# Image Classification on ImageNet 1k (OTS Acc., Trained Acc.)

Source: [Image Classification on ImageNet 1k (OTS Acc., Trained Acc.)](<https://www.sota2.com/research/sota/image-classification-on-imagenet-1k-ots-acc-trained-acc>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-02-26T06:07:33.702Z

Evaluation of Image Classification performance on the ImageNet-1k dataset, reporting custom metrics including 'OTS Acc.' and 'Trained Acc.'.

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Image Classification |
| Dataset (including any stated split) | ImageNet 1k |
| Metric | Trained Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 32 of 32 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Trained Accuracy](<https://www.sota2.com/research/sota/image-classification-on-imagenet-1k-ots-acc-trained-acc.md?metric=Trained%20Accuracy>): Higher is better; unit: PERCENT.

- [OTS Accuracy](<https://www.sota2.com/research/sota/image-classification-on-imagenet-1k-ots-acc-trained-acc.md?metric=OTS%20Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Trained Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| ViT | 86.9 | Not recorded | 68fa75761e209131cea681c4 |
| PiToMe | 86.7 | Not recorded | 68fa75761e209131cea681c4 |
| DiffRate | 86.6 | Not recorded | 68fa75761e209131cea681c4 |
| ToMe | 86.4 | Not recorded | 68fa75761e209131cea681c4 |
| ToFu | 86.4 | Not recorded | 68fa75761e209131cea681c4 |
| DCT | 86 | Not recorded | 68fa75761e209131cea681c4 |
| ViT | 85.7 | Not recorded | 68fa75761e209131cea681c4 |
| MViTv2-L | 85.3 | Not recorded | 68fa75761e209131cea681c4 |
| DiffRate | 85.3 | Not recorded | 68fa75761e209131cea681c4 |
| PiToMe | 85.3 | Not recorded | 68fa75761e209131cea681c4 |
| ToFu | 85.1 | Not recorded | 68fa75761e209131cea681c4 |
| ToMe | 85 | Not recorded | 68fa75761e209131cea681c4 |
| MViTv2-B | 84.4 | Not recorded | 68fa75761e209131cea681c4 |
| DCT | 84.4 | Not recorded | 68fa75761e209131cea681c4 |
| CSWin-B | 84.2 | Not recorded | 68fa75761e209131cea681c4 |
| Swin-B | 84 | Not recorded | 68fa75761e209131cea681c4 |
| PiToMe | 79.8 | Not recorded | 68fa75761e209131cea681c4 |
| ViT | 79.8 | Not recorded | 68fa75761e209131cea681c4 |
| ToFu | 79.6 | Not recorded | 68fa75761e209131cea681c4 |
| E-ViT | 79.5 | Not recorded | 68fa75761e209131cea681c4 |
| DiffRate | 79.5 | Not recorded | 68fa75761e209131cea681c4 |
| ToMe | 79.4 | Not recorded | 68fa75761e209131cea681c4 |
| Dynamic-ViT | 79.3 | Not recorded | 68fa75761e209131cea681c4 |
| SP-ViT | 79.3 | Not recorded | 68fa75761e209131cea681c4 |
| A-ViT | 78.6 | Not recorded | 68fa75761e209131cea681c4 |
| DCT | 78.6 | Not recorded | 68fa75761e209131cea681c4 |
| ViT | 72.3 | Not recorded | 68fa75761e209131cea681c4 |
| PiToMe | 71.6 | Not recorded | 68fa75761e209131cea681c4 |
| DiffRate | 70.7 | Not recorded | 68fa75761e209131cea681c4 |
| ToFu | 70.5 | Not recorded | 68fa75761e209131cea681c4 |
| ToMe | 70 | Not recorded | 68fa75761e209131cea681c4 |
| DCT | 68.7 | Not recorded | 68fa75761e209131cea681c4 |

### ViT

Trained Accuracy: 86.9. Result ID: 6993b1903a6c31484917e0d6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a0b1e209131cea25b04.md>)

[Implementation](<https://github.com/google-research/vision_transformer>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"MAE-H","Flops":"167.4","Train speed up":false}. This record does not assert independent reproduction.

### PiToMe

Trained Accuracy: 86.7. Result ID: 6993b1903a6c31484917e0d5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"MAE-H","Flops":"92.8","Train speed up":true}. This record does not assert independent reproduction.

### DiffRate

Trained Accuracy: 86.6. Result ID: 6993b1903a6c31484917e0d4.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f7a1e209131cea45d68.md>)

[Implementation](<https://github.com/OpenGVLab/DiffRate>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"MAE-H","Flops":"92.8","Train speed up":true}. This record does not assert independent reproduction.

### ToMe

Trained Accuracy: 86.4. Result ID: 6993b1903a6c31484917e0d1.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ea91e209131cea4130c.md>)

[Implementation](<http://github.com/facebookresearch/ToMe>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"MAE-H","Flops":"92.8","Train speed up":true}. This record does not assert independent reproduction.

### ToFu

Trained Accuracy: 86.4. Result ID: 6993b1903a6c31484917e0d2.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71561e209131cea509f8.md>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"MAE-H","Flops":"92.8","Train speed up":true}. This record does not assert independent reproduction.

### DCT

Trained Accuracy: 86. Result ID: 6993b1903a6c31484917e0d3.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78a01e209131cea796d3.md>)

[Implementation](<https://github.com/LUMIA-Group/FourierTransformer>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"MAE-H","Flops":"92.8","Train speed up":true}. This record does not assert independent reproduction.

### ViT

Trained Accuracy: 85.7. Result ID: 6993b1903a6c31484917e0d0.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a0b1e209131cea25b04.md>)

[Implementation](<https://github.com/google-research/vision_transformer>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"MAE-L","Flops":"61.6","Train speed up":false}. This record does not assert independent reproduction.

### MViTv2-L

Trained Accuracy: 85.3. Result ID: 6993b1903a6c31484917e0ba.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c2f1e209131cea32a41.md>)

[Implementation](<https://github.com/facebookresearch/mvit>)

Recorded evaluation configuration: {"Type":"Other","Flops":"42.1","Train speed up":false}. This record does not assert independent reproduction.

### DiffRate

Trained Accuracy: 85.3. Result ID: 6993b1903a6c31484917e0ce.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f7a1e209131cea45d68.md>)

[Implementation](<https://github.com/OpenGVLab/DiffRate>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"MAE-L","Flops":"31.0","Train speed up":true}. This record does not assert independent reproduction.

### PiToMe

Trained Accuracy: 85.3. Result ID: 6993b1903a6c31484917e0cf.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"MAE-L","Flops":"31.0","Train speed up":true}. This record does not assert independent reproduction.

### ToFu

Trained Accuracy: 85.1. Result ID: 6993b1903a6c31484917e0cc.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71561e209131cea509f8.md>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"MAE-L","Flops":"31.0","Train speed up":true}. This record does not assert independent reproduction.

### ToMe

Trained Accuracy: 85. Result ID: 6993b1903a6c31484917e0cb.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ea91e209131cea4130c.md>)

[Implementation](<http://github.com/facebookresearch/ToMe>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"MAE-L","Flops":"31.0","Train speed up":true}. This record does not assert independent reproduction.

### MViTv2-B

Trained Accuracy: 84.4. Result ID: 6993b1903a6c31484917e0b9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c2f1e209131cea32a41.md>)

[Implementation](<https://github.com/facebookresearch/mvit>)

Recorded evaluation configuration: {"Type":"Other","Flops":"10.2","Train speed up":false}. This record does not assert independent reproduction.

### DCT

Trained Accuracy: 84.4. Result ID: 6993b1903a6c31484917e0cd.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78a01e209131cea796d3.md>)

[Implementation](<https://github.com/LUMIA-Group/FourierTransformer>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"MAE-L","Flops":"31.0","Train speed up":true}. This record does not assert independent reproduction.

### CSWin-B

Trained Accuracy: 84.2. Result ID: 6993b1903a6c31484917e0b8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6b991e209131cea2f30f.md>)

Recorded evaluation configuration: {"Type":"Other","Flops":"15.0","Train speed up":false}. This record does not assert independent reproduction.

### Swin-B

Trained Accuracy: 84. Result ID: 6993b1903a6c31484917e0b7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a8c1e209131cea28b62.md>)

[Implementation](<https://github.com/microsoft/Swin-Transformer>)

Recorded evaluation configuration: {"Type":"Other","Flops":"15.4","Train speed up":false}. This record does not assert independent reproduction.

### PiToMe

Trained Accuracy: 79.8. Result ID: 6993b1903a6c31484917e0c9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"DeiT-S","Flops":"2.9","Train speed up":true}. This record does not assert independent reproduction.

### ViT

Trained Accuracy: 79.8. Result ID: 6993b1903a6c31484917e0ca.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a0b1e209131cea25b04.md>)

[Implementation](<https://github.com/google-research/vision_transformer>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"DeiT-S","Flops":"4.6","Train speed up":false}. This record does not assert independent reproduction.

### ToFu

Trained Accuracy: 79.6. Result ID: 6993b1903a6c31484917e0c6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71561e209131cea509f8.md>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"DeiT-S","Flops":"2.9","Train speed up":true}. This record does not assert independent reproduction.

### E-ViT

Trained Accuracy: 79.5. Result ID: 6993b1903a6c31484917e0c4.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

Recorded evaluation configuration: {"Type":"prune","Backbone":"DeiT-S","Flops":"2.9","Train speed up":false}. This record does not assert independent reproduction.

### DiffRate

Trained Accuracy: 79.5. Result ID: 6993b1903a6c31484917e0c8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f7a1e209131cea45d68.md>)

[Implementation](<https://github.com/OpenGVLab/DiffRate>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"DeiT-S","Flops":"2.9","Train speed up":true}. This record does not assert independent reproduction.

### ToMe

Trained Accuracy: 79.4. Result ID: 6993b1903a6c31484917e0c5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ea91e209131cea4130c.md>)

[Implementation](<http://github.com/facebookresearch/ToMe>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"DeiT-S","Flops":"2.9","Train speed up":true}. This record does not assert independent reproduction.

### Dynamic-ViT

Trained Accuracy: 79.3. Result ID: 6993b1903a6c31484917e0c2.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6b1b1e209131cea2c45a.md>)

Recorded evaluation configuration: {"Type":"prune","Backbone":"DeiT-S","Flops":"2.9","Train speed up":false}. This record does not assert independent reproduction.

### SP-ViT

Trained Accuracy: 79.3. Result ID: 6993b1903a6c31484917e0c3.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

Recorded evaluation configuration: {"Type":"prune","Backbone":"DeiT-S","Flops":"2.6","Train speed up":false}. This record does not assert independent reproduction.

### A-ViT

Trained Accuracy: 78.6. Result ID: 6993b1903a6c31484917e0c1.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6da21e209131cea3b029.md>)

[Implementation](<https://a-vit.github.io/>)

Recorded evaluation configuration: {"Type":"prune","Backbone":"DeiT-S","Flops":"2.9","Train speed up":false}. This record does not assert independent reproduction.

### DCT

Trained Accuracy: 78.6. Result ID: 6993b1903a6c31484917e0c7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78a01e209131cea796d3.md>)

[Implementation](<https://github.com/LUMIA-Group/FourierTransformer>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"DeiT-S","Flops":"2.9","Train speed up":true}. This record does not assert independent reproduction.

### ViT

Trained Accuracy: 72.3. Result ID: 6993b1903a6c31484917e0c0.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6a0b1e209131cea25b04.md>)

[Implementation](<https://github.com/google-research/vision_transformer>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"DeiT-T","Flops":"1.2","Train speed up":false}. This record does not assert independent reproduction.

### PiToMe

Trained Accuracy: 71.6. Result ID: 6993b1903a6c31484917e0bf.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"DeiT-T","Flops":"0.79","Train speed up":true}. This record does not assert independent reproduction.

### DiffRate

Trained Accuracy: 70.7. Result ID: 6993b1903a6c31484917e0be.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f7a1e209131cea45d68.md>)

[Implementation](<https://github.com/OpenGVLab/DiffRate>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"DeiT-T","Flops":"0.79","Train speed up":true}. This record does not assert independent reproduction.

### ToFu

Trained Accuracy: 70.5. Result ID: 6993b1903a6c31484917e0bc.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71561e209131cea509f8.md>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"DeiT-T","Flops":"0.79","Train speed up":true}. This record does not assert independent reproduction.

### ToMe

Trained Accuracy: 70. Result ID: 6993b1903a6c31484917e0bb.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ea91e209131cea4130c.md>)

[Implementation](<http://github.com/facebookresearch/ToMe>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"DeiT-T","Flops":"0.79","Train speed up":true}. This record does not assert independent reproduction.

### DCT

Trained Accuracy: 68.7. Result ID: 6993b1903a6c31484917e0bd.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75761e209131cea681c4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78a01e209131cea796d3.md>)

[Implementation](<https://github.com/LUMIA-Group/FourierTransformer>)

Recorded evaluation configuration: {"Type":"merge","Backbone":"DeiT-T","Flops":"0.79","Train speed up":true}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.