Application Details

Information about the platform's technical infrastructure and API endpoints.

API Endpoints

The platform provides the following RESTful API endpoints:

GET /models
GET /models/{id}
GET /models/{id}/score
POST /admin/scan

Static Analysis Rules

Our static analysis engine looks for specific patterns in model files:

Usage of dangerous functions (e.g., eval, exec).
Insecure deserialization (e.g., use of pickle).
Code that makes outbound network calls.
Suspicious file system operations.

Database Schema

The system uses the following basic schema to store model and score data:

CREATE TABLE models (
   id TEXT PRIMARY KEY,
   name TEXT,
   owner TEXT,
   hf_url TEXT,
   license TEXT,
   last_updated TIMESTAMP
);

CREATE TABLE scores (
   model_id TEXT REFERENCES models(id),
   computed_at TIMESTAMP DEFAULT now(),
   overall_score NUMERIC(5,2),
   PRIMARY KEY (model_id, computed_at)
);

Scoring Function (Example)

The overall score is calculated by taking a weighted average of different metrics:

def compute_overall_score(metrics):
    weights = {
      "provenance": 20, "usage": 20, "maintenance": 15, 
      "evaluation": 20, "safety": 15, "code_quality": 10
    }
    
    weighted_sum = 0
    total_weight = sum(weights.values())

    for k, w in weights.items():
        v = metrics.get(k, 0)
        weighted_sum += v * w

    return round(weighted_sum / total_weight, 2)

Dynamic Tests (Expanded)

Dynamic analysis uses various inputs to test the model behavior.

Technical Examples Home