A single calibration that marginalizes layer distortion over random quantized upstream contexts yields budget-agnostic bit allocations that beat FP16-scored adaptive baselines across Llama-3.2-3B, Llama-2-7B, and Mistral-7B.
Title resolution pending
1 Pith paper cite this work, alongside 9 external citations. Polarity classification is still indexing.
1
Pith paper citing it
9
external citations · OpenAlex
fields
cs.LG 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
MixQuant: Adaptive Mixed-Precision Quantization for Large Language Models
A single calibration that marginalizes layer distortion over random quantized upstream contexts yields budget-agnostic bit allocations that beat FP16-scored adaptive baselines across Llama-3.2-3B, Llama-2-7B, and Mistral-7B.