FactBench: A Dynamic Benchmark for In-the-Wild Language Model Factuality Evaluation

Documenti analoghi

• VeriFact: Enhancing Long-Form Factuality Evaluation with Refined Fact Extraction and Reference Facts
  di: Liu, Xin, et al.
  Pubblicazione: (2025)
• Enhancing Language Model Factuality via Activation-Based Confidence Calibration and Guided Decoding
  di: Liu, Xin, et al.
  Pubblicazione: (2024)
• From Proof to Program: Characterizing Tool-Induced Reasoning Hallucinations in Large Language Models
  di: Bayat, Farima Fatahi, et al.
  Pubblicazione: (2025)
• Enhanced Language Model Truthfulness with Learnable Intervention and Uncertainty Expression
  di: Bayat, Farima Fatahi, et al.
  Pubblicazione: (2024)
• Logit Arithmetic Elicits Long Reasoning Capabilities Without Training
  di: Zhang, Yunxiang, et al.
  Pubblicazione: (2025)