A Data-Efficient Path to Multilingual LLMs: Language Expansion via Post-training PARAM$Δ$ Integration into Upcycled MoE

Documenti analoghi

• MoE-LPR: Multilingual Extension of Large Language Models through Mixture-of-Experts with Language Priors Routing
  di: Zhou, Hao, et al.
  Pubblicazione: (2024)
• Multilingual Contrastive Decoding via Language-Agnostic Layers Skipping
  di: Zhu, Wenhao, et al.
  Pubblicazione: (2024)
• How Does Alignment Enhance LLMs' Multilingual Capabilities? A Language Neurons Perspective
  di: Zhang, Shimao, et al.
  Pubblicazione: (2025)
• Question Translation Training for Better Multilingual Reasoning
  di: Zhu, Wenhao, et al.
  Pubblicazione: (2024)
• MAPO: Advancing Multilingual Reasoning through Multilingual Alignment-as-Preference Optimization
  di: She, Shuaijie, et al.
  Pubblicazione: (2024)