# TinyViT

**Category:** Hybrid · **Code:**
[`timm/models/tiny_vit.py`](https://github.com/huggingface/pytorch-image-models/blob/main/timm/models/tiny_vit.py) ·
**Architectures:** 5 · **Pretrained weights:** 11 (8 with ImageNet-1k results)

TinyViT is a family of small hierarchical vision transformers (a convolutional first stage, then local window
attention), pretrained by fast distillation from large teacher models.

## Papers

* [TinyViT: Fast Pretraining Distillation for Small Vision Transformers](https://arxiv.org/abs/2207.10666)

## 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('tiny_vit_21m_512.dist_in22k_ft_in1k', pretrained=True)
>>> data_config = timm.data.resolve_model_data_config(model)
```

## Model API support

| Feature | `TinyVit` |
|---|---|
| [`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 (`TinyVit` with `tiny_vit_5m_224` 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 |
|---|---:|---:|---:|---:|
| `tiny_vit_5m_224` | 12.08 | 1.27 | 224 | 3 |
| `tiny_vit_11m_224` | 20.35 | 2.03 | 224 | 3 |
| `tiny_vit_21m_224` | 33.21 | 4.27 | 224 | 3 |
| `tiny_vit_21m_384` | 21.22 | 13.72 | 384 | 1 |
| `tiny_vit_21m_512` | 21.26 | 26.93 | 512 | 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 |
|---|---:|---:|---:|---:|---|---|
| `tiny_vit_5m_224.dist_in22k` | 224 | - | - | 21841 | apache-2.0 | [timm](https://huggingface.co/timm/tiny_vit_5m_224.dist_in22k) |
| `tiny_vit_5m_224.dist_in22k_ft_in1k` | 224 | 80.874 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tiny_vit_5m_224.dist_in22k_ft_in1k) |
| `tiny_vit_5m_224.in1k` | 224 | 79.196 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tiny_vit_5m_224.in1k) |
| `tiny_vit_11m_224.dist_in22k` | 224 | - | - | 21841 | apache-2.0 | [timm](https://huggingface.co/timm/tiny_vit_11m_224.dist_in22k) |
| `tiny_vit_11m_224.dist_in22k_ft_in1k` | 224 | 83.232 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tiny_vit_11m_224.dist_in22k_ft_in1k) |
| `tiny_vit_11m_224.in1k` | 224 | 81.538 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tiny_vit_11m_224.in1k) |
| `tiny_vit_21m_224.dist_in22k` | 224 | - | - | 21841 | apache-2.0 | [timm](https://huggingface.co/timm/tiny_vit_21m_224.dist_in22k) |
| `tiny_vit_21m_224.dist_in22k_ft_in1k` | 224 | 85.088 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tiny_vit_21m_224.dist_in22k_ft_in1k) |
| `tiny_vit_21m_224.in1k` | 224 | 83.268 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tiny_vit_21m_224.in1k) |
| `tiny_vit_21m_384.dist_in22k_ft_in1k` | 384 | 86.088 | 384 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tiny_vit_21m_384.dist_in22k_ft_in1k) |
| `tiny_vit_21m_512.dist_in22k_ft_in1k` | 512 | 86.444 | 512 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tiny_vit_21m_512.dist_in22k_ft_in1k) |