# Next-ViT

**Category:** Hybrid · **Code:**
[`timm/models/nextvit.py`](https://github.com/huggingface/pytorch-image-models/blob/main/timm/models/nextvit.py) ·
**Architectures:** 3 · **Pretrained weights:** 12 (12 with ImageNet-1k results)

Next-ViT is a hybrid CNN / transformer for efficient deployment, stacking Next Convolution Blocks (NCB) and Next
Transformer Blocks (NTB).

## Papers

* [Next-ViT: Next Generation Vision Transformer for Efficient Deployment in Realistic Industrial
  Scenarios](https://arxiv.org/abs/2207.05501)

## 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('nextvit_small.bd_in1k', pretrained=True)
>>> data_config = timm.data.resolve_model_data_config(model)
```

## Model API support

| Feature | `NextViT` |
|---|---|
| [`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 (`NextViT` with `nextvit_small` 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 |
|---|---:|---:|---:|---:|
| `nextvit_base` | 44.79 | 8.29 | 224 | 4 |
| `nextvit_large` | 57.83 | 10.77 | 224 | 4 |
| `nextvit_small` | 31.74 | 5.80 | 224 | 4 |

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 |
|---|---:|---:|---:|---:|---|---|
| `nextvit_base.bd_in1k` | 224 | 83.498 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/nextvit_base.bd_in1k) |
| `nextvit_base.bd_in1k_384` | 384 | 84.706 | 384 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/nextvit_base.bd_in1k_384) |
| `nextvit_base.bd_ssld_6m_in1k` | 224 | 85.186 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/nextvit_base.bd_ssld_6m_in1k) |
| `nextvit_base.bd_ssld_6m_in1k_384` | 384 | 86.372 | 384 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/nextvit_base.bd_ssld_6m_in1k_384) |
| `nextvit_large.bd_in1k` | 224 | 83.654 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/nextvit_large.bd_in1k) |
| `nextvit_large.bd_in1k_384` | 384 | 84.930 | 384 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/nextvit_large.bd_in1k_384) |
| `nextvit_large.bd_ssld_6m_in1k` | 224 | 85.514 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/nextvit_large.bd_ssld_6m_in1k) |
| `nextvit_large.bd_ssld_6m_in1k_384` | 384 | 86.544 | 384 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/nextvit_large.bd_ssld_6m_in1k_384) |
| `nextvit_small.bd_in1k` | 224 | 82.626 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/nextvit_small.bd_in1k) |
| `nextvit_small.bd_in1k_384` | 384 | 84.036 | 384 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/nextvit_small.bd_in1k_384) |
| `nextvit_small.bd_ssld_6m_in1k` | 224 | 84.872 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/nextvit_small.bd_ssld_6m_in1k) |
| `nextvit_small.bd_ssld_6m_in1k_384` | 384 | 85.966 | 384 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/nextvit_small.bd_ssld_6m_in1k_384) |