# Gradient Inversion on ImageNet 64x64

Source: [Gradient Inversion on ImageNet 64x64](<https://www.sota2.com/research/sota/gradient-inversion-on-imagenet-64x64>)

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

Data updated: 2026-02-26T05:55:30.467Z

Evaluation of models performing gradient inversion or reconstruction tasks on the ImageNet dataset downsampled to 64x64 resolution, typically measured using fidelity metrics like PSNR, SSIM, and LPIPS.

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 | Gradient Inversion |
| Dataset (including any stated split) | ImageNet 64x64 |
| Metric | PSNR |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 6 of 6 |
| 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

- [PSNR](<https://www.sota2.com/research/sota/gradient-inversion-on-imagenet-64x64.md?metric=PSNR>): Higher is better; unit: SCALAR.

- [SSIM](<https://www.sota2.com/research/sota/gradient-inversion-on-imagenet-64x64.md?metric=SSIM>): Higher is better; unit: SCALAR.

- [LPIPS](<https://www.sota2.com/research/sota/gradient-inversion-on-imagenet-64x64.md?metric=LPIPS>): Lower is better; unit: SCALAR.

## Recorded results

| Method | PSNR | Result date | Report paper |
| --- | --- | --- | --- |
| GI-z/w+BN | 16.31 | Not recorded | 68fa6b2f1e209131cea2cb02 |
| GI-z/w (GIAS) | 15.58 | Not recorded | 68fa6b2f1e209131cea2cb02 |
| GI-x+BN | 15.52 | Not recorded | 68fa6b2f1e209131cea2cb02 |
| GI-w | 14.7 | Not recorded | 68fa6b2f1e209131cea2cb02 |
| GI-z | 14.27 | Not recorded | 68fa6b2f1e209131cea2cb02 |
| GI-x | 13.78 | Not recorded | 68fa6b2f1e209131cea2cb02 |

### GI-z/w+BN

PSNR: 16.31. Result ID: 698e6b81ffc3901de09d73c1.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6b2f1e209131cea2cb02.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6b2f1e209131cea2cb02.md>)

[Implementation](<https://github.com/ml-postech/gradient-inversion-generative-image-prior>)

Recorded evaluation configuration: {"batch\_normalization":true}. This record does not assert independent reproduction.

### GI-z/w (GIAS)

PSNR: 15.58. Result ID: 698e6b81ffc3901de09d73bf.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6b2f1e209131cea2cb02.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6b2f1e209131cea2cb02.md>)

[Implementation](<https://github.com/ml-postech/gradient-inversion-generative-image-prior>)

Recorded evaluation configuration: {"batch\_normalization":false}. This record does not assert independent reproduction.

### GI-x+BN

PSNR: 15.52. Result ID: 698e6b81ffc3901de09d73c0.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6b2f1e209131cea2cb02.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69ce1e209131cea24381.md>)

Recorded evaluation configuration: {"batch\_normalization":true}. This record does not assert independent reproduction.

### GI-w

PSNR: 14.7. Result ID: 698e6b81ffc3901de09d73be.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6b2f1e209131cea2cb02.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6b2f1e209131cea2cb02.md>)

[Implementation](<https://github.com/ml-postech/gradient-inversion-generative-image-prior>)

Recorded evaluation configuration: {"batch\_normalization":false}. This record does not assert independent reproduction.

### GI-z

PSNR: 14.27. Result ID: 698e6b81ffc3901de09d73bd.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6b2f1e209131cea2cb02.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6b2f1e209131cea2cb02.md>)

[Implementation](<https://github.com/ml-postech/gradient-inversion-generative-image-prior>)

Recorded evaluation configuration: {"batch\_normalization":false}. This record does not assert independent reproduction.

### GI-x

PSNR: 13.78. Result ID: 698e6b81ffc3901de09d73bc.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6b2f1e209131cea2cb02.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa68511e209131cea1b662.md>)

Recorded evaluation configuration: {"batch\_normalization":false}. This record does not assert independent reproduction.

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