Geometric Deviation as an Unsupervised Pre-Generation Reliability Signal: Probing LLM Representations for Answerability

Documenti analoghi

• WSM: Decay-Free Learning Rate Schedule via Checkpoint Merging for LLM Pre-training
  di: Tian, Changxin, et al.
  Pubblicazione: (2025)
• The Geometry of Harmful Intent: Training-Free Anomaly Detection via Angular Deviation in LLM Residual Streams
  di: Llorente-Saguer, Isaac
  Pubblicazione: (2026)
• $S^3$ -- Semantic Signal Separation
  di: Kardos, Márton, et al.
  Pubblicazione: (2024)
• Evaluating the Efficacy of Hybrid Deep Learning Models in Distinguishing AI-Generated Text
  di: Oketunji, Abiodun Finbarrs
  Pubblicazione: (2023)
• Large Language Model (LLM) Bias Index -- LLMBI
  di: Oketunji, Abiodun Finbarrs, et al.
  Pubblicazione: (2023)