Per-option encoders beat joint scoring on closed-set decisions
At 32 million parameters, per-option scoring gains 5.8 skill points but needs 5 to 13 times more passes.
40
AI model
DeBERTa-v3
- Mentions · 7d0
- Mentions · 30d1
- All mentions2
- First seenSep 10, 2026
- Last mentioned9d ago
Timeline
2 articles · newest firstDomain-Specific Hallucination Detection in Large Language Models
A multi-signal pipeline detects LLM hallucinations, reaching F1 0.915 on HaluEval and cutting Qwen2.5-0.5B hallucination rates from 85.5% to 37.7% via DPO.
32
Appears with
- DeBERTa-v3product · 1
- Qwen2.5-0.5Bai model · 1
- PubMedBERTproduct · 1
- arXivorganization · 1
- PubMedBERTai model · 1
- HaluEvalproduct · 1
- System1-Arenaproduct · 1
- SciFactproduct · 1
Entities are extracted by the model from each article. First seen , last mentioned . Watching keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.