# Text-to-Image Generation on MS COCO 5,000 samples

Source: [Text-to-Image Generation on MS COCO 5,000 samples](<https://www.sota2.com/research/sota/text-to-image-generation-on-ms-coco-5-000-samples>)

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

Data updated: 2026-03-06T07:08:17.723Z

Evaluation of Text-to-Image Generation models using the MS COCO 5,000 sample subset, assessed using FID, CLIP, and SSIM metrics.

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 | Text-to-Image Generation |
| Dataset (including any stated split) | MS COCO 5,000 samples |
| Metric | FID |
| Unit | SCALAR |
| Direction | Lower is better |
| Results loaded | 14 of 14 |
| 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](<https://www.sota2.com/research/sota/text-to-image-generation-on-ms-coco-5-000-samples.md?metric=FID>): Lower is better; unit: SCALAR.

- [CLIP Score](<https://www.sota2.com/research/sota/text-to-image-generation-on-ms-coco-5-000-samples.md?metric=CLIP%20Score>): Higher is better; unit: SCALAR.

- [SSIM](<https://www.sota2.com/research/sota/text-to-image-generation-on-ms-coco-5-000-samples.md?metric=SSIM>): Higher is better; unit: SCALAR.

## Recorded results

| Method | FID | Result date | Report paper |
| --- | --- | --- | --- |
| Original (SDXL) | 27.43 | Not recorded | 68fa75fe1e209131cea6acad |
| Dev (FLUX) | 28.47 | Not recorded | 68fa75fe1e209131cea6acad |
| FLUX-Lite | 29.36 | Not recorded | 68fa75fe1e209131cea6acad |
| EcoDiff | 30.81 | Not recorded | 68fa75fe1e209131cea6acad |
| Schnell (FLUX) | 30.99 | Not recorded | 68fa75fe1e209131cea6acad |
| EcoDiff | 31.76 | Not recorded | 68fa75fe1e209131cea6acad |
| EcoDiff | 32.19 | Not recorded | 68fa75fe1e209131cea6acad |
| DiffPruning | 40.84 | Not recorded | 68fa75fe1e209131cea6acad |
| DiffPruning | 42.36 | Not recorded | 68fa75fe1e209131cea6acad |
| BK-SDM | 42.87 | Not recorded | 68fa75fe1e209131cea6acad |
| Per-Step Loss | 50.56 | Not recorded | 68fa75fe1e209131cea6acad |
| Per-Step Loss | 52.61 | Not recorded | 68fa75fe1e209131cea6acad |
| DiffPruning | 83.81 | Not recorded | 68fa75fe1e209131cea6acad |
| Per-Step Loss | 97.36 | Not recorded | 68fa75fe1e209131cea6acad |

### Original (SDXL)

FID: 27.43. Result ID: 69aa7c1e6dbb7075b953b6e1.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75fe1e209131cea6acad.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa70011e209131cea487fb.md>)

Recorded evaluation configuration: {"Architecture":"SDXL Diffusion U-Net","Sparsity":"0%","#Param":"2.6B","Speedup":"1x","Compute Budget":"-"}. This record does not assert independent reproduction.

### Dev (FLUX)

FID: 28.47. Result ID: 69aa7c1e6dbb7075b953b6e6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75fe1e209131cea6acad.md>)

Recorded evaluation configuration: {"Architecture":"FLUX Flow Matching DiT","Sparsity":"0%","#Param":"11.9B","Speedup":"1x","Compute Budget":"-"}. This record does not assert independent reproduction.

### FLUX-Lite

FID: 29.36. Result ID: 69aa7c1e6dbb7075b953b6e8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75fe1e209131cea6acad.md>)

Recorded evaluation configuration: {"Architecture":"FLUX Flow Matching DiT","Sparsity":"33%","#Param":"8B","Speedup":"1.49x","Compute Budget":"1120hr"}. This record does not assert independent reproduction.

### EcoDiff

FID: 30.81. Result ID: 69aa7c1e6dbb7075b953b6ea.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75fe1e209131cea6acad.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75fe1e209131cea6acad.md>)

Recorded evaluation configuration: {"Architecture":"FLUX Flow Matching DiT (Dev)","Sparsity":"20%","#Param":"9.6B","Speedup":"1.25x","Compute Budget":"10hr"}. This record does not assert independent reproduction.

### Schnell (FLUX)

FID: 30.99. Result ID: 69aa7c1e6dbb7075b953b6eb.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75fe1e209131cea6acad.md>)

Recorded evaluation configuration: {"Architecture":"FLUX Flow Matching DiT","Sparsity":"0%","#Param":"11.9B","Speedup":"7x","Compute Budget":"-"}. This record does not assert independent reproduction.

### EcoDiff

FID: 31.76. Result ID: 69aa7c1e6dbb7075b953b6ee.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75fe1e209131cea6acad.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75fe1e209131cea6acad.md>)

Recorded evaluation configuration: {"Architecture":"FLUX Flow Matching DiT (Schnell)","Sparsity":"20%","#Param":"9.6B","Speedup":"8.75x","Compute Budget":"10hr"}. This record does not assert independent reproduction.

### EcoDiff

FID: 32.19. Result ID: 69aa7c1e6dbb7075b953b6e5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75fe1e209131cea6acad.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75fe1e209131cea6acad.md>)

Recorded evaluation configuration: {"Architecture":"SDXL Diffusion U-Net","Sparsity":"20%","#Param":"2.1B","Speedup":"1.25x","Compute Budget":"10hr"}. This record does not assert independent reproduction.

### DiffPruning

FID: 40.84. Result ID: 69aa7c1e6dbb7075b953b6e7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75fe1e209131cea6acad.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa70bd1e209131cea4cf9d.md>)

Recorded evaluation configuration: {"Architecture":"FLUX Flow Matching DiT (Dev)","Sparsity":"20%","#Param":"9.6B","Speedup":"1.25x","Compute Budget":"10hr"}. This record does not assert independent reproduction.

### DiffPruning

FID: 42.36. Result ID: 69aa7c1e6dbb7075b953b6ec.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75fe1e209131cea6acad.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa70bd1e209131cea4cf9d.md>)

Recorded evaluation configuration: {"Architecture":"FLUX Flow Matching DiT (Schnell)","Sparsity":"20%","#Param":"9.6B","Speedup":"8.75x","Compute Budget":"10hr"}. This record does not assert independent reproduction.

### BK-SDM

FID: 42.87. Result ID: 69aa7c1e6dbb7075b953b6e2.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75fe1e209131cea6acad.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76071e209131cea6b026.md>)

Recorded evaluation configuration: {"Architecture":"SDXL Diffusion U-Net","Sparsity":"20%","#Param":"2.1B","Speedup":"1.25x","Compute Budget":"10hr"}. This record does not assert independent reproduction.

### Per-Step Loss

FID: 50.56. Result ID: 69aa7c1e6dbb7075b953b6e9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75fe1e209131cea6acad.md>)

Recorded evaluation configuration: {"Architecture":"FLUX Flow Matching DiT (Dev)","Sparsity":"20%","#Param":"9.6B","Speedup":"1.25x","Compute Budget":"10hr"}. This record does not assert independent reproduction.

### Per-Step Loss

FID: 52.61. Result ID: 69aa7c1e6dbb7075b953b6ed.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75fe1e209131cea6acad.md>)

Recorded evaluation configuration: {"Architecture":"FLUX Flow Matching DiT (Schnell)","Sparsity":"20%","#Param":"9.6B","Speedup":"8.75x","Compute Budget":"10hr"}. This record does not assert independent reproduction.

### DiffPruning

FID: 83.81. Result ID: 69aa7c1e6dbb7075b953b6e3.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75fe1e209131cea6acad.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa70bd1e209131cea4cf9d.md>)

Recorded evaluation configuration: {"Architecture":"SDXL Diffusion U-Net","Sparsity":"20%","#Param":"2.1B","Speedup":"1.25x","Compute Budget":"10hr"}. This record does not assert independent reproduction.

### Per-Step Loss

FID: 97.36. Result ID: 69aa7c1e6dbb7075b953b6e4.

[Reporting paper](<https://www.sota2.com/research/paper/68fa75fe1e209131cea6acad.md>)

Recorded evaluation configuration: {"Architecture":"SDXL Diffusion U-Net","Sparsity":"20%","#Param":"2.1B","Speedup":"1.25x","Compute Budget":"10hr"}. This record does not assert independent reproduction.

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