# Harmful prompt detection on WJB Source: [Harmful prompt detection on WJB]() Collection index: [llms.txt]() Data updated: 2026-06-02T11:47:12.429Z Evaluation of model performance on the WJB dataset for harmful prompt detection, measured using F1 score. 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 | Harmful prompt detection | | Dataset (including any stated split) | WJB | | Metric | F1 Score | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 17 of 17 | | 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 - [F1 Score](): Higher is better; unit: PERCENT. ## Recorded results | Method | F1 Score | Result date | Report paper | | --- | --- | --- | --- | | MLPM | 97.55 | Not recorded | 68fa79d81e209131cea7fb79 | | WildGuard | 97.1 | Not recorded | 68fa79d81e209131cea7fb79 | | Ayub & Majumdar | 96.84 | Not recorded | 68fa79d81e209131cea7fb79 | | GraniteGuardian-3-1-8B | 96.75 | Not recorded | 68fa79d81e209131cea7fb79 | | MLPM | 94.69 | Not recorded | 68fa79d81e209131cea7fb79 | | MLPM | 93.65 | Not recorded | 68fa79d81e209131cea7fb79 | | Abdelnabi et al. | 93.48 | Not recorded | 68fa79d81e209131cea7fb79 | | Abdelnabi et al. | 93.27 | Not recorded | 68fa79d81e209131cea7fb79 | | MLPM | 92 | Not recorded | 68fa79d81e209131cea7fb79 | | Abdelnabi et al. | 91.13 | Not recorded | 68fa79d81e209131cea7fb79 | | Abdelnabi et al. | 90.44 | Not recorded | 68fa79d81e209131cea7fb79 | | Ayub & Majumdar | 87.53 | Not recorded | 68fa79d81e209131cea7fb79 | | Ayub & Majumdar | 82.85 | Not recorded | 68fa79d81e209131cea7fb79 | | Ayub & Majumdar | 81.31 | Not recorded | 68fa79d81e209131cea7fb79 | | Aegis-Guard-D | 75.44 | Not recorded | 68fa79d81e209131cea7fb79 | | LlamaGuard3 | 67.83 | Not recorded | 68fa79d81e209131cea7fb79 | | ShieldGemma-9B | 59.94 | Not recorded | 68fa79d81e209131cea7fb79 | ### MLPM F1 Score: 97.55. Result ID: 6a1ec1e0fbc373592c1a1f5f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"OLMo2-7B-Inst","Methodology":"Latent-Based"}. This record does not assert independent reproduction. ### WildGuard F1 Score: 97.1. Result ID: 6a1ec1e0fbc373592c1a1f64. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Methodology":"Guard Model"}. This record does not assert independent reproduction. ### Ayub & Majumdar F1 Score: 96.84. Result ID: 6a1ec1e0fbc373592c1a1f5d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"OLMo2-7B-Inst","Methodology":"Latent-Based"}. This record does not assert independent reproduction. ### GraniteGuardian-3-1-8B F1 Score: 96.75. Result ID: 6a1ec1e0fbc373592c1a1f62. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Methodology":"Guard Model"}. This record does not assert independent reproduction. ### MLPM F1 Score: 94.69. Result ID: 6a1ec1e0fbc373592c1a1f56. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Llama-8B-Inst","Methodology":"Latent-Based"}. This record does not assert independent reproduction. ### MLPM F1 Score: 93.65. Result ID: 6a1ec1e0fbc373592c1a1f5c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Mistral-7B-Inst","Methodology":"Latent-Based"}. This record does not assert independent reproduction. ### Abdelnabi et al. F1 Score: 93.48. Result ID: 6a1ec1e0fbc373592c1a1f5e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"OLMo2-7B-Inst","Methodology":"Latent-Based"}. This record does not assert independent reproduction. ### Abdelnabi et al. F1 Score: 93.27. Result ID: 6a1ec1e0fbc373592c1a1f55. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Llama-8B-Inst","Methodology":"Latent-Based"}. This record does not assert independent reproduction. ### MLPM F1 Score: 92. Result ID: 6a1ec1e0fbc373592c1a1f59. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-8B-Inst","Methodology":"Latent-Based"}. This record does not assert independent reproduction. ### Abdelnabi et al. F1 Score: 91.13. Result ID: 6a1ec1e0fbc373592c1a1f5b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Mistral-7B-Inst","Methodology":"Latent-Based"}. This record does not assert independent reproduction. ### Abdelnabi et al. F1 Score: 90.44. Result ID: 6a1ec1e0fbc373592c1a1f58. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-8B-Inst","Methodology":"Latent-Based"}. This record does not assert independent reproduction. ### Ayub & Majumdar F1 Score: 87.53. Result ID: 6a1ec1e0fbc373592c1a1f5a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Mistral-7B-Inst","Methodology":"Latent-Based"}. This record does not assert independent reproduction. ### Ayub & Majumdar F1 Score: 82.85. Result ID: 6a1ec1e0fbc373592c1a1f54. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Llama-8B-Inst","Methodology":"Latent-Based"}. This record does not assert independent reproduction. ### Ayub & Majumdar F1 Score: 81.31. Result ID: 6a1ec1e0fbc373592c1a1f57. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-8B-Inst","Methodology":"Latent-Based"}. This record does not assert independent reproduction. ### Aegis-Guard-D F1 Score: 75.44. Result ID: 6a1ec1e0fbc373592c1a1f60. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Methodology":"Guard Model"}. This record does not assert independent reproduction. ### LlamaGuard3 F1 Score: 67.83. Result ID: 6a1ec1e0fbc373592c1a1f61. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Methodology":"Guard Model"}. This record does not assert independent reproduction. ### ShieldGemma-9B F1 Score: 59.94. Result ID: 6a1ec1e0fbc373592c1a1f63. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Methodology":"Guard Model"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.