You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
ColossalAI/tests/test_legacy/test_layers/test_3d/test_3d.py

65 lines
1.7 KiB

3 years ago
#!/usr/bin/env python
# -*- encoding: utf-8 -*-
import pytest
import torch
from checks_3d.check_layer_3d import (
check_classifier_no_given_weight,
check_embed,
check_layernorm,
check_linear,
check_loss,
check_patch_embed,
check_vocab_parallel_classifier_given_embed_weight,
check_vocab_parallel_classifier_no_given_weight,
check_vocab_parallel_embed,
check_vocab_parallel_loss,
)
from colossalai.legacy.core import global_context as gpc
from colossalai.legacy.initialize import launch
from colossalai.logging import disable_existing_loggers
from colossalai.testing import rerun_if_address_is_in_use, skip_if_not_enough_gpus, spawn
3 years ago
CONFIG = dict(
parallel=dict(
pipeline=1,
tensor=dict(mode="3d", size=8),
),
seed=42,
)
3 years ago
def check_layer():
check_linear()
check_layernorm()
check_classifier_no_given_weight()
check_vocab_parallel_classifier_no_given_weight()
check_vocab_parallel_classifier_given_embed_weight()
check_embed()
check_patch_embed()
check_vocab_parallel_embed()
check_loss()
check_vocab_parallel_loss()
3 years ago
def check_layer_and_operation(rank, world_size, port):
disable_existing_loggers()
launch(rank=rank, world_size=world_size, host="localhost", port=port, backend="nccl")
torch.backends.cuda.matmul.allow_tf32 = False
torch.backends.cudnn.allow_tf32 = False
torch.backends.cudnn.deterministic = True
3 years ago
check_layer()
gpc.destroy()
torch.cuda.empty_cache()
@pytest.mark.dist
@skip_if_not_enough_gpus(min_gpus=8)
@rerun_if_address_is_in_use()
def test_3d():
spawn(check_layer_and_operation, 8)
3 years ago
if __name__ == "__main__":
test_3d()