# Saliency-grounded Question Answering on SalBench p3 (test) Source: [Saliency-grounded Question Answering on SalBench p3 (test)]() Collection index: [llms.txt]() Data updated: 2026-04-07T10:32:08.085Z Evaluation of model performance on the test split of the SalBench p3 dataset for Saliency-grounded Question Answering, reporting accuracy. 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 | Saliency-grounded Question Answering | | Dataset (including any stated split) | SalBench p3 (test) | | Metric | Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 8 of 8 | | 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](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy | Result date | Report paper | | --- | --- | --- | --- | | GPT-4o | 89.2 | Not recorded | 69d48fb37d0172140ec744be | | Claude-3.5 Sonnet | 86.7 | Not recorded | 69d48fb37d0172140ec744be | | Qwen2.5-VL-7B + Saliency-R1 | 63.7 | Not recorded | 69d48fb37d0172140ec744be | | Qwen2.5-VL-3B + Saliency-R1 | 59.6 | Not recorded | 69d48fb37d0172140ec744be | | Qwen2.5-VL-7B + SFT | 57.1 | Not recorded | 69d48fb37d0172140ec744be | | Qwen2.5-VL-3B + SFT | 51.7 | Not recorded | 69d48fb37d0172140ec744be | | Qwen2.5-VL-7B | 49.1 | Not recorded | 69d48fb37d0172140ec744be | | Qwen2.5-VL-3B | 40.4 | Not recorded | 69d48fb37d0172140ec744be | ### GPT-4o Accuracy: 89.2. Result ID: 69d4dd2447f34ce924bd5501. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Claude-3.5 Sonnet Accuracy: 86.7. Result ID: 69d4dd2447f34ce924bd5502. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Qwen2.5-VL-7B + Saliency-R1 Accuracy: 63.7. Result ID: 69d4dd2447f34ce924bd5508. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Stage":"Saliency-R1"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B + Saliency-R1 Accuracy: 59.6. Result ID: 69d4dd2447f34ce924bd5505. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Stage":"Saliency-R1"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B + SFT Accuracy: 57.1. Result ID: 69d4dd2447f34ce924bd5507. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Stage":"SFT"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B + SFT Accuracy: 51.7. Result ID: 69d4dd2447f34ce924bd5504. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Stage":"SFT"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B Accuracy: 49.1. Result ID: 69d4dd2447f34ce924bd5506. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Stage":"Base"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B Accuracy: 40.4. Result ID: 69d4dd2447f34ce924bd5503. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Stage":"Base"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.