# Vision-Language Question Answering on IconQA txt Source: [Vision-Language Question Answering on IconQA txt]() Collection index: [llms.txt]() Data updated: 2026-06-11T07:55:24.682Z Evaluation of model performance on the IconQA txt variant/split for Vision-Language Question Answering tasks, typically measured by 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 | Vision-Language Question Answering | | Dataset (including any stated split) | IconQA txt | | 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 | | --- | --- | --- | --- | | LoRA | 93.7 | Not recorded | 6a28f32790966ade919c168e | | LoRA | 92.9 | Not recorded | 6a28f32790966ade919c168e | | LoRA | 92.7 | Not recorded | 6a28f32790966ade919c168e | | LoRA | 91.5 | Not recorded | 6a28f32790966ade919c168e | | Prefix Tuning | 91.5 | Not recorded | 6a28f32790966ade919c168e | | Prefix Tuning | 91.2 | Not recorded | 6a28f32790966ade919c168e | | Prefix Tuning | 91 | Not recorded | 6a28f32790966ade919c168e | | Prefix Tuning | 89.5 | Not recorded | 6a28f32790966ade919c168e | ### LoRA Accuracy: 93.7. Result ID: 6a2a68a0fbc373592c274155. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-VL-3B","Optimization":"FlatPO"}. This record does not assert independent reproduction. ### LoRA Accuracy: 92.9. Result ID: 6a2a68a0fbc373592c274154. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-VL-3B","Optimization":"GAM"}. This record does not assert independent reproduction. ### LoRA Accuracy: 92.7. Result ID: 6a2a68a0fbc373592c274153. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-VL-3B","Optimization":"SAM"}. This record does not assert independent reproduction. ### LoRA Accuracy: 91.5. Result ID: 6a2a68a0fbc373592c274152. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-VL-3B","Optimization":"Base"}. This record does not assert independent reproduction. ### Prefix Tuning Accuracy: 91.5. Result ID: 6a2a68a0fbc373592c274159. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-VL-3B","Optimization":"FlatPO"}. This record does not assert independent reproduction. ### Prefix Tuning Accuracy: 91.2. Result ID: 6a2a68a0fbc373592c274158. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-VL-3B","Optimization":"GAM"}. This record does not assert independent reproduction. ### Prefix Tuning Accuracy: 91. Result ID: 6a2a68a0fbc373592c274157. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-VL-3B","Optimization":"SAM"}. This record does not assert independent reproduction. ### Prefix Tuning Accuracy: 89.5. Result ID: 6a2a68a0fbc373592c274156. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-VL-3B","Optimization":"Base"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.