# Jailbreak Attack Success Evaluation on RedTeam2K SD+TYPO Source: [Jailbreak Attack Success Evaluation on RedTeam2K SD+TYPO]() Collection index: [llms.txt]() Data updated: 2026-03-19T09:10:07.442Z Evaluation of model performance against adversarial jailbreak prompts using the RedTeam2K SD+TYPO dataset, measured by Attack Success Rate (ASR). 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 Attack Success Evaluation | | Dataset (including any stated split) | RedTeam2K SD+TYPO | | Metric | Attack Success Rate (ASR) | | 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 - [Attack Success Rate (ASR)](): Higher is better; unit: PERCENT. ## Recorded results | Method | Attack Success Rate (ASR) | Result date | Report paper | | --- | --- | --- | --- | | CMRM | 51 | Not recorded | 69bb831d0636e5bf4afc1af1 | | LLaVA-1.5-7B (Baseline) | 49.8 | Not recorded | 69bb831d0636e5bf4afc1af1 | | InternVL-Chat-19B (Baseline) | 43.6 | Not recorded | 69bb831d0636e5bf4afc1af1 | | ShiftDC | 42.5 | Not recorded | 69bb831d0636e5bf4afc1af1 | | ShareGPT4V-7B (Baseline) | 39.2 | Not recorded | 69bb831d0636e5bf4afc1af1 | | ECSO | 28.6 | Not recorded | 69bb831d0636e5bf4afc1af1 | | ShiftDC | 27.6 | Not recorded | 69bb831d0636e5bf4afc1af1 | | ECSO | 26.1 | Not recorded | 69bb831d0636e5bf4afc1af1 | | AdaShield | 22.2 | Not recorded | 69bb831d0636e5bf4afc1af1 | | JRS-Rem | 21.9 | Not recorded | 69bb831d0636e5bf4afc1af1 | | CMRM | 21.4 | Not recorded | 69bb831d0636e5bf4afc1af1 | | ShiftDC | 20.1 | Not recorded | 69bb831d0636e5bf4afc1af1 | | ECSO | 19.8 | Not recorded | 69bb831d0636e5bf4afc1af1 | | AdaShield | 19.6 | Not recorded | 69bb831d0636e5bf4afc1af1 | | AdaShield | 15 | Not recorded | 69bb831d0636e5bf4afc1af1 | | CMRM | 12.1 | Not recorded | 69bb831d0636e5bf4afc1af1 | | JRS-Rem | 9.6 | Not recorded | 69bb831d0636e5bf4afc1af1 | | JRS-Rem | 9.1 | Not recorded | 69bb831d0636e5bf4afc1af1 | ### CMRM Attack Success Rate (ASR): 51. Result ID: 69bbbd6a47f34ce924a5d1e5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VLM":"LLaVA-1.5-7B"}. This record does not assert independent reproduction. ### LLaVA-1.5-7B (Baseline) Attack Success Rate (ASR): 49.8. Result ID: 69bbbd6a47f34ce924a5d1e1. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"VLM":"LLaVA-1.5-7B"}. This record does not assert independent reproduction. ### InternVL-Chat-19B (Baseline) Attack Success Rate (ASR): 43.6. Result ID: 69bbbd6a47f34ce924a5d1ed. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"VLM":"InternVL-Chat-19B"}. This record does not assert independent reproduction. ### ShiftDC Attack Success Rate (ASR): 42.5. Result ID: 69bbbd6a47f34ce924a5d1e4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VLM":"LLaVA-1.5-7B"}. This record does not assert independent reproduction. ### ShareGPT4V-7B (Baseline) Attack Success Rate (ASR): 39.2. Result ID: 69bbbd6a47f34ce924a5d1e7. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"VLM":"ShareGPT4V-7B"}. This record does not assert independent reproduction. ### ECSO Attack Success Rate (ASR): 28.6. Result ID: 69bbbd6a47f34ce924a5d1e3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VLM":"LLaVA-1.5-7B"}. This record does not assert independent reproduction. ### ShiftDC Attack Success Rate (ASR): 27.6. Result ID: 69bbbd6a47f34ce924a5d1f0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VLM":"InternVL-Chat-19B"}. This record does not assert independent reproduction. ### ECSO Attack Success Rate (ASR): 26.1. Result ID: 69bbbd6a47f34ce924a5d1e9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VLM":"ShareGPT4V-7B"}. This record does not assert independent reproduction. ### AdaShield Attack Success Rate (ASR): 22.2. Result ID: 69bbbd6a47f34ce924a5d1e2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VLM":"LLaVA-1.5-7B"}. This record does not assert independent reproduction. ### JRS-Rem Attack Success Rate (ASR): 21.9. Result ID: 69bbbd6a47f34ce924a5d1e6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VLM":"LLaVA-1.5-7B"}. This record does not assert independent reproduction. ### CMRM Attack Success Rate (ASR): 21.4. Result ID: 69bbbd6a47f34ce924a5d1f1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VLM":"InternVL-Chat-19B"}. This record does not assert independent reproduction. ### ShiftDC Attack Success Rate (ASR): 20.1. Result ID: 69bbbd6a47f34ce924a5d1ea. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VLM":"ShareGPT4V-7B"}. This record does not assert independent reproduction. ### ECSO Attack Success Rate (ASR): 19.8. Result ID: 69bbbd6a47f34ce924a5d1ef. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VLM":"InternVL-Chat-19B"}. This record does not assert independent reproduction. ### AdaShield Attack Success Rate (ASR): 19.6. Result ID: 69bbbd6a47f34ce924a5d1ee. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VLM":"InternVL-Chat-19B"}. This record does not assert independent reproduction. ### AdaShield Attack Success Rate (ASR): 15. Result ID: 69bbbd6a47f34ce924a5d1e8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VLM":"ShareGPT4V-7B"}. This record does not assert independent reproduction. ### CMRM Attack Success Rate (ASR): 12.1. Result ID: 69bbbd6a47f34ce924a5d1eb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VLM":"ShareGPT4V-7B"}. This record does not assert independent reproduction. ### JRS-Rem Attack Success Rate (ASR): 9.6. Result ID: 69bbbd6a47f34ce924a5d1f2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VLM":"InternVL-Chat-19B"}. This record does not assert independent reproduction. ### JRS-Rem Attack Success Rate (ASR): 9.1. Result ID: 69bbbd6a47f34ce924a5d1ec. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"VLM":"ShareGPT4V-7B"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.