-
Notifications
You must be signed in to change notification settings - Fork 1
/
pretrain_vit.py
91 lines (70 loc) · 2.68 KB
/
pretrain_vit.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
# coding=utf-8
# Copyright (c) 2020, NVIDIA CORPORATION. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Pretrain VIT"""
import torch
import torch.nn.functional as F
from megatron import get_args, get_timers, mpu, print_rank_0
from megatron.data.vit_dataset import build_train_valid_datasets
from megatron.model.vit_model import VitModel
from megatron.training import pretrain
from megatron.utils import average_losses_across_data_parallel_group
def model_provider():
"""Build the model."""
print_rank_0("building VIT model ...")
args = get_args()
model = VitModel(num_classes=args.num_classes)
return model
def get_batch(data_iterator):
"""Build the batch."""
data = next(data_iterator)
# only data parallelism; no need for broadcast
images = data[0].to(get_accelerator().device_name())
labels = data[1].to(get_accelerator().device_name())
return images, labels
def forward_step(data_iterator, model, input_tensor):
"""Forward step."""
timers = get_timers()
assert input_tensor is None
# Get the batch.
timers("batch-generator").start()
(
images,
labels,
) = get_batch(data_iterator)
timers("batch-generator").stop()
# Forward model. lm_labels
logits = model(images).contiguous().float()
loss = F.cross_entropy(logits, labels)
outputs = torch.argmax(logits, -1)
correct = (outputs == labels).float()
accuracy = torch.mean(correct)
averaged_loss = average_losses_across_data_parallel_group([loss, accuracy])
return loss, {"loss": averaged_loss[0], "accuracy": averaged_loss[1]}
def train_valid_test_datasets_provider(train_val_test_num_samples):
"""Build train, valid, and test datasets."""
args = get_args()
print_rank_0(
"> building train, validation, and test datasets " "for VIT ..."
)
train_ds, valid_ds = build_train_valid_datasets(data_path=args.data_path)
print_rank_0("> finished creating VIT datasets ...")
return train_ds, valid_ds, None
if __name__ == "__main__":
pretrain(
train_valid_test_datasets_provider,
model_provider,
forward_step,
args_defaults={'dataloader_type': 'cyclic'}
)