# Hallucination Detection on SQuAD (ROC, RAC, F1@B, RA@50)

Source: [Hallucination Detection on SQuAD (ROC, RAC, F1@B, RA@50)](<https://www.sota2.com/research/sota/hallucination-detection-on-squad-roc-rac-f1-at-b-ra-at-50>)

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

Data updated: 2026-02-26T06:00:19.613Z

Evaluation of model performance on detecting hallucinations within the SQuAD dataset, measured using ROC, RAC, F1@B, and RA@50 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 Detection |
| Dataset (including any stated split) | SQuAD |
| Metric | ROC |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 18 of 18 |
| 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

- [ROC](<https://www.sota2.com/research/sota/hallucination-detection-on-squad-roc-rac-f1-at-b-ra-at-50.md?metric=ROC>): Higher is better; unit: PERCENT.

- [RAC](<https://www.sota2.com/research/sota/hallucination-detection-on-squad-roc-rac-f1-at-b-ra-at-50.md?metric=RAC>): Higher is better; unit: PERCENT.

- [F1@B](<https://www.sota2.com/research/sota/hallucination-detection-on-squad-roc-rac-f1-at-b-ra-at-50.md?metric=F1%40B>): Higher is better; unit: PERCENT.

- [RA@50](<https://www.sota2.com/research/sota/hallucination-detection-on-squad-roc-rac-f1-at-b-ra-at-50.md?metric=RA%4050>): Higher is better; unit: PERCENT.

## Recorded results

| Method | ROC | Result date | Report paper |
| --- | --- | --- | --- |
| HaluNet | 83.9 | Not recorded | 6955ff9053fdd5c36173e03f |
| HaluNet | 78.1 | Not recorded | 6955ff9053fdd5c36173e03f |
| SelfCheckGPT | 77.2 | Not recorded | 6955ff9053fdd5c36173e03f |
| SEU | 76.3 | Not recorded | 6955ff9053fdd5c36173e03f |
| EmbVar | 75.7 | Not recorded | 6955ff9053fdd5c36173e03f |
| SEU | 74.8 | Not recorded | 6955ff9053fdd5c36173e03f |
| P(True) | 74.8 | Not recorded | 6955ff9053fdd5c36173e03f |
| Logistic | 73.5 | Not recorded | 6955ff9053fdd5c36173e03f |
| PE | 73.2 | Not recorded | 6955ff9053fdd5c36173e03f |
| EmbVar | 72.6 | Not recorded | 6955ff9053fdd5c36173e03f |
| PE | 70.5 | Not recorded | 6955ff9053fdd5c36173e03f |
| Logistic | 69.9 | Not recorded | 6955ff9053fdd5c36173e03f |
| P(True) | 69.2 | Not recorded | 6955ff9053fdd5c36173e03f |
| SE | 66.5 | Not recorded | 6955ff9053fdd5c36173e03f |
| SelfCheckGPT | 65.2 | Not recorded | 6955ff9053fdd5c36173e03f |
| T-NLL | 62.8 | Not recorded | 6955ff9053fdd5c36173e03f |
| SE | 61.9 | Not recorded | 6955ff9053fdd5c36173e03f |
| T-NLL | 57.7 | Not recorded | 6955ff9053fdd5c36173e03f |

### HaluNet

ROC: 83.9. Result ID: 6996a9485698665ec4f32b27.

[Reporting paper](<https://www.sota2.com/research/paper/6955ff9053fdd5c36173e03f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6955ff9053fdd5c36173e03f.md>)

Recorded evaluation configuration: {"Backbone":"Llama3-8B","CR":1,"Sample Latency":"97.663","Inference Latency":"0.124"}. This record does not assert independent reproduction.

### HaluNet

ROC: 78.1. Result ID: 6996a9485698665ec4f32b30.

[Reporting paper](<https://www.sota2.com/research/paper/6955ff9053fdd5c36173e03f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6955ff9053fdd5c36173e03f.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-14B","CR":1,"Sample Latency":"128.006","Inference Latency":"0.119"}. This record does not assert independent reproduction.

### SelfCheckGPT

ROC: 77.2. Result ID: 6996a9485698665ec4f32b24.

[Reporting paper](<https://www.sota2.com/research/paper/6955ff9053fdd5c36173e03f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa70d71e209131cea4d943.md>)

Recorded evaluation configuration: {"Backbone":"Llama3-8B","CR":1,"Sample Latency":"760.002","Inference Latency":"87.161"}. This record does not assert independent reproduction.

### SEU

ROC: 76.3. Result ID: 6996a9485698665ec4f32b23.

[Reporting paper](<https://www.sota2.com/research/paper/6955ff9053fdd5c36173e03f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75781e209131cea682be.md>)

Recorded evaluation configuration: {"Backbone":"Llama3-8B","CR":1,"Sample Latency":"731.105","Inference Latency":"2.068"}. This record does not assert independent reproduction.

### EmbVar

ROC: 75.7. Result ID: 6996a9485698665ec4f32b21.

[Reporting paper](<https://www.sota2.com/research/paper/6955ff9053fdd5c36173e03f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75781e209131cea682be.md>)

Recorded evaluation configuration: {"Backbone":"Llama3-8B","CR":1,"Sample Latency":"727.163","Inference Latency":"0.109"}. This record does not assert independent reproduction.

### SEU

ROC: 74.8. Result ID: 6996a9485698665ec4f32b2c.

[Reporting paper](<https://www.sota2.com/research/paper/6955ff9053fdd5c36173e03f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75781e209131cea682be.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-14B","CR":1,"Sample Latency":"1302.565","Inference Latency":"1.948"}. This record does not assert independent reproduction.

### P(True)

ROC: 74.8. Result ID: 6996a9485698665ec4f32b2e.

[Reporting paper](<https://www.sota2.com/research/paper/6955ff9053fdd5c36173e03f.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-14B","CR":1,"Sample Latency":"118.871","Inference Latency":"53.593"}. This record does not assert independent reproduction.

### Logistic

ROC: 73.5. Result ID: 6996a9485698665ec4f32b26.

[Reporting paper](<https://www.sota2.com/research/paper/6955ff9053fdd5c36173e03f.md>)

Recorded evaluation configuration: {"Backbone":"Llama3-8B","CR":1,"Sample Latency":"98.849","Inference Latency":"0.028"}. This record does not assert independent reproduction.

### PE

ROC: 73.2. Result ID: 6996a9485698665ec4f32b1f.

[Reporting paper](<https://www.sota2.com/research/paper/6955ff9053fdd5c36173e03f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa67041e209131cea13c72.md>)

Recorded evaluation configuration: {"Backbone":"Llama3-8B","CR":1,"Sample Latency":"105.435","Inference Latency":"0.001"}. This record does not assert independent reproduction.

### EmbVar

ROC: 72.6. Result ID: 6996a9485698665ec4f32b2a.

[Reporting paper](<https://www.sota2.com/research/paper/6955ff9053fdd5c36173e03f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75781e209131cea682be.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-14B","CR":1,"Sample Latency":"1289.101","Inference Latency":"0.126"}. This record does not assert independent reproduction.

### PE

ROC: 70.5. Result ID: 6996a9485698665ec4f32b28.

[Reporting paper](<https://www.sota2.com/research/paper/6955ff9053fdd5c36173e03f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa67041e209131cea13c72.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-14B","CR":1,"Sample Latency":"127.035","Inference Latency":"0.001"}. This record does not assert independent reproduction.

### Logistic

ROC: 69.9. Result ID: 6996a9485698665ec4f32b2f.

[Reporting paper](<https://www.sota2.com/research/paper/6955ff9053fdd5c36173e03f.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-14B","CR":1,"Sample Latency":"121.403","Inference Latency":"0.042"}. This record does not assert independent reproduction.

### P(True)

ROC: 69.2. Result ID: 6996a9485698665ec4f32b25.

[Reporting paper](<https://www.sota2.com/research/paper/6955ff9053fdd5c36173e03f.md>)

Recorded evaluation configuration: {"Backbone":"Llama3-8B","CR":1,"Sample Latency":"81.557","Inference Latency":"30.413"}. This record does not assert independent reproduction.

### SE

ROC: 66.5. Result ID: 6996a9485698665ec4f32b2b.

[Reporting paper](<https://www.sota2.com/research/paper/6955ff9053fdd5c36173e03f.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-14B","CR":1,"Sample Latency":"1253.741","Inference Latency":"24.100"}. This record does not assert independent reproduction.

### SelfCheckGPT

ROC: 65.2. Result ID: 6996a9485698665ec4f32b2d.

[Reporting paper](<https://www.sota2.com/research/paper/6955ff9053fdd5c36173e03f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa70d71e209131cea4d943.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-14B","CR":1,"Sample Latency":"1267.408","Inference Latency":"74.015"}. This record does not assert independent reproduction.

### T-NLL

ROC: 62.8. Result ID: 6996a9485698665ec4f32b20.

[Reporting paper](<https://www.sota2.com/research/paper/6955ff9053fdd5c36173e03f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa713d1e209131cea50031.md>)

Recorded evaluation configuration: {"Backbone":"Llama3-8B","CR":1,"Sample Latency":"80.780","Inference Latency":"0.001"}. This record does not assert independent reproduction.

### SE

ROC: 61.9. Result ID: 6996a9485698665ec4f32b22.

[Reporting paper](<https://www.sota2.com/research/paper/6955ff9053fdd5c36173e03f.md>)

Recorded evaluation configuration: {"Backbone":"Llama3-8B","CR":1,"Sample Latency":"743.258","Inference Latency":"29.927"}. This record does not assert independent reproduction.

### T-NLL

ROC: 57.7. Result ID: 6996a9485698665ec4f32b29.

[Reporting paper](<https://www.sota2.com/research/paper/6955ff9053fdd5c36173e03f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa713d1e209131cea50031.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-14B","CR":1,"Sample Latency":"131.44","Inference Latency":"0.001"}. This record does not assert independent reproduction.

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