mirror of https://github.com/hpcaitech/ColossalAI
[zero] add unit test for low-level zero init (#2474)
parent
f525d1f528
commit
21c88220ce
@ -0,0 +1,61 @@
|
|||||||
|
from functools import partial
|
||||||
|
|
||||||
|
import pytest
|
||||||
|
import torch
|
||||||
|
import torch.distributed as dist
|
||||||
|
import torch.multiprocessing as mp
|
||||||
|
import torch.nn as nn
|
||||||
|
|
||||||
|
import colossalai
|
||||||
|
from colossalai.tensor import ProcessGroup
|
||||||
|
from colossalai.utils import free_port, get_current_device
|
||||||
|
from colossalai.utils.model.colo_init_context import ColoInitContext
|
||||||
|
from colossalai.zero import LowLevelZeroOptimizer
|
||||||
|
|
||||||
|
|
||||||
|
class TestModel(nn.Module):
|
||||||
|
|
||||||
|
def __init__(self):
|
||||||
|
super(TestModel, self).__init__()
|
||||||
|
self.linear1 = nn.Linear(128, 256)
|
||||||
|
self.linear2 = nn.Linear(256, 512)
|
||||||
|
|
||||||
|
def forward(self, x):
|
||||||
|
x = self.linear1(x)
|
||||||
|
x = self.linear2(x)
|
||||||
|
return x
|
||||||
|
|
||||||
|
|
||||||
|
def exam_zero_init():
|
||||||
|
dp_2_tp_2_pg = ProcessGroup(dp_degree=2, tp_degree=2)
|
||||||
|
model1 = TestModel().cuda()
|
||||||
|
with ColoInitContext(device=get_current_device(), default_pg=dp_2_tp_2_pg):
|
||||||
|
model2 = TestModel()
|
||||||
|
optimizer1 = LowLevelZeroOptimizer(torch.optim.Adam(model1.parameters(), lr=1))
|
||||||
|
optimizer2 = LowLevelZeroOptimizer(torch.optim.Adam(model2.parameters(), lr=1))
|
||||||
|
|
||||||
|
assert optimizer1._local_rank == optimizer2._local_rank
|
||||||
|
assert optimizer1._world_size == optimizer2._world_size
|
||||||
|
assert optimizer1._dp_global_ranks == optimizer2._dp_global_ranks
|
||||||
|
|
||||||
|
mp_group1 = optimizer1._mp_torch_group
|
||||||
|
mp_group2 = optimizer2._mp_torch_group
|
||||||
|
assert dist.get_world_size(mp_group1) == dist.get_world_size(mp_group2)
|
||||||
|
assert dist.get_rank(mp_group1) == dist.get_rank(mp_group2)
|
||||||
|
|
||||||
|
|
||||||
|
def run_dist(rank, world_size, port):
|
||||||
|
config_dict = dict(parallel=dict(data=2, tensor=dict(size=2, mode='1d')))
|
||||||
|
colossalai.launch(config=config_dict, rank=rank, world_size=world_size, port=port, host='localhost')
|
||||||
|
exam_zero_init()
|
||||||
|
|
||||||
|
|
||||||
|
@pytest.mark.dist
|
||||||
|
def test_zero_init():
|
||||||
|
world_size = 4
|
||||||
|
run_func = partial(run_dist, world_size=world_size, port=free_port())
|
||||||
|
mp.spawn(run_func, nprocs=world_size)
|
||||||
|
|
||||||
|
|
||||||
|
if __name__ == '__main__':
|
||||||
|
test_zero_init()
|
Loading…
Reference in new issue