# EfficientNet

**Category:** Efficient/Mobile · **Code:**
[`timm/models/efficientnet.py`](https://github.com/huggingface/pytorch-image-models/blob/main/timm/models/efficientnet.py)
· **Architectures:** 115 · **Pretrained weights:** 119 (113 with ImageNet-1k results)

EfficientNet and EfficientNetV2, and the related mobile architectures built from the same depthwise separable, inverted
residual (MBConv) and fused MBConv blocks: MixNet, MnasNet, FBNet, Single-Path NAS, MobileNet-V1 / V2,
EfficientNet-Lite, EdgeTPU and TinyNet.

## Papers

* [EfficientNetV2: Smaller Models and Faster Training](https://arxiv.org/abs/2104.00298)
* [EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks](https://arxiv.org/abs/1905.11946)
* [CondConv: Conditionally Parameterized Convolutions for Efficient Inference](https://arxiv.org/abs/1904.04971)
* [Adversarial Examples Improve Image Recognition](https://arxiv.org/abs/1911.09665)
* [Self-training with Noisy Student improves ImageNet classification](https://arxiv.org/abs/1911.04252)
* [Knapsack Pruning with Inner Distillation](https://arxiv.org/abs/2002.08258)
* [Evolving Normalization-Activation Layers](https://arxiv.org/abs/2004.02967)
* [Searching for Fast Model Families on Datacenter Accelerators](https://arxiv.org/abs/2102.05610)
* [MixConv: Mixed Depthwise Convolutional Kernels](https://arxiv.org/abs/1907.09595)
* [MnasNet: Platform-Aware Neural Architecture Search for Mobile](https://arxiv.org/abs/1807.11626)
* [MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications](https://arxiv.org/abs/1704.04861)
* [MobileNetV2: Inverted Residuals and Linear Bottlenecks](https://arxiv.org/abs/1801.04381)
* [Searching for Efficient Neural Architectures for On-Device ML on Edge TPUs](https://arxiv.org/abs/2204.14007)
* [FBNet: Hardware-Aware Efficient ConvNet Design via Differentiable Neural Architecture
  Search](https://arxiv.org/abs/1812.03443)
* [Single-Path NAS: Designing Hardware-Efficient ConvNets in less than 4 Hours](https://arxiv.org/abs/1904.02877)
* [Model Rubik's Cube: Twisting Resolution, Depth and Width for TinyNets](https://arxiv.org/abs/2010.14819)

## 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('efficientnet_b0.ra_in1k', pretrained=True)
>>> data_config = timm.data.resolve_model_data_config(model)
```

## Model API support

| Feature | `EfficientNet` |
|---|---|
| [`features_only=True` feature maps](../model_api#intermediate-features) | ✓ (strides 2, 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 (`EfficientNet` with `mobilenetv2_035` 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 |
|---|---:|---:|---:|---:|
| `efficientnet_b0` | 5.29 | 0.40 | 224 | 2 |
| `efficientnet_b0_g8_gn` | 6.56 | 0.66 | 224 | 0 |
| `efficientnet_b0_g16_evos` | 8.11 | 1.01 | 224 | 0 |
| `efficientnet_b0_gn` | 5.29 | 0.42 | 224 | 0 |
| `efficientnet_b1` | 7.79 | 0.71 | 240 | 2 |
| `efficientnet_b1_pruned` | 6.33 | 0.40 | 240 | 1 |
| `efficientnet_b2` | 9.11 | 0.89 | 256 | 1 |
| `efficientnet_b2_pruned` | 8.31 | 0.73 | 260 | 1 |
| `efficientnet_b3` | 12.23 | 1.63 | 288 | 1 |
| `efficientnet_b3_g8_gn` | 14.25 | 2.59 | 288 | 0 |
| `efficientnet_b3_gn` | 11.73 | 1.74 | 288 | 0 |
| `efficientnet_b3_pruned` | 9.86 | 1.04 | 300 | 1 |
| `efficientnet_b4` | 19.34 | 3.13 | 320 | 1 |
| `efficientnet_b5` | 30.39 | 9.59 | 448 | 2 |
| `efficientnet_b6` | 43.04 | 19.40 | 528 | 0 |
| `efficientnet_b7` | 66.35 | 38.33 | 600 | 0 |
| `efficientnet_b8` | 87.41 | 63.48 | 672 | 0 |
| `efficientnet_blur_b0` | 5.29 | 0.43 | 224 | 0 |
| `efficientnet_cc_b0_4e` | 13.31 | 0.41 | 224 | 0 |
| `efficientnet_cc_b0_8e` | 24.01 | 0.42 | 224 | 0 |
| `efficientnet_cc_b1_8e` | 39.72 | 0.75 | 240 | 0 |
| `efficientnet_el` | 10.59 | 8.00 | 300 | 1 |
| `efficientnet_el_pruned` | 10.59 | 8.00 | 300 | 1 |
| `efficientnet_em` | 6.90 | 3.04 | 240 | 1 |
| `efficientnet_es` | 5.44 | 1.81 | 224 | 1 |
| `efficientnet_es_pruned` | 5.44 | 1.81 | 224 | 1 |
| `efficientnet_h_b5` | 45.88 | 27.16 | 448 | 1 |
| `efficientnet_l2` | 480.31 | 479.12 | 800 | 0 |
| `efficientnet_lite0` | 4.65 | 0.40 | 224 | 1 |
| `efficientnet_lite1` | 5.42 | 0.62 | 240 | 0 |
| `efficientnet_lite2` | 6.09 | 0.89 | 260 | 0 |
| `efficientnet_lite3` | 8.20 | 1.65 | 300 | 0 |
| `efficientnet_lite4` | 13.01 | 4.04 | 380 | 0 |
| `efficientnet_x_b3` | 13.30 | 3.91 | 288 | 0 |
| `efficientnet_x_b5` | 33.44 | 23.35 | 448 | 1 |
| `efficientnetv2_l` | 118.52 | 36.10 | 384 | 0 |
| `efficientnetv2_m` | 54.14 | 11.01 | 320 | 0 |
| `efficientnetv2_rw_m` | 53.24 | 12.72 | 320 | 1 |
| `efficientnetv2_rw_s` | 23.94 | 4.91 | 288 | 1 |
| `efficientnetv2_rw_t` | 13.65 | 1.93 | 224 | 1 |
| `efficientnetv2_s` | 21.46 | 4.75 | 288 | 0 |
| `efficientnetv2_xl` | 208.12 | 52.81 | 384 | 0 |
| `fbnetc_100` | 5.57 | 0.40 | 224 | 1 |
| `gc_efficientnetv2_rw_t` | 13.68 | 1.94 | 224 | 1 |
| `mixnet_l` | 7.33 | 0.58 | 224 | 1 |
| `mixnet_m` | 5.01 | 0.36 | 224 | 1 |
| `mixnet_s` | 4.13 | 0.25 | 224 | 1 |
| `mixnet_xl` | 11.90 | 0.93 | 224 | 1 |
| `mixnet_xxl` | 23.96 | 2.04 | 224 | 0 |
| `mnasnet_050` | 2.22 | 0.11 | 224 | 0 |
| `mnasnet_075` | 3.17 | 0.23 | 224 | 0 |
| `mnasnet_100` | 4.38 | 0.33 | 224 | 1 |
| `mnasnet_140` | 7.12 | 0.60 | 224 | 0 |
| `mnasnet_small` | 2.03 | 0.07 | 224 | 1 |
| `mobilenet_edgetpu_100` | 4.09 | 1.00 | 224 | 0 |
| `mobilenet_edgetpu_v2_l` | 10.92 | 2.55 | 224 | 0 |
| `mobilenet_edgetpu_v2_m` | 8.46 | 1.85 | 224 | 1 |
| `mobilenet_edgetpu_v2_s` | 5.99 | 1.21 | 224 | 0 |
| `mobilenet_edgetpu_v2_xs` | 4.46 | 0.70 | 224 | 0 |
| `mobilenetv1_100` | 4.23 | 0.58 | 224 | 1 |
| `mobilenetv1_100h` | 5.28 | 0.63 | 224 | 1 |
| `mobilenetv1_125` | 6.27 | 0.89 | 224 | 1 |
| `mobilenetv2_035` | 1.68 | 0.07 | 224 | 0 |
| `mobilenetv2_050` | 1.97 | 0.10 | 224 | 1 |
| `mobilenetv2_075` | 2.64 | 0.22 | 224 | 0 |
| `mobilenetv2_100` | 3.50 | 0.31 | 224 | 1 |
| `mobilenetv2_110d` | 4.52 | 0.45 | 224 | 1 |
| `mobilenetv2_120d` | 5.83 | 0.69 | 224 | 1 |
| `mobilenetv2_140` | 6.11 | 0.60 | 224 | 1 |
| `semnasnet_050` | 2.08 | 0.11 | 224 | 0 |
| `semnasnet_075` | 2.91 | 0.23 | 224 | 1 |
| `semnasnet_100` | 3.89 | 0.32 | 224 | 1 |
| `semnasnet_140` | 6.11 | 0.60 | 224 | 0 |
| `spnasnet_100` | 4.42 | 0.35 | 224 | 1 |
| `test_efficientnet` | 0.36 | 0.06 | 160 | 1 |
| `test_efficientnet_evos` | 0.36 | 0.06 | 160 | 1 |
| `test_efficientnet_gn` | 0.36 | 0.06 | 160 | 1 |
| `test_efficientnet_ln` | 0.36 | 0.06 | 160 | 1 |
| `tf_efficientnet_b0` | 5.29 | 0.40 | 224 | 4 |
| `tf_efficientnet_b1` | 7.79 | 0.71 | 240 | 4 |
| `tf_efficientnet_b2` | 9.11 | 1.02 | 260 | 4 |
| `tf_efficientnet_b3` | 12.23 | 1.87 | 300 | 4 |
| `tf_efficientnet_b4` | 19.34 | 4.49 | 380 | 4 |
| `tf_efficientnet_b5` | 30.39 | 10.46 | 456 | 5 |
| `tf_efficientnet_b6` | 43.04 | 19.40 | 528 | 3 |
| `tf_efficientnet_b7` | 66.35 | 38.33 | 600 | 4 |
| `tf_efficientnet_b8` | 87.41 | 63.48 | 672 | 2 |
| `tf_efficientnet_cc_b0_4e` | 13.31 | 0.41 | 224 | 1 |
| `tf_efficientnet_cc_b0_8e` | 24.01 | 0.42 | 224 | 1 |
| `tf_efficientnet_cc_b1_8e` | 39.72 | 0.75 | 240 | 1 |
| `tf_efficientnet_el` | 10.59 | 8.00 | 300 | 1 |
| `tf_efficientnet_em` | 6.90 | 3.04 | 240 | 1 |
| `tf_efficientnet_es` | 5.44 | 1.81 | 224 | 1 |
| `tf_efficientnet_l2` | 480.31 | 172.11 | 475 | 2 |
| `tf_efficientnet_lite0` | 4.65 | 0.40 | 224 | 1 |
| `tf_efficientnet_lite1` | 5.42 | 0.62 | 240 | 1 |
| `tf_efficientnet_lite2` | 6.09 | 0.89 | 260 | 1 |
| `tf_efficientnet_lite3` | 8.20 | 1.65 | 300 | 1 |
| `tf_efficientnet_lite4` | 13.01 | 4.04 | 380 | 1 |
| `tf_efficientnetv2_b0` | 7.14 | 0.54 | 192 | 1 |
| `tf_efficientnetv2_b1` | 8.14 | 0.76 | 192 | 1 |
| `tf_efficientnetv2_b2` | 10.10 | 1.06 | 208 | 1 |
| `tf_efficientnetv2_b3` | 14.36 | 1.93 | 240 | 3 |
| `tf_efficientnetv2_l` | 118.52 | 36.10 | 384 | 3 |
| `tf_efficientnetv2_m` | 54.14 | 15.85 | 384 | 3 |
| `tf_efficientnetv2_s` | 21.46 | 5.35 | 300 | 3 |
| `tf_efficientnetv2_xl` | 208.12 | 52.81 | 384 | 2 |
| `tf_mixnet_l` | 7.33 | 0.58 | 224 | 1 |
| `tf_mixnet_m` | 5.01 | 0.36 | 224 | 1 |
| `tf_mixnet_s` | 4.13 | 0.25 | 224 | 1 |
| `tinynet_a` | 6.19 | 0.35 | 192 | 1 |
| `tinynet_b` | 3.73 | 0.21 | 188 | 1 |
| `tinynet_c` | 2.46 | 0.11 | 184 | 1 |
| `tinynet_d` | 2.34 | 0.05 | 152 | 1 |
| `tinynet_e` | 2.04 | 0.03 | 106 | 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.

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.

### `ra` tags

| Weight | Input size | Top-1 (%) | Eval size | Classes | License | Hub |
|---|---:|---:|---:|---:|---|---|
| `efficientnet_b0.ra_in1k` | 224 | 77.698 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/efficientnet_b0.ra_in1k) |
| `efficientnet_b2.ra_in1k` | 256 (test 288) | 80.612 | 288 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/efficientnet_b2.ra_in1k) |
| `efficientnet_el.ra_in1k` | 300 | 81.308 | 300 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/efficientnet_el.ra_in1k) |
| `efficientnet_es.ra_in1k` | 224 | 78.100 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/efficientnet_es.ra_in1k) |
| `efficientnet_lite0.ra_in1k` | 224 | 75.488 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/efficientnet_lite0.ra_in1k) |
| `mixnet_xl.ra_in1k` | 224 | 80.494 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/mixnet_xl.ra_in1k) |
| `mobilenetv2_100.ra_in1k` | 224 | 72.910 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/mobilenetv2_100.ra_in1k) |
| `mobilenetv2_110d.ra_in1k` | 224 | 75.056 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/mobilenetv2_110d.ra_in1k) |
| `mobilenetv2_120d.ra_in1k` | 224 | 77.300 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/mobilenetv2_120d.ra_in1k) |
| `mobilenetv2_140.ra_in1k` | 224 | 76.528 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/mobilenetv2_140.ra_in1k) |
| `tf_efficientnet_b5.ra_in1k` | 456 | 83.834 | 456 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b5.ra_in1k) |
| `tf_efficientnet_b7.ra_in1k` | 600 | 84.926 | 600 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b7.ra_in1k) |
| `tf_efficientnet_b8.ra_in1k` | 672 | 85.344 | 672 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b8.ra_in1k) |

### `ns` tags

| Weight | Input size | Top-1 (%) | Eval size | Classes | License | Hub |
|---|---:|---:|---:|---:|---|---|
| `tf_efficientnet_b0.ns_jft_in1k` | 224 | 78.680 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b0.ns_jft_in1k) |
| `tf_efficientnet_b1.ns_jft_in1k` | 240 | 81.388 | 240 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b1.ns_jft_in1k) |
| `tf_efficientnet_b2.ns_jft_in1k` | 260 | 82.412 | 260 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b2.ns_jft_in1k) |
| `tf_efficientnet_b3.ns_jft_in1k` | 300 | 84.054 | 300 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b3.ns_jft_in1k) |
| `tf_efficientnet_b4.ns_jft_in1k` | 380 | 85.156 | 380 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b4.ns_jft_in1k) |
| `tf_efficientnet_b5.ns_jft_in1k` | 456 | 86.096 | 456 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b5.ns_jft_in1k) |
| `tf_efficientnet_b6.ns_jft_in1k` | 528 | 86.476 | 528 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b6.ns_jft_in1k) |
| `tf_efficientnet_b7.ns_jft_in1k` | 600 | 86.850 | 600 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b7.ns_jft_in1k) |
| `tf_efficientnet_l2.ns_jft_in1k` | 800 | 88.360 | 800 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_l2.ns_jft_in1k) |
| `tf_efficientnet_l2.ns_jft_in1k_475` | 475 | 88.246 | 475 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_l2.ns_jft_in1k_475) |

### `ap` tags

| Weight | Input size | Top-1 (%) | Eval size | Classes | License | Hub |
|---|---:|---:|---:|---:|---|---|
| `tf_efficientnet_b0.ap_in1k` | 224 | 77.100 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b0.ap_in1k) |
| `tf_efficientnet_b1.ap_in1k` | 240 | 79.288 | 240 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b1.ap_in1k) |
| `tf_efficientnet_b2.ap_in1k` | 260 | 80.300 | 260 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b2.ap_in1k) |
| `tf_efficientnet_b3.ap_in1k` | 300 | 81.842 | 300 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b3.ap_in1k) |
| `tf_efficientnet_b4.ap_in1k` | 380 | 83.246 | 380 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b4.ap_in1k) |
| `tf_efficientnet_b5.ap_in1k` | 456 | 84.258 | 456 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b5.ap_in1k) |
| `tf_efficientnet_b6.ap_in1k` | 528 | 84.794 | 528 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b6.ap_in1k) |
| `tf_efficientnet_b7.ap_in1k` | 600 | 85.132 | 600 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b7.ap_in1k) |
| `tf_efficientnet_b8.ap_in1k` | 672 | 85.358 | 672 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b8.ap_in1k) |

### `aa` tags

| Weight | Input size | Top-1 (%) | Eval size | Classes | License | Hub |
|---|---:|---:|---:|---:|---|---|
| `tf_efficientnet_b0.aa_in1k` | 224 | 76.882 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b0.aa_in1k) |
| `tf_efficientnet_b1.aa_in1k` | 240 | 78.840 | 240 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b1.aa_in1k) |
| `tf_efficientnet_b2.aa_in1k` | 260 | 80.074 | 260 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b2.aa_in1k) |
| `tf_efficientnet_b3.aa_in1k` | 300 | 81.648 | 300 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b3.aa_in1k) |
| `tf_efficientnet_b4.aa_in1k` | 380 | 83.040 | 380 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b4.aa_in1k) |
| `tf_efficientnet_b5.aa_in1k` | 456 | 83.696 | 456 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b5.aa_in1k) |
| `tf_efficientnet_b6.aa_in1k` | 528 | 84.124 | 528 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b6.aa_in1k) |
| `tf_efficientnet_b7.aa_in1k` | 600 | 84.424 | 600 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b7.aa_in1k) |

### `ra4` tags

| Weight | Input size | Top-1 (%) | Eval size | Classes | License | Hub |
|---|---:|---:|---:|---:|---|---|
| `efficientnet_b0.ra4_e3600_r224_in1k` | 224 (test 256) | 79.364 | 256 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/efficientnet_b0.ra4_e3600_r224_in1k) |
| `efficientnet_b1.ra4_e3600_r240_in1k` | 240 (test 288) | 81.444 | 288 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/efficientnet_b1.ra4_e3600_r240_in1k) |
| `mobilenet_edgetpu_v2_m.ra4_e3600_r224_in1k` | 224 (test 256) | 80.684 | 256 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/mobilenet_edgetpu_v2_m.ra4_e3600_r224_in1k) |
| `mobilenetv1_100.ra4_e3600_r224_in1k` | 224 (test 256) | 76.088 | 256 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/mobilenetv1_100.ra4_e3600_r224_in1k) |
| `mobilenetv1_100h.ra4_e3600_r224_in1k` | 224 (test 256) | 76.598 | 256 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/mobilenetv1_100h.ra4_e3600_r224_in1k) |
| `mobilenetv1_125.ra4_e3600_r224_in1k` | 224 (test 256) | 77.612 | 256 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/mobilenetv1_125.ra4_e3600_r224_in1k) |

### `ra2` tags

| Weight | Input size | Top-1 (%) | Eval size | Classes | License | Hub |
|---|---:|---:|---:|---:|---|---|
| `efficientnet_b3.ra2_in1k` | 288 (test 320) | 82.296 | 320 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/efficientnet_b3.ra2_in1k) |
| `efficientnet_b4.ra2_in1k` | 320 (test 384) | 83.468 | 384 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/efficientnet_b4.ra2_in1k) |
| `efficientnet_em.ra2_in1k` | 240 | 79.256 | 240 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/efficientnet_em.ra2_in1k) |
| `efficientnetv2_rw_s.ra2_in1k` | 288 (test 384) | 83.830 | 384 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/efficientnetv2_rw_s.ra2_in1k) |
| `efficientnetv2_rw_t.ra2_in1k` | 224 (test 288) | 82.354 | 288 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/efficientnetv2_rw_t.ra2_in1k) |

### `rmsp` tags

| Weight | Input size | Top-1 (%) | Eval size | Classes | License | Hub |
|---|---:|---:|---:|---:|---|---|
| `fbnetc_100.rmsp_in1k` | 224 | 75.120 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/fbnetc_100.rmsp_in1k) |
| `mnasnet_100.rmsp_in1k` | 224 | 74.672 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/mnasnet_100.rmsp_in1k) |
| `semnasnet_075.rmsp_in1k` | 224 | 73.000 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/semnasnet_075.rmsp_in1k) |
| `semnasnet_100.rmsp_in1k` | 224 | 75.452 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/semnasnet_100.rmsp_in1k) |
| `spnasnet_100.rmsp_in1k` | 224 | 74.086 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/spnasnet_100.rmsp_in1k) |

### `sw` tags

| Weight | Input size | Top-1 (%) | Eval size | Classes | License | Hub |
|---|---:|---:|---:|---:|---|---|
| `efficientnet_b5.sw_in12k` | 416 | - | - | 11821 | apache-2.0 | [timm](https://huggingface.co/timm/efficientnet_b5.sw_in12k) |
| `efficientnet_b5.sw_in12k_ft_in1k` | 448 | 85.894 | 448 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/efficientnet_b5.sw_in12k_ft_in1k) |
| `efficientnet_h_b5.sw_r448_e450_in1k` | 448 (test 576) | 84.942 | 576 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/efficientnet_h_b5.sw_r448_e450_in1k) |
| `efficientnet_x_b5.sw_r448_e450_in1k` | 448 (test 576) | 84.928 | 576 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/efficientnet_x_b5.sw_r448_e450_in1k) |

### `agc` tags

| Weight | Input size | Top-1 (%) | Eval size | Classes | License | Hub |
|---|---:|---:|---:|---:|---|---|
| `efficientnetv2_rw_m.agc_in1k` | 320 (test 416) | 84.830 | 416 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/efficientnetv2_rw_m.agc_in1k) |
| `gc_efficientnetv2_rw_t.agc_in1k` | 224 (test 288) | 82.486 | 288 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/gc_efficientnetv2_rw_t.agc_in1k) |

### `lamb` tags

| Weight | Input size | Top-1 (%) | Eval size | Classes | License | Hub |
|---|---:|---:|---:|---:|---|---|
| `mnasnet_small.lamb_in1k` | 224 | 66.204 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/mnasnet_small.lamb_in1k) |
| `mobilenetv2_050.lamb_in1k` | 224 | 65.928 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/mobilenetv2_050.lamb_in1k) |

### Other

| Weight | Input size | Top-1 (%) | Eval size | Classes | License | Hub |
|---|---:|---:|---:|---:|---|---|
| `efficientnet_b1.ft_in1k` | 224 (test 256) | 78.806 | 256 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/efficientnet_b1.ft_in1k) |
| `efficientnet_b1_pruned.in1k` | 240 | 78.252 | 240 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/efficientnet_b1_pruned.in1k) |
| `efficientnet_b2_pruned.in1k` | 260 | 79.904 | 260 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/efficientnet_b2_pruned.in1k) |
| `efficientnet_b3_pruned.in1k` | 300 | 80.864 | 300 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/efficientnet_b3_pruned.in1k) |
| `efficientnet_el_pruned.in1k` | 300 | 80.284 | 300 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/efficientnet_el_pruned.in1k) |
| `efficientnet_es_pruned.in1k` | 224 | 75.010 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/efficientnet_es_pruned.in1k) |
| `mixnet_l.ft_in1k` | 224 | 78.970 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/mixnet_l.ft_in1k) |
| `mixnet_m.ft_in1k` | 224 | 77.268 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/mixnet_m.ft_in1k) |
| `mixnet_s.ft_in1k` | 224 | 76.000 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/mixnet_s.ft_in1k) |
| `test_efficientnet.r160_in1k` | 160 | 46.506 | 160 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/test_efficientnet.r160_in1k) |
| `test_efficientnet_evos.r160_in1k` | 160 | 46.504 | 160 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/test_efficientnet_evos.r160_in1k) |
| `test_efficientnet_gn.r160_in1k` | 160 | 43.888 | 160 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/test_efficientnet_gn.r160_in1k) |
| `test_efficientnet_ln.r160_in1k` | 160 | 43.958 | 160 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/test_efficientnet_ln.r160_in1k) |
| `tf_efficientnet_b0.in1k` | 224 | 76.542 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b0.in1k) |
| `tf_efficientnet_b1.in1k` | 240 | 78.546 | 240 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b1.in1k) |
| `tf_efficientnet_b2.in1k` | 260 | 79.612 | 260 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b2.in1k) |
| `tf_efficientnet_b3.in1k` | 300 | 80.878 | 300 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b3.in1k) |
| `tf_efficientnet_b4.in1k` | 380 | 82.598 | 380 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b4.in1k) |
| `tf_efficientnet_b5.in1k` | 456 | 83.170 | 456 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_b5.in1k) |
| `tf_efficientnet_cc_b0_4e.in1k` | 224 | 77.324 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_cc_b0_4e.in1k) |
| `tf_efficientnet_cc_b0_8e.in1k` | 224 | 77.956 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_cc_b0_8e.in1k) |
| `tf_efficientnet_cc_b1_8e.in1k` | 240 | 79.318 | 240 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_cc_b1_8e.in1k) |
| `tf_efficientnet_el.in1k` | 300 | 80.246 | 300 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_el.in1k) |
| `tf_efficientnet_em.in1k` | 240 | 78.128 | 240 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_em.in1k) |
| `tf_efficientnet_es.in1k` | 224 | 76.608 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_es.in1k) |
| `tf_efficientnet_lite0.in1k` | 224 | 74.846 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_lite0.in1k) |
| `tf_efficientnet_lite1.in1k` | 240 | 76.674 | 240 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_lite1.in1k) |
| `tf_efficientnet_lite2.in1k` | 260 | 77.470 | 260 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_lite2.in1k) |
| `tf_efficientnet_lite3.in1k` | 300 | 79.808 | 300 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_lite3.in1k) |
| `tf_efficientnet_lite4.in1k` | 380 | 81.540 | 380 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnet_lite4.in1k) |
| `tf_efficientnetv2_b0.in1k` | 192 (test 224) | 78.386 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnetv2_b0.in1k) |
| `tf_efficientnetv2_b1.in1k` | 192 (test 240) | 79.518 | 240 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnetv2_b1.in1k) |
| `tf_efficientnetv2_b2.in1k` | 208 (test 260) | 80.226 | 260 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnetv2_b2.in1k) |
| `tf_efficientnetv2_b3.in1k` | 240 (test 300) | 81.960 | 300 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnetv2_b3.in1k) |
| `tf_efficientnetv2_b3.in21k` | 240 (test 300) | - | - | 21843 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnetv2_b3.in21k) |
| `tf_efficientnetv2_b3.in21k_ft_in1k` | 240 (test 300) | 82.634 | 300 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnetv2_b3.in21k_ft_in1k) |
| `tf_efficientnetv2_l.in1k` | 384 (test 480) | 85.660 | 480 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnetv2_l.in1k) |
| `tf_efficientnetv2_l.in21k` | 384 (test 480) | - | - | 21843 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnetv2_l.in21k) |
| `tf_efficientnetv2_l.in21k_ft_in1k` | 384 (test 480) | 86.802 | 480 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnetv2_l.in21k_ft_in1k) |
| `tf_efficientnetv2_m.in1k` | 384 (test 480) | 85.208 | 480 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnetv2_m.in1k) |
| `tf_efficientnetv2_m.in21k` | 384 (test 480) | - | - | 21843 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnetv2_m.in21k) |
| `tf_efficientnetv2_m.in21k_ft_in1k` | 384 (test 480) | 86.014 | 480 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnetv2_m.in21k_ft_in1k) |
| `tf_efficientnetv2_s.in1k` | 300 (test 384) | 83.900 | 384 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnetv2_s.in1k) |
| `tf_efficientnetv2_s.in21k` | 300 (test 384) | - | - | 21843 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnetv2_s.in21k) |
| `tf_efficientnetv2_s.in21k_ft_in1k` | 300 (test 384) | 84.332 | 384 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnetv2_s.in21k_ft_in1k) |
| `tf_efficientnetv2_xl.in21k` | 384 (test 512) | - | - | 21843 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnetv2_xl.in21k) |
| `tf_efficientnetv2_xl.in21k_ft_in1k` | 384 (test 512) | 86.754 | 512 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_efficientnetv2_xl.in21k_ft_in1k) |
| `tf_mixnet_l.in1k` | 224 | 78.774 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_mixnet_l.in1k) |
| `tf_mixnet_m.in1k` | 224 | 76.952 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_mixnet_m.in1k) |
| `tf_mixnet_s.in1k` | 224 | 75.656 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tf_mixnet_s.in1k) |
| `tinynet_a.in1k` | 192 | 77.666 | 192 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tinynet_a.in1k) |
| `tinynet_b.in1k` | 188 | 74.946 | 188 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tinynet_b.in1k) |
| `tinynet_c.in1k` | 184 | 71.214 | 184 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tinynet_c.in1k) |
| `tinynet_d.in1k` | 152 | 66.936 | 152 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tinynet_d.in1k) |
| `tinynet_e.in1k` | 106 | 59.874 | 106 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/tinynet_e.in1k) |