# Image Captioning Hallucination Detection on NoCaps Near-Domain (val)

Source: [Image Captioning Hallucination Detection on NoCaps Near-Domain (val)](<https://www.sota2.com/research/sota/image-captioning-hallucination-detection-on-nocaps-near-domain-val>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-03-30T07:49:42.886Z

Evaluation of model performance on detecting caption hallucinations specifically within the near-domain subset of the NoCaps validation set, measured using CHAIR metrics.

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 | Image Captioning Hallucination Detection |
| Dataset (including any stated split) | NoCaps Near-Domain (val) |
| Metric | CHAIR Score S |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 4 of 4 |
| 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

- [CHAIR Score S](<https://www.sota2.com/research/sota/image-captioning-hallucination-detection-on-nocaps-near-domain-val.md?metric=CHAIR%20Score%20S>): Higher is better; unit: PERCENT.

- [CHAIR Score I](<https://www.sota2.com/research/sota/image-captioning-hallucination-detection-on-nocaps-near-domain-val.md?metric=CHAIR%20Score%20I>): Higher is better; unit: PERCENT.

## Recorded results

| Method | CHAIR Score S | Result date | Report paper |
| --- | --- | --- | --- |
| Greedy Decoding | 63.3 | Not recorded | 69ca037b0636e5bf4afc933b |
| VCD | 51.8 | Not recorded | 69ca037b0636e5bf4afc933b |
| VAP | 51.5 | Not recorded | 69ca037b0636e5bf4afc933b |
| ClipTTT | 50 | Not recorded | 69ca037b0636e5bf4afc933b |

### Greedy Decoding

CHAIR Score S: 63.3. Result ID: 69ca2abc47f34ce924b28c0d.

[Reporting paper](<https://www.sota2.com/research/paper/69ca037b0636e5bf4afc933b.md>)

Recorded evaluation configuration: {"Backbone":"LLaVA-v1.5-7B"}. This record does not assert independent reproduction.

### VCD

CHAIR Score S: 51.8. Result ID: 69ca2abc47f34ce924b28c0f.

[Reporting paper](<https://www.sota2.com/research/paper/69ca037b0636e5bf4afc933b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71461e209131cea5053d.md>)

Recorded evaluation configuration: {"Backbone":"LLaVA-v1.5-7B"}. This record does not assert independent reproduction.

### VAP

CHAIR Score S: 51.5. Result ID: 69ca2abc47f34ce924b28c0e.

[Reporting paper](<https://www.sota2.com/research/paper/69ca037b0636e5bf4afc933b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa770b1e209131cea70b60.md>)

Recorded evaluation configuration: {"Backbone":"LLaVA-v1.5-7B"}. This record does not assert independent reproduction.

### ClipTTT

CHAIR Score S: 50. Result ID: 69ca2abc47f34ce924b28c10.

[Reporting paper](<https://www.sota2.com/research/paper/69ca037b0636e5bf4afc933b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ca037b0636e5bf4afc933b.md>)

Recorded evaluation configuration: {"Backbone":"LLaVA-v1.5-7B"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.
