# High-Resolution Visual Question Answering on HR-4k Source: [High-Resolution Visual Question Answering on HR-4k]() Collection index: [llms.txt]() Data updated: 2026-04-15T07:32:53.257Z Evaluation of model performance on the HR-4k dataset specifically for High-Resolution Visual Question Answering tasks, 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 | High-Resolution Visual Question Answering | | Dataset (including any stated split) | HR-4k | | Metric | Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 10 of 10 | | 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 | | --- | --- | --- | --- | | Qwen 2.5 VL + CropVLM | 65.13 | Not recorded | 69268982d6b0dd3342930319 | | Qwen 2.5 VL + ViCrop (rel-att) | 56.25 | Not recorded | 69268982d6b0dd3342930319 | | Qwen 2.5 VL + ViCrop (grad-att) | 54.38 | Not recorded | 69268982d6b0dd3342930319 | | Qwen 2.5 VL + UV-CoT | 53.62 | Not recorded | 69268982d6b0dd3342930319 | | Qwen 2.5 VL | 51.88 | Not recorded | 69268982d6b0dd3342930319 | | LLaVA 1.5 + ViCrop (rel-att) | 44.25 | Not recorded | 69268982d6b0dd3342930319 | | LLaVA 1.5 + ViCrop (grad-att) | 43.62 | Not recorded | 69268982d6b0dd3342930319 | | LLaVA 1.5 + CropVLM | 41.38 | Not recorded | 69268982d6b0dd3342930319 | | LLaVA 1.5 + UV-CoT | 37.88 | Not recorded | 69268982d6b0dd3342930319 | | LLaVA 1.5 | 35.25 | Not recorded | 69268982d6b0dd3342930319 | ### Qwen 2.5 VL + CropVLM Accuracy: 65.13. Result ID: 69df3f1647f34ce924c7ae03. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen 2.5 VL","Cropping Strategy":"CropVLM","Input Resolution":"2048x2048"}. This record does not assert independent reproduction. ### Qwen 2.5 VL + ViCrop (rel-att) Accuracy: 56.25. Result ID: 69df3f1647f34ce924c7ae00. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen 2.5 VL","Cropping Strategy":"ViCrop (rel-att)","Input Resolution":"448x448"}. This record does not assert independent reproduction. ### Qwen 2.5 VL + ViCrop (grad-att) Accuracy: 54.38. Result ID: 69df3f1647f34ce924c7ae01. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen 2.5 VL","Cropping Strategy":"ViCrop (grad-att)","Input Resolution":"448x448"}. This record does not assert independent reproduction. ### Qwen 2.5 VL + UV-CoT Accuracy: 53.62. Result ID: 69df3f1647f34ce924c7ae02. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen 2.5 VL","Cropping Strategy":"UV-CoT","Input Resolution":"336x336"}. This record does not assert independent reproduction. ### Qwen 2.5 VL Accuracy: 51.88. Result ID: 69df3f1647f34ce924c7adff. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen 2.5 VL","Cropping Strategy":"None","Input Resolution":"448x448"}. This record does not assert independent reproduction. ### LLaVA 1.5 + ViCrop (rel-att) Accuracy: 44.25. Result ID: 69df3f1647f34ce924c7adfb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"LLaVA 1.5","Cropping Strategy":"ViCrop (rel-att)","Input Resolution":"336x336"}. This record does not assert independent reproduction. ### LLaVA 1.5 + ViCrop (grad-att) Accuracy: 43.62. Result ID: 69df3f1647f34ce924c7adfc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"LLaVA 1.5","Cropping Strategy":"ViCrop (grad-att)","Input Resolution":"336x336"}. This record does not assert independent reproduction. ### LLaVA 1.5 + CropVLM Accuracy: 41.38. Result ID: 69df3f1647f34ce924c7adfe. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"LLaVA 1.5","Cropping Strategy":"CropVLM","Input Resolution":"2048x2048"}. This record does not assert independent reproduction. ### LLaVA 1.5 + UV-CoT Accuracy: 37.88. Result ID: 69df3f1647f34ce924c7adfd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"LLaVA 1.5","Cropping Strategy":"UV-CoT","Input Resolution":"336x336"}. This record does not assert independent reproduction. ### LLaVA 1.5 Accuracy: 35.25. Result ID: 69df3f1647f34ce924c7adfa. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Base Model":"LLaVA 1.5","Cropping Strategy":"None","Input Resolution":"336x336"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.