Quantization Error Is Spectrally Flat: A Single Random Probe Is a Calibrated, Data-Free Sensitivity Estimator, with Application to Budget-Targeted Mixed-Precision Quantization

arXiv:2609.33923 · cs.CL, cs.AI, cs.LG · Submitted 2026-09-27 · Read on arXiv

cs.CL, cs.AI, cs.LG

Submitted: 2026-09-27

Updated: 2026-09-29

Code: https://github.com/mobiusml/hqq

Terminology

• Quantization — 74× in this paper · explained in 12 episodes
• Calibration — 56× in this paper · explained in 7 episodes
• Perplexity — 49× in this paper · explained in 3 episodes
• Allocator — 26× in this paper · explained in 1 episode
• Adaptation — 14× in this paper · explained in 1 episode
• Trace — 10× in this paper · explained in 2 episodes
• Ablation — 5× in this paper · explained in 1 episode
• Embeddings — 5× in this paper · explained in 2 episodes
• Accuracy — 4× in this paper · explained in 1 episode
• Elias — 3× in this paper · explained in 2 episodes
• Latent Space — 3× in this paper · explained in 7 episodes
• Core — 2× in this paper · explained in 2 episodes
• Half-Quadratic Quantization — 2× in this paper
• Multi-head Latent Attention — 2× in this paper
• Resource-Aware Mixed-precision — 2× in this paper
• Transformer — 2× in this paper · explained in 2 episodes
• Alignment — 1× in this paper · explained in 1 episode
• Ambient Dimension — 1× in this paper · explained in 1 episode
• B-parameter Mixture-of-Experts — 1× in this paper
• Fidelity — 1× in this paper · explained in 2 episodes
• Fisher Information — 1× in this paper · explained in 3 episodes
• Large Language Models (LLMs) — 1× in this paper · explained in 6 episodes
• Mixture-of-Experts (MoE) — 1× in this paper · explained in 7 episodes
• Pruning — 1× in this paper · explained in 4 episodes

Sources

• KurTail : Kurtosis-based LLM Quantization
• Cherry on Top: Parameter Heterogeneity and Quantization in Large Language Models
• DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model
• Extreme Compression of Large Language Models via Additive Quantization
• Mixture Compressor for Mixture-of-Experts LLMs Gains More
• SliM-LLM: Salience-Driven Mixed-Precision Quantization for Large Language Models
• Fidelity Is Not Safety: Gently-Compressed LLMs Pass Every Data-Free Quality Guard Yet Invent Procedure Steps in Agentic Execution
• Q-Strata: Hierarchical Bit Allocation for Mixed-Precision Quantization of Mixture-of-Experts LLMs
• Data-free mixed-precision quantization using novel sensitivity metric
• Pushing the Limits of Large Language Model Quantization via the Linearity Theorem
• MixQuant: Adaptive Mixed-Precision Quantization for Large Language Models
• AlphaQ: Calibration-Free Bit Allocation for Mixture-of-Experts Quantization
• GAMMA: Global Bit Allocation for Mixed-Precision Models under Arbitrary Budgets
• Beyond Scalar Sensitivity: Activation-Aware Mixed-Precision LLM Quantization with Cross-Layer Refinement
• LQER: Low-Rank Quantization Error Reconstruction for LLMs
• Beyond Outliers: A Data-Free Layer-wise Mixed-Precision Quantization Approach Driven by Numerical and Structural Dual-Sensitivity
• BitsMoE: Cost-Aware Bit Allocation in Spectral Space for MoE LLM Quantization

Related papers

• Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
• Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
• Subliminal Steering: Stronger Encoding of Hidden Signals
• MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
• The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
• Untangling the Mechanisms of Misleading Context in Medical Question Answering