# Multi-modal Representation Learning on MMEB OOD 1.0

Source: [Multi-modal Representation Learning on MMEB OOD 1.0](<https://www.sota2.com/research/sota/multi-modal-representation-learning-on-mmeb-ood-1-0>)

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

Data updated: 2026-02-26T05:54:05.720Z

Evaluation of model performance on the MMEB OOD 1.0 benchmark, focusing on multi-modal representation learning capabilities measured by Out-of-Distribution Precision@1 scores.

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 | Multi-modal Representation Learning |
| Dataset (including any stated split) | MMEB OOD 1.0 |
| Metric | OOD Precision@1 |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 18 of 18 |
| 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

- [OOD Precision@1](<https://www.sota2.com/research/sota/multi-modal-representation-learning-on-mmeb-ood-1-0.md?metric=OOD%20Precision%401>): Higher is better; unit: PERCENT.

## Recorded results

| Method | OOD Precision@1 | Result date | Report paper |
| --- | --- | --- | --- |
| MegaPairs | 59.1 | Not recorded | 698c0cecc97a0b3e3b5e103f |
| SDE | 58.1 | Not recorded | 698c0cecc97a0b3e3b5e103f |
| UniME | 57.9 | Not recorded | 698c0cecc97a0b3e3b5e103f |
| VLM2Vec | 57.1 | Not recorded | 698c0cecc97a0b3e3b5e103f |
| UniME | 52.7 | Not recorded | 698c0cecc97a0b3e3b5e103f |
| SDE | 52.7 | Not recorded | 698c0cecc97a0b3e3b5e103f |
| VLM2Vec | 52.6 | Not recorded | 698c0cecc97a0b3e3b5e103f |
| VLM2Vec | 52 | Not recorded | 698c0cecc97a0b3e3b5e103f |
| SDE | 51.6 | Not recorded | 698c0cecc97a0b3e3b5e103f |
| SDE | 49.8 | Not recorded | 698c0cecc97a0b3e3b5e103f |
| VLM2Vec | 47.5 | Not recorded | 698c0cecc97a0b3e3b5e103f |
| OpenCLIP-FFT | 43.1 | Not recorded | 698c0cecc97a0b3e3b5e103f |
| CLIP-FFT | 42.8 | Not recorded | 698c0cecc97a0b3e3b5e103f |
| UniIR | 41.7 | Not recorded | 698c0cecc97a0b3e3b5e103f |
| UniIR | 40.4 | Not recorded | 698c0cecc97a0b3e3b5e103f |
| CLIP | 38.7 | Not recorded | 698c0cecc97a0b3e3b5e103f |
| BLIP2 | 25.1 | Not recorded | 698c0cecc97a0b3e3b5e103f |
| Magiclens | 23.7 | Not recorded | 698c0cecc97a0b3e3b5e103f |

### MegaPairs

OOD Precision@1: 59.1. Result ID: 6998daf95698665ec43f0188.

[Reporting paper](<https://www.sota2.com/research/paper/698c0cecc97a0b3e3b5e103f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76481e209131cea6c211.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"Fine-tuning","Model Backbone":"LLaVA-1.6"}. This record does not assert independent reproduction.

### SDE

OOD Precision@1: 58.1. Result ID: 6998daf95698665ec43f0190.

[Reporting paper](<https://www.sota2.com/research/paper/698c0cecc97a0b3e3b5e103f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/698c0cecc97a0b3e3b5e103f.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"Fine-tuning","Model Backbone":"LLaVA-1.6","Resolution":"1344x1344 (HR)"}. This record does not assert independent reproduction.

### UniME

OOD Precision@1: 57.9. Result ID: 6998daf95698665ec43f0187.

[Reporting paper](<https://www.sota2.com/research/paper/698c0cecc97a0b3e3b5e103f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a711e209131cea83422.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"Fine-tuning","Model Backbone":"LLaVA-1.6"}. This record does not assert independent reproduction.

### VLM2Vec

OOD Precision@1: 57.1. Result ID: 6998daf95698665ec43f018f.

[Reporting paper](<https://www.sota2.com/research/paper/698c0cecc97a0b3e3b5e103f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa766b1e209131cea6cd6f.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"Fine-tuning","Model Backbone":"LLaVA-1.6","Resolution":"1344x1344 (HR)"}. This record does not assert independent reproduction.

### UniME

OOD Precision@1: 52.7. Result ID: 6998daf95698665ec43f0186.

[Reporting paper](<https://www.sota2.com/research/paper/698c0cecc97a0b3e3b5e103f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a711e209131cea83422.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"Fine-tuning","Model Backbone":"Phi3.5-V"}. This record does not assert independent reproduction.

### SDE

OOD Precision@1: 52.7. Result ID: 6998daf95698665ec43f018c.

[Reporting paper](<https://www.sota2.com/research/paper/698c0cecc97a0b3e3b5e103f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/698c0cecc97a0b3e3b5e103f.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"Fine-tuning","Model Backbone":"Qwen2-VL-2B"}. This record does not assert independent reproduction.

### VLM2Vec

OOD Precision@1: 52.6. Result ID: 6998daf95698665ec43f018b.

[Reporting paper](<https://www.sota2.com/research/paper/698c0cecc97a0b3e3b5e103f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa766b1e209131cea6cd6f.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"Fine-tuning","Model Backbone":"Qwen2-VL-2B"}. This record does not assert independent reproduction.

### VLM2Vec

OOD Precision@1: 52. Result ID: 6998daf95698665ec43f0189.

[Reporting paper](<https://www.sota2.com/research/paper/698c0cecc97a0b3e3b5e103f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa766b1e209131cea6cd6f.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"Fine-tuning","Model Backbone":"Phi3.5-V"}. This record does not assert independent reproduction.

### SDE

OOD Precision@1: 51.6. Result ID: 6998daf95698665ec43f018a.

[Reporting paper](<https://www.sota2.com/research/paper/698c0cecc97a0b3e3b5e103f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/698c0cecc97a0b3e3b5e103f.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"Fine-tuning","Model Backbone":"Phi3.5-V"}. This record does not assert independent reproduction.

### SDE

OOD Precision@1: 49.8. Result ID: 6998daf95698665ec43f018e.

[Reporting paper](<https://www.sota2.com/research/paper/698c0cecc97a0b3e3b5e103f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/698c0cecc97a0b3e3b5e103f.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"Fine-tuning","Model Backbone":"LLaVA-1.6","Resolution":"336x336 (LR)"}. This record does not assert independent reproduction.

### VLM2Vec

OOD Precision@1: 47.5. Result ID: 6998daf95698665ec43f018d.

[Reporting paper](<https://www.sota2.com/research/paper/698c0cecc97a0b3e3b5e103f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa766b1e209131cea6cd6f.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"Fine-tuning","Model Backbone":"LLaVA-1.6","Resolution":"336x336 (LR)"}. This record does not assert independent reproduction.

### OpenCLIP-FFT

OOD Precision@1: 43.1. Result ID: 6998daf95698665ec43f0185.

[Reporting paper](<https://www.sota2.com/research/paper/698c0cecc97a0b3e3b5e103f.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"Fine-tuning"}. This record does not assert independent reproduction.

### CLIP-FFT

OOD Precision@1: 42.8. Result ID: 6998daf95698665ec43f0184.

[Reporting paper](<https://www.sota2.com/research/paper/698c0cecc97a0b3e3b5e103f.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"Fine-tuning"}. This record does not assert independent reproduction.

### UniIR

OOD Precision@1: 41.7. Result ID: 6998daf95698665ec43f0182.

[Reporting paper](<https://www.sota2.com/research/paper/698c0cecc97a0b3e3b5e103f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71471e209131cea505a6.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"Zero-shot","Fusion Strategy":"score-level","Backbone":"CLIP"}. This record does not assert independent reproduction.

### UniIR

OOD Precision@1: 40.4. Result ID: 6998daf95698665ec43f0181.

[Reporting paper](<https://www.sota2.com/research/paper/698c0cecc97a0b3e3b5e103f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71471e209131cea505a6.md>)

Recorded evaluation configuration: {"Evaluation Protocol":"Zero-shot","Fusion Strategy":"feature-level","Backbone":"BLIP"}. This record does not assert independent reproduction.

### CLIP

OOD Precision@1: 38.7. Result ID: 6998daf95698665ec43f017f.

[Reporting paper](<https://www.sota2.com/research/paper/698c0cecc97a0b3e3b5e103f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69581e209131cea21ad0.md>)

[Implementation](<https://github.com/OpenAI/CLIP>)

Recorded evaluation configuration: {"Evaluation Protocol":"Zero-shot"}. This record does not assert independent reproduction.

### BLIP2

OOD Precision@1: 25.1. Result ID: 6998daf95698665ec43f0180.

[Reporting paper](<https://www.sota2.com/research/paper/698c0cecc97a0b3e3b5e103f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6fb71e209131cea46fe6.md>)

[Implementation](<https://github.com/salesforce/LAVIS/tree/main/projects/blip2>)

Recorded evaluation configuration: {"Evaluation Protocol":"Zero-shot"}. This record does not assert independent reproduction.

### Magiclens

OOD Precision@1: 23.7. Result ID: 6998daf95698665ec43f0183.

[Reporting paper](<https://www.sota2.com/research/paper/698c0cecc97a0b3e3b5e103f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73e01e209131cea5e9da.md>)

[Implementation](<https://open-vision-language.github.io/MagicLens/>)

Recorded evaluation configuration: {"Evaluation Protocol":"Zero-shot"}. This record does not assert independent reproduction.

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