# EfficientViM

**Category:** Efficient/Mobile · **Code:**
[`timm/models/efficientvim.py`](https://github.com/huggingface/pytorch-image-models/blob/main/timm/models/efficientvim.py)
· **Architectures:** 8 · **Pretrained weights:** 12 (0 with ImageNet-1k results)

Efficient Vision Mamba (EfficientViM) is built around the Hidden State Mixer based State Space Duality (HSM-SSD) layer,
implemented here in plain PyTorch. The `_dist` models add a distillation head.

## Papers

* [EfficientViM: Efficient Vision Mamba with Hidden State Mixer based State Space
  Duality](https://arxiv.org/abs/2411.15241)

## 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('efficientvim_m1.e300_in1k', pretrained=True)
>>> data_config = timm.data.resolve_model_data_config(model)
```

## Model API support

| Feature | `EfficientViM` | `EfficientViMDistilled` |
|---|---|---|
| [`features_only=True` feature maps](../model_api#intermediate-features) | ✓ (strides 16, 32, 64) | ✓ (strides 16, 32, 64) |
| [`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 (`EfficientViM` with `efficientvim_m1` at 224,
`EfficientViMDistilled` with `efficientvim_m1_dist` 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 | Class | Params (M) | GMACs | Input size | Pretrained weights |
|---|---|---:|---:|---:|---:|
| `efficientvim_m1` | `EfficientViM` | 6.68 | - | 224 | 2 |
| `efficientvim_m1_dist` | `EfficientViMDistilled` | 7.64 | - | 224 | 1 |
| `efficientvim_m2` | `EfficientViM` | 13.90 | - | 224 | 2 |
| `efficientvim_m2_dist` | `EfficientViMDistilled` | 15.32 | - | 224 | 1 |
| `efficientvim_m3` | `EfficientViM` | 16.60 | - | 224 | 2 |
| `efficientvim_m3_dist` | `EfficientViMDistilled` | 18.18 | - | 224 | 1 |
| `efficientvim_m4` | `EfficientViM` | 19.61 | - | 256 | 2 |
| `efficientvim_m4_dist` | `EfficientViMDistilled` | 21.18 | - | 256 | 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 |
|---|---:|---:|---:|---:|---|---|
| `efficientvim_m1.e300_in1k` | 224 | - | - | 1000 | mit | [timm](https://huggingface.co/timm/efficientvim_m1.e300_in1k) |
| `efficientvim_m1.e450_in1k` | 224 | - | - | 1000 | mit | [timm](https://huggingface.co/timm/efficientvim_m1.e450_in1k) |
| `efficientvim_m1_dist.in1k` | 224 | - | - | 1000 | mit | [timm](https://huggingface.co/timm/efficientvim_m1_dist.in1k) |
| `efficientvim_m2.e300_in1k` | 224 | - | - | 1000 | mit | [timm](https://huggingface.co/timm/efficientvim_m2.e300_in1k) |
| `efficientvim_m2.e450_in1k` | 224 | - | - | 1000 | mit | [timm](https://huggingface.co/timm/efficientvim_m2.e450_in1k) |
| `efficientvim_m2_dist.in1k` | 224 | - | - | 1000 | mit | [timm](https://huggingface.co/timm/efficientvim_m2_dist.in1k) |
| `efficientvim_m3.e300_in1k` | 224 | - | - | 1000 | mit | [timm](https://huggingface.co/timm/efficientvim_m3.e300_in1k) |
| `efficientvim_m3.e450_in1k` | 224 | - | - | 1000 | mit | [timm](https://huggingface.co/timm/efficientvim_m3.e450_in1k) |
| `efficientvim_m3_dist.in1k` | 224 | - | - | 1000 | mit | [timm](https://huggingface.co/timm/efficientvim_m3_dist.in1k) |
| `efficientvim_m4.e300_in1k` | 256 | - | - | 1000 | mit | [timm](https://huggingface.co/timm/efficientvim_m4.e300_in1k) |
| `efficientvim_m4.e450_in1k` | 256 | - | - | 1000 | mit | [timm](https://huggingface.co/timm/efficientvim_m4.e450_in1k) |
| `efficientvim_m4_dist.in1k` | 256 | - | - | 1000 | mit | [timm](https://huggingface.co/timm/efficientvim_m4_dist.in1k) |