# Visual Question Answering on TextVQA (F1, Hallucination Rate, Budget Usage)

Source: [Visual Question Answering on TextVQA (F1, Hallucination Rate, Budget Usage)](<https://www.sota2.com/research/sota/visual-question-answering-on-textvqa-f1-hallucination-rate-budget-usage>)

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

Data updated: 2026-03-04T09:24:22.816Z

Evaluation of model performance on the TextVQA dataset for Visual Question Answering, focusing on F1 score, Hallucination Rate, and average budget usage.

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 | F1 Score |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 5 of 5 |
| 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

- [F1 Score](<https://www.sota2.com/research/sota/visual-question-answering-on-textvqa-f1-hallucination-rate-budget-usage.md?metric=F1%20Score>): Higher is better; unit: PERCENT.

- [Hallucination Rate](<https://www.sota2.com/research/sota/visual-question-answering-on-textvqa-f1-hallucination-rate-budget-usage.md?metric=Hallucination%20Rate>): Lower is better; unit: PERCENT.

- [Average Budget Usage (B-bar)](<https://www.sota2.com/research/sota/visual-question-answering-on-textvqa-f1-hallucination-rate-budget-usage.md?metric=Average%20Budget%20Usage%20(B-bar)>): Lower is better; unit: OTHER.

## Recorded results

| Method | F1 Score | Result date | Report paper |
| --- | --- | --- | --- |
| PoP | 72.9 | Not recorded | 69a66bc12d603e6de3e3caa0 |
| ProgVLM | 69.7 | Not recorded | 69a66bc12d603e6de3e3caa0 |
| MM-ReAct | 69.2 | Not recorded | 69a66bc12d603e6de3e3caa0 |
| M-CoT | 66.9 | Not recorded | 69a66bc12d603e6de3e3caa0 |
| Direct | 64.7 | Not recorded | 69a66bc12d603e6de3e3caa0 |

### PoP

F1 Score: 72.9. Result ID: 69a7f9da6dbb7075b94d1520.

[Reporting paper](<https://www.sota2.com/research/paper/69a66bc12d603e6de3e3caa0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69a66bc12d603e6de3e3caa0.md>)

Recorded evaluation configuration: {"Reasoning protocol":"PoP"}. This record does not assert independent reproduction.

### ProgVLM

F1 Score: 69.7. Result ID: 69a7f9da6dbb7075b94d151f.

[Reporting paper](<https://www.sota2.com/research/paper/69a66bc12d603e6de3e3caa0.md>)

Recorded evaluation configuration: {"Reasoning protocol":"ProgVLM"}. This record does not assert independent reproduction.

### MM-ReAct

F1 Score: 69.2. Result ID: 69a7f9da6dbb7075b94d151e.

[Reporting paper](<https://www.sota2.com/research/paper/69a66bc12d603e6de3e3caa0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ed31e209131cea4272b.md>)

Recorded evaluation configuration: {"Reasoning protocol":"MM-ReAct"}. This record does not assert independent reproduction.

### M-CoT

F1 Score: 66.9. Result ID: 69a7f9da6dbb7075b94d151d.

[Reporting paper](<https://www.sota2.com/research/paper/69a66bc12d603e6de3e3caa0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73421e209131cea5b5b0.md>)

[Implementation](<https://github.com/amazon-science/mm-cot>)

Recorded evaluation configuration: {"Reasoning protocol":"M-CoT"}. This record does not assert independent reproduction.

### Direct

F1 Score: 64.7. Result ID: 69a7f9da6dbb7075b94d151c.

[Reporting paper](<https://www.sota2.com/research/paper/69a66bc12d603e6de3e3caa0.md>)

Recorded evaluation configuration: {"Reasoning protocol":"Direct"}. This record does not assert independent reproduction.

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