch collection.
Defaults to ``False``.
postproc (Callable, optional): A post-processing transform, such as
a :class:`~torchrl.envs.Transform` or a :class:`~torchrl.data.postprocs.MultiStep`
instance.
Defaults to ``None``.
split_trajs (bool, optional): Boolean indicating whether the resulting
TensorDict should be split according to the trajectories.
See :func:`~torchrl.collectors.utils.split_trajectories` for more
information.
Defaults to ``False``.
exploration_type (ExplorationType, optional): interaction mode to be used when
collecting data. Must be one of ``torchrl.envs.utils.ExplorationType.RANDOM``,
``torchrl.envs.utils.ExplorationType.MODE`` or ``torchrl.envs.utils.ExplorationType.MEAN``.
Defaults to ``torchrl.envs.utils.ExplorationType.RANDOM``.
collector_class (type or str, optional): a collector class for the remote node. Can be
:class:`~torchrl.collectors.SyncDataCollector`,
:class:`~torchrl.collectors.MultiSyncDataCollector`,
:class:`~torchrl.collectors.MultiaSyncDataCollector`
or a derived class of these. The strings "single", "sync" and
"async" correspond to respective class.
Defaults to :class:`~torchrl.collectors.SyncDataCollector`.
collector_kwargs (dict or list, optional): a dictionary of parameters to be passed to the
remote data-collector. If a list is provided, each element will
correspond to an individual set of keyword arguments for the
dedicated collector.
num_workers_per_collector (int, optional): the number of copies of the
env constructor that is to be used on the remote nodes.
Defaults to 1 (a single env per collector).
On a single worker node all the sub-workers will be
executing the same environment. If different environments need to
be executed, they should be dispatched across worker nodes, not
subnodes.
slurm_kwargs (dict): a dictionary of parameters to be passed to the
submitit executor.
backend (str, optional): must a string "<distributed_backed>" where
<distributed_backed> is one of ``"gloo"``, ``"mpi"``, ``"nccl"`` or ``"ucc"``. See
the torch.distributed documentation for more information.
Defaults to ``"gloo"``.
max_weight_update_interval (int, optional): the maximum number of
batches that can be collected before the policy weights of a worker
is updated.
For sync collections, this parameter is overwritten by ``update_after_each_batch``.
For async collections, it may be that one worker has not seen its
parameters being updated for a certain time even if ``update_after_each_batch``
is turned on.
Defaults to -1 (no forced update).
update_interval (int, optional): the frequency at which the policy is
updated. Defaults to 1.
launcher (str, optional): how jobs should be launched.
Can be one of "submitit" or "mp" for multiprocessing. The former
can launch jobs across multiple nodes, whilst the latter will only
launch jobs on a single machine. "submitit" requires the homonymous
library to be installed.
To find more about submitit, visit
https://github.com/facebookincubator/submitit
Defaults to "submitit".
tcp_port (int, optional): the TCP port to be used. Defaults to 10003.
"""
def __init__(
self,
create_env_fn,
policy,
*,
frames_per_batch: int,
total_frames: int = -1,
device: torch.device | List[torch.device] = None,
storing_device: torch.device | List[torch.device] = None,
env_device: torch.device | List[torch.device] = None,
policy_device: torch.device | List[torch.device] = None,
max_frames_per_traj: int = -1,
init_random_frames: int = -1,
reset_at_each_iter: bool = False,
postproc: Callable | None = None,
split_trajs: bool = False,
exploration_type: "ExporationType" = DEFAULT_EXPLORATION_TYPE, # noqa
exploration_mode: str = None,
collector_class=SyncDataCollector,
collector_kwargs=None,
num_workers_per_collector=1,
slurm_kwargs=None,
backend="gloo",
max_weight_update_interval=-1,
update_interval=1,
launcher="submitit",
tcp_port=None,
):
exploration_type = _convert_exploration_type(
exploration_mode=exploration_mode, exploration_type=exploration_type
)
if collector_class == "async":
collector_class = MultiaSyncDataCollector
elif collector_class == "sync":
collector_class = MultiSyncDataCollector
elif collector_class == "single":
collector_class = SyncDataCollector
self.collector_class = collector_class
self.env_constructors = create_env_fn
self.policy = policy
if isinstance(policy, nn.Module):
policy_weights = TensorDict(dict(policy.named_parameters()), [])
policy_weights = policy_weights.apply(lambda x: x.data)
else:
policy_weights = TensorDict({}, [])
self.policy_weights = policy_weights
self.num_workers = len(create_env_fn)
self.frames_per_batch = frames_per_batch
self.device = device
self.storing_device = storing_device
self.env_device = env_device
self.policy_device = policy_device
self.storing_device = storing_device
# make private to avoid changes from users during collection
self.update_interval = update_interval
self.total_frames_per_collector = total_frames // self.num_workers
if self.total_frames_per_collector * self.num_workers != total_frames:
raise RuntimeError(
f"Cannot dispatch {total_frames} frames across {self.num_workers}. "
f"Consider using a number of frames that is divisible by the number of workers."
)
self.max_weight_update_interval = max_weight_update_interval
self.launcher = launcher
self._batches_since_weight_update = [0 for _ in range(self.num_workers)]
if tcp_port is None:
self.tcp_port = os.environ.get("TCP_PORT", "10003")
else:
self.tcp_port = str(tcp_port)
if self.frames_per_batch % self.num_workers != 0:
raise RuntimeError(
f"Cannot dispatch {self.frames_per_batch} frames across {self.num_workers}. "
f"Consider using a number of frames per batch that is divisible by the number of workers."
)
self._frames_per_batch_corrected = self.frames_per_batch // self.num_workers
self.num_workers_per_collector = num_workers_per_collector
self.total_frames = total_frames
self.slurm_kwargs = copy(DEFAULT_SLURM_CONF)
if slurm_kwargs is not None:
self.slurm_kwargs.update(slurm_kwargs)
collector_kwargs = collector_kwargs if collector_kwargs is not None else {}
self.collector_kwargs = (
deepcopy(collector_kwargs)
if isinstance(collector_kwargs, (list, tuple))
else [copy(collector_kwargs) for _ in range(self.num_workers)]
)
# update collector kwargs
for i, collector_kwarg in enumerate(self.collector_kwargs):
collector_kwarg["max_frames_per_traj"] = max_frames_per_traj
collector_kwarg["init_random_frames"] = (
init_random_frames // self.num_workers
)
collector_kwarg["reset_at_each_iter"] = reset_at_each_iter
collector_kwarg["exploration_type"] = exploration_type
collector_kwarg["device"] = self.device[i]
collector_kwarg["storing_device"] = self.storing_device[i]
collector_kwarg["env_device"] = self.env_device[i]
collector_kwarg["policy_device"] = self.policy_device[i]
self.postproc = postproc
self.split_trajs = split_trajs
self.backend = backend
# os.environ['TP_SOCKET_IFNAME'] = 'lo'
self._init_workers()
self._make_container()
@property
def device(self) -> List[torch.device]:
return self._device
@property
def storing_device(self) -> List[torch.device]:
return self._storing_device
@property
def env_device(self) -> List[torch.device]:
return self._env_device
@property
def policy_device(self) -> List[torch.device]:
return self._policy_device
@device.setter
def device(self, value):
if isinstance(value, (tuple, list)):
if len(value) != self.num_workers:
raise RuntimeError(
"The number of devices passed to the collector must match the number of workers."
)
self._device = value
else:
self._device = [value] * self.num_workers
@storing_device.setter
def storing_device(self, value):
if isinstance(value, (tuple, list)):
if len(value) != self.num_workers:
raise RuntimeError(
"The number of devices passed to the collector must match the number of workers."
)
self._storing_device = value
else:
self._storing_device = [value] * self.num_workers
@env_device.setter
def env_device(self, value):
if isinstance(value, (tuple, list)):
if len(value) != self.num_workers:
raise RuntimeError(
"The number of devices passed to the collector must match the number of workers."
)
self._env_device = value
else:
self._env_device = [value] * self.num_workers
@policy_device.setter
def policy_device(self, value):
if isinstance(value, (tuple, list)):
if len(value) != self.num_workers:
raise RuntimeError(
"The number of devices passed to the collector must match the number of workers."
)
self._policy_device = value
else:
self._policy_device = [value] * self.num_workers
def _init_master_dist(
self,
world_size,
backend,
):
TCP_PORT = self.tcp_port
torchrl_logger.info("init master...")
torch.distributed.init_process_group(
backend,
rank=0,
world_size=world_size,
timeout=timedelta(MAX_TIME_TO_CONNECT),
init_method=f"tcp://{self.IPAddr}:{TCP_PORT}",
)
torchrl_logger.info("done")
def _make_container(self):
env_constructor = self.env_constructors[0]
pseudo_collector = SyncDataCollector(
env_constructor,
self.policy,
frames_per_batch=self._frames_per_batch_corrected,
total_frames=self.total_frames,
split_trajs=False,
)
for _data in pseudo_collector:
break
self._tensordict_out = _data.expand((self.num_workers, *_data.shape))
self._single_tds = self._tensordict_out.unbind(0)
self._tensordict_out.lock_()
pseudo_collector.shutdown()
del pseudo_collector
def _init_worker_dist_submitit(self, executor, i):
TCP_PORT = self.tcp_port
env_make = self.env_constructors[i]
if not isinstance(env_make, (EnvBase, EnvCreator)):
env_make = CloudpickleWrapper(env_make)
job = executor.submit(
_distributed_init_collection_node,
i + 1,
self.IPAddr,
int(TCP_PORT),
self.num_workers + 1,
self.backend,
self.collector_class,
self.num_workers_per_collector,
env_make,
self.policy,
self._frames_per_batch_corrected,
self.collector_kwargs[i],
self.update_interval,
self.total_frames_per_collector,
)
return job
def _init_worker_dist_mp(self, i):
TCP_PORT = self.tcp_port
env_make = self.env_constructors[i]
if not isinstance(env_make, (EnvBase, EnvCreator)):
env_make = CloudpickleWrapper(env_make)
job = _ProcessNoWarn(
target=_distributed_init_collection_node,
args=(
i + 1,
self.IPAddr,
int(TCP_PORT),
self.num_workers + 1,
self.backend,
self.collector_class,
self.num_workers_per_collector,
env_make,
self.policy,
self._frames_per_batch_corrected,
self.collector_kwargs[i],
self.update_interval,
self.total_frames_per_collector,
),
)
job.start()
return job
def _init_workers(self):
hostname = socket.gethostname()
IPAddr = socket.gethostbyname(hostname)
torchrl_logger.info(f"Server IP address: {IPAddr}")
self.IPAddr = IPAddr
os.environ["MASTER_ADDR"] = str(self.IPAddr)
os.environ["MASTER_PORT"] = str(self.tcp_port)
self.jobs = []
if self.launcher == "submitit":
if not _has_submitit:
raise ImportError("submitit not found.") from SUBMITIT_ERR
executor = submitit.AutoExecutor(folder="log_test")
executor.update_parameters(**self.slurm_kwargs)
for i in range(self.num_workers):
torchrl_logger.info("Submitting job")
if self.launcher == "submitit":
job = self._init_worker_dist_submitit(
executor,
i,
)
torchrl_logger.info(f"job id {job.job_id}") # ID of your job
elif self.launcher == "mp":
job = self._init_worker_dist_mp(
i,
)
torchrl_logger.info("job launched")
self.jobs.append(job)
self._init_master_dist(self.num_workers + 1, self.backend)
def iterator(self):
yield from self._iterator_dist()
def _iterator_dist(self):
total_frames = 0
j = -1
while total_frames < self.total_frames:
j += 1
if j % self.update_interval == 0 and not self.policy_weights.is_empty():
for i in range(self.num_workers):
rank = i + 1
self.policy_weights.isend(rank)
trackers = []
for i in range(self.num_workers):
rank = i + 1
trackers.append(
self._single_tds[i].irecv(src=rank, return_premature=True)
)
for tracker in trackers:
for _tracker in tracker:
_tracker.wait()
data = self._tensordict_out.clone()
traj_ids = data.get(("collector", "traj_ids"), None)
if traj_ids is not None:
for i in range(1, self.num_workers):
traj_ids[i] += traj_ids[i - 1].max()
data.set_(("collector", "traj_ids"), traj_ids)
total_frames += data.numel()
if self.split_trajs:
data = split_trajectories(data)
if self.postproc is not None:
data = self.postproc(data)
yield data
def set_seed(self, seed: int, static_seed: bool = False) -> int:
raise NotImplementedError
def state_dict(self) -> OrderedDict:
raise NotImplementedError
def load_state_dict(self, state_dict: OrderedDict) -> None:
raise NotImplementedError
def shutdown(self):
pass
