# Image + Text to Text Generation on ImageNet (val)

Source: [Image + Text to Text Generation on ImageNet (val)](<https://www.sota2.com/research/sota/image-plus-text-to-text-generation-on-imagenet-val>)

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

Data updated: 2026-02-26T05:38:11.602Z

Evaluation of models performing multimodal input processing (Image and Text) to generate textual output, using the ImageNet validation set.

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 + Text to Text Generation |
| Dataset (including any stated split) | ImageNet (val) |
| Metric | CD Score |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 4 of 4 |
| 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

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

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

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

- [White Image Error Rate](<https://www.sota2.com/research/sota/image-plus-text-to-text-generation-on-imagenet-val.md?metric=White%20Image%20Error%20Rate>): Lower is better; unit: PERCENT.

## Recorded results

| Method | CD Score | Result date | Report paper |
| --- | --- | --- | --- |
| CB-SAE | 0.291 | Not recorded | 693ba17b6cfd488761a551b7 |
| CB-SAE | 0.244 | Not recorded | 693ba17b6cfd488761a551b7 |
| SAE | 0.194 | Not recorded | 693ba17b6cfd488761a551b7 |
| SAE | 0.154 | Not recorded | 693ba17b6cfd488761a551b7 |

### CB-SAE

CD Score: 0.291. Result ID: 69996d905698665ec4db63d8.

[Reporting paper](<https://www.sota2.com/research/paper/693ba17b6cfd488761a551b7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/693ba17b6cfd488761a551b7.md>)

Recorded evaluation configuration: {"Steered Model":"LLaVA-MORE (DINOv2-L + Gemma2-9B)"}. This record does not assert independent reproduction.

### CB-SAE

CD Score: 0.244. Result ID: 69996d905698665ec4db63d6.

[Reporting paper](<https://www.sota2.com/research/paper/693ba17b6cfd488761a551b7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/693ba17b6cfd488761a551b7.md>)

Recorded evaluation configuration: {"Steered Model":"LLaVA-1.5-7B (CLIP-ViT-L + Vicuna-7B)"}. This record does not assert independent reproduction.

### SAE

CD Score: 0.194. Result ID: 69996d905698665ec4db63d7.

[Reporting paper](<https://www.sota2.com/research/paper/693ba17b6cfd488761a551b7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79291e209131cea7c552.md>)

Recorded evaluation configuration: {"Steered Model":"LLaVA-MORE (DINOv2-L + Gemma2-9B)"}. This record does not assert independent reproduction.

### SAE

CD Score: 0.154. Result ID: 69996d905698665ec4db63d5.

[Reporting paper](<https://www.sota2.com/research/paper/693ba17b6cfd488761a551b7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79291e209131cea7c552.md>)

Recorded evaluation configuration: {"Steered Model":"LLaVA-1.5-7B (CLIP-ViT-L + Vicuna-7B)"}. This record does not assert independent reproduction.

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