# Hallucination Attribution on POPE Source: [Hallucination Attribution on POPE]() Collection index: [llms.txt]() Data updated: 2026-07-09T06:50:15.909Z Evaluation of model performance on attributing the cause of object hallucinations detected using the POPE methodology, utilizing metrics focused on attribution quality (e.g., Ins, Del, High, MEC). 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 | Hallucination Attribution | | Dataset (including any stated split) | POPE | | Metric | Ins Score | | 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 - [Ins Score](): Higher is better; unit: PERCENT. - [Del Score](): Higher is better; unit: PERCENT. - [High Score](): Higher is better; unit: PERCENT. - [MEC(k)](): Lower is better; unit: SCALAR. ## Recorded results | Method | Ins Score | Result date | Report paper | | --- | --- | --- | --- | | TRACE+G | 93.53 | Not recorded | 6a4f2ae532fb937314875a6c | | COPAIR+G | 93.47 | Not recorded | 6a4f2ae532fb937314875a6c | | COPAIR+PW | 93.08 | Not recorded | 6a4f2ae532fb937314875a6c | | Greedy (G) | 93.05 | Not recorded | 6a4f2ae532fb937314875a6c | | TRACE+PW | 92.99 | Not recorded | 6a4f2ae532fb937314875a6c | | PhaseWin (PW) | 92.53 | Not recorded | 6a4f2ae532fb937314875a6c | | TRACE+G | 91.65 | Not recorded | 6a4f2ae532fb937314875a6c | | COPAIR+G | 91.61 | Not recorded | 6a4f2ae532fb937314875a6c | | Greedy (G) | 91.52 | Not recorded | 6a4f2ae532fb937314875a6c | | COPAIR+PW | 91.04 | Not recorded | 6a4f2ae532fb937314875a6c | | TRACE+PW | 91.02 | Not recorded | 6a4f2ae532fb937314875a6c | | PhaseWin (PW) | 90.9 | Not recorded | 6a4f2ae532fb937314875a6c | | IGOS++ | 78.79 | Not recorded | 6a4f2ae532fb937314875a6c | | IGOS++ | 77.54 | Not recorded | 6a4f2ae532fb937314875a6c | | LLaVA-CAM | 76.52 | Not recorded | 6a4f2ae532fb937314875a6c | | LLaVA-CAM | 75.63 | Not recorded | 6a4f2ae532fb937314875a6c | ### TRACE+G Ins Score: 93.53. Result ID: 6a4f449960b88bb339e44678. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-VL-3B"}. This record does not assert independent reproduction. ### COPAIR+G Ins Score: 93.47. Result ID: 6a4f449960b88bb339e44676. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-VL-3B"}. This record does not assert independent reproduction. ### COPAIR+PW Ins Score: 93.08. Result ID: 6a4f449960b88bb339e44672. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-VL-3B"}. This record does not assert independent reproduction. ### Greedy (G) Ins Score: 93.05. Result ID: 6a4f449960b88bb339e4466e. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model":"Qwen2.5-VL-3B"}. This record does not assert independent reproduction. ### TRACE+PW Ins Score: 92.99. Result ID: 6a4f449960b88bb339e44674. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-VL-3B"}. This record does not assert independent reproduction. ### PhaseWin (PW) Ins Score: 92.53. Result ID: 6a4f449960b88bb339e44670. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-VL-3B"}. This record does not assert independent reproduction. ### TRACE+G Ins Score: 91.65. Result ID: 6a4f449960b88bb339e44679. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LLaVA-v1.5-7B"}. This record does not assert independent reproduction. ### COPAIR+G Ins Score: 91.61. Result ID: 6a4f449960b88bb339e44677. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LLaVA-v1.5-7B"}. This record does not assert independent reproduction. ### Greedy (G) Ins Score: 91.52. Result ID: 6a4f449960b88bb339e4466f. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model":"LLaVA-v1.5-7B"}. This record does not assert independent reproduction. ### COPAIR+PW Ins Score: 91.04. Result ID: 6a4f449960b88bb339e44673. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LLaVA-v1.5-7B"}. This record does not assert independent reproduction. ### TRACE+PW Ins Score: 91.02. Result ID: 6a4f449960b88bb339e44675. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LLaVA-v1.5-7B"}. This record does not assert independent reproduction. ### PhaseWin (PW) Ins Score: 90.9. Result ID: 6a4f449960b88bb339e44671. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LLaVA-v1.5-7B"}. This record does not assert independent reproduction. ### IGOS++ Ins Score: 78.79. Result ID: 6a4f449960b88bb339e4466d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LLaVA-v1.5-7B"}. This record does not assert independent reproduction. ### IGOS++ Ins Score: 77.54. Result ID: 6a4f449960b88bb339e4466c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-VL-3B"}. This record does not assert independent reproduction. ### LLaVA-CAM Ins Score: 76.52. Result ID: 6a4f449960b88bb339e4466a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-VL-3B"}. This record does not assert independent reproduction. ### LLaVA-CAM Ins Score: 75.63. Result ID: 6a4f449960b88bb339e4466b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LLaVA-v1.5-7B"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.