# GCViT

**Category:** ViT · **Code:**
[`timm/models/gcvit.py`](https://github.com/huggingface/pytorch-image-models/blob/main/timm/models/gcvit.py) ·
**Architectures:** 5 · **Pretrained weights:** 5 (5 with ImageNet-1k results)

Global Context Vision Transformers (GCViT) alternate local window attention with global context attention, whose queries
are generated from the whole feature map by a convolutional module.

## Papers

* [Global Context Vision Transformers](https://arxiv.org/abs/2206.09959)

## Usage

See the [Quickstart](../quickstart) for loading pretrained weights, image preprocessing and fine-tuning, and [Feature
Extraction](../feature_extraction) for using the models as backbones.

```py
>>> import timm
>>> model = timm.create_model('gcvit_tiny.in1k', pretrained=True)
>>> data_config = timm.data.resolve_model_data_config(model)
```

## Model API support

| Feature | `GlobalContextVit` |
|---|---|
| [`features_only=True` feature maps](../model_api#intermediate-features) | ✓ (strides 4, 8, 16, 32) |
| [`forward_intermediates()`](../model_api#intermediate-features) | ✓ |
| [`prune_intermediate_layers()`](../model_api#intermediate-features) | ✓ |
| [Any input size, no changes needed](../model_api#input-size) | ✗ |
| [`set_input_size()`](../model_api#input-size) | ✗ |
| [`dynamic_img_size=True`](../model_api#input-size) | ✗ |
| [`set_grad_checkpointing()`](../model_api#training-hooks) | ✓ |
| [Layer-wise LR decay (`group_matcher()`)](../model_api#training-hooks) | ✓ |
| [`reset_classifier()`](../model_api#classifier) | ✓ |

✓ supported, ✓ (square) works for square sizes only, ✗ not supported, – not needed (the model already accepts any input
size). Checked by running each feature on a randomly initialized model (`GlobalContextVit` with `gcvit_xxtiny` at 224),
other architectures of the same class are expected to behave the same. See the [Model API](../model_api) page for
details.

## Architectures

| Architecture | Params (M) | GMACs | Input size | Pretrained weights |
|---|---:|---:|---:|---:|
| `gcvit_base` | 90.32 | 14.87 | 224 | 1 |
| `gcvit_small` | 51.09 | 8.57 | 224 | 1 |
| `gcvit_tiny` | 28.22 | 4.79 | 224 | 1 |
| `gcvit_xtiny` | 19.98 | 2.93 | 224 | 1 |
| `gcvit_xxtiny` | 12.00 | 2.14 | 224 | 1 |

GMACs and parameter counts are for the default input size and classifier of each architecture, from the [benchmark
results](https://github.com/huggingface/pytorch-image-models/tree/main/results).

## Pretrained weights

Top-1 is ImageNet-1k validation accuracy from
[`results-imagenet.csv`](https://github.com/huggingface/pytorch-image-models/blob/main/results/results-imagenet.csv) at
the listed eval size, - if the weight has no ImageNet-1k result (e.g. pretraining or image-text weights). Classes is the
size of the pretrained classifier, none for weights without one.

| Weight | Input size | Top-1 (%) | Eval size | Classes | License | Hub |
|---|---:|---:|---:|---:|---|---|
| `gcvit_base.in1k` | 224 | 84.456 | 224 | 1000 | apache-2.0 | [original](https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights-morevit/gcvit_base_224_nvidia-f009139b.pth) |
| `gcvit_small.in1k` | 224 | 83.904 | 224 | 1000 | apache-2.0 | [original](https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights-morevit/gcvit_small_224_nvidia-4e98afa2.pth) |
| `gcvit_tiny.in1k` | 224 | 83.400 | 224 | 1000 | apache-2.0 | [original](https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights-morevit/gcvit_tiny_224_nvidia-ac783954.pth) |
| `gcvit_xtiny.in1k` | 224 | 81.964 | 224 | 1000 | apache-2.0 | [original](https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights-morevit/gcvit_xtiny_224_nvidia-274b92b7.pth) |
| `gcvit_xxtiny.in1k` | 224 | 79.756 | 224 | 1000 | apache-2.0 | [original](https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights-morevit/gcvit_xxtiny_224_nvidia-d1d86009.pth) |

## Citation

```bibtex
@article{hatamizadeh2022global,
  title={Global Context Vision Transformers},
  author={Hatamizadeh, Ali and Yin, Hongxu and Kautz, Jan and Molchanov, Pavlo},
  journal={arXiv preprint arXiv:2206.09959},
  year={2022}
}
```