# Personalized Understanding on OmniPBench

Source: [Personalized Understanding on OmniPBench](<https://www.sota2.com/research/sota/personalized-understanding-on-omnipbench>)

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

Data updated: 2026-02-26T06:01:10.824Z

Evaluation of Personalized Understanding capabilities using the OmniPBench dataset, reporting metrics such as Rec. Weight, VQA BLEU, and QA GPT.

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 | Personalized Understanding |
| Dataset (including any stated split) | OmniPBench |
| Metric | Rec Weight |
| Unit | SCALAR |
| Direction | Higher 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

- [Rec Weight](<https://www.sota2.com/research/sota/personalized-understanding-on-omnipbench.md?metric=Rec%20Weight>): Higher is better; unit: SCALAR.

- [VQA BLEU](<https://www.sota2.com/research/sota/personalized-understanding-on-omnipbench.md?metric=VQA%20BLEU>): Higher is better; unit: PERCENT.

- [VQA GPT Score](<https://www.sota2.com/research/sota/personalized-understanding-on-omnipbench.md?metric=VQA%20GPT%20Score>): Higher is better; unit: PERCENT.

- [QA BLEU](<https://www.sota2.com/research/sota/personalized-understanding-on-omnipbench.md?metric=QA%20BLEU>): Higher is better; unit: PERCENT.

- [QA GPT Score](<https://www.sota2.com/research/sota/personalized-understanding-on-omnipbench.md?metric=QA%20GPT%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Rec Weight | Result date | Report paper |
| --- | --- | --- | --- |
| RAP-MLLM | 0.94 | Not recorded | 6965d1ac6343fa78c832383e |
| MC-LLaVA | 0.924 | Not recorded | 6965d1ac6343fa78c832383e |
| Yo'LLaVA | 0.919 | Not recorded | 6965d1ac6343fa78c832383e |
| OmniPersona | 0.852 | Not recorded | 6965d1ac6343fa78c832383e |
| Unictoken | 0.79 | Not recorded | 6965d1ac6343fa78c832383e |
| Bagel+TP | 0.788 | Not recorded | 6965d1ac6343fa78c832383e |
| GPT-4o+IP | 0.773 | Not recorded | 6965d1ac6343fa78c832383e |
| Yo'LLaVA(Phi-1.5) | 0.765 | Not recorded | 6965d1ac6343fa78c832383e |
| Yo'Chameleon | 0.764 | Not recorded | 6965d1ac6343fa78c832383e |
| GPT-4o+TP | 0.742 | Not recorded | 6965d1ac6343fa78c832383e |
| Chameleon+TP | 0.69 | Not recorded | 6965d1ac6343fa78c832383e |
| Qwen2.5-VL + TP | 0.66 | Not recorded | 6965d1ac6343fa78c832383e |
| Show-o+TP | 0.566 | Not recorded | 6965d1ac6343fa78c832383e |
| Chameleon+IP | 0.493 | Not recorded | 6965d1ac6343fa78c832383e |

### RAP-MLLM

Rec Weight: 0.94. Result ID: 6997cca55698665ec423e586.

[Reporting paper](<https://www.sota2.com/research/paper/6965d1ac6343fa78c832383e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77ba1e209131cea74fc1.md>)

Recorded evaluation configuration: {"Token":"~1,000","Training Images":null}. This record does not assert independent reproduction.

### MC-LLaVA

Rec Weight: 0.924. Result ID: 6997cca55698665ec423e585.

[Reporting paper](<https://www.sota2.com/research/paper/6965d1ac6343fa78c832383e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77a01e209131cea7475f.md>)

Recorded evaluation configuration: {"Token":"16","Training Images":"~10"}. This record does not assert independent reproduction.

### Yo'LLaVA

Rec Weight: 0.919. Result ID: 6997cca55698665ec423e584.

[Reporting paper](<https://www.sota2.com/research/paper/6965d1ac6343fa78c832383e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75fa1e209131cea6ac29.md>)

[Implementation](<https://thaoshibe.github.io/YoLLaVA/>)

Recorded evaluation configuration: {"Token":"16","Training Images":"~100"}. This record does not assert independent reproduction.

### OmniPersona

Rec Weight: 0.852. Result ID: 6997cca55698665ec423e58f.

[Reporting paper](<https://www.sota2.com/research/paper/6965d1ac6343fa78c832383e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6965d1ac6343fa78c832383e.md>)

Recorded evaluation configuration: {"Token":"32","Training Images":"~10"}. This record does not assert independent reproduction.

### Unictoken

Rec Weight: 0.79. Result ID: 6997cca55698665ec423e58e.

[Reporting paper](<https://www.sota2.com/research/paper/6965d1ac6343fa78c832383e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c401e209131cea8d27b.md>)

Recorded evaluation configuration: {"Token":"32","Training Images":"~10"}. This record does not assert independent reproduction.

### Bagel+TP

Rec Weight: 0.788. Result ID: 6997cca55698665ec423e58c.

[Reporting paper](<https://www.sota2.com/research/paper/6965d1ac6343fa78c832383e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a311e209131cea81aad.md>)

[Implementation](<https://bagel-ai.org/>)

Recorded evaluation configuration: {"Token":"~100","Training Images":null}. This record does not assert independent reproduction.

### GPT-4o+IP

Rec Weight: 0.773. Result ID: 6997cca55698665ec423e583.

[Reporting paper](<https://www.sota2.com/research/paper/6965d1ac6343fa78c832383e.md>)

Recorded evaluation configuration: {"Token":"~1,000","Training Images":null}. This record does not assert independent reproduction.

### Yo'LLaVA(Phi-1.5)

Rec Weight: 0.765. Result ID: 6997cca55698665ec423e588.

[Reporting paper](<https://www.sota2.com/research/paper/6965d1ac6343fa78c832383e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75fa1e209131cea6ac29.md>)

[Implementation](<https://thaoshibe.github.io/YoLLaVA/>)

Recorded evaluation configuration: {"Token":"16","Training Images":"~100"}. This record does not assert independent reproduction.

### Yo'Chameleon

Rec Weight: 0.764. Result ID: 6997cca55698665ec423e58d.

[Reporting paper](<https://www.sota2.com/research/paper/6965d1ac6343fa78c832383e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa785a1e209131cea77ea1.md>)

[Implementation](<https://thaoshibe.github.io/YoChameleon>)

Recorded evaluation configuration: {"Token":"32","Training Images":"~1,000"}. This record does not assert independent reproduction.

### GPT-4o+TP

Rec Weight: 0.742. Result ID: 6997cca55698665ec423e582.

[Reporting paper](<https://www.sota2.com/research/paper/6965d1ac6343fa78c832383e.md>)

Recorded evaluation configuration: {"Token":"~100","Training Images":null}. This record does not assert independent reproduction.

### Chameleon+TP

Rec Weight: 0.69. Result ID: 6997cca55698665ec423e589.

[Reporting paper](<https://www.sota2.com/research/paper/6965d1ac6343fa78c832383e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77971e209131cea74348.md>)

Recorded evaluation configuration: {"Token":"~100","Training Images":null}. This record does not assert independent reproduction.

### Qwen2.5-VL + TP

Rec Weight: 0.66. Result ID: 6997cca55698665ec423e587.

[Reporting paper](<https://www.sota2.com/research/paper/6965d1ac6343fa78c832383e.md>)

Recorded evaluation configuration: {"Token":"~100","Training Images":null}. This record does not assert independent reproduction.

### Show-o+TP

Rec Weight: 0.566. Result ID: 6997cca55698665ec423e58b.

[Reporting paper](<https://www.sota2.com/research/paper/6965d1ac6343fa78c832383e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7b761e209131cea88a1f.md>)

Recorded evaluation configuration: {"Token":"~100","Training Images":null}. This record does not assert independent reproduction.

### Chameleon+IP

Rec Weight: 0.493. Result ID: 6997cca55698665ec423e58a.

[Reporting paper](<https://www.sota2.com/research/paper/6965d1ac6343fa78c832383e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77971e209131cea74348.md>)

Recorded evaluation configuration: {"Token":"~1,000","Training Images":null}. This record does not assert independent reproduction.

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