# RAG Question Answering on SCORE domain-specific hazard response dataset (test)

Source: [RAG Question Answering on SCORE domain-specific hazard response dataset (test)](<https://www.sota2.com/research/sota/rag-question-answering-on-score-domain-specific-hazard-response-dataset-test>)

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

Data updated: 2026-02-26T06:04:39.166Z

Evaluation of RAG Question Answering performance on the test split of the SCORE domain-specific hazard response dataset, measured using metrics like Specificity, Robustness, and Answer Relevance.

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 | RAG Question Answering |
| Dataset (including any stated split) | SCORE domain-specific hazard response dataset (test) |
| Metric | Specificity |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 4 of 4 |
| 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

- [Specificity](<https://www.sota2.com/research/sota/rag-question-answering-on-score-domain-specific-hazard-response-dataset-test.md?metric=Specificity>): Higher is better; unit: PERCENT.

- [Robustness (Para.)](<https://www.sota2.com/research/sota/rag-question-answering-on-score-domain-specific-hazard-response-dataset-test.md?metric=Robustness%20(Para.)>): Higher is better; unit: PERCENT.

- [Robustness (Pert.)](<https://www.sota2.com/research/sota/rag-question-answering-on-score-domain-specific-hazard-response-dataset-test.md?metric=Robustness%20(Pert.)>): Higher is better; unit: PERCENT.

- [Answer Relevance](<https://www.sota2.com/research/sota/rag-question-answering-on-score-domain-specific-hazard-response-dataset-test.md?metric=Answer%20Relevance>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Specificity | Result date | Report paper |
| --- | --- | --- | --- |
| GPT-4o | 61 | Not recorded | 698c0d2fc97a0b3e3b5e1276 |
| GPT-4o | 58 | Not recorded | 698c0d2fc97a0b3e3b5e1276 |
| Gemini 2.5 | 55 | Not recorded | 698c0d2fc97a0b3e3b5e1276 |
| Gemini 2.5 | 45 | Not recorded | 698c0d2fc97a0b3e3b5e1276 |

### GPT-4o

Specificity: 61. Result ID: 69963c455698665ec47f40d2.

[Reporting paper](<https://www.sota2.com/research/paper/698c0d2fc97a0b3e3b5e1276.md>)

Recorded evaluation configuration: {"Evaluator":"GPT-4o"}. This record does not assert independent reproduction.

### GPT-4o

Specificity: 58. Result ID: 69963c455698665ec47f40d3.

[Reporting paper](<https://www.sota2.com/research/paper/698c0d2fc97a0b3e3b5e1276.md>)

Recorded evaluation configuration: {"Evaluator":"Qwen3-8B"}. This record does not assert independent reproduction.

### Gemini 2.5

Specificity: 55. Result ID: 69963c455698665ec47f40d4.

[Reporting paper](<https://www.sota2.com/research/paper/698c0d2fc97a0b3e3b5e1276.md>)

Recorded evaluation configuration: {"Evaluator":"GPT-4o"}. This record does not assert independent reproduction.

### Gemini 2.5

Specificity: 45. Result ID: 69963c455698665ec47f40d5.

[Reporting paper](<https://www.sota2.com/research/paper/698c0d2fc97a0b3e3b5e1276.md>)

Recorded evaluation configuration: {"Evaluator":"Qwen3-8B"}. This record does not assert independent reproduction.

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