Dissociating Decodability and Causal Use in Bracket-Sequence Transformers

Documenti analoghi

• H-Probes: Extracting Hierarchical Structures From Latent Representations of Language Models
  di: Dawes, Cutter, et al.
  Pubblicazione: (2026)
• Linear probes rely on textual evidence: Results from leakage mitigation studies in language models
  di: Boxo, Gerard, et al.
  Pubblicazione: (2025)
• Causal Reflection with Language Models
  di: Aryan, Abi, et al.
  Pubblicazione: (2025)
• Transforming Chatbot Text: A Sequence-to-Sequence Approach
  di: Reddy, Natesh, et al.
  Pubblicazione: (2025)
• Causal Reasoning Favors Encoders: On The Limits of Decoder-Only Models
  di: Roy, Amartya, et al.
  Pubblicazione: (2025)