eromang/cyberscale-technical-v1

Model

0

stars

4

commits

1

linked in READMEs

Apr 1, 2026

updated

cyber-blueprint
cybersecurity
endpoints_compatible
incident-classification
model-index
modernbert
nis2
safetensors
technical-severity
text-classification
text-embeddings-inference
transformers
Browse cluster: Transformer Text Classification & Embeddings

README

CyberScale Technical Severity v1

Incident technical severity classifier (T1-T4). Assesses observable technical impact from a CSIRT perspective based on structured incident fields.

Model Description

  • Architecture: ModernBERT-base with 4-class classification head
  • Training: 8,000 parametric incident scenarios (50 templates × field combinations)
  • Confidence: Monte Carlo dropout (20 passes) maps variance to high/medium/low
  • Labels: T1 (minor) → T4 (catastrophic)

Intended Use

Classify the technical severity of cyber incidents based on service disruption, affected entities, cascading effects, and data compromise. Part of the CyberScale dual-scale incident classification system (T-level + O-level → Blueprint matrix).

Input format: <description> [SEP] disruption: <level> entities: <N> sectors: <N> cascading: <level> data_compromise: <level>

Metrics

MetricValueTarget
Accuracy1.0> 75%
Macro F11.0> 75%

Citation

Part of the CyberScale project — multi-phase cyber severity assessment MCP server.

Contributors

eromang

4 commits

eromang/cyberscale-technical-v1

Model

0

stars

4

commits

1

linked in READMEs

Apr 1, 2026

updated

cyber-blueprint
cybersecurity
endpoints_compatible
incident-classification
model-index
modernbert
nis2
safetensors
technical-severity
text-classification
text-embeddings-inference
transformers
Browse cluster: Transformer Text Classification & Embeddings

README

CyberScale Technical Severity v1

Incident technical severity classifier (T1-T4). Assesses observable technical impact from a CSIRT perspective based on structured incident fields.

Model Description

  • Architecture: ModernBERT-base with 4-class classification head
  • Training: 8,000 parametric incident scenarios (50 templates × field combinations)
  • Confidence: Monte Carlo dropout (20 passes) maps variance to high/medium/low
  • Labels: T1 (minor) → T4 (catastrophic)

Intended Use

Classify the technical severity of cyber incidents based on service disruption, affected entities, cascading effects, and data compromise. Part of the CyberScale dual-scale incident classification system (T-level + O-level → Blueprint matrix).

Input format: <description> [SEP] disruption: <level> entities: <N> sectors: <N> cascading: <level> data_compromise: <level>

Metrics

MetricValueTarget
Accuracy1.0> 75%
Macro F11.0> 75%

Citation

Part of the CyberScale project — multi-phase cyber severity assessment MCP server.

Contributors

eromang

4 commits