Skip to content
GitLab
Menu
Projects
Groups
Snippets
Loading...
Help
Help
Support
Community forum
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in / Register
Toggle navigation
Menu
Open sidebar
ModelZoo
ViT_pytorch
Commits
08b7113e
Commit
08b7113e
authored
Sep 22, 2023
by
sunxx1
Browse files
updata readme
parent
19ee37d6
Changes
4
Hide whitespace changes
Inline
Side-by-side
Showing
4 changed files
with
19 additions
and
5 deletions
+19
-5
README.md
README.md
+14
-2
image/README/1695380482110.png
image/README/1695380482110.png
+0
-0
image/README/1695381570003.png
image/README/1695381570003.png
+0
-0
model.properties
model.properties
+5
-3
No files found.
README.md
View file @
08b7113e
# Vi
sion Transformer(ViT)
# Vi
T
## 论文
...
...
@@ -61,6 +61,17 @@ cifar10
链接:https://pan.baidu.com/s/1ZFMQVBGQZI6UWZKJcTYPAQ?pwd=fq3l 提取码:fq3l
```
├── batches.meta
├── data_batch_1
├── data_batch_2
├── data_batch_3
├── data_batch_4
├── data_batch_5
├── readme.html
└── test_batch
```
## 训练
下载预训练模型放在checkpoint目录下:
...
...
@@ -78,7 +89,8 @@ wget https://storage.googleapis.com/vit_models/imagenet21k/ViT-B_16.npz
```
python3 -m torch.distributed.launch --nproc_per_node=8 train.py --name cifar10-100_500 --dataset cifar10 --model_type ViT-B_16 --pretrained_dir checkpoint/ViT-B_16.npz --train_batch_size 64 --num_steps 500
```
## result

## 精度
测试数据使用的是cifar10,使用的加速卡是DCU Z100L。
...
...
image/README/1695380482110.png
0 → 100644
View file @
08b7113e
5.31 KB
image/README/1695381570003.png
0 → 100644
View file @
08b7113e
258 KB
model.properties
View file @
08b7113e
# 模型唯一标识
modelCode
=
105
# 模型名称
modelName
=
ViT-PyT
orch
modelName
=
vit_pyt
orch
# 模型描述
modelDescription
=
ViT是一种新的基于Transformer应用于计算机视觉领域的神经网络模型,基于PyTorch实现测试
# 应用场景
appScenario
=
训练
,
图像分类,
目标检测
appScenario
=
训练
,
图像分类,
制造,交通,网安
# 框架类型
frameType
=
PyT
orch
frameType
=
pyt
orch
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
.
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment