| license: apache-2.0 | |
| task_categories: | |
| - text-generation | |
| language: | |
| - en | |
| tags: | |
| - cybersecurity | |
| - vulnerability | |
| - exploit | |
| size_categories: | |
| - 1K<n<10K | |
| # LXM-GENTLEMAN Dataset | |
| This is the official fine-tuning dataset used to train the **LXM-GENTLEMAN** model, a cybersecurity-focused assistant from the LACUNEX (LXM) family by Shasradha Karmakar. | |
| ## Dataset Structure | |
| * **File:** `lxm-gentleman-dataset.jsonl` | |
| * **Format:** JSON Lines (JSONL) with `instruction` and `output` keys. | |
| * **Total Rows:** 4,636 unique examples. | |
| ## Data Sources | |
| The dataset is a curated mix of the following security datasets: | |
| 1. Hand-crafted identity Q&A pairs. | |
| 2. `darkknight25/software_vulnerabilities` | |
| 3. `darkknight25/Exploit_Database` | |
| 4. `AlicanKiraz0/All-CVE-Records` | |
| 5. `WaiperOK/exploitdb-dataset` | |
| All data has been parsed, mapped to instruction format, and deduplicated. | |