# Visual Question Answering on TextVQA with OCR hints (val)

Source: [Visual Question Answering on TextVQA with OCR hints (val)](<https://www.sota2.com/research/sota/visual-question-answering-on-textvqa-with-ocr-hints-val>)

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

Data updated: 2026-02-27T10:00:48.673Z

Evaluation of model performance on the validation split of the TextVQA dataset, specifically utilizing the provided OCR hints, for the Visual Question Answering task.

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 with OCR hints (val) |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 7 of 7 |
| 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

- [Accuracy](<https://www.sota2.com/research/sota/visual-question-answering-on-textvqa-with-ocr-hints-val.md?metric=Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| RADIOv2.5-H | 71.5 | Not recorded | 68fa76e01e209131cea6f909 |
| RADIOv2.5-H | 70.8 | Not recorded | 68fa76e01e209131cea6f909 |
| RADIOv2.5-H | 70.2 | Not recorded | 68fa76e01e209131cea6f909 |
| SigLIP-SO400M | 69.7 | Not recorded | 68fa76e01e209131cea6f909 |
| SigLIP-SO400M | 67.6 | Not recorded | 68fa76e01e209131cea6f909 |
| OpenAI-CLIP | 63.8 | Not recorded | 68fa76e01e209131cea6f909 |
| AM-RADIO-H | 55.9 | Not recorded | 68fa76e01e209131cea6f909 |

### RADIOv2.5-H

Accuracy: 71.5. Result ID: 69a16b3d35dfcabb00e6ec81.

[Reporting paper](<https://www.sota2.com/research/paper/68fa76e01e209131cea6f909.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76e01e209131cea6f909.md>)

Recorded evaluation configuration: {"Vision Encoder":"RADIOv2.5-H","LLM":"Qwen2-7B-Instruct","SFT Data Mixture":"9.8M samples","Tokens per image":"512"}. This record does not assert independent reproduction.

### RADIOv2.5-H

Accuracy: 70.8. Result ID: 69a16b3d35dfcabb00e6ece8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa76e01e209131cea6f909.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76e01e209131cea6f909.md>)

Recorded evaluation configuration: {"Resolution":"768²","Tokens per image":"196","LLM":"Qwen2-7B-Instruct","Training Data":"ShareGPT4v and VFLAN"}. This record does not assert independent reproduction.

### RADIOv2.5-H

Accuracy: 70.2. Result ID: 69a16b3d35dfcabb00e6ece9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa76e01e209131cea6f909.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76e01e209131cea6f909.md>)

Recorded evaluation configuration: {"Resolution":"768²","Tokens per image":"512","LLM":"Qwen2-7B-Instruct","Training Data":"ShareGPT4v and VFLAN"}. This record does not assert independent reproduction.

### SigLIP-SO400M

Accuracy: 69.7. Result ID: 69a16b3d35dfcabb00e6ec80.

[Reporting paper](<https://www.sota2.com/research/paper/68fa76e01e209131cea6f909.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa70ae1e209131cea4c7a4.md>)

Recorded evaluation configuration: {"Vision Encoder":"SigLIP-SO400M","LLM":"Qwen2-7B-Instruct","SFT Data Mixture":"9.8M samples","Tokens per image":"512"}. This record does not assert independent reproduction.

### SigLIP-SO400M

Accuracy: 67.6. Result ID: 69a16b3d35dfcabb00e6ece7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa76e01e209131cea6f909.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa70ae1e209131cea4c7a4.md>)

Recorded evaluation configuration: {"Resolution":"384²","Tokens per image":"196","LLM":"Qwen2-7B-Instruct","Training Data":"ShareGPT4v and VFLAN"}. This record does not assert independent reproduction.

### OpenAI-CLIP

Accuracy: 63.8. Result ID: 69a16b3d35dfcabb00e6ece5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa76e01e209131cea6f909.md>)

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

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

Recorded evaluation configuration: {"Resolution":"336²","Tokens per image":"144","LLM":"Qwen2-7B-Instruct","Training Data":"ShareGPT4v and VFLAN"}. This record does not assert independent reproduction.

### AM-RADIO-H

Accuracy: 55.9. Result ID: 69a16b3d35dfcabb00e6ece6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa76e01e209131cea6f909.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74541e209131cea611e9.md>)

[Implementation](<https://github.com/NVlabs/RADIO>)

Recorded evaluation configuration: {"Resolution":"512²","Tokens per image":"256","LLM":"Qwen2-7B-Instruct","Training Data":"ShareGPT4v and VFLAN"}. This record does not assert independent reproduction.

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