# Text-to-Image Generation on Parti-Prompts

Source: [Text-to-Image Generation on Parti-Prompts](<https://www.sota2.com/research/sota/text-to-image-generation-on-parti-prompts>)

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

Data updated: 2026-06-02T13:42:08.210Z

Evaluation of Text-to-Image generation models using the Parti-Prompts dataset, assessed using a suite of automated metrics reported both as raw scores and win rates (PickScore, HPS, CLIP, ImageReward, AES).

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) | Parti-Prompts |
| Metric | PickScore |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 35 of 35 |
| 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

- [PickScore](<https://www.sota2.com/research/sota/text-to-image-generation-on-parti-prompts.md?metric=PickScore>): Higher is better; unit: SCALAR.

- [PickScore Winrate](<https://www.sota2.com/research/sota/text-to-image-generation-on-parti-prompts.md?metric=PickScore%20Winrate>): Higher is better; unit: PERCENT.

- [HPS Winrate](<https://www.sota2.com/research/sota/text-to-image-generation-on-parti-prompts.md?metric=HPS%20Winrate>): Higher is better; unit: PERCENT.

- [CLIP Winrate](<https://www.sota2.com/research/sota/text-to-image-generation-on-parti-prompts.md?metric=CLIP%20Winrate>): Higher is better; unit: PERCENT.

- [ImageReward Winrate](<https://www.sota2.com/research/sota/text-to-image-generation-on-parti-prompts.md?metric=ImageReward%20Winrate>): Higher is better; unit: PERCENT.

- [AES Winrate](<https://www.sota2.com/research/sota/text-to-image-generation-on-parti-prompts.md?metric=AES%20Winrate>): Higher is better; unit: PERCENT.

- [Average Winrate](<https://www.sota2.com/research/sota/text-to-image-generation-on-parti-prompts.md?metric=Average%20Winrate>): Higher is better; unit: PERCENT.

- [HPS Score](<https://www.sota2.com/research/sota/text-to-image-generation-on-parti-prompts.md?metric=HPS%20Score>): Higher is better; unit: SCALAR.

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

- [ImageReward Score](<https://www.sota2.com/research/sota/text-to-image-generation-on-parti-prompts.md?metric=ImageReward%20Score>): Higher is better; unit: SCALAR.

- [AES Score](<https://www.sota2.com/research/sota/text-to-image-generation-on-parti-prompts.md?metric=AES%20Score>): Higher is better; unit: SCALAR.

- [UniRwd Score](<https://www.sota2.com/research/sota/text-to-image-generation-on-parti-prompts.md?metric=UniRwd%20Score>): Higher is better; unit: SCALAR.

- [HPSv3 Score](<https://www.sota2.com/research/sota/text-to-image-generation-on-parti-prompts.md?metric=HPSv3%20Score>): Higher is better; unit: SCALAR.

- [DeQA Score](<https://www.sota2.com/research/sota/text-to-image-generation-on-parti-prompts.md?metric=DeQA%20Score>): Higher is better; unit: SCALAR.

## Recorded results

| Method | PickScore | Result date | Report paper |
| --- | --- | --- | --- |
| DRaFT | 24.34 | Not recorded | 6a1e6454dbd7150506f859f9 |
| VGG-Flow1-step | 23.98 | Not recorded | 6a1e6454dbd7150506f859f9 |
| DrPO | 23.71 | Not recorded | 6a1e6454dbd7150506f859f9 |
| Diff.-NPO | 23.26 | Not recorded | 69fc1cac5497b3b670c88614 |
| Online-DPO | 23.07 | Not recorded | 69fc1cac5497b3b670c88614 |
| DRaFT | 23.07 | Not recorded | 6a1e6454dbd7150506f859f9 |
| VGG-Flow1-step | 22.99 | Not recorded | 6a1e6454dbd7150506f859f9 |
| DrPO | 22.99 | Not recorded | 6a1e6454dbd7150506f859f9 |
| SDXL-DPO | 22.95 | Not recorded | 6a1e6454dbd7150506f859f9 |
| RealAlign | 22.9 | Not recorded | 6a0d402af984739ed67378f8 |
| PSO | 22.87 | Not recorded | 6a1e6454dbd7150506f859f9 |
| DPO1-step | 22.85 | Not recorded | 6a1e6454dbd7150506f859f9 |
| GRPO1-step | 22.8 | Not recorded | 6a1e6454dbd7150506f859f9 |
| SDXL-Turbo (base) | 22.77 | Not recorded | 6a1e6454dbd7150506f859f9 |
| Diffusion-DPO | 22.75 | Not recorded | 6a0d402af984739ed67378f8 |
| SDXL base | 22.64 | Not recorded | 6a1e6454dbd7150506f859f9 |
| SD-3.5-M | 22.54 | Not recorded | 6a0d402af984739ed67378f8 |
| Original (SDXL) | 22.51 | Not recorded | 69fc1cac5497b3b670c88614 |
| DPO1-step | 22.39 | Not recorded | 6a1e6454dbd7150506f859f9 |
| GRPO1-step | 22.35 | Not recorded | 6a1e6454dbd7150506f859f9 |
| SD-Turbo (base) | 22.29 | Not recorded | 6a1e6454dbd7150506f859f9 |
| PSO | 22.29 | Not recorded | 6a1e6454dbd7150506f859f9 |
| Online-SFT | 22.21 | Not recorded | 69fc1cac5497b3b670c88614 |
| Diff.-NPO | 22.07 | Not recorded | 69fc1cac5497b3b670c88614 |
| SPIN | 21.98 | Not recorded | 69fc1cac5497b3b670c88614 |
| Seppo | 21.97 | Not recorded | 69fc1cac5497b3b670c88614 |
| Online-DPO | 21.95 | Not recorded | 69fc1cac5497b3b670c88614 |
| SD2.1 | 21.77 | Not recorded | 6a1e6454dbd7150506f859f9 |
| RealAlign | 21.64 | Not recorded | 6a0d402af984739ed67378f8 |
| Diffusion-DPO | 21.58 | Not recorded | 6a0d402af984739ed67378f8 |
| Original | 21.53 | Not recorded | 69fc1cac5497b3b670c88614 |
| SD1.5 | 21.49 | Not recorded | 6a1e6454dbd7150506f859f9 |
| Online-SFT | 21.35 | Not recorded | 69fc1cac5497b3b670c88614 |
| SD-1.5 | 21.34 | Not recorded | 6a0d402af984739ed67378f8 |
| LCM-SD1.5 | 21.15 | Not recorded | 6a1e6454dbd7150506f859f9 |

### DRaFT

PickScore: 24.34. Result ID: 6a1edd93fbc373592c1b09d4.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6454dbd7150506f859f9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73c11e209131cea5e470.md>)

Recorded evaluation configuration: {"Inference Steps":"1","Reward Grad":"Yes"}. This record does not assert independent reproduction.

### VGG-Flow1-step

PickScore: 23.98. Result ID: 6a1edd93fbc373592c1b09d5.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6454dbd7150506f859f9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6932675fd6b0dd33429345a3.md>)

Recorded evaluation configuration: {"Inference Steps":"1","Reward Grad":"Yes"}. This record does not assert independent reproduction.

### DrPO

PickScore: 23.71. Result ID: 6a1edd93fbc373592c1b09d9.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6454dbd7150506f859f9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a1e6454dbd7150506f859f9.md>)

Recorded evaluation configuration: {"Inference Steps":"1","Reward Grad":"No"}. This record does not assert independent reproduction.

### Diff.-NPO

PickScore: 23.26. Result ID: 69fc4cdb47f34ce924defed7.

[Reporting paper](<https://www.sota2.com/research/paper/69fc1cac5497b3b670c88614.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69fc1cac5497b3b670c88614.md>)

Recorded evaluation configuration: {"Base Model":"SDXL"}. This record does not assert independent reproduction.

### Online-DPO

PickScore: 23.07. Result ID: 69fc4cdb47f34ce924defed6.

[Reporting paper](<https://www.sota2.com/research/paper/69fc1cac5497b3b670c88614.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa713d1e209131cea4ffe9.md>)

Recorded evaluation configuration: {"Base Model":"SDXL"}. This record does not assert independent reproduction.

### DRaFT

PickScore: 23.07. Result ID: 6a1edd93fbc373592c1b09e8.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6454dbd7150506f859f9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73c11e209131cea5e470.md>)

Recorded evaluation configuration: {"Inference Steps":"1","Reward Grad":"Yes"}. This record does not assert independent reproduction.

### VGG-Flow1-step

PickScore: 22.99. Result ID: 6a1edd93fbc373592c1b09e9.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6454dbd7150506f859f9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6932675fd6b0dd33429345a3.md>)

Recorded evaluation configuration: {"Inference Steps":"1","Reward Grad":"Yes"}. This record does not assert independent reproduction.

### DrPO

PickScore: 22.99. Result ID: 6a1edd93fbc373592c1b09ed.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6454dbd7150506f859f9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a1e6454dbd7150506f859f9.md>)

Recorded evaluation configuration: {"Inference Steps":"1","Reward Grad":"No"}. This record does not assert independent reproduction.

### SDXL-DPO

PickScore: 22.95. Result ID: 6a1edd93fbc373592c1b09d2.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6454dbd7150506f859f9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa713d1e209131cea4ffe9.md>)

Recorded evaluation configuration: {"Inference Steps":"50"}. This record does not assert independent reproduction.

### RealAlign

PickScore: 22.9. Result ID: 6a0d7c81fbc373592c05cb72.

[Reporting paper](<https://www.sota2.com/research/paper/6a0d402af984739ed67378f8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a0d402af984739ed67378f8.md>)

Recorded evaluation configuration: {"Base Model":"SD-3.5-M","Train Data":"512 pairs"}. This record does not assert independent reproduction.

### PSO

PickScore: 22.87. Result ID: 6a1edd93fbc373592c1b09d7.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6454dbd7150506f859f9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa713d1e209131cea4ffe9.md>)

Recorded evaluation configuration: {"Inference Steps":"1","Reward Grad":"No"}. This record does not assert independent reproduction.

### DPO1-step

PickScore: 22.85. Result ID: 6a1edd93fbc373592c1b09d6.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6454dbd7150506f859f9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74511e209131cea6105d.md>)

Recorded evaluation configuration: {"Inference Steps":"1","Reward Grad":"No"}. This record does not assert independent reproduction.

### GRPO1-step

PickScore: 22.8. Result ID: 6a1edd93fbc373592c1b09d8.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6454dbd7150506f859f9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73101e209131cea5a395.md>)

[Implementation](<https://github.com/deepseek-ai/DeepSeek-Math>)

Recorded evaluation configuration: {"Inference Steps":"1","Reward Grad":"No"}. This record does not assert independent reproduction.

### SDXL-Turbo (base)

PickScore: 22.77. Result ID: 6a1edd93fbc373592c1b09d3.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6454dbd7150506f859f9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71461e209131cea50558.md>)

Recorded evaluation configuration: {"Inference Steps":"1"}. This record does not assert independent reproduction.

### Diffusion-DPO

PickScore: 22.75. Result ID: 6a0d7c81fbc373592c05cb71.

[Reporting paper](<https://www.sota2.com/research/paper/6a0d402af984739ed67378f8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c521e209131cea8d8b4.md>)

Recorded evaluation configuration: {"Base Model":"SD-3.5-M","Train Data":"851k pairs","re-implementation":"true"}. This record does not assert independent reproduction.

### SDXL base

PickScore: 22.64. Result ID: 6a1edd93fbc373592c1b09d1.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6454dbd7150506f859f9.md>)

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

Recorded evaluation configuration: {"Inference Steps":"50"}. This record does not assert independent reproduction.

### SD-3.5-M

PickScore: 22.54. Result ID: 6a0d7c81fbc373592c05cb70.

[Reporting paper](<https://www.sota2.com/research/paper/6a0d402af984739ed67378f8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72621e209131cea56163.md>)

Recorded evaluation configuration: {"Base Model":"SD-3.5-M","Train Data":"/"}. This record does not assert independent reproduction.

### Original (SDXL)

PickScore: 22.51. Result ID: 69fc4cdb47f34ce924defed4.

[Reporting paper](<https://www.sota2.com/research/paper/69fc1cac5497b3b670c88614.md>)

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

Recorded evaluation configuration: {"Base Model":"SDXL"}. This record does not assert independent reproduction.

### DPO1-step

PickScore: 22.39. Result ID: 6a1edd93fbc373592c1b09ea.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6454dbd7150506f859f9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74511e209131cea6105d.md>)

Recorded evaluation configuration: {"Inference Steps":"1","Reward Grad":"No"}. This record does not assert independent reproduction.

### GRPO1-step

PickScore: 22.35. Result ID: 6a1edd93fbc373592c1b09ec.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6454dbd7150506f859f9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73101e209131cea5a395.md>)

[Implementation](<https://github.com/deepseek-ai/DeepSeek-Math>)

Recorded evaluation configuration: {"Inference Steps":"1","Reward Grad":"No"}. This record does not assert independent reproduction.

### SD-Turbo (base)

PickScore: 22.29. Result ID: 6a1edd93fbc373592c1b09e7.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6454dbd7150506f859f9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71461e209131cea50558.md>)

Recorded evaluation configuration: {"Inference Steps":"1","Reward Grad":"–"}. This record does not assert independent reproduction.

### PSO

PickScore: 22.29. Result ID: 6a1edd93fbc373592c1b09eb.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6454dbd7150506f859f9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa713d1e209131cea4ffe9.md>)

Recorded evaluation configuration: {"Inference Steps":"1","Reward Grad":"No"}. This record does not assert independent reproduction.

### Online-SFT

PickScore: 22.21. Result ID: 69fc4cdb47f34ce924defed5.

[Reporting paper](<https://www.sota2.com/research/paper/69fc1cac5497b3b670c88614.md>)

Recorded evaluation configuration: {"Base Model":"SDXL"}. This record does not assert independent reproduction.

### Diff.-NPO

PickScore: 22.07. Result ID: 69fc4cdb47f34ce924defee1.

[Reporting paper](<https://www.sota2.com/research/paper/69fc1cac5497b3b670c88614.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69fc1cac5497b3b670c88614.md>)

Recorded evaluation configuration: {"Backbone":"SD1.5"}. This record does not assert independent reproduction.

### SPIN

PickScore: 21.98. Result ID: 69fc4cdb47f34ce924defee0.

[Reporting paper](<https://www.sota2.com/research/paper/69fc1cac5497b3b670c88614.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72351e209131cea54edd.md>)

[Implementation](<https://huggingface.co/runwayml/stable-diffusion-v1-5>)

Recorded evaluation configuration: {"Backbone":"SD1.5"}. This record does not assert independent reproduction.

### Seppo

PickScore: 21.97. Result ID: 69fc4cdb47f34ce924defedf.

[Reporting paper](<https://www.sota2.com/research/paper/69fc1cac5497b3b670c88614.md>)

Recorded evaluation configuration: {"Backbone":"SD1.5"}. This record does not assert independent reproduction.

### Online-DPO

PickScore: 21.95. Result ID: 69fc4cdb47f34ce924defede.

[Reporting paper](<https://www.sota2.com/research/paper/69fc1cac5497b3b670c88614.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa713d1e209131cea4ffe9.md>)

Recorded evaluation configuration: {"Backbone":"SD1.5"}. This record does not assert independent reproduction.

### SD2.1

PickScore: 21.77. Result ID: 6a1edd93fbc373592c1b09e6.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6454dbd7150506f859f9.md>)

Recorded evaluation configuration: {"Inference Steps":"50","Reward Grad":"–"}. This record does not assert independent reproduction.

### RealAlign

PickScore: 21.64. Result ID: 6a0d7c81fbc373592c05cb6f.

[Reporting paper](<https://www.sota2.com/research/paper/6a0d402af984739ed67378f8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a0d402af984739ed67378f8.md>)

Recorded evaluation configuration: {"Base Model":"SD-1.5","Train Data":"512 pairs"}. This record does not assert independent reproduction.

### Diffusion-DPO

PickScore: 21.58. Result ID: 6a0d7c81fbc373592c05cb6e.

[Reporting paper](<https://www.sota2.com/research/paper/6a0d402af984739ed67378f8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c521e209131cea8d8b4.md>)

Recorded evaluation configuration: {"Base Model":"SD-1.5","Train Data":"851k pairs"}. This record does not assert independent reproduction.

### Original

PickScore: 21.53. Result ID: 69fc4cdb47f34ce924defedc.

[Reporting paper](<https://www.sota2.com/research/paper/69fc1cac5497b3b670c88614.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c391e209131cea32e4a.md>)

[Implementation](<https://github.com/CompVis/latent-diffusion>)

Recorded evaluation configuration: {"Backbone":"SD1.5"}. This record does not assert independent reproduction.

### SD1.5

PickScore: 21.49. Result ID: 6a1edd93fbc373592c1b09e4.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6454dbd7150506f859f9.md>)

Recorded evaluation configuration: {"Inference Steps":"50","Reward Grad":"–"}. This record does not assert independent reproduction.

### Online-SFT

PickScore: 21.35. Result ID: 69fc4cdb47f34ce924defedd.

[Reporting paper](<https://www.sota2.com/research/paper/69fc1cac5497b3b670c88614.md>)

Recorded evaluation configuration: {"Backbone":"SD1.5"}. This record does not assert independent reproduction.

### SD-1.5

PickScore: 21.34. Result ID: 6a0d7c81fbc373592c05cb6d.

[Reporting paper](<https://www.sota2.com/research/paper/6a0d402af984739ed67378f8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c391e209131cea32e4a.md>)

[Implementation](<https://github.com/CompVis/latent-diffusion>)

Recorded evaluation configuration: {"Base Model":"SD-1.5","Train Data":"/"}. This record does not assert independent reproduction.

### LCM-SD1.5

PickScore: 21.15. Result ID: 6a1edd93fbc373592c1b09e5.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e6454dbd7150506f859f9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa70c81e209131cea4d0a2.md>)

[Implementation](<https://latent-consistency-models.github.io/>)

Recorded evaluation configuration: {"Inference Steps":"4","Reward Grad":"–"}. This record does not assert independent reproduction.

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