# Visual Question Answering on LUCID-VQA (test)

Source: [Visual Question Answering on LUCID-VQA (test)](<https://www.sota2.com/research/sota/visual-question-answering-on-lucid-vqa-test>)

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

Data updated: 2026-03-27T12:28:22.608Z

Evaluation of model performance on the test split of the LUCID-VQA dataset for visual question answering, utilizing GPT and Gemini based judges for detailed, conversational, and reasoning 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 | Visual Question Answering |
| Dataset (including any stated split) | LUCID-VQA (test) |
| Metric | GPT Judge Detailed Score |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 3 of 3 |
| 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

- [GPT Judge Detailed Score](<https://www.sota2.com/research/sota/visual-question-answering-on-lucid-vqa-test.md?metric=GPT%20Judge%20Detailed%20Score>): Higher is better; unit: SCALAR.

- [GPT Judge Conversation Score](<https://www.sota2.com/research/sota/visual-question-answering-on-lucid-vqa-test.md?metric=GPT%20Judge%20Conversation%20Score>): Higher is better; unit: SCALAR.

- [GPT Judge Reasoning Score](<https://www.sota2.com/research/sota/visual-question-answering-on-lucid-vqa-test.md?metric=GPT%20Judge%20Reasoning%20Score>): Higher is better; unit: SCALAR.

- [GPT Judge Overall Score](<https://www.sota2.com/research/sota/visual-question-answering-on-lucid-vqa-test.md?metric=GPT%20Judge%20Overall%20Score>): Higher is better; unit: SCALAR.

- [Gemini Judge Detailed Score](<https://www.sota2.com/research/sota/visual-question-answering-on-lucid-vqa-test.md?metric=Gemini%20Judge%20Detailed%20Score>): Higher is better; unit: SCALAR.

- [Gemini Judge Conversation Score](<https://www.sota2.com/research/sota/visual-question-answering-on-lucid-vqa-test.md?metric=Gemini%20Judge%20Conversation%20Score>): Higher is better; unit: SCALAR.

- [Gemini Judge Reasoning Score](<https://www.sota2.com/research/sota/visual-question-answering-on-lucid-vqa-test.md?metric=Gemini%20Judge%20Reasoning%20Score>): Higher is better; unit: SCALAR.

- [Gemini Judge Overall Score](<https://www.sota2.com/research/sota/visual-question-answering-on-lucid-vqa-test.md?metric=Gemini%20Judge%20Overall%20Score>): Higher is better; unit: SCALAR.

## Recorded results

| Method | GPT Judge Detailed Score | Result date | Report paper |
| --- | --- | --- | --- |
| LLaVA-LE | 0.952 | Not recorded | 69c60f180636e5bf4afc8582 |
| LLaVA-LE | 0.433 | Not recorded | 69c60f180636e5bf4afc8582 |
| Base LLaVA | 0.319 | Not recorded | 69c60f180636e5bf4afc8582 |

### LLaVA-LE

GPT Judge Detailed Score: 0.952. Result ID: 69c6769747f34ce924b03bf3.

[Reporting paper](<https://www.sota2.com/research/paper/69c60f180636e5bf4afc8582.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c60f180636e5bf4afc8582.md>)

Recorded evaluation configuration: {"Stage":"Stage 2"}. This record does not assert independent reproduction.

### LLaVA-LE

GPT Judge Detailed Score: 0.433. Result ID: 69c6769747f34ce924b03bf4.

[Reporting paper](<https://www.sota2.com/research/paper/69c60f180636e5bf4afc8582.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c60f180636e5bf4afc8582.md>)

Recorded evaluation configuration: {"Stage":"Stage 1"}. This record does not assert independent reproduction.

### Base LLaVA

GPT Judge Detailed Score: 0.319. Result ID: 69c6769747f34ce924b03bf5.

[Reporting paper](<https://www.sota2.com/research/paper/69c60f180636e5bf4afc8582.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa717d1e209131cea51970.md>)

[Implementation](<https://llava-vl.github.io/>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

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