Scaling Up, Speeding Up: A Benchmark of Speculative Decoding for Efficient LLM Test-Time Scaling

Documenti analoghi

• PSD: Pushing the Pareto Frontier of Diffusion LLMs via Parallel Speculative Decoding
  di: Sun, Shengyin, et al.
  Pubblicazione: (2026)
• Revisiting Judge Decoding from First Principles via Training-Free Distributional Divergence
  di: Sun, Shengyin, et al.
  Pubblicazione: (2026)
• TrimR: Verifier-based Training-Free Thinking Compression for Efficient Test-Time Scaling
  di: Lin, Weizhe, et al.
  Pubblicazione: (2025)
• DFlare: Scaling Up Draft Capacity for Block Diffusion Speculative Decoding
  di: Zhang, Jiebin, et al.
  Pubblicazione: (2026)
• Iterative Deepening Sampling as Efficient Test-Time Scaling
  di: Chen, Weizhe, et al.
  Pubblicazione: (2025)