| language: [en] | |
| license: mit | |
| tags: [cybersecurity, vision, multimodal, fine-tuned, amd-mi300x, rocm, lora, qwen2.5] | |
| base_model: Qwen/Qwen2.5-7B-Instruct | |
| pipeline_tag: text-generation | |
| # IMMUNIS-Vision | |
| The Visual Threat Analyst of IMMUNIS ACIN — detects QR phishing, deepfakes, document forgery, and steganography. | |
| ## Training | |
| | Property | Value | | |
| |----------|-------| | |
| | Base Model | Qwen/Qwen2.5-7B-Instruct | | |
| | Method | bf16 LoRA (rank 64, alpha 128, 2.08% params) | | |
| | Data | 18,000 visual threat classification examples (6 types) | | |
| | Hardware | AMD Instinct MI300X (192GB HBM3) + ROCm 7.0 | | |
| | Training Time | ~55 minutes | | |
| ## Threat Types | |
| QR Phishing, Deepfake, Document Forgery, Steganography, Screenshot Phishing, Brand Impersonation. | |
| ## Part of IMMUNIS ACIN | |
| Agent 8 (Visual Threat Analyst) in a 12-agent adversarial coevolutionary immune network. AMD Developer Hackathon — Track 3 (Vision & Multimodal). | |