Distributed Training for A Simple Network by Distributed RPC Framework#
The example shows how to combine Distributed DataParallel with the Distributed RPC Framework. There are two trainer nodes, 1 master node and 1 parameter server in the example.
The master node creates an embedding table on the parameter server and drives the training loop on the trainers. The model consists of a dense part (cdopt.nn.Linear_cdopt) replicated on the trainers via Distributed DataParallel and a sparse part (nn.EmbeddingBag) which resides on the parameter server. Each trainer performs an embedding lookup on the parameter server (using the Distributed RPC Framework) and then executes its local nn.Linear module. During the backward pass, the gradients for the dense part are aggregated via allreduce by DDP and the distributed backward pass updates the parameters for the embedding table on the parameter server.
To run the code, just save the following code as main.py and use the command python main.py.
import random
import torch
import torch.distributed as dist
import torch.distributed.autograd as dist_autograd
import torch.distributed.rpc as rpc
import torch.multiprocessing as mp
import torch.optim as optim
from torch.distributed.nn import RemoteModule
from torch.distributed.optim import DistributedOptimizer
from torch.distributed.rpc import RRef
from torch.distributed.rpc import TensorPipeRpcBackendOptions
from torch.nn.parallel import DistributedDataParallel as DDP
import cdopt
from cdopt.manifold_torch import stiefel_torch, oblique_torch
NUM_EMBEDDINGS = 100
EMBEDDING_DIM = 16
class HybridModel(torch.nn.Module):
r"""
The model consists of a sparse part and a dense part.
1) The dense part is an nn.Linear module that is replicated across all trainers using DistributedDataParallel.
2) The sparse part is a Remote Module that holds an nn.EmbeddingBag on the parameter server.
This remote model can get a Remote Reference to the embedding table on the parameter server.
"""
def __init__(self, remote_emb_module, device):
super(HybridModel, self).__init__()
self.remote_emb_module = remote_emb_module
self.fc = DDP(cdopt.nn.Linear_cdopt(16, 8, manifold_class=stiefel_torch, penalty_param = 0.5).cuda(device), device_ids=[device])
self.device = device
def forward(self, indices, offsets):
emb_lookup = self.remote_emb_module.forward(indices, offsets)
return self.fc(emb_lookup.cuda(self.device))
def _run_trainer(remote_emb_module, rank):
r"""
Each trainer runs a forward pass which involves an embedding lookup on the
parameter server and running nn.Linear locally. During the backward pass,
DDP is responsible for aggregating the gradients for the dense part
(nn.Linear) and distributed autograd ensures gradients updates are
propagated to the parameter server.
"""
# Setup the model.
model = HybridModel(remote_emb_module, rank)
# Retrieve all model parameters as rrefs for DistributedOptimizer.
# Retrieve parameters for embedding table.
model_parameter_rrefs = model.remote_emb_module.remote_parameters()
# model.fc.parameters() only includes local parameters.
# NOTE: Cannot call model.parameters() here,
# because this will call remote_emb_module.parameters(),
# which supports remote_parameters() but not parameters().
for param in model.fc.parameters():
model_parameter_rrefs.append(RRef(param))
# Setup distributed optimizer
opt = DistributedOptimizer(
optim.SGD,
model_parameter_rrefs,
lr=0.05,
)
criterion = torch.nn.CrossEntropyLoss()
def get_next_batch(rank):
for _ in range(10):
num_indices = random.randint(20, 50)
indices = torch.LongTensor(num_indices).random_(0, NUM_EMBEDDINGS)
# Generate offsets.
offsets = []
start = 0
batch_size = 0
while start < num_indices:
offsets.append(start)
start += random.randint(1, 10)
batch_size += 1
offsets_tensor = torch.LongTensor(offsets)
target = torch.LongTensor(batch_size).random_(8).cuda(rank)
yield indices, offsets_tensor, target
# Train for 100 epochs
for epoch in range(100):
# create distributed autograd context
for indices, offsets, target in get_next_batch(rank):
with dist_autograd.context() as context_id:
output = model(indices, offsets)
loss = criterion(output, target) + get_quad_penalty(model)
# Run distributed backward pass
dist_autograd.backward(context_id, [loss])
# Tun distributed optimizer
opt.step(context_id)
# Not necessary to zero grads as each iteration creates a different
# distributed autograd context which hosts different grads
print("Training epoch: {}, feas: {}".format(epoch, model.fc.module.quad_penalty()))
def run_worker(rank, world_size):
r"""
A wrapper function that initializes RPC, calls the function, and shuts down
RPC.
"""
# We need to use different port numbers in TCP init_method for init_rpc and
# init_process_group to avoid port conflicts.
rpc_backend_options = TensorPipeRpcBackendOptions()
rpc_backend_options.init_method = "tcp://localhost:29501"
# Rank 2 is master, 3 is ps and 0 and 1 are trainers.
if rank == 2:
rpc.init_rpc(
"master",
rank=rank,
world_size=world_size,
rpc_backend_options=rpc_backend_options,
)
remote_emb_module = RemoteModule(
"ps",
torch.nn.EmbeddingBag,
args=(NUM_EMBEDDINGS, EMBEDDING_DIM),
kwargs={"mode": "sum"},
)
# Run the training loop on trainers.
futs = []
for trainer_rank in [0, 1]:
trainer_name = "trainer{}".format(trainer_rank)
fut = rpc.rpc_async(
trainer_name, _run_trainer, args=(remote_emb_module, trainer_rank)
)
futs.append(fut)
# Wait for all training to finish.
for fut in futs:
fut.wait()
elif rank <= 1:
# Initialize process group for Distributed DataParallel on trainers.
dist.init_process_group(
backend="gloo", rank=rank, world_size=2, init_method="tcp://localhost:29500"
)
# Initialize RPC.
trainer_name = "trainer{}".format(rank)
rpc.init_rpc(
trainer_name,
rank=rank,
world_size=world_size,
rpc_backend_options=rpc_backend_options,
)
# Trainer just waits for RPCs from master.
else:
rpc.init_rpc(
"ps",
rank=rank,
world_size=world_size,
rpc_backend_options=rpc_backend_options,
)
# parameter server do nothing
pass
# block until all rpcs finish
rpc.shutdown()
if __name__ == "__main__":
# 2 trainers, 1 parameter server, 1 master.
world_size = 4
mp.spawn(run_worker, args=(world_size,), nprocs=world_size, join=True)