Lightweight hallucination and grounding detection model. Checks whether a claim is supported by the given context.
Built on
ModernBERT-base (149M params), fine-tuned on 1M+ NLI pairs from SNLI, MultiNLI, and ANLI.
1from transformers import pipeline
2
3classifier = pipeline("text-classification", model="ENTUM-AI/FactGuard")
4
5result = classifier({
6 "text": "Apple reported revenue of $94.8 billion in Q1 2024.",
7 "text_pair": "Apple's Q1 2024 revenue was $94.8 billion."
8})
9# [{'label': 'Supported', 'score': 0.99}]
10
11result = classifier({
12 "text": "Apple reported revenue of $94.8 billion in Q1 2024.",
13 "text_pair": "Apple's revenue exceeded $100 billion."
14})
15# [{'label': 'Not Supported', 'score': 0.97}]
1M+ NLI pairs mapped to binary grounding labels.
