# Jailbreak Defense on AdvBench PAIR attack

Source: [Jailbreak Defense on AdvBench PAIR attack](<https://www.sota2.com/research/sota/jailbreak-defense-on-advbench-pair-attack>)

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

Data updated: 2026-02-26T05:40:31.953Z

Evaluation of model robustness against adversarial jailbreak prompts derived from the AdvBench PAIR attack dataset, measured by Defense Success Rate (DSR).

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 | Jailbreak Defense |
| Dataset (including any stated split) | AdvBench PAIR attack |
| Metric | DSR |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 35 of 35 |
| 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

- [DSR](<https://www.sota2.com/research/sota/jailbreak-defense-on-advbench-pair-attack.md?metric=DSR>): Higher is better; unit: PERCENT.

## Recorded results

| Method | DSR | Result date | Report paper |
| --- | --- | --- | --- |
| SmoothLLM | 98 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Backtranslation | 98 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Backtranslation | 94 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Backtranslation | 94 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Paraphrasing | 90 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Backtranslation | 88 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Response check | 82 | Not recorded | 68fa738f1e209131cea5d3a7 |
| SmoothLLM | 82 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Paraphrasing | 80 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Paraphrasing | 78 | Not recorded | 68fa738f1e209131cea5d3a7 |
| SmoothLLM | 76 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Backtranslation | 76 | Not recorded | 68fa738f1e209131cea5d3a7 |
| SmoothLLM | 70 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Backtranslation | 70 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Response check | 68 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Response check | 66 | Not recorded | 68fa738f1e209131cea5d3a7 |
| No defense | 64 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Paraphrasing | 64 | Not recorded | 68fa738f1e209131cea5d3a7 |
| No defense | 64 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Backtranslation | 56 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Paraphrasing | 54 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Response check | 46 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Paraphrasing | 38 | Not recorded | 68fa738f1e209131cea5d3a7 |
| No defense | 36 | Not recorded | 68fa738f1e209131cea5d3a7 |
| No defense | 36 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Response check | 32 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Response check | 30 | Not recorded | 68fa738f1e209131cea5d3a7 |
| SmoothLLM | 28 | Not recorded | 68fa738f1e209131cea5d3a7 |
| SmoothLLM | 20 | Not recorded | 68fa738f1e209131cea5d3a7 |
| No defense | 8 | Not recorded | 68fa738f1e209131cea5d3a7 |
| No defense | 8 | Not recorded | 68fa738f1e209131cea5d3a7 |
| No defense | 8 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Response check | 6 | Not recorded | 68fa738f1e209131cea5d3a7 |
| Paraphrasing | 4 | Not recorded | 68fa738f1e209131cea5d3a7 |
| SmoothLLM | 2 | Not recorded | 68fa738f1e209131cea5d3a7 |

### SmoothLLM

DSR: 98. Result ID: 699300653a6c31484912f789.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73961e209131cea5d658.md>)

Recorded evaluation configuration: {"Target Model":"Llama-2-13B","Attack Setting":"w/o defense"}. This record does not assert independent reproduction.

### Backtranslation

DSR: 98. Result ID: 699300653a6c31484912f78c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Implementation](<https://github.com/YihanWang617/LLM-Jailbreaking-Defense-Backtranslation>)

Recorded evaluation configuration: {"Target Model":"Llama-2-13B","Attack Setting":"w/o defense"}. This record does not assert independent reproduction.

### Backtranslation

DSR: 94. Result ID: 699300653a6c31484912f791.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Implementation](<https://github.com/YihanWang617/LLM-Jailbreaking-Defense-Backtranslation>)

Recorded evaluation configuration: {"Target Model":"Vicuna-13B","Attack Setting":"w/o defense"}. This record does not assert independent reproduction.

### Backtranslation

DSR: 94. Result ID: 699300653a6c31484912f79b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Implementation](<https://github.com/YihanWang617/LLM-Jailbreaking-Defense-Backtranslation>)

Recorded evaluation configuration: {"Target Model":"Llama-2-13B","Attack Setting":"w/ defense"}. This record does not assert independent reproduction.

### Paraphrasing

DSR: 90. Result ID: 699300653a6c31484912f78a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa707f1e209131cea4b7c6.md>)

Recorded evaluation configuration: {"Target Model":"Llama-2-13B","Attack Setting":"w/o defense"}. This record does not assert independent reproduction.

### Backtranslation

DSR: 88. Result ID: 699300653a6c31484912f787.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Implementation](<https://github.com/YihanWang617/LLM-Jailbreaking-Defense-Backtranslation>)

Recorded evaluation configuration: {"Target Model":"GPT-3.5-turbo","Attack Setting":"w/o defense"}. This record does not assert independent reproduction.

### Response check

DSR: 82. Result ID: 699300653a6c31484912f78b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

Recorded evaluation configuration: {"Target Model":"Llama-2-13B","Attack Setting":"w/o defense"}. This record does not assert independent reproduction.

### SmoothLLM

DSR: 82. Result ID: 699300653a6c31484912f798.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73961e209131cea5d658.md>)

Recorded evaluation configuration: {"Target Model":"Llama-2-13B","Attack Setting":"w/ defense"}. This record does not assert independent reproduction.

### Paraphrasing

DSR: 80. Result ID: 699300653a6c31484912f78f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa707f1e209131cea4b7c6.md>)

Recorded evaluation configuration: {"Target Model":"Vicuna-13B","Attack Setting":"w/o defense"}. This record does not assert independent reproduction.

### Paraphrasing

DSR: 78. Result ID: 699300653a6c31484912f785.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa707f1e209131cea4b7c6.md>)

Recorded evaluation configuration: {"Target Model":"GPT-3.5-turbo","Attack Setting":"w/o defense"}. This record does not assert independent reproduction.

### SmoothLLM

DSR: 76. Result ID: 699300653a6c31484912f78e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73961e209131cea5d658.md>)

Recorded evaluation configuration: {"Target Model":"Vicuna-13B","Attack Setting":"w/o defense"}. This record does not assert independent reproduction.

### Backtranslation

DSR: 76. Result ID: 699300653a6c31484912f796.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Implementation](<https://github.com/YihanWang617/LLM-Jailbreaking-Defense-Backtranslation>)

Recorded evaluation configuration: {"Target Model":"GPT-3.5-turbo","Attack Setting":"w/ defense"}. This record does not assert independent reproduction.

### SmoothLLM

DSR: 70. Result ID: 699300653a6c31484912f784.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73961e209131cea5d658.md>)

Recorded evaluation configuration: {"Target Model":"GPT-3.5-turbo","Attack Setting":"w/o defense"}. This record does not assert independent reproduction.

### Backtranslation

DSR: 70. Result ID: 699300653a6c31484912f7b4.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Implementation](<https://github.com/YihanWang617/LLM-Jailbreaking-Defense-Backtranslation>)

Recorded evaluation configuration: {"Target Model":"GPT-3.5-turbo"}. This record does not assert independent reproduction.

### Response check

DSR: 68. Result ID: 699300653a6c31484912f79a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

Recorded evaluation configuration: {"Target Model":"Llama-2-13B","Attack Setting":"w/ defense"}. This record does not assert independent reproduction.

### Response check

DSR: 66. Result ID: 699300653a6c31484912f786.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

Recorded evaluation configuration: {"Target Model":"GPT-3.5-turbo","Attack Setting":"w/o defense"}. This record does not assert independent reproduction.

### No defense

DSR: 64. Result ID: 699300653a6c31484912f788.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

Recorded evaluation configuration: {"Target Model":"Llama-2-13B","Attack Setting":"w/o defense"}. This record does not assert independent reproduction.

### Paraphrasing

DSR: 64. Result ID: 699300653a6c31484912f794.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa707f1e209131cea4b7c6.md>)

Recorded evaluation configuration: {"Target Model":"GPT-3.5-turbo","Attack Setting":"w/ defense"}. This record does not assert independent reproduction.

### No defense

DSR: 64. Result ID: 699300653a6c31484912f797.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

Recorded evaluation configuration: {"Target Model":"Llama-2-13B","Attack Setting":"w/ defense"}. This record does not assert independent reproduction.

### Backtranslation

DSR: 56. Result ID: 699300653a6c31484912f7a0.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Implementation](<https://github.com/YihanWang617/LLM-Jailbreaking-Defense-Backtranslation>)

Recorded evaluation configuration: {"Target Model":"Vicuna-13B","Attack Setting":"w/ defense"}. This record does not assert independent reproduction.

### Paraphrasing

DSR: 54. Result ID: 699300653a6c31484912f799.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa707f1e209131cea4b7c6.md>)

Recorded evaluation configuration: {"Target Model":"Llama-2-13B","Attack Setting":"w/ defense"}. This record does not assert independent reproduction.

### Response check

DSR: 46. Result ID: 699300653a6c31484912f795.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

Recorded evaluation configuration: {"Target Model":"GPT-3.5-turbo","Attack Setting":"w/ defense"}. This record does not assert independent reproduction.

### Paraphrasing

DSR: 38. Result ID: 699300653a6c31484912f7b2.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa707f1e209131cea4b7c6.md>)

Recorded evaluation configuration: {"Target Model":"GPT-3.5-turbo"}. This record does not assert independent reproduction.

### No defense

DSR: 36. Result ID: 699300653a6c31484912f783.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

Recorded evaluation configuration: {"Target Model":"GPT-3.5-turbo","Attack Setting":"w/o defense"}. This record does not assert independent reproduction.

### No defense

DSR: 36. Result ID: 699300653a6c31484912f792.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

Recorded evaluation configuration: {"Target Model":"GPT-3.5-turbo","Attack Setting":"w/ defense"}. This record does not assert independent reproduction.

### Response check

DSR: 32. Result ID: 699300653a6c31484912f790.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

Recorded evaluation configuration: {"Target Model":"Vicuna-13B","Attack Setting":"w/o defense"}. This record does not assert independent reproduction.

### Response check

DSR: 30. Result ID: 699300653a6c31484912f7b3.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

Recorded evaluation configuration: {"Target Model":"GPT-3.5-turbo"}. This record does not assert independent reproduction.

### SmoothLLM

DSR: 28. Result ID: 699300653a6c31484912f793.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73961e209131cea5d658.md>)

Recorded evaluation configuration: {"Target Model":"GPT-3.5-turbo","Attack Setting":"w/ defense"}. This record does not assert independent reproduction.

### SmoothLLM

DSR: 20. Result ID: 699300653a6c31484912f7b1.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73961e209131cea5d658.md>)

Recorded evaluation configuration: {"Target Model":"GPT-3.5-turbo"}. This record does not assert independent reproduction.

### No defense

DSR: 8. Result ID: 699300653a6c31484912f78d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

Recorded evaluation configuration: {"Target Model":"Vicuna-13B","Attack Setting":"w/o defense"}. This record does not assert independent reproduction.

### No defense

DSR: 8. Result ID: 699300653a6c31484912f79c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

Recorded evaluation configuration: {"Target Model":"Vicuna-13B","Attack Setting":"w/ defense"}. This record does not assert independent reproduction.

### No defense

DSR: 8. Result ID: 699300653a6c31484912f7b0.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

Recorded evaluation configuration: {"Target Model":"GPT-3.5-turbo"}. This record does not assert independent reproduction.

### Response check

DSR: 6. Result ID: 699300653a6c31484912f79f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

Recorded evaluation configuration: {"Target Model":"Vicuna-13B","Attack Setting":"w/ defense"}. This record does not assert independent reproduction.

### Paraphrasing

DSR: 4. Result ID: 699300653a6c31484912f79e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa707f1e209131cea4b7c6.md>)

Recorded evaluation configuration: {"Target Model":"Vicuna-13B","Attack Setting":"w/ defense"}. This record does not assert independent reproduction.

### SmoothLLM

DSR: 2. Result ID: 699300653a6c31484912f79d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa738f1e209131cea5d3a7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73961e209131cea5d658.md>)

Recorded evaluation configuration: {"Target Model":"Vicuna-13B","Attack Setting":"w/ defense"}. This record does not assert independent reproduction.

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