pith. sign in

Quicksilver – speeding up LLM inference through dynamic token halting, KV skipping, contextual token fusion, and adaptive matryoshka quantization, 2025

Danush Khanna, Aditya Kumar Guru, Srivarshinee Sridhar, Zidan Ahmed, Rubhav Bahirwani, Meetu Malhotra, Vinija Jain, Aman Chadha, Amitava Das, Kripabandhu Ghosh · 2025 · arXiv 2506.22396

1 Pith paper cite this work. Polarity classification is still indexing.

1 Pith paper citing it
read on arXiv browse 1 citing papers

fields

cs.CL 1

years

2026 1

verdicts

UNVERDICTED 1

representative citing papers

Two-dimensional early exit optimisation of LLM inference

cs.CL · 2026-03-27 · unverdicted · novelty 7.0

Coordinating layer-wise and sentence-wise early exits in LLMs produces multiplicative speedups of 1.4-2.3x over single-dimension early exit on sentiment classification tasks.

citing papers explorer

Showing 1 of 1 citing paper.

  • Two-dimensional early exit optimisation of LLM inference cs.CL · 2026-03-27 · unverdicted · none · ref 11

    Coordinating layer-wise and sentence-wise early exits in LLMs produces multiplicative speedups of 1.4-2.3x over single-dimension early exit on sentiment classification tasks.