# Vision-Language Modeling on MVision Source: [Vision-Language Modeling on MVision]() Collection index: [llms.txt]() Data updated: 2026-06-17T08:11:51.681Z Evaluation of model performance on the MVision dataset for Vision-Language Modeling tasks, measured using 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 Modeling | | Dataset (including any stated split) | MVision | | Metric | Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 16 of 16 | | 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 | | --- | --- | --- | --- | | ZPPO | 50.5 | Not recorded | 6a322a62d33dfb23e783ee2d | | GRPO | 43.4 | Not recorded | 6a322a62d33dfb23e783ee2d | | GRPO | 43.1 | Not recorded | 6a322a62d33dfb23e783ee2d | | On-Distill | 42.7 | Not recorded | 6a322a62d33dfb23e783ee2d | | On-Distill | 41.9 | Not recorded | 6a322a62d33dfb23e783ee2d | | Off-Distill | 41.8 | Not recorded | 6a322a62d33dfb23e783ee2d | | Off-Distill | 40.4 | Not recorded | 6a322a62d33dfb23e783ee2d | | Qwen3.5 | 38.4 | Not recorded | 6a322a62d33dfb23e783ee2d | | ZPPO | 29.2 | Not recorded | 6a322a62d33dfb23e783ee2d | | GRPO | 20.9 | Not recorded | 6a322a62d33dfb23e783ee2d | | GRPO | 20.1 | Not recorded | 6a322a62d33dfb23e783ee2d | | On-Distill | 19.4 | Not recorded | 6a322a62d33dfb23e783ee2d | | On-Distill | 18.7 | Not recorded | 6a322a62d33dfb23e783ee2d | | Off-Distill | 17.9 | Not recorded | 6a322a62d33dfb23e783ee2d | | Off-Distill | 17.3 | Not recorded | 6a322a62d33dfb23e783ee2d | | Qwen3.5 | 16.4 | Not recorded | 6a322a62d33dfb23e783ee2d | ### ZPPO Accuracy: 50.5. Result ID: 6a3255a6fbc373592c2fc243. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"2B","Prompt Replay Buffer Augmentation":true}. This record does not assert independent reproduction. ### GRPO Accuracy: 43.4. Result ID: 6a3255a6fbc373592c2fc242. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"2B","Prompt Replay Buffer Augmentation":true}. This record does not assert independent reproduction. ### GRPO Accuracy: 43.1. Result ID: 6a3255a6fbc373592c2fc241. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"2B","Prompt Replay Buffer Augmentation":false}. This record does not assert independent reproduction. ### On-Distill Accuracy: 42.7. Result ID: 6a3255a6fbc373592c2fc240. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"2B","Prompt Replay Buffer Augmentation":true}. This record does not assert independent reproduction. ### On-Distill Accuracy: 41.9. Result ID: 6a3255a6fbc373592c2fc23f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"2B","Prompt Replay Buffer Augmentation":false}. This record does not assert independent reproduction. ### Off-Distill Accuracy: 41.8. Result ID: 6a3255a6fbc373592c2fc23e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"2B","Prompt Replay Buffer Augmentation":true}. This record does not assert independent reproduction. ### Off-Distill Accuracy: 40.4. Result ID: 6a3255a6fbc373592c2fc23d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"2B","Prompt Replay Buffer Augmentation":false}. This record does not assert independent reproduction. ### Qwen3.5 Accuracy: 38.4. Result ID: 6a3255a6fbc373592c2fc23c. [Reporting paper]() Recorded evaluation configuration: {"Model Scale":"2B","Prompt Replay Buffer Augmentation":false}. This record does not assert independent reproduction. ### ZPPO Accuracy: 29.2. Result ID: 6a3255a6fbc373592c2fc23b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"0.8B","Prompt Replay Buffer Augmentation":true}. This record does not assert independent reproduction. ### GRPO Accuracy: 20.9. Result ID: 6a3255a6fbc373592c2fc23a. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"0.8B","Prompt Replay Buffer Augmentation":true}. This record does not assert independent reproduction. ### GRPO Accuracy: 20.1. Result ID: 6a3255a6fbc373592c2fc239. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"0.8B","Prompt Replay Buffer Augmentation":false}. This record does not assert independent reproduction. ### On-Distill Accuracy: 19.4. Result ID: 6a3255a6fbc373592c2fc237. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"0.8B","Prompt Replay Buffer Augmentation":false}. This record does not assert independent reproduction. ### On-Distill Accuracy: 18.7. Result ID: 6a3255a6fbc373592c2fc238. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"0.8B","Prompt Replay Buffer Augmentation":true}. This record does not assert independent reproduction. ### Off-Distill Accuracy: 17.9. Result ID: 6a3255a6fbc373592c2fc236. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"0.8B","Prompt Replay Buffer Augmentation":true}. This record does not assert independent reproduction. ### Off-Distill Accuracy: 17.3. Result ID: 6a3255a6fbc373592c2fc235. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"0.8B","Prompt Replay Buffer Augmentation":false}. This record does not assert independent reproduction. ### Qwen3.5 Accuracy: 16.4. Result ID: 6a3255a6fbc373592c2fc234. [Reporting paper]() Recorded evaluation configuration: {"Model Scale":"0.8B","Prompt Replay Buffer Augmentation":false}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.