A Silver Bullet or a Compromise for Full Attention? A Comprehensive Study of Gist Token-based Context Compression

Documenti analoghi

• UniGist: Towards General and Hardware-aligned Sequence-level Long Context Compression
  di: Deng, Chenlong, et al.
  Pubblicazione: (2025)
• Attention Entropy is a Key Factor: An Analysis of Parallel Context Encoding with Full-attention-based Pre-trained Language Models
  di: Zhang, Zhisong, et al.
  Pubblicazione: (2024)
• Learning Interpretable Legal Case Retrieval via Knowledge-Guided Case Reformulation
  di: Deng, Chenlong, et al.
  Pubblicazione: (2024)
• Enabling Discriminative Reasoning in LLMs for Legal Judgment Prediction
  di: Deng, Chenlong, et al.
  Pubblicazione: (2024)
• Compressing Lengthy Context With UltraGist
  di: Zhang, Peitian, et al.
  Pubblicazione: (2024)