# Multimodal Watermarking on MS-COCO 14 Source: [Multimodal Watermarking on MS-COCO 14]() Collection index: [llms.txt]() Data updated: 2026-02-27T11:24:25.972Z Evaluation of Multimodal Watermarking techniques using the MS-COCO 14 dataset, assessed using metrics including AUC, Perplexity (PPL), BertScore, and Chair-I score. 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 | Multimodal Watermarking | | Dataset (including any stated split) | MS-COCO 14 | | Metric | PPL | | Unit | SCALAR | | Direction | Lower is better | | Results loaded | 14 of 14 | | 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 - [PPL](): Lower is better; unit: SCALAR. - [AUC](): Higher is better; unit: SCALAR. - [BertScore](): Higher is better; unit: SCALAR. - [Chair-I Score](): Higher is better; unit: SCALAR. ## Recorded results | Method | PPL | Result date | Report paper | | --- | --- | --- | --- | | NW | 3.01 | Not recorded | 6965d19a6343fa78c8323688 | | VISA-M | 3.04 | Not recorded | 6965d19a6343fa78c8323688 | | VLA | 3.05 | Not recorded | 6965d19a6343fa78c8323688 | | KGW | 3.08 | Not recorded | 6965d19a6343fa78c8323688 | | SWEET | 3.15 | Not recorded | 6965d19a6343fa78c8323688 | | DiP | 3.15 | Not recorded | 6965d19a6343fa78c8323688 | | Unbiased | 3.16 | Not recorded | 6965d19a6343fa78c8323688 | | NW | 5.24 | Not recorded | 6965d19a6343fa78c8323688 | | VISA-M | 5.52 | Not recorded | 6965d19a6343fa78c8323688 | | SWEET | 5.74 | Not recorded | 6965d19a6343fa78c8323688 | | VLA | 5.8 | Not recorded | 6965d19a6343fa78c8323688 | | KGW | 5.83 | Not recorded | 6965d19a6343fa78c8323688 | | DiP | 5.92 | Not recorded | 6965d19a6343fa78c8323688 | | Unbiased | 5.94 | Not recorded | 6965d19a6343fa78c8323688 | ### NW PPL: 3.01. Result ID: 69a17edd35dfcabb00e76e8e. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen"}. This record does not assert independent reproduction. ### VISA-M PPL: 3.04. Result ID: 69a17edd35dfcabb00e76e94. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen"}. This record does not assert independent reproduction. ### VLA PPL: 3.05. Result ID: 69a17edd35dfcabb00e76e8f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen"}. This record does not assert independent reproduction. ### KGW PPL: 3.08. Result ID: 69a17edd35dfcabb00e76e90. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen"}. This record does not assert independent reproduction. ### SWEET PPL: 3.15. Result ID: 69a17edd35dfcabb00e76e91. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model":"Qwen"}. This record does not assert independent reproduction. ### DiP PPL: 3.15. Result ID: 69a17edd35dfcabb00e76e92. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen"}. This record does not assert independent reproduction. ### Unbiased PPL: 3.16. Result ID: 69a17edd35dfcabb00e76e93. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen"}. This record does not assert independent reproduction. ### NW PPL: 5.24. Result ID: 69a17edd35dfcabb00e76e87. [Reporting paper]() Recorded evaluation configuration: {"Model":"LLaVA"}. This record does not assert independent reproduction. ### VISA-M PPL: 5.52. Result ID: 69a17edd35dfcabb00e76e8d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LLaVA"}. This record does not assert independent reproduction. ### SWEET PPL: 5.74. Result ID: 69a17edd35dfcabb00e76e8a. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model":"LLaVA"}. This record does not assert independent reproduction. ### VLA PPL: 5.8. Result ID: 69a17edd35dfcabb00e76e88. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LLaVA"}. This record does not assert independent reproduction. ### KGW PPL: 5.83. Result ID: 69a17edd35dfcabb00e76e89. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LLaVA"}. This record does not assert independent reproduction. ### DiP PPL: 5.92. Result ID: 69a17edd35dfcabb00e76e8b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LLaVA"}. This record does not assert independent reproduction. ### Unbiased PPL: 5.94. Result ID: 69a17edd35dfcabb00e76e8c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LLaVA"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.