End of training

README.md

@@ -1,69 +1,50 @@
 ---
 library_name: transformers
-license: cc-by-4.0
+license: mit
 base_model: roberta-base
-metrics:
-- accuracy
 tags:
 - generated_from_trainer
-- text-classification
-- classification
-- nlp
-- vulnerability
+metrics:
+- accuracy
 model-index:
 - name: vulnerability-severity-classification-roberta-base
   results: []
-datasets:
-- CIRCL/vulnerability-scores
 ---
 
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
 
-# 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.
+# vulnerability-severity-classification-roberta-base
 
+This model is a fine-tuned version of [roberta-base](https://huggingface.co/roberta-base) on an unknown dataset.
+It achieves the following results on the evaluation set:
+- Loss: 2.0683
+- Accuracy: 0.8139
+- F1 Macro: 0.7422
+- Low Precision: 0.6402
+- Low Recall: 0.4896
+- Low F1: 0.5548
+- Medium Precision: 0.8440
+- Medium Recall: 0.8665
+- Medium F1: 0.8551
+- High Precision: 0.8090
+- High Recall: 0.8118
+- High F1: 0.8104
+- Critical Precision: 0.7586
+- Critical Recall: 0.7390
+- Critical F1: 0.7487
 
 ## 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
+More information needed
 
-labels = ["low", "medium", "high", "critical"]
+## Intended uses & limitations
 
-model_name = "CIRCL/vulnerability-severity-classification-roberta-base"
-tokenizer = AutoTokenizer.from_pretrained(model_name)
-model = AutoModelForSequenceClassification.from_pretrained(model_name)
-model.eval()
+More information needed
 
-print("Model revision:", model.config._commit_hash)
+## Training and evaluation data
 
-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])
-```
+More information needed
 
 ## Training procedure
 
@@ -78,37 +59,20 @@ The following hyperparameters were used during training:
 - lr_scheduler_type: linear
 - num_epochs: 5
 
-It achieves the following results on the evaluation set:
-- Loss: 2.0190
-- Accuracy: 0.8181
-- F1 Macro: 0.7449
-- Low Precision: 0.6507
-- Low Recall: 0.4837
-- Low F1: 0.5549
-- Medium Precision: 0.8435
-- Medium Recall: 0.8746
-- Medium F1: 0.8588
-- High Precision: 0.8174
-- High Recall: 0.8112
-- High F1: 0.8143
-- Critical Precision: 0.7620
-- Critical Recall: 0.7419
-- Critical F1: 0.7518
-
 ### 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 |
 |:-------------:|:-----:|:-----:|:---------------:|:--------:|:--------:|:-------------:|:----------:|:------:|:----------------:|:-------------:|:---------:|:--------------:|:-----------:|:-------:|:------------------:|:---------------:|:-----------:|
-| 1.9719        | 1.0   | 16879 | 2.6228          | 0.7372   | 0.6018   | 0.7267        | 0.1845     | 0.2943 | 0.7425           | 0.8797        | 0.8053    | 0.7324         | 0.7008      | 0.7163  | 0.7245             | 0.4996          | 0.5914      |
-| 1.8998        | 2.0   | 33758 | 2.3195          | 0.7712   | 0.6818   | 0.6525        | 0.3581     | 0.4625 | 0.7943           | 0.8589        | 0.8253    | 0.7740         | 0.7409      | 0.7571  | 0.6889             | 0.6756          | 0.6822      |
-| 1.9809        | 3.0   | 50637 | 2.1185          | 0.7922   | 0.7137   | 0.6561        | 0.4318     | 0.5208 | 0.8192           | 0.8621        | 0.8401    | 0.7874         | 0.7779      | 0.7826  | 0.7267             | 0.6962          | 0.7111      |
-| 1.8121        | 4.0   | 67516 | 2.0117          | 0.8098   | 0.7325   | 0.6624        | 0.4442     | 0.5318 | 0.8380           | 0.8675        | 0.8525    | 0.8108         | 0.8004      | 0.8055  | 0.7321             | 0.7483          | 0.7401      |
-| 1.4412        | 5.0   | 84395 | 2.0190          | 0.8181   | 0.7449   | 0.6507        | 0.4837     | 0.5549 | 0.8435           | 0.8746        | 0.8588    | 0.8174         | 0.8112      | 0.8143  | 0.7620             | 0.7419          | 0.7518      |
+| 2.6670        | 1.0   | 16995 | 2.5369          | 0.7400   | 0.6442   | 0.5747        | 0.3270     | 0.4168 | 0.7828           | 0.8259        | 0.8038    | 0.7294         | 0.7183      | 0.7238  | 0.6342             | 0.6303          | 0.6323      |
+| 2.1703        | 2.0   | 33990 | 2.3324          | 0.7666   | 0.6816   | 0.5869        | 0.4116     | 0.4839 | 0.8101           | 0.8316        | 0.8207    | 0.7373         | 0.7808      | 0.7584  | 0.7315             | 0.6068          | 0.6633      |
+| 2.2587        | 3.0   | 50985 | 2.1425          | 0.7882   | 0.7101   | 0.6348        | 0.4258     | 0.5097 | 0.8243           | 0.8477        | 0.8358    | 0.7800         | 0.7806      | 0.7803  | 0.7114             | 0.7179          | 0.7146      |
+| 1.7962        | 4.0   | 67980 | 2.0750          | 0.8019   | 0.7345   | 0.5742        | 0.5436     | 0.5585 | 0.8288           | 0.8637        | 0.8459    | 0.8168         | 0.7744      | 0.7950  | 0.7324             | 0.7452          | 0.7387      |
+| 1.2919        | 5.0   | 84975 | 2.0683          | 0.8139   | 0.7422   | 0.6402        | 0.4896     | 0.5548 | 0.8440           | 0.8665        | 0.8551    | 0.8090         | 0.8118      | 0.8104  | 0.7586             | 0.7390          | 0.7487      |
 
 
 ### Framework versions
 
 - Transformers 5.12.1
-- Pytorch 2.12.0+cu130
+- Pytorch 2.12.1+cu130
 - Datasets 4.8.5
 - Tokenizers 0.22.2

emissions.csv

@@ -1,2 +1,2 @@
 timestamp,project_name,run_id,experiment_id,duration,emissions,emissions_rate,cpu_power,gpu_power,ram_power,cpu_energy,gpu_energy,ram_energy,energy_consumed,water_consumed,country_name,country_iso_code,region,cloud_provider,cloud_region,os,python_version,codecarbon_version,cpu_count,cpu_model,gpu_count,gpu_model,longitude,latitude,ram_total_size,tracking_mode,cpu_utilization_percent,gpu_utilization_percent,ram_utilization_percent,ram_used_gb,on_cloud,pue,wue
-2026-06-16T11:05:44,VulnTrain,0e1d4869-c876-4473-8f9b-413b084ab3dc,5b0fa12a-3dd7-45bb-9766-cc326314d9f1,16266.829393087,0.47444260579462283,2.916626186515789e-05,70.0001854599049,863.9267400007767,70.0,0.3051255646602946,3.8969895462002526,0.3050964050056411,4.507211515866192,0.0,Luxembourg,LUX,luxembourg,,,Linux-6.8.0-124-generic-x86_64-with-glibc2.39,3.12.3,3.2.8,224,Intel(R) Xeon(R) Platinum 8480+,4,4 x NVIDIA L40S,6.1327,49.6098,2015.33540725708,machine,0.9380928853754941,71.58030200098814,1.0,20.170921206709895,N,1.0,0.0
+2026-06-24T20:39:06,VulnTrain,c534907f-9272-4c7c-8403-493913a349bf,5b0fa12a-3dd7-45bb-9766-cc326314d9f1,16390.004414127998,0.4770782180269855,2.910787611599113e-05,70.00017948888741,862.2472512914511,70.0,0.3071031554245731,3.9180739047343964,0.30707280947219734,4.532249869631166,0.0,Luxembourg,LUX,luxembourg,,,Linux-6.8.0-124-generic-x86_64-with-glibc2.39,3.12.3,3.2.8,224,Intel(R) Xeon(R) Platinum 8480+,4,4 x NVIDIA L40S,6.1327,49.6098,2015.3354606628418,machine,0.9469711833231147,71.61883813611281,1.0,20.26325034136278,N,1.0,0.0

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4923900e61732349fa7d3769c6135c3f37b84fdf2adb21d42e457f8b2ea87b0e
+oid sha256:ebb71e2c67d02e572cbe1e6893104ffe5b7d5bdb188be3a7c1fd1add54d079f0
 size 498618976