# Structural Erasure on WMDP-cyber Source: [Structural Erasure on WMDP-cyber]() Collection index: [llms.txt]() Data updated: 2026-05-15T09:59:37.238Z Evaluation of Structural Erasure performance on the WMDP-cyber dataset, reporting metrics including CAD, AS-C, and AS-NC. 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 | Structural Erasure | | Dataset (including any stated split) | WMDP-cyber | | Metric | CAD | | Unit | SCALAR | | Direction | Lower is better | | Results loaded | 16 of 16 | | 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 - [CAD](): Lower is better; unit: SCALAR. - [AS-C](): Lower is better; unit: SCALAR. - [AS-NC](): Lower is better; unit: SCALAR. ## Recorded results | Method | CAD | Result date | Report paper | | --- | --- | --- | --- | | Zero-shot | 0 | Not recorded | 6a06a8c9dbc836b87d82a365 | | Zero-shot | 0 | Not recorded | 6a06a8c9dbc836b87d82a365 | | LUNAR | 0.027 | Not recorded | 6a06a8c9dbc836b87d82a365 | | LUNAR | 0.029 | Not recorded | 6a06a8c9dbc836b87d82a365 | | Surgical GA | 0.248 | Not recorded | 6a06a8c9dbc836b87d82a365 | | Surgical GA | 0.262 | Not recorded | 6a06a8c9dbc836b87d82a365 | | NPO | 0.356 | Not recorded | 6a06a8c9dbc836b87d82a365 | | Global GA | 0.44 | Not recorded | 6a06a8c9dbc836b87d82a365 | | NPO | 0.624 | Not recorded | 6a06a8c9dbc836b87d82a365 | | SimNPO | 0.63 | Not recorded | 6a06a8c9dbc836b87d82a365 | | GU+SimNPO | 0.771 | Not recorded | 6a06a8c9dbc836b87d82a365 | | Global GA | 0.782 | Not recorded | 6a06a8c9dbc836b87d82a365 | | MANSU | 1.053 | Not recorded | 6a06a8c9dbc836b87d82a365 | | MANSU | 1.118 | Not recorded | 6a06a8c9dbc836b87d82a365 | | GU+SimNPO | 1.366 | Not recorded | 6a06a8c9dbc836b87d82a365 | | SimNPO | 1.523 | Not recorded | 6a06a8c9dbc836b87d82a365 | ### Zero-shot CAD: 0. Result ID: 6a06ee7247f34ce924ef1bef. [Reporting paper]() Recorded evaluation configuration: {"Model":"Llama-3.1-8B-Instruct"}. This record does not assert independent reproduction. ### Zero-shot CAD: 0. Result ID: 6a06ee7247f34ce924ef1bf7. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen-3-8B"}. This record does not assert independent reproduction. ### LUNAR CAD: 0.027. Result ID: 6a06ee7247f34ce924ef1bfd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen-3-8B"}. This record does not assert independent reproduction. ### LUNAR CAD: 0.029. Result ID: 6a06ee7247f34ce924ef1bf5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Llama-3.1-8B-Instruct"}. This record does not assert independent reproduction. ### Surgical GA CAD: 0.248. Result ID: 6a06ee7247f34ce924ef1bf1. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model":"Llama-3.1-8B-Instruct"}. This record does not assert independent reproduction. ### Surgical GA CAD: 0.262. Result ID: 6a06ee7247f34ce924ef1bf9. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model":"Qwen-3-8B"}. This record does not assert independent reproduction. ### NPO CAD: 0.356. Result ID: 6a06ee7247f34ce924ef1bf2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Llama-3.1-8B-Instruct"}. This record does not assert independent reproduction. ### Global GA CAD: 0.44. Result ID: 6a06ee7247f34ce924ef1bf0. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model":"Llama-3.1-8B-Instruct"}. This record does not assert independent reproduction. ### NPO CAD: 0.624. Result ID: 6a06ee7247f34ce924ef1bfa. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen-3-8B"}. This record does not assert independent reproduction. ### SimNPO CAD: 0.63. Result ID: 6a06ee7247f34ce924ef1bfb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen-3-8B"}. This record does not assert independent reproduction. ### GU+SimNPO CAD: 0.771. Result ID: 6a06ee7247f34ce924ef1bfc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen-3-8B"}. This record does not assert independent reproduction. ### Global GA CAD: 0.782. Result ID: 6a06ee7247f34ce924ef1bf8. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model":"Qwen-3-8B"}. This record does not assert independent reproduction. ### MANSU CAD: 1.053. Result ID: 6a06ee7247f34ce924ef1bfe. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen-3-8B"}. This record does not assert independent reproduction. ### MANSU CAD: 1.118. Result ID: 6a06ee7247f34ce924ef1bf6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Llama-3.1-8B-Instruct"}. This record does not assert independent reproduction. ### GU+SimNPO CAD: 1.366. Result ID: 6a06ee7247f34ce924ef1bf4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Llama-3.1-8B-Instruct"}. This record does not assert independent reproduction. ### SimNPO CAD: 1.523. Result ID: 6a06ee7247f34ce924ef1bf3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Llama-3.1-8B-Instruct"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.