# Image Generation on ImageNet 256 x 256 (FID U/G)

Source: [Image Generation on ImageNet 256 x 256 (FID U/G)](<https://www.sota2.com/research/sota/image-generation-on-imagenet-256-x-256-fid-u-g>)

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

Data updated: 2026-04-24T07:01:50.022Z

Evaluation of image generation models on ImageNet 256x256, reporting FID results specifically labeled as FID (U) and FID (G).

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 Generation |
| Dataset (including any stated split) | ImageNet 256 x 256 |
| Metric | FID (U) |
| Unit | SCALAR |
| Direction | Lower is better |
| Results loaded | 10 of 10 |
| 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

- [FID (U)](<https://www.sota2.com/research/sota/image-generation-on-imagenet-256-x-256-fid-u-g.md?metric=FID%20(U)>): Lower is better; unit: SCALAR.

- [FID (G)](<https://www.sota2.com/research/sota/image-generation-on-imagenet-256-x-256-fid-u-g.md?metric=FID%20(G)>): Lower is better; unit: SCALAR.

## Recorded results

| Method | FID (U) | Result date | Report paper |
| --- | --- | --- | --- |
| 3-Stream (+Semantic) | 6.99 | Not recorded | 69eaf90555a3c32af89f88c0 |
| LF-DiT-L (Semantic Only) | 7.2 | Not recorded | 69eaf90555a3c32af89f88c0 |
| SiT-XL/2 | 8.3 | Not recorded | 69eaf90555a3c32af89f88c0 |
| DiT-XL/2 | 9.62 | Not recorded | 69eaf90555a3c32af89f88c0 |
| JiT-L | 16.21 | Not recorded | 69eaf90555a3c32af89f88c0 |
| 2-Stream (Learnable Wavelet) | 17.08 | Not recorded | 69eaf90555a3c32af89f88c0 |
| 2-Stream (Laplacian) | 20.7 | Not recorded | 69eaf90555a3c32af89f88c0 |
| 2-Stream (Haar) | 22.84 | Not recorded | 69eaf90555a3c32af89f88c0 |
| 1-Stream + REPA | 36.01 | Not recorded | 69eaf90555a3c32af89f88c0 |
| 1-Stream (Pixel DiT Baseline) | 37.03 | Not recorded | 69eaf90555a3c32af89f88c0 |

### 3-Stream (+Semantic)

FID (U): 6.99. Result ID: 69eb155447f34ce924d2bcdc.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90555a3c32af89f88c0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69eaf90555a3c32af89f88c0.md>)

Recorded evaluation configuration: {"Params":"466M","Epochs":"400"}. This record does not assert independent reproduction.

### LF-DiT-L (Semantic Only)

FID (U): 7.2. Result ID: 69eb155447f34ce924d2bcd8.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90555a3c32af89f88c0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/698eafe8c97a0b3e3b5e659d.md>)

Recorded evaluation configuration: {"Params":"465M","Epochs":"400"}. This record does not assert independent reproduction.

### SiT-XL/2

FID (U): 8.3. Result ID: 69eb155447f34ce924d2bcd4.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90555a3c32af89f88c0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74e71e209131cea64974.md>)

Recorded evaluation configuration: {"Params":"675M","Epochs":"1400"}. This record does not assert independent reproduction.

### DiT-XL/2

FID (U): 9.62. Result ID: 69eb155447f34ce924d2bcd3.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90555a3c32af89f88c0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ea91e209131cea41328.md>)

Recorded evaluation configuration: {"Params":"675M","Epochs":"1400"}. This record does not assert independent reproduction.

### JiT-L

FID (U): 16.21. Result ID: 69eb155447f34ce924d2bcd5.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90555a3c32af89f88c0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/691bfe976750be83fe5cf517.md>)

[Implementation](<https://github.com/LTH14/JiT>)

Recorded evaluation configuration: {"Params":"459M","Epochs":"200"}. This record does not assert independent reproduction.

### 2-Stream (Learnable Wavelet)

FID (U): 17.08. Result ID: 69eb155447f34ce924d2bcdb.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90555a3c32af89f88c0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69eaf90555a3c32af89f88c0.md>)

Recorded evaluation configuration: {"Params":"461M","Epochs":"400"}. This record does not assert independent reproduction.

### 2-Stream (Laplacian)

FID (U): 20.7. Result ID: 69eb155447f34ce924d2bcd9.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90555a3c32af89f88c0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69eaf90555a3c32af89f88c0.md>)

Recorded evaluation configuration: {"Params":"461M","Epochs":"400"}. This record does not assert independent reproduction.

### 2-Stream (Haar)

FID (U): 22.84. Result ID: 69eb155447f34ce924d2bcda.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90555a3c32af89f88c0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69eaf90555a3c32af89f88c0.md>)

Recorded evaluation configuration: {"Params":"461M","Epochs":"400"}. This record does not assert independent reproduction.

### 1-Stream + REPA

FID (U): 36.01. Result ID: 69eb155447f34ce924d2bcd7.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90555a3c32af89f88c0.md>)

Recorded evaluation configuration: {"Epochs":"200"}. This record does not assert independent reproduction.

### 1-Stream (Pixel DiT Baseline)

FID (U): 37.03. Result ID: 69eb155447f34ce924d2bcd6.

[Reporting paper](<https://www.sota2.com/research/paper/69eaf90555a3c32af89f88c0.md>)

Recorded evaluation configuration: {"Epochs":"200"}. This record does not assert independent reproduction.

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