---
library_name: transformers
license: cc-by-4.0
base_model: roberta-base
metrics:
- accuracy
tags:
- generated_from_trainer
- text-classification
- classification
- nlp
- vulnerability
model-index:
- name: vulnerability-severity-classification-roberta-base
  results: []
datasets:
- CIRCL/vulnerability-scores
---


# VLAI: A RoBERTa-Based Model for Automated Vulnerability Severity Classification

# Severity classification

This model is a fine-tuned version of [roberta-base](https://huggingface.co/roberta-base) on the dataset [CIRCL/vulnerability-scores](https://huggingface.co/datasets/CIRCL/vulnerability-scores).

The model was presented in the paper [VLAI: A RoBERTa-Based Model for Automated Vulnerability Severity Classification](https://huggingface.co/papers/2507.03607) [[arXiv](https://arxiv.org/abs/2507.03607)].

**Abstract:** VLAI is a transformer-based model that predicts software vulnerability severity levels directly from text descriptions. Built on RoBERTa, VLAI is fine-tuned on over 600,000 real-world vulnerabilities and achieves over 82% accuracy in predicting severity categories, enabling faster and more consistent triage ahead of manual CVSS scoring. The model and dataset are open-source and integrated into the Vulnerability-Lookup service.

You can read [this page](https://www.vulnerability-lookup.org/user-manual/ai/) for more information.


## Model description

It is a classification model and is aimed to assist in classifying vulnerabilities by severity based on their descriptions.

## How to get started with the model

```python
from transformers import AutoModelForSequenceClassification, AutoTokenizer
import torch

labels = ["low", "medium", "high", "critical"]

model_name = "CIRCL/vulnerability-severity-classification-roberta-base"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSequenceClassification.from_pretrained(model_name)
model.eval()

print("Model revision:", model.config._commit_hash)

test_description = "SAP NetWeaver Visual Composer Metadata Uploader is not protected with a proper authorization, allowing unauthenticated agent to upload potentially malicious executable binaries \
that could severely harm the host system. This could significantly affect the confidentiality, integrity, and availability of the targeted system."
inputs = tokenizer(test_description, return_tensors="pt", truncation=True, padding=True)

# Run inference
with torch.no_grad():
    outputs = model(**inputs)
    predictions = torch.nn.functional.softmax(outputs.logits, dim=-1)

# Print results
print("Predictions:", predictions)
predicted_class = torch.argmax(predictions, dim=-1).item()
print("Predicted severity:", labels[predicted_class])
```

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 3e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- num_epochs: 5

It achieves the following results on the evaluation set:
- Loss: 2.0373
- Accuracy: 0.8137
- F1 Macro: 0.7424
- Low Precision: 0.6660
- Low Recall: 0.4757
- Low F1: 0.5550
- Medium Precision: 0.8348
- Medium Recall: 0.8692
- Medium F1: 0.8516
- High Precision: 0.8163
- High Recall: 0.8102
- High F1: 0.8132
- Critical Precision: 0.7602
- Critical Recall: 0.7395
- Critical F1: 0.7497

### Training results

| Training Loss | Epoch | Step  | Validation Loss | Accuracy | F1 Macro | Low Precision | Low Recall | Low F1 | Medium Precision | Medium Recall | Medium F1 | High Precision | High Recall | High F1 | Critical Precision | Critical Recall | Critical F1 |
|:-------------:|:-----:|:-----:|:---------------:|:--------:|:--------:|:-------------:|:----------:|:------:|:----------------:|:-------------:|:---------:|:--------------:|:-----------:|:-------:|:------------------:|:---------------:|:-----------:|
| 2.6518        | 1.0   | 19049 | 2.6882          | 0.7235   | 0.6396   | 0.4169        | 0.4805     | 0.4464 | 0.7792           | 0.7764        | 0.7778    | 0.7148         | 0.7409      | 0.7276  | 0.6632             | 0.5592          | 0.6068      |
| 2.5168        | 2.0   | 38098 | 2.3584          | 0.7578   | 0.6749   | 0.6350        | 0.3686     | 0.4664 | 0.8087           | 0.8128        | 0.8108    | 0.7431         | 0.7566      | 0.7498  | 0.6498             | 0.6971          | 0.6726      |
| 1.9209        | 3.0   | 57147 | 2.1623          | 0.7839   | 0.6967   | 0.7067        | 0.3504     | 0.4685 | 0.8078           | 0.8563        | 0.8314    | 0.7898         | 0.7654      | 0.7774  | 0.6907             | 0.7288          | 0.7093      |
| 1.4313        | 4.0   | 76196 | 2.1291          | 0.8001   | 0.7300   | 0.6197        | 0.4962     | 0.5511 | 0.8262           | 0.8599        | 0.8427    | 0.8112         | 0.7838      | 0.7973  | 0.7193             | 0.7391          | 0.7290      |
| 1.6812        | 5.0   | 95245 | 2.0373          | 0.8137   | 0.7424   | 0.6660        | 0.4757     | 0.5550 | 0.8348           | 0.8692        | 0.8516    | 0.8163         | 0.8102      | 0.8132  | 0.7602             | 0.7395          | 0.7497      |


### Framework versions

- Transformers 5.17.0
- Pytorch 2.14.0+cu130
- Datasets 4.8.5
- Tokenizers 0.23.2