mirror of
https://github.com/huggingface/candle.git
synced 2025-06-18 11:37:11 +00:00
21 lines
610 B
Markdown
21 lines
610 B
Markdown
# candle-vit
|
|
|
|
Vision Transformer (ViT) model implementation following the lines of
|
|
[vit-base-patch16-224](https://huggingface.co/google/vit-base-patch16-224)
|
|
This uses a classification head trained on the ImageNet dataset and returns the
|
|
probabilities for the top-5 classes.
|
|
|
|
## Running an example
|
|
|
|
```
|
|
$ cargo run --example vit --release -- --image tiger.jpg
|
|
|
|
loaded image Tensor[dims 3, 224, 224; f32]
|
|
model built
|
|
tiger, Panthera tigris : 100.00%
|
|
tiger cat : 0.00%
|
|
jaguar, panther, Panthera onca, Felis onca: 0.00%
|
|
leopard, Panthera pardus: 0.00%
|
|
lion, king of beasts, Panthera leo: 0.00%
|
|
```
|