huggingface-vit-finetune
所属分类：语音合成
开发工具：Python
文件大小：4KB
下载次数：0
上传日期：2021-04-04 22:14:42
上 传 者：
sh-1993
说明： 从HuggingFace的模型中心对谷歌预先训练的ViT模型进行微调。
(Finetune Google s pre-trained ViT models from HuggingFace s model hub.)
文件列表:
img_model.py (3490, 2021-04-05)
requirements.txt (35, 2021-04-05)
run.py (2488, 2021-04-05)
# huggingface-vit-finetune
Huggingface does images now!
Well...they will soon. For now we gotta install `transformers` from master.
```
pip install -r requirements.txt
pip install git+https://github.com/huggingface/transformers.git@master --upgrade
python run.py
```
## Using trained models w/ `transformers`
Currently, the following models are available:
- nateraw/vit-base-patch16-224-cifar10
```python
from transformers import ViTFeatureExtractor, ViTForImageClassification
from PIL import Image
import requests
url = 'https://www.cs.toronto.edu/~kriz/cifar-10-sample/dog10.png'
image = Image.open(requests.get(url, stream=True).raw)
feature_extractor = ViTFeatureExtractor.from_pretrained('nateraw/vit-base-patch16-224-cifar10')
model = ViTForImageClassification.from_pretrained('nateraw/vit-base-patch16-224-cifar10')
inputs = feature_extractor(images=image, return_tensors="pt")
outputs = model(**inputs)
preds = outputs.logits.argmax(dim=1)
classes = [
'airplane', 'automobile', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse', 'ship', 'truck'
]
classes[preds[0]]
```
近期下载者：
相关文件：
收藏者：