# Stealthiness Evaluation on LLaMA Guard 2 8B

Source: [Stealthiness Evaluation on LLaMA Guard 2 8B](<https://www.sota2.com/research/sota/stealthiness-evaluation-on-llama-guard-2-8b>)

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

Data updated: 2026-05-18T09:28:34.720Z

Evaluation of model stealthiness performance using the LLaMA Guard 2 8B dataset, measured by Perplexity metrics (PPL Mean and PPL Std).

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 | Stealthiness Evaluation |
| Dataset (including any stated split) | LLaMA Guard 2 8B |
| Metric | PPL Mean |
| Unit | SCALAR |
| Direction | Lower is better |
| Results loaded | 10 of 10 |
| 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

- [PPL Mean](<https://www.sota2.com/research/sota/stealthiness-evaluation-on-llama-guard-2-8b.md?metric=PPL%20Mean>): Lower is better; unit: SCALAR.

- [PPL Std](<https://www.sota2.com/research/sota/stealthiness-evaluation-on-llama-guard-2-8b.md?metric=PPL%20Std>): Lower is better; unit: SCALAR.

## Recorded results

| Method | PPL Mean | Result date | Report paper |
| --- | --- | --- | --- |
| ArtPrompt | 2.34 | Not recorded | 68fa751a1e209131cea65ca8 |
| Ascii | 7.25 | Not recorded | 68fa751a1e209131cea65ca8 |
| Base64 | 9.78 | Not recorded | 68fa751a1e209131cea65ca8 |
| Morse Cipher | 16.78 | Not recorded | 68fa751a1e209131cea65ca8 |
| ReNeLLM | 21.13 | Not recorded | 68fa751a1e209131cea65ca8 |
| Unicode | 35.25 | Not recorded | 68fa751a1e209131cea65ca8 |
| UTF-8 | 35.25 | Not recorded | 68fa751a1e209131cea65ca8 |
| Origin | 41.81 | Not recorded | 68fa751a1e209131cea65ca8 |
| Caesar Cipher | 192.17 | Not recorded | 68fa751a1e209131cea65ca8 |
| FlipAttack | 909.42 | Not recorded | 68fa751a1e209131cea65ca8 |

### ArtPrompt

PPL Mean: 2.34. Result ID: 6a0adb6a47f34ce924f14eed.

[Reporting paper](<https://www.sota2.com/research/paper/68fa751a1e209131cea65ca8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d39c.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Ascii

PPL Mean: 7.25. Result ID: 6a0adb6a47f34ce924f14eeb.

[Reporting paper](<https://www.sota2.com/research/paper/68fa751a1e209131cea65ca8.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Base64

PPL Mean: 9.78. Result ID: 6a0adb6a47f34ce924f14eec.

[Reporting paper](<https://www.sota2.com/research/paper/68fa751a1e209131cea65ca8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa70021e209131cea4887c.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Morse Cipher

PPL Mean: 16.78. Result ID: 6a0adb6a47f34ce924f14ee9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa751a1e209131cea65ca8.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### ReNeLLM

PPL Mean: 21.13. Result ID: 6a0adb6a47f34ce924f14eee.

[Reporting paper](<https://www.sota2.com/research/paper/68fa751a1e209131cea65ca8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72d71e209131cea58d97.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Unicode

PPL Mean: 35.25. Result ID: 6a0adb6a47f34ce924f14ee8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa751a1e209131cea65ca8.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### UTF-8

PPL Mean: 35.25. Result ID: 6a0adb6a47f34ce924f14eea.

[Reporting paper](<https://www.sota2.com/research/paper/68fa751a1e209131cea65ca8.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Origin

PPL Mean: 41.81. Result ID: 6a0adb6a47f34ce924f14ee6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa751a1e209131cea65ca8.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Caesar Cipher

PPL Mean: 192.17. Result ID: 6a0adb6a47f34ce924f14ee7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa751a1e209131cea65ca8.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### FlipAttack

PPL Mean: 909.42. Result ID: 6a0adb6a47f34ce924f14eef.

[Reporting paper](<https://www.sota2.com/research/paper/68fa751a1e209131cea65ca8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa751a1e209131cea65ca8.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

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