# DeepSeek ViT

**Category:** Encoder · **Code:**
[`timm/models/deepseek_vit.py`](https://github.com/huggingface/pytorch-image-models/blob/main/timm/models/deepseek_vit.py)
· **Architectures:** 3 · **Pretrained weights:** 6 (0 with ImageNet-1k results)

The vision encoder of the DeepSeek-V4 / V4.1 multimodal models, a pre-norm ViT with RMSNorm, a SwiGLU MLP and axial 2D
RoPE as the only position encoding. Classifier variants add pooling and a linear head, the `_enc` variants keep the
native projector that maps the tokens to the language model width.

## Papers

* [DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression](https://arxiv.org/abs/2609.19969)

## 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('deepseek_vit_412m.deepseek_v4_1_flash', pretrained=True)
>>> data_config = timm.data.resolve_model_data_config(model)
```

## Model API support

| Feature | `DeepseekVitClassifier` | `DeepseekVitEncoder` |
|---|---|---|
| [`features_only=True` feature maps](../model_api#intermediate-features) | ✓ (strides 14, 14, 14) | ✓ (strides 14, 14, 14) |
| [`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 (`DeepseekVitClassifier` with `deepseek_vit_412m`
at 252, `DeepseekVitEncoder` with `deepseek_vit_412m_enc` at 252), other architectures of the same class are expected to
behave the same. See the [Model API](../model_api) page for details.

## Architectures

| Architecture | Class | Params (M) | GMACs | Input size | Pretrained weights |
|---|---|---:|---:|---:|---:|
| `deepseek_vit_412m` | `DeepseekVitClassifier` | 411.84 | - | 546 | 2 |
| `deepseek_vit_412m_align` | `DeepseekVitClassifier` | 485.25 | - | 546 | 2 |
| `deepseek_vit_412m_enc` | `DeepseekVitEncoder` | 485.25 | - | 546 | 2 |

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.

Weights are grouped by source: the paper or origin given in the pretrained config, else a Hub organization other than
`timm`, else the first part of the pretrained tag (`architecture.tag`). Sources with a single weight are listed under
other.

### deepseek-ai/DeepSeek-V4-Flash-Vision-Exp

Origin:
[huggingface.co/deepseek-ai/DeepSeek-V4-Flash-Vision-Exp](https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-Vision-Exp)

| Weight | Input size | Top-1 (%) | Eval size | Classes | License | Hub |
|---|---:|---:|---:|---:|---|---|
| `deepseek_vit_412m.deepseek_v4_flash_vision_exp` | 392 | - | - | none | mit | [timm](https://huggingface.co/timm/deepseek_vit_412m.deepseek_v4_flash_vision_exp) |
| `deepseek_vit_412m_align.deepseek_v4_flash_vision_exp` | 392 | - | - | none | mit | [timm](https://huggingface.co/timm/deepseek_vit_412m_align.deepseek_v4_flash_vision_exp) |
| `deepseek_vit_412m_enc.deepseek_v4_flash_vision_exp` | 392 | - | - | none | mit | [timm](https://huggingface.co/timm/deepseek_vit_412m_enc.deepseek_v4_flash_vision_exp) |

### deepseek-ai/DeepSeek-V4.1-Flash

Origin: [huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash](https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash)

| Weight | Input size | Top-1 (%) | Eval size | Classes | License | Hub |
|---|---:|---:|---:|---:|---|---|
| `deepseek_vit_412m.deepseek_v4_1_flash` | 546 | - | - | none | mit | [timm](https://huggingface.co/timm/deepseek_vit_412m.deepseek_v4_1_flash) |
| `deepseek_vit_412m_align.deepseek_v4_1_flash` | 546 | - | - | none | mit | [timm](https://huggingface.co/timm/deepseek_vit_412m_align.deepseek_v4_1_flash) |
| `deepseek_vit_412m_enc.deepseek_v4_1_flash` | 546 | - | - | none | mit | [timm](https://huggingface.co/timm/deepseek_vit_412m_enc.deepseek_v4_1_flash) |