<!-- # hard line break macro for HTML -->

<a id="model-zoo-zero-shot-classification-transformer-torch"></a>

# zero-shot-classification-transformer-torch

Finds any object you name in images without requiring training on those specific items.

**Details**

- Model name: `zero-shot-classification-transformer-torch`
- Model source: [https://huggingface.co/docs/transformers/tasks/zero_shot_image_classification](https://huggingface.co/docs/transformers/tasks/zero_shot_image_classification)
- Model author: Thomas Wolf, et al.
- Model license: Apache 2.0
- Exposes embeddings? yes
- Tags: `text-embeddings, classification, logits, embeddings, torch, transformers, zero-shot, official`

**Requirements**

- Packages: `torch, torchvision, transformers`
- CPU support
  - yes
- GPU support
  - yes

**Example usage**

```python
import fiftyone as fo
import fiftyone.zoo as foz

dataset = foz.load_zoo_dataset(
    "coco-2017",
    split="validation",
    dataset_name=fo.get_default_dataset_name(),
    max_samples=50,
    shuffle=True,
)

classes = ["person", "dog", "cat", "bird", "car", "tree", "chair"]

model = foz.load_zoo_model(
    "zero-shot-classification-transformer-torch",
    classes=classes,
)

dataset.apply_model(model, label_field="predictions")

session = fo.launch_app(dataset)

# some models make require additional arguments
# check the Hugging Face docs to see if any are needed

# for example, AltCLIP requires `padding=True` in its processor
model = foz.load_zoo_model(
    "zero-shot-classification-transformer-torch",
    classes=classes,
    name_or_path="BAAI/AltCLIP",
    transformers_processor_kwargs={
        "padding": True,
    }
)

dataset.apply_model(model, label_field="predictions")

session = fo.launch_app(dataset)
```