# Visual Question Answering on TextVQA (UPR, CAE, CSG)

Source: [Visual Question Answering on TextVQA (UPR, CAE, CSG)](<https://www.sota2.com/research/sota/visual-question-answering-on-textvqa-upr-cae-csg>)

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

Data updated: 2026-02-26T06:07:50.490Z

Evaluation of model performance on the TextVQA dataset for Visual Question Answering, utilizing UPR, CAE, and CSG 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 | Visual Question Answering |
| Dataset (including any stated split) | TextVQA |
| Metric | UPR |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 6 of 6 |
| 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

- [UPR](<https://www.sota2.com/research/sota/visual-question-answering-on-textvqa-upr-cae-csg.md?metric=UPR>): Higher is better; unit: PERCENT.

- [CAE](<https://www.sota2.com/research/sota/visual-question-answering-on-textvqa-upr-cae-csg.md?metric=CAE>): Higher is better; unit: PERCENT.

- [CSG](<https://www.sota2.com/research/sota/visual-question-answering-on-textvqa-upr-cae-csg.md?metric=CSG>): Higher is better; unit: SCALAR.

## Recorded results

| Method | UPR | Result date | Report paper |
| --- | --- | --- | --- |
| RayS | 97.94 | Not recorded | 69705d5de25ad2fa9d1ece20 |
| T-CAA | 92.56 | Not recorded | 69705d5de25ad2fa9d1ece20 |
| T-CAA | 87.98 | Not recorded | 69705d5de25ad2fa9d1ece20 |
| HSJA | 71.46 | Not recorded | 69705d5de25ad2fa9d1ece20 |
| RayS | 66.89 | Not recorded | 69705d5de25ad2fa9d1ece20 |
| HSJA | 57.64 | Not recorded | 69705d5de25ad2fa9d1ece20 |

### RayS

UPR: 97.94. Result ID: 6996ec5f5698665ec438e3a3.

[Reporting paper](<https://www.sota2.com/research/paper/69705d5de25ad2fa9d1ece20.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa68461e209131cea1af79.md>)

Recorded evaluation configuration: {"Target Model":"Qwen-VL"}. This record does not assert independent reproduction.

### T-CAA

UPR: 92.56. Result ID: 6996ec5f5698665ec438e3a1.

[Reporting paper](<https://www.sota2.com/research/paper/69705d5de25ad2fa9d1ece20.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69705d5de25ad2fa9d1ece20.md>)

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

### T-CAA

UPR: 87.98. Result ID: 6996ec5f5698665ec438e3a4.

[Reporting paper](<https://www.sota2.com/research/paper/69705d5de25ad2fa9d1ece20.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69705d5de25ad2fa9d1ece20.md>)

Recorded evaluation configuration: {"Target Model":"Qwen-VL"}. This record does not assert independent reproduction.

### HSJA

UPR: 71.46. Result ID: 6996ec5f5698665ec438e39f.

[Reporting paper](<https://www.sota2.com/research/paper/69705d5de25ad2fa9d1ece20.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa56db1e209131ce9a05a4.md>)

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

### RayS

UPR: 66.89. Result ID: 6996ec5f5698665ec438e3a0.

[Reporting paper](<https://www.sota2.com/research/paper/69705d5de25ad2fa9d1ece20.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa68461e209131cea1af79.md>)

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

### HSJA

UPR: 57.64. Result ID: 6996ec5f5698665ec438e3a2.

[Reporting paper](<https://www.sota2.com/research/paper/69705d5de25ad2fa9d1ece20.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa56db1e209131ce9a05a4.md>)

Recorded evaluation configuration: {"Target Model":"Qwen-VL"}. This record does not assert independent reproduction.

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