Views
No views yet
microsoft/deberta-v3-large for claim checkworthiness detection as part of the ExplainableACD project for IJCAI 2026.1from transformers import AutoTokenizer, AutoModelForSequenceClassification
2import torch
3
4# Load model and tokenizer
5model_name = "sergiopinto/deberta-v3-large-claim-checkworthiness-seed456"
6tokenizer = AutoTokenizer.from_pretrained(model_name)
7model = AutoModelForSequenceClassification.from_pretrained(model_name)
8
9# Example claim
10claim = "The president announced a new economic policy yesterday."
11
12# Tokenize and predict
13inputs = tokenizer(claim, return_tensors="pt", max_length=128, truncation=True)
14with torch.no_grad():
15 outputs = model(**inputs)
16 probs = torch.softmax(outputs.logits, dim=1)
17
18# Get prediction
19checkworthy_prob = probs[0][1].item()
20is_checkworthy = checkworthy_prob > 0.40
21
22print(f"Checkworthy probability: {checkworthy_prob:.4f}")
23print(f"Is checkworthy: {is_checkworthy}")sergiopinto/deberta-v3-large-claim-checkworthiness-seed0sergiopinto/deberta-v3-large-claim-checkworthiness-seed421@inproceedings{pinto2026explainableacd,
2 title={Explainable Automatic Claim Detection for Real-Time Fact-Checking},
3 author={Pinto, Sérgio and [Co-authors]},
4 booktitle={Proceedings of the 35th International Joint Conference on Artificial Intelligence (IJCAI)},
5 year={2026}
6}