Near-Oracle KV Selection via Pre-hoc Sparsity for Long-Context Inference

Documenti analoghi

• Is (Selective) Round-To-Nearest Quantization All You Need?
  di: Kogan, Alex
  Pubblicazione: (2025)
• Large Language Models as Oracles for Ontology Alignment
  di: Lushnei, Sviatoslav, et al.
  Pubblicazione: (2025)
• KV-Fold: One-Step KV-Cache Recurrence for Long-Context Inference
  di: Nadali, Alireza, et al.
  Pubblicazione: (2026)
• Homogeneous Keys, Heterogeneous Values: Exploiting Local KV Cache Asymmetry for Long-Context LLMs
  di: Cui, Wanyun, et al.
  Pubblicazione: (2025)
• Compensate Quantization Errors+: Quantized Models Are Inquisitive Learners
  di: Gao, Yifei, et al.
  Pubblicazione: (2024)