Narrative Flattening: How Post-Training Compresses Thematic, Affective, and Stylistic Variation in LLM Fiction

Documenti analoghi

• Missing vs. Unused Knowledge Hypothesis for Language Model Bottlenecks in Patent Understanding
  di: Wu, Siyang, et al.
  Pubblicazione: (2025)
• Mapping Overlaps in Benchmarks through Perplexity in the Wild
  di: Wu, Siyang, et al.
  Pubblicazione: (2025)
• Language Models Should be Used to Surface the Unwritten Code of Science and Society
  di: Bao, Honglin, et al.
  Pubblicazione: (2025)
• Interpretable Stylistic Variation in Human and LLM Writing Across Genres, Models, and Decoding Strategies
  di: Rallapalli, Swati, et al.
  Pubblicazione: (2026)
• Stylistic Evolution and LLM Neutrality in Singlish Language
  di: Foo, Linus Tze En, et al.
  Pubblicazione: (2026)