trinadhsriram02/ModelSentinel Security Scan
Scan PyTorch models for backdoors before deployment
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Stale
- License
- None
Inputs
| name | description | required | default |
|---|---|---|---|
| model_path | Path to the model file to scan | yes | — |
| risk_threshold | Maximum acceptable risk score (0-100) | no | 40 |
| num_classes | Number of output classes in the model | no | 10 |
| fail_on_detection | Fail the CI pipeline if backdoor is detected | no | true |
Outputs
| name | description |
|---|---|
| verdict | Scan verdict CLEAN SUSPICIOUS or BACKDOORED |
| risk_score | Risk score 0-100 |
| safe_to_deploy | Whether the model is safe to deploy |