# Visual Document Question Answering on REAL-MM-RAG Source: [Visual Document Question Answering on REAL-MM-RAG]() Collection index: [llms.txt]() Data updated: 2026-03-09T08:54:31.575Z Evaluation of Visual Document Question Answering performance on the REAL-MM-RAG dataset, reporting accuracy across financial report, financial slide, technical report, and technical slide subsets. 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 Document Question Answering | | Dataset (including any stated split) | REAL-MM-RAG | | Metric | Accuracy (Financial Report) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 5 of 5 | | 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 (Financial Report)](): Higher is better; unit: PERCENT. - [Accuracy (Financial Slide)](): Higher is better; unit: PERCENT. - [Accuracy (Technical Report)](): Higher is better; unit: PERCENT. - [Accuracy (Technical Slide)](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy (Financial Report) | Result date | Report paper | | --- | --- | --- | --- | | Oracle | 55.45 | Not recorded | 68fa7ba71e209131cea89ab7 | | Oracle | 53.11 | Not recorded | 68fa7ba71e209131cea89ab7 | | CMRAG | 32.13 | Not recorded | 68fa7ba71e209131cea89ab7 | | SigLIP | 29.54 | Not recorded | 68fa7ba71e209131cea89ab7 | | CLIP-L/14-336 | 27.9 | Not recorded | 68fa7ba71e209131cea89ab7 | ### Oracle Accuracy (Financial Report): 55.45. Result ID: 69ae8ab06dbb7075b956cedf. [Reporting paper]() Recorded evaluation configuration: {"Top-k retrieval":"3","Input modality":"I + T","Judge model":"Qwen2.5-7B-Instruct","Backbone Vision Language Model":"Qwen2.5-VL-7B-Instruct"}. This record does not assert independent reproduction. ### Oracle Accuracy (Financial Report): 53.11. Result ID: 69ae8ab06dbb7075b956cede. [Reporting paper]() Recorded evaluation configuration: {"Top-k retrieval":"3","Input modality":"I","Judge model":"Qwen2.5-7B-Instruct","Backbone Vision Language Model":"Qwen2.5-VL-7B-Instruct"}. This record does not assert independent reproduction. ### CMRAG Accuracy (Financial Report): 32.13. Result ID: 69ae8ab06dbb7075b956cedd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Top-k retrieval":"3","Judge model":"Qwen2.5-7B-Instruct","Backbone Vision Language Model":"Qwen2.5-VL-7B-Instruct"}. This record does not assert independent reproduction. ### SigLIP Accuracy (Financial Report): 29.54. Result ID: 69ae8ab06dbb7075b956cedc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Top-k retrieval":"3","Judge model":"Qwen2.5-7B-Instruct","Backbone Vision Language Model":"Qwen2.5-VL-7B-Instruct"}. This record does not assert independent reproduction. ### CLIP-L/14-336 Accuracy (Financial Report): 27.9. Result ID: 69ae8ab06dbb7075b956cedb. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Top-k retrieval":"3","Judge model":"Qwen2.5-7B-Instruct","Backbone Vision Language Model":"Qwen2.5-VL-7B-Instruct"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.