# Hallucination Evaluation on CHAIR MSCOCO 2014 Source: [Hallucination Evaluation on CHAIR MSCOCO 2014]() Collection index: [llms.txt]() Data updated: 2026-04-01T08:10:33.608Z Evaluation of model hallucination performance using CHAIRs and CHAIRi metrics on the CHAIR MSCOCO 2014 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 | Hallucination Evaluation | | Dataset (including any stated split) | CHAIR MSCOCO 2014 | | Metric | CHAIRs Score | | Unit | SCALAR | | Direction | Higher is better | | Results loaded | 28 of 28 | | 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 - [CHAIRs Score](): Higher is better; unit: SCALAR. - [CHAIRi Score](): Higher is better; unit: SCALAR. ## Recorded results | Method | CHAIRs Score | Result date | Report paper | | --- | --- | --- | --- | | baseline | 51.3 | Not recorded | 69cca6c30636e5bf4afcbd91 | | baseline | 51 | Not recorded | 69cca6c30636e5bf4afcbd91 | | M3ID | 48.3 | Not recorded | 69cca6c30636e5bf4afcbd91 | | ICD | 47.4 | Not recorded | 69cca6c30636e5bf4afcbd91 | | OPERA | 47.4 | Not recorded | 69cca6c30636e5bf4afcbd91 | | VCD | 46.8 | Not recorded | 69cca6c30636e5bf4afcbd91 | | ICD | 46.3 | Not recorded | 69cca6c30636e5bf4afcbd91 | | OPERA | 45.2 | Not recorded | 69cca6c30636e5bf4afcbd91 | | AVISC | 45.2 | Not recorded | 69cca6c30636e5bf4afcbd91 | | M3ID | 45.2 | Not recorded | 69cca6c30636e5bf4afcbd91 | | SID | 44.2 | Not recorded | 69cca6c30636e5bf4afcbd91 | | VCD | 44 | Not recorded | 69cca6c30636e5bf4afcbd91 | | AVISC | 43.9 | Not recorded | 69cca6c30636e5bf4afcbd91 | | Projectaway | 43.8 | Not recorded | 69cca6c30636e5bf4afcbd91 | | VTI | 43.4 | Not recorded | 69cca6c30636e5bf4afcbd91 | | SID | 42.3 | Not recorded | 69cca6c30636e5bf4afcbd91 | | Projectaway | 42 | Not recorded | 69cca6c30636e5bf4afcbd91 | | Octopus | 40.6 | Not recorded | 69cca6c30636e5bf4afcbd91 | | Octopus | 39.2 | Not recorded | 69cca6c30636e5bf4afcbd91 | | HIRE | 39 | Not recorded | 69cca6c30636e5bf4afcbd91 | | VTI | 35.8 | Not recorded | 69cca6c30636e5bf4afcbd91 | | HIRE | 30.2 | Not recorded | 69cca6c30636e5bf4afcbd91 | | baseline | 25.4 | Not recorded | 69cca6c30636e5bf4afcbd91 | | baseline | 20.4 | Not recorded | 69cca6c30636e5bf4afcbd91 | | Nullu | 17 | Not recorded | 69cca6c30636e5bf4afcbd91 | | HIRE | 15.2 | Not recorded | 69cca6c30636e5bf4afcbd91 | | HIRE | 14.8 | Not recorded | 69cca6c30636e5bf4afcbd91 | | M3ID+DPO | 13.5 | Not recorded | 69cca6c30636e5bf4afcbd91 | ### baseline CHAIRs Score: 51.3. Result ID: 69ccd2ea47f34ce924b6653d. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"LLaVA-1.5","Max new tokens":"512","TFLOPs":"10.23"}. This record does not assert independent reproduction. ### baseline CHAIRs Score: 51. Result ID: 69ccd2ea47f34ce924b66548. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"InstructBLIP","Max new tokens":"512","TFLOPs":"2.77"}. This record does not assert independent reproduction. ### M3ID CHAIRs Score: 48.3. Result ID: 69ccd2ea47f34ce924b66542. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"LLaVA-1.5","Max new tokens":"512","TFLOPs":"20.71"}. This record does not assert independent reproduction. ### ICD CHAIRs Score: 47.4. Result ID: 69ccd2ea47f34ce924b66540. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"LLaVA-1.5","Max new tokens":"512","TFLOPs":"20.63"}. This record does not assert independent reproduction. ### OPERA CHAIRs Score: 47.4. Result ID: 69ccd2ea47f34ce924b66549. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"InstructBLIP","Max new tokens":"512"}. This record does not assert independent reproduction. ### VCD CHAIRs Score: 46.8. Result ID: 69ccd2ea47f34ce924b66541. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"LLaVA-1.5","Max new tokens":"512","TFLOPs":"20.46"}. This record does not assert independent reproduction. ### ICD CHAIRs Score: 46.3. Result ID: 69ccd2ea47f34ce924b6654b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"InstructBLIP","Max new tokens":"512","TFLOPs":"5.64"}. This record does not assert independent reproduction. ### OPERA CHAIRs Score: 45.2. Result ID: 69ccd2ea47f34ce924b6653e. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"LLaVA-1.5","Max new tokens":"512"}. This record does not assert independent reproduction. ### AVISC CHAIRs Score: 45.2. Result ID: 69ccd2ea47f34ce924b66543. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"LLaVA-1.5","Max new tokens":"512","TFLOPs":"20.08"}. This record does not assert independent reproduction. ### M3ID CHAIRs Score: 45.2. Result ID: 69ccd2ea47f34ce924b6654d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"InstructBLIP","Max new tokens":"512","TFLOPs":"5.70"}. This record does not assert independent reproduction. ### SID CHAIRs Score: 44.2. Result ID: 69ccd2ea47f34ce924b6653f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"LLaVA-1.5","Max new tokens":"512"}. This record does not assert independent reproduction. ### VCD CHAIRs Score: 44. Result ID: 69ccd2ea47f34ce924b6654c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"InstructBLIP","Max new tokens":"512","TFLOPs":"5.61"}. This record does not assert independent reproduction. ### AVISC CHAIRs Score: 43.9. Result ID: 69ccd2ea47f34ce924b6654e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"InstructBLIP","Max new tokens":"512","TFLOPs":"5.63"}. This record does not assert independent reproduction. ### Projectaway CHAIRs Score: 43.8. Result ID: 69ccd2ea47f34ce924b66550. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"InstructBLIP","Max new tokens":"512"}. This record does not assert independent reproduction. ### VTI CHAIRs Score: 43.4. Result ID: 69ccd2ea47f34ce924b66551. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"InstructBLIP","Max new tokens":"512"}. This record does not assert independent reproduction. ### SID CHAIRs Score: 42.3. Result ID: 69ccd2ea47f34ce924b6654a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"InstructBLIP","Max new tokens":"512"}. This record does not assert independent reproduction. ### Projectaway CHAIRs Score: 42. Result ID: 69ccd2ea47f34ce924b66545. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"LLaVA-1.5","Max new tokens":"512"}. This record does not assert independent reproduction. ### Octopus CHAIRs Score: 40.6. Result ID: 69ccd2ea47f34ce924b6654f. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"InstructBLIP","Max new tokens":"512","TFLOPs":"5.87"}. This record does not assert independent reproduction. ### Octopus CHAIRs Score: 39.2. Result ID: 69ccd2ea47f34ce924b66544. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"LLaVA-1.5","Max new tokens":"512","TFLOPs":"21.39"}. This record does not assert independent reproduction. ### HIRE CHAIRs Score: 39. Result ID: 69ccd2ea47f34ce924b66552. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"InstructBLIP","Max new tokens":"512","TFLOPs":"3.45"}. This record does not assert independent reproduction. ### VTI CHAIRs Score: 35.8. Result ID: 69ccd2ea47f34ce924b66546. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"LLaVA-1.5","Max new tokens":"512"}. This record does not assert independent reproduction. ### HIRE CHAIRs Score: 30.2. Result ID: 69ccd2ea47f34ce924b66547. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"LLaVA-1.5","Max new tokens":"512","TFLOPs":"11.81"}. This record does not assert independent reproduction. ### baseline CHAIRs Score: 25.4. Result ID: 69ccd2ea47f34ce924b66557. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"InstructBLIP","Max new tokens":"64","TFLOPs":"1.89"}. This record does not assert independent reproduction. ### baseline CHAIRs Score: 20.4. Result ID: 69ccd2ea47f34ce924b66553. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"LLaVA-1.5","Max new tokens":"64","TFLOPs":"8.91"}. This record does not assert independent reproduction. ### Nullu CHAIRs Score: 17. Result ID: 69ccd2ea47f34ce924b66555. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"LLaVA-1.5","Max new tokens":"64"}. This record does not assert independent reproduction. ### HIRE CHAIRs Score: 15.2. Result ID: 69ccd2ea47f34ce924b66556. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"LLaVA-1.5","Max new tokens":"64","TFLOPs":"9.18"}. This record does not assert independent reproduction. ### HIRE CHAIRs Score: 14.8. Result ID: 69ccd2ea47f34ce924b66558. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"InstructBLIP","Max new tokens":"64","TFLOPs":"2.32"}. This record does not assert independent reproduction. ### M3ID+DPO CHAIRs Score: 13.5. Result ID: 69ccd2ea47f34ce924b66554. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"LLaVA-1.5","Max new tokens":"64"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.