ColossalAI/tests/test_legacy/test_amp/test_naive_fp16.py

93 lines
3.2 KiB
Python

import copy
import pytest
import torch
import colossalai
from colossalai.legacy.amp import convert_to_apex_amp, convert_to_naive_amp
from colossalai.testing import assert_close_loose, clear_cache_before_run, rerun_if_address_is_in_use, spawn
from tests.kit.model_zoo import model_zoo
def check_equal(a, b):
"""
This function checks if two tensors are equal within tolerance
"""
assert torch.allclose(a.float(), b.float(), rtol=1e-4, atol=1e-3), f"a = {a}, b = {b}"
def run_naive_amp():
"""
In this test, we compare the naive fp16 optimizer implemented in colossalai
and fp32 torch optimizer
"""
torch.backends.cudnn.benchmark = False
torch.backends.cudnn.deterministic = True
# create layer
test_models = ["custom_repeated_computed_layers", "custom_nested_model", "torchvision_resnet18"]
for test_name in test_models:
model_builder, data_gen_fn, *_ = next(iter(model_zoo.get_sub_registry(test_name).values()))
# create model
naive_amp_model = model_builder().cuda()
apex_amp_model = copy.deepcopy(naive_amp_model)
# create optimizer
# we use SGD here, since the correctness of gradient clipping can't be tested with Adam
naive_amp_optimizer = torch.optim.SGD(naive_amp_model.parameters(), lr=1e-3)
apex_amp_optimizer = torch.optim.SGD(apex_amp_model.parameters(), lr=1e-3)
# inject naive and apex amp
naive_amp_config = dict(initial_scale=128, clip_grad_norm=1.0)
naive_amp_model, naive_amp_optimizer = convert_to_naive_amp(
naive_amp_model, naive_amp_optimizer, naive_amp_config
)
apex_amp_config = dict(opt_level="O2", loss_scale=128, keep_batchnorm_fp32=False)
apex_amp_model, apex_amp_optimizer = convert_to_apex_amp(apex_amp_model, apex_amp_optimizer, apex_amp_config)
# create data
data = data_gen_fn()
data = {k: v.cuda() if isinstance(v, torch.Tensor) else v for k, v in data.items()}
# forward pass
naive_amp_output = naive_amp_model(**data)
apex_amp_output = apex_amp_model(**data)
assert_close_loose(naive_amp_output, apex_amp_output)
# backward
# use sum() to get big gradient
naive_amp_optimizer.backward(naive_amp_output.sum())
apex_amp_optimizer.backward(apex_amp_output.sum())
# check grad
for naive_amp_param, apex_amp_param in zip(naive_amp_model.parameters(), apex_amp_model.parameters()):
assert_close_loose(naive_amp_param.grad, apex_amp_param.grad)
# clip gradient
apex_amp_optimizer.clip_grad_norm(model=apex_amp_model, max_norm=1.0)
# step
naive_amp_optimizer.step()
apex_amp_optimizer.step()
# check updated param
for naive_amp_param, apex_amp_param in zip(naive_amp_model.parameters(), apex_amp_model.parameters()):
assert_close_loose(naive_amp_param, apex_amp_param)
def run_dist(rank, world_size, port):
colossalai.legacy.launch(rank=rank, world_size=world_size, port=port, host="localhost")
run_naive_amp()
@pytest.mark.dist
@rerun_if_address_is_in_use()
@clear_cache_before_run()
def test_naive_amp():
spawn(run_dist, 1)
if __name__ == "__main__":
test_naive_amp()