# Hallucination Rate Assessment on HEAL Probes SceneMod - SceneTaskCon

Source: [Hallucination Rate Assessment on HEAL Probes SceneMod - SceneTaskCon](<https://www.sota2.com/research/sota/hallucination-rate-assessment-on-heal-probes-scenemod-scenetaskcon>)

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

Data updated: 2026-04-13T07:51:35.226Z

Evaluation of model performance on hallucination rate assessment using the HEAL Probes SceneMod - SceneTaskCon dataset, measured by CO and CS 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 | Hallucination Rate Assessment |
| Dataset (including any stated split) | HEAL Probes SceneMod - SceneTaskCon |
| Metric | CO Score |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 5 of 5 |
| 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

- [CO Score](<https://www.sota2.com/research/sota/hallucination-rate-assessment-on-heal-probes-scenemod-scenetaskcon.md?metric=CO%20Score>): Higher is better; unit: SCALAR.

- [CS Score](<https://www.sota2.com/research/sota/hallucination-rate-assessment-on-heal-probes-scenemod-scenetaskcon.md?metric=CS%20Score>): Higher is better; unit: SCALAR.

## Recorded results

| Method | CO Score | Result date | Report paper |
| --- | --- | --- | --- |
| Gemma-2-9b-it | 73.1 | Not recorded | 69dc788d7d0172140ec78ada |
| DS-R1-Distil-LLaMA-8B | 60 | Not recorded | 69dc788d7d0172140ec78ada |
| Qwen-14B-Instruct | 53.9 | Not recorded | 69dc788d7d0172140ec78ada |
| Llama-3-8B-Instruct | 17.8 | Not recorded | 69dc788d7d0172140ec78ada |
| Qwen-14B-Instruct-VCD | 1.5 | Not recorded | 69dc788d7d0172140ec78ada |

### Gemma-2-9b-it

CO Score: 73.1. Result ID: 69dca07647f34ce924c35e49.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ada.md>)

Recorded evaluation configuration: {"Env":"BEHAVIOR"}. This record does not assert independent reproduction.

### DS-R1-Distil-LLaMA-8B

CO Score: 60. Result ID: 69dca07647f34ce924c35e4a.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ada.md>)

Recorded evaluation configuration: {"Env":"BEHAVIOR"}. This record does not assert independent reproduction.

### Qwen-14B-Instruct

CO Score: 53.9. Result ID: 69dca07647f34ce924c35e4c.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ada.md>)

Recorded evaluation configuration: {"Env":"BEHAVIOR"}. This record does not assert independent reproduction.

### Llama-3-8B-Instruct

CO Score: 17.8. Result ID: 69dca07647f34ce924c35e4b.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ada.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa701f1e209131cea492fe.md>)

[Implementation](<https://github.com/facebookresearch/llama>)

Recorded evaluation configuration: {"Env":"BEHAVIOR"}. This record does not assert independent reproduction.

### Qwen-14B-Instruct-VCD

CO Score: 1.5. Result ID: 69dca07647f34ce924c35e4d.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ada.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ada.md>)

Recorded evaluation configuration: {"Env":"BEHAVIOR"}. This record does not assert independent reproduction.

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