Cybersecurity Models
Collection
5 items • Updated
Configuration Parsing Warning:In config.json: "quantization_config.modules_to_not_convert" must be an array
Foundation-Sec-8B-Instruct-AWQ
4-bit AWQ quantization of fdtn-ai/Foundation-Sec-8B-Instruct.
Evaluation
Matching Cisco Foundation-Sec-8B-Instruct paper protocol (arXiv:2508.01059)
CTI-MCQA (CTIBench, 2500 questions)
| Model | Accuracy |
|---|---|
| Foundation-Sec-8B-Instruct (Cisco reported) | 0.644 |
| Foundation-Sec-8B-Instruct (BF16, our re-run) | 0.650 |
| Foundation-Sec-8B-Instruct-AWQ (this model) | 0.628 |
CTI-RCM (CTIBench, 1000 CVE→CWE)
| Model | Accuracy |
|---|---|
| Foundation-Sec-8B-Instruct (Cisco reported) | 0.692 |
| Foundation-Sec-8B-Instruct-AWQ (this model) | 0.649 |
CTI-VSP (CTIBench, 1000 CVE→CVSS)
| Model | CVSS Score |
|---|---|
| Foundation-Sec-8B-Instruct (Cisco reported) | 0.802 |
| Foundation-Sec-8B-Instruct-AWQ (this model) | 0.831 |
- Downloads last month
- 12
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for Varadrajan/Foundation-Sec-8B-Instruct-AWQ
Base model
meta-llama/Llama-3.1-8B Finetuned
fdtn-ai/Foundation-Sec-8B Finetuned
fdtn-ai/Foundation-Sec-8B-Instruct