Downloads · 30 days
0
alirezaaminzadeh/cve-riskrank-baseline
cve-riskrank-baseline is a text classification model from alirezaaminzadeh. Use it when you need a label for a piece of text. It is set up for sklearn. The card lists the license as mit.
Calibrated classical baselines for CVE exploitation-priority ranking on alirezaaminzadeh/cve-riskrank.
Downloads · 30 days
0
Access
Public
Updated Aug 11, 2026
Repo size
3.4 MB
Likes
0
Public
Click a slice to open those files.
.joblib3.1 MB · 91%
From the Hugging Face model README
Calibrated classical baselines for CVE exploitation-priority ranking on
alirezaaminzadeh/cve-riskrank.
| Key | Algorithm | Features |
|---|---|---|
logistic_regression | Logistic Regression | TF-IDF (description) + tabular |
lightgbm | LightGBM classifier | TF-IDF (description) + tabular |
Both models are wrapped with isotonic calibration fitted on the temporal
validation split only. A CVSS-score-only ranker is reported in metrics.json
as the reference every model must beat.
| File | Description |
|---|---|
model.joblib | Preprocessor, models, calibrators, feature column lists |
metrics.json | Validation / test metrics (raw + calibrated) |
test_predictions.parquet | Per-CVE scores on the temporal test split |
import joblib
from huggingface_hub import hf_hub_download
import pandas as pd
path = hf_hub_download("alirezaaminzadeh/cve-riskrank-baseline", "model.joblib")
bundle = joblib.load(path)
row = pd.DataFrame([{
"description": "A remote code execution vulnerability in ...",
**{c: 0 for c in bundle["numeric_cols"]},
**{c: 0 for c in bundle["bool_cols"]},
**{c: "UNKNOWN" for c in bundle["cat_cols"]},
}])
X = bundle["preprocessor"].transform(row)
raw = bundle["models"]["lightgbm"].predict_proba(X)[:, 1]
prob = bundle["calibrators"]["lightgbm"].predict(raw)
print(float(prob[0]))
MIT