# Prompt-embedding Optimization on Parti Prompts (36)

Source: [Prompt-embedding Optimization on Parti Prompts (36)](<https://www.sota2.com/research/sota/prompt-embedding-optimization-on-parti-prompts-36>)

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

Data updated: 2026-04-14T07:37:52.588Z

Evaluation of prompt-embedding optimization techniques using the Parti Prompts (36) dataset, assessed via LAION Aesthetic V2 scores, CLIPScore, and Fitness 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 | Prompt-embedding Optimization |
| Dataset (including any stated split) | Parti Prompts (36) |
| Metric | LAION Aesthetic V2 Average Score |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 15 of 15 |
| 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

- [LAION Aesthetic V2 Average Score](<https://www.sota2.com/research/sota/prompt-embedding-optimization-on-parti-prompts-36.md?metric=LAION%20Aesthetic%20V2%20Average%20Score>): Higher is better; unit: SCALAR.

- [LAION Aesthetic V2 Standard Deviation](<https://www.sota2.com/research/sota/prompt-embedding-optimization-on-parti-prompts-36.md?metric=LAION%20Aesthetic%20V2%20Standard%20Deviation>): Lower is better; unit: SCALAR.

- [LAION Aesthetic V2 Maximum Score](<https://www.sota2.com/research/sota/prompt-embedding-optimization-on-parti-prompts-36.md?metric=LAION%20Aesthetic%20V2%20Maximum%20Score>): Higher is better; unit: SCALAR.

- [LAION Aesthetic V2 Change vs Base (%)](<https://www.sota2.com/research/sota/prompt-embedding-optimization-on-parti-prompts-36.md?metric=LAION%20Aesthetic%20V2%20Change%20vs%20Base%20(%25)>): Higher is better; unit: SCALAR.

- [CLIPScore Average](<https://www.sota2.com/research/sota/prompt-embedding-optimization-on-parti-prompts-36.md?metric=CLIPScore%20Average>): Higher is better; unit: SCALAR.

- [CLIPScore Standard Deviation](<https://www.sota2.com/research/sota/prompt-embedding-optimization-on-parti-prompts-36.md?metric=CLIPScore%20Standard%20Deviation>): Lower is better; unit: SCALAR.

- [CLIPScore Maximum](<https://www.sota2.com/research/sota/prompt-embedding-optimization-on-parti-prompts-36.md?metric=CLIPScore%20Maximum>): Higher is better; unit: SCALAR.

- [CLIPScore Change vs Base (%)](<https://www.sota2.com/research/sota/prompt-embedding-optimization-on-parti-prompts-36.md?metric=CLIPScore%20Change%20vs%20Base%20(%25)>): Higher is better; unit: SCALAR.

- [Fitness Average Score](<https://www.sota2.com/research/sota/prompt-embedding-optimization-on-parti-prompts-36.md?metric=Fitness%20Average%20Score>): Higher is better; unit: SCALAR.

- [Fitness Standard Deviation](<https://www.sota2.com/research/sota/prompt-embedding-optimization-on-parti-prompts-36.md?metric=Fitness%20Standard%20Deviation>): Lower is better; unit: SCALAR.

- [Fitness Maximum Score](<https://www.sota2.com/research/sota/prompt-embedding-optimization-on-parti-prompts-36.md?metric=Fitness%20Maximum%20Score>): Higher is better; unit: SCALAR.

- [Fitness Change vs Base (%)](<https://www.sota2.com/research/sota/prompt-embedding-optimization-on-parti-prompts-36.md?metric=Fitness%20Change%20vs%20Base%20(%25)>): Higher is better; unit: SCALAR.

- [Wins Count](<https://www.sota2.com/research/sota/prompt-embedding-optimization-on-parti-prompts-36.md?metric=Wins%20Count>): Higher is better; unit: SCALAR.

## Recorded results

| Method | LAION Aesthetic V2 Average Score | Result date | Report paper |
| --- | --- | --- | --- |
| sep-CMA-ES | 8.32 | Not recorded | 690d7d0a2828b7cc6937ce16 |
| GA Empty | 7.45 | Not recorded | 69ddc9e07d0172140ec797b6 |
| GA Random | 7.39 | Not recorded | 69ddc9e07d0172140ec797b6 |
| GA Mutated | 7.3 | Not recorded | 69ddc9e07d0172140ec797b6 |
| sep-CMA-ES | 7.27 | Not recorded | 690d7d0a2828b7cc6937ce16 |
| Adam | 7.12 | Not recorded | 690d7d0a2828b7cc6937ce16 |
| Random Search | 6.93 | Not recorded | 69ddc9e07d0172140ec797b6 |
| Promptist | 6.43 | Not recorded | 69ddc9e07d0172140ec797b6 |
| Adam | 6.16 | Not recorded | 690d7d0a2828b7cc6937ce16 |
| SDXL Turbo (no optimization) | 5.78 | Not recorded | 69ddc9e07d0172140ec797b6 |
| SDXL Turbo (no optimization) | 5.75 | Not recorded | 690d7d0a2828b7cc6937ce16 |
| SDXL Turbo (no optimization) | 5.75 | Not recorded | 690d7d0a2828b7cc6937ce16 |
| SDXL Turbo (no optimization) | 5.75 | Not recorded | 690d7d0a2828b7cc6937ce16 |
| Adam | 5.7 | Not recorded | 690d7d0a2828b7cc6937ce16 |
| sep-CMA-ES | 5.62 | Not recorded | 690d7d0a2828b7cc6937ce16 |

### sep-CMA-ES

LAION Aesthetic V2 Average Score: 8.32. Result ID: 69dc997e47f34ce924c3322d.

[Reporting paper](<https://www.sota2.com/research/paper/690d7d0a2828b7cc6937ce16.md>)

[Method origin paper](<https://www.sota2.com/research/paper/690d7d0a2828b7cc6937ce16.md>)

Recorded evaluation configuration: {"a":"1","b":"0"}. This record does not assert independent reproduction.

### GA Empty

LAION Aesthetic V2 Average Score: 7.45. Result ID: 69ddeeb547f34ce924c50a11.

[Reporting paper](<https://www.sota2.com/research/paper/69ddc9e07d0172140ec797b6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ddc9e07d0172140ec797b6.md>)

Recorded evaluation configuration: {"Algorithm":"Evolutionary Token-Level Prompt Optimization","Population initialization":"Empty"}. This record does not assert independent reproduction.

### GA Random

LAION Aesthetic V2 Average Score: 7.39. Result ID: 69ddeeb547f34ce924c50a12.

[Reporting paper](<https://www.sota2.com/research/paper/69ddc9e07d0172140ec797b6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ddc9e07d0172140ec797b6.md>)

Recorded evaluation configuration: {"Algorithm":"Evolutionary Token-Level Prompt Optimization","Population initialization":"Random"}. This record does not assert independent reproduction.

### GA Mutated

LAION Aesthetic V2 Average Score: 7.3. Result ID: 69ddeeb547f34ce924c50a10.

[Reporting paper](<https://www.sota2.com/research/paper/69ddc9e07d0172140ec797b6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ddc9e07d0172140ec797b6.md>)

Recorded evaluation configuration: {"Algorithm":"Evolutionary Token-Level Prompt Optimization","Population initialization":"Mutated"}. This record does not assert independent reproduction.

### sep-CMA-ES

LAION Aesthetic V2 Average Score: 7.27. Result ID: 69dc997e47f34ce924c33230.

[Reporting paper](<https://www.sota2.com/research/paper/690d7d0a2828b7cc6937ce16.md>)

[Method origin paper](<https://www.sota2.com/research/paper/690d7d0a2828b7cc6937ce16.md>)

Recorded evaluation configuration: {"a":"0.5","b":"0.5"}. This record does not assert independent reproduction.

### Adam

LAION Aesthetic V2 Average Score: 7.12. Result ID: 69dc997e47f34ce924c3322c.

[Reporting paper](<https://www.sota2.com/research/paper/690d7d0a2828b7cc6937ce16.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63171e209131ce9fca0b.md>)

Recorded evaluation configuration: {"a":"1","b":"0"}. This record does not assert independent reproduction.

### Random Search

LAION Aesthetic V2 Average Score: 6.93. Result ID: 69ddeeb547f34ce924c50a13.

[Reporting paper](<https://www.sota2.com/research/paper/69ddc9e07d0172140ec797b6.md>)

Recorded evaluation configuration: {"Algorithm":"Random Search"}. This record does not assert independent reproduction.

### Promptist

LAION Aesthetic V2 Average Score: 6.43. Result ID: 69ddeeb547f34ce924c50a14.

[Reporting paper](<https://www.sota2.com/research/paper/69ddc9e07d0172140ec797b6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71a41e209131cea523e3.md>)

Recorded evaluation configuration: {"Algorithm":"Promptist"}. This record does not assert independent reproduction.

### Adam

LAION Aesthetic V2 Average Score: 6.16. Result ID: 69dc997e47f34ce924c3322f.

[Reporting paper](<https://www.sota2.com/research/paper/690d7d0a2828b7cc6937ce16.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63171e209131ce9fca0b.md>)

Recorded evaluation configuration: {"a":"0.5","b":"0.5"}. This record does not assert independent reproduction.

### SDXL Turbo (no optimization)

LAION Aesthetic V2 Average Score: 5.78. Result ID: 69ddeeb547f34ce924c50a0f.

[Reporting paper](<https://www.sota2.com/research/paper/69ddc9e07d0172140ec797b6.md>)

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

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

### SDXL Turbo (no optimization)

LAION Aesthetic V2 Average Score: 5.75. Result ID: 69dc997e47f34ce924c3322b.

[Reporting paper](<https://www.sota2.com/research/paper/690d7d0a2828b7cc6937ce16.md>)

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

Recorded evaluation configuration: {"a":"1","b":"0"}. This record does not assert independent reproduction.

### SDXL Turbo (no optimization)

LAION Aesthetic V2 Average Score: 5.75. Result ID: 69dc997e47f34ce924c3322e.

[Reporting paper](<https://www.sota2.com/research/paper/690d7d0a2828b7cc6937ce16.md>)

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

Recorded evaluation configuration: {"a":"0.5","b":"0.5"}. This record does not assert independent reproduction.

### SDXL Turbo (no optimization)

LAION Aesthetic V2 Average Score: 5.75. Result ID: 69dc997e47f34ce924c33231.

[Reporting paper](<https://www.sota2.com/research/paper/690d7d0a2828b7cc6937ce16.md>)

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

Recorded evaluation configuration: {"a":"0","b":"1"}. This record does not assert independent reproduction.

### Adam

LAION Aesthetic V2 Average Score: 5.7. Result ID: 69dc997e47f34ce924c33232.

[Reporting paper](<https://www.sota2.com/research/paper/690d7d0a2828b7cc6937ce16.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63171e209131ce9fca0b.md>)

Recorded evaluation configuration: {"a":"0","b":"1"}. This record does not assert independent reproduction.

### sep-CMA-ES

LAION Aesthetic V2 Average Score: 5.62. Result ID: 69dc997e47f34ce924c33233.

[Reporting paper](<https://www.sota2.com/research/paper/690d7d0a2828b7cc6937ce16.md>)

[Method origin paper](<https://www.sota2.com/research/paper/690d7d0a2828b7cc6937ce16.md>)

Recorded evaluation configuration: {"a":"0","b":"1"}. This record does not assert independent reproduction.

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