Multilingual Pretraining and Instruction Tuning Improve Cross-Lingual Knowledge Alignment, But Only Shallowly

Documenti analoghi

• Measuring Meaning Composition in the Human Brain with Composition Scores from Large Language Models
  di: Gao, Changjiang, et al.
  Pubblicazione: (2024)
• PreAlign: Boosting Cross-Lingual Transfer by Early Establishment of Multilingual Alignment
  di: Li, Jiahuan, et al.
  Pubblicazione: (2024)
• ExpLang: Improved Exploration and Exploitation in LLM Reasoning with On-Policy Thinking Language Selection
  di: Gao, Changjiang, et al.
  Pubblicazione: (2026)
• Large Language Models Are Cross-Lingual Knowledge-Free Reasoners
  di: Hu, Peng, et al.
  Pubblicazione: (2024)
• Large Language Models are Limited in Out-of-Context Knowledge Reasoning
  di: Hu, Peng, et al.
  Pubblicazione: (2024)