# Difference Caption Generation on Imagen edits 3 Source: [Difference Caption Generation on Imagen edits 3]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:51:35.735Z Evaluation of model performance on generating captions describing the differences between image pairs using the Imagen edits 3 dataset. 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 | Difference Caption Generation | | Dataset (including any stated split) | Imagen edits 3 | | Metric | Main Difference Score | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 7 of 7 | | 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 - [Main Difference Score](): Higher is better; unit: PERCENT. - [MP Score](): Higher is better; unit: PERCENT. - [MP Soft Score](): Higher is better; unit: PERCENT. - [HR Score](): Higher is better; unit: PERCENT. - [HR Soft Score](): Higher is better; unit: PERCENT. - [Average Difference](): Higher is better; unit: SCALAR. - [No Diffs Rate](): Higher is better; unit: PERCENT. ## Recorded results | Method | Main Difference Score | Result date | Report paper | | --- | --- | --- | --- | | GPT-4o | 44 | Not recorded | 68fa79651e209131cea7d6a6 | | GPT-4 | 29 | Not recorded | 68fa79651e209131cea7d6a6 | | Qwen2.5 VL | 29 | Not recorded | 68fa79651e209131cea7d6a6 | | GPT-4 Turbo | 27 | Not recorded | 68fa79651e209131cea7d6a6 | | InternVL3 | 13 | Not recorded | 68fa79651e209131cea7d6a6 | | LLaVA | 11 | Not recorded | 68fa79651e209131cea7d6a6 | | LLaVA | 4 | Not recorded | 68fa79651e209131cea7d6a6 | ### GPT-4o Main Difference Score: 44. Result ID: 698b4b8e8d67a41d0a6ce169. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### GPT-4 Main Difference Score: 29. Result ID: 698b4b8e8d67a41d0a6ce168. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Qwen2.5 VL Main Difference Score: 29. Result ID: 698b4b8e8d67a41d0a6ce16b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### GPT-4 Turbo Main Difference Score: 27. Result ID: 698b4b8e8d67a41d0a6ce16a. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### InternVL3 Main Difference Score: 13. Result ID: 698b4b8e8d67a41d0a6ce16c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### LLaVA Main Difference Score: 11. Result ID: 698b4b8e8d67a41d0a6ce16e. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"training":"supervised"}. This record does not assert independent reproduction. ### LLaVA Main Difference Score: 4. Result ID: 698b4b8e8d67a41d0a6ce16d. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.