language:
- de
- en
license: mit
base_model: deepset/gbert-base
pipeline_tag: text-classification
library_name: transformers
tags:
- klarki
- eu-ai-act
- compliance
- german
- text-classification
- bert
model-index:
- name: klarki-actor-classifier
results:
- task:
type: text-classification
name: Text Classification
dataset:
name: KlarKI EU AI Act Regulatory Training Data
type: custom
metrics:
- type: f1
value: 1
name: Macro F1
verified: false
KlarKI — EU AI Act Article 3 Actor Classifier
4-class text classification — identifies the Article 3 actor role of an organisation relative to an AI system
Part of KlarKI — a local-first EU AI Act + GDPR compliance auditor for German SMEs. All inference runs on-device. No data leaves your machine.
Model Overview
| Property | Value |
|---|---|
| Base model | deepset/gbert-base |
| Architecture | Transformers — BertForSequenceClassification |
| Parameters | ~110M parameters |
| Languages | German (primary), English |
| Training samples | 2767 train / 491 validation |
| License | MIT |
| Part of | KlarKI audit pipeline |
Quickstart
Option A — Via KlarKI (recommended)
Use this if you want the full audit pipeline. The download script places all 5 models exactly where KlarKI expects them.
git clone https://github.com/s4nkar/KlarKI-EU-AI-Act-compliance-auditor.git
cd KlarKI-EU-AI-Act-compliance-auditor
pip install huggingface-hub>=0.26.0
python scripts/download_pretrained.py --model actor
./run.sh up
Option B — Direct usage
from transformers import pipeline
classifier = pipeline("text-classification", model="s4nkar/klarki-actor-classifier")
result = classifier("We developed and placed the AI system on the market under our own name and brand.")
# Output: [{'label': 'provider', 'score': 0.99}]
Labels
| Label | Description |
|---|---|
provider |
Developed / placed the AI system on the market (Art. 3(3)) |
deployer |
Uses the AI system under its authority (Art. 3(4)) |
importer |
Places a third-country AI system on the EU market (Art. 3(6)) |
distributor |
Makes the AI system available without modifying it (Art. 3(7)) |
Evaluation Results
Overall
| Macro F1 | Val samples |
|---|---|
| 1.0000 | 491 |
Per-Class
| Class | Precision | Recall | F1 | Support |
|---|---|---|---|---|
provider |
1.0000 | 1.0000 | 1.0000 | 125 |
deployer |
1.0000 | 1.0000 | 1.0000 | 122 |
importer |
1.0000 | 1.0000 | 1.0000 | 122 |
distributor |
1.0000 | 1.0000 | 1.0000 | 122 |
Training Details
| Property | Value |
|---|---|
| Base model | deepset/gbert-base |
| Training epochs | 5 (AdamW, early stopping) |
| Batch size | 16 |
| Data split | 85% train / 15% validation, stratified, seed=42 |
| Data generation | Async Ollama-grounded synthesis (phi3:mini) + real regulatory text |
| Optimiser | AdamW |
| Training framework | Docker container (Python 3.11, isolated from host) |
Intended Use
Determining whether an organisation is acting as a provider, deployer, importer, or distributor under EU AI Act Article 3. Runs before the applicability gate in KlarKI's legal decision hierarchy.
This model is a decision-support tool, not a substitute for qualified legal advice. EU AI Act compliance determinations should always be reviewed by a legal professional.
Limitations
- Outputs a single role; does not detect organisations with multiple concurrent roles.
- Confidence threshold in KlarKI is 0.80; below that, a 39-pattern regex fallback is used.
- Performance degrades on very short texts (< 50 tokens).
Citation
@software{klarki2026,
author = {Sankar},
title = {KlarKI: Local-First EU AI Act and GDPR Compliance Auditor},
year = {2026},
url = {https://github.com/s4nkar/KlarKI-EU-AI-Act-compliance-auditor},
note = {Open-source compliance tooling for German SMEs}
}
About KlarKI
KlarKI is an open-source, local-first EU AI Act + GDPR compliance auditor built for German SMEs. Upload a policy document and receive a scored gap analysis against Articles 9–15 entirely on your own hardware.
Key features:
- Deterministic legal decision hierarchy (actor detection, Annex III applicability gate)
- Hybrid RAG retrieval (BM25 + ChromaDB vector + cross-encoder re-ranking)
- LangGraph multi-agent gap analysis (3-node per applicable article)
- Bilingual EN/DE support — all inference runs locally, no external API calls