# Evasion Detection (Subtask 2) on Official 308-sample (test) Source: [Evasion Detection (Subtask 2) on Official 308-sample (test)]() Collection index: [llms.txt]() Data updated: 2026-06-16T08:49:50.975Z Evaluation of Evasion Detection (Subtask 2) performance on the test split of the Official 308-sample dataset, measured by Macro F1. 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 | Evasion Detection (Subtask 2) | | Dataset (including any stated split) | Official 308-sample (test) | | Metric | Macro F1 | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 6 of 6 | | 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 - [Macro F1](): Higher is better; unit: PERCENT. ## Recorded results | Method | Macro F1 | Result date | Report paper | | --- | --- | --- | --- | | Organizer revised baseline | 57 | Not recorded | 6a30d8a1d33dfb23e783ccab | | Official submission | 56 | Not recorded | 6a30d8a1d33dfb23e783ccab | | Grok Rsn. Few-shot Hier. | 51.47 | Not recorded | 6a30d8a1d33dfb23e783ccab | | Grok Rsn. Zero-shot Flat | 51.26 | Not recorded | 6a30d8a1d33dfb23e783ccab | | DeepSeek Rsn. Few-shot Hier. | 51.15 | Not recorded | 6a30d8a1d33dfb23e783ccab | | Best Qwen3 PEFT | 36.3 | Not recorded | 6a30d8a1d33dfb23e783ccab | ### Organizer revised baseline Macro F1: 57. Result ID: 6a310dd4fbc373592c2d6e5f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Status":"Blind/private leaderboard"}. This record does not assert independent reproduction. ### Official submission Macro F1: 56. Result ID: 6a310dd4fbc373592c2d6e5e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Team":"ttda704","Status":"Blind/private leaderboard"}. This record does not assert independent reproduction. ### Grok Rsn. Few-shot Hier. Macro F1: 51.47. Result ID: 6a310dd4fbc373592c2d6e5d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Grok-4-Fast","Mode":"Reasoning","In-context learning":"Few-shot","Taxonomy":"Hierarchical"}. This record does not assert independent reproduction. ### Grok Rsn. Zero-shot Flat Macro F1: 51.26. Result ID: 6a310dd4fbc373592c2d6e5c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Grok-4-Fast","Mode":"Reasoning","In-context learning":"Zero-shot","Taxonomy":"Flat"}. This record does not assert independent reproduction. ### DeepSeek Rsn. Few-shot Hier. Macro F1: 51.15. Result ID: 6a310dd4fbc373592c2d6e5b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"DeepSeek-V3.2","Mode":"Reasoning","In-context learning":"Few-shot","Taxonomy":"Hierarchical"}. This record does not assert independent reproduction. ### Best Qwen3 PEFT Macro F1: 36.3. Result ID: 6a310dd4fbc373592c2d6e5a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen3-32B","In-context learning":"Few-shot CoT"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.