Q-MINO: A Minimal-Norm Method for Quantization-Aware Training

arXiv:2610.00738 · math.OC, cs.AR, cs.LG · Submitted 2026-09-30 · Read on arXiv

math.OC, cs.AR, cs.LG

Submitted: 2026-09-30

Updated: 2026-09-30

Terminology

• Quantization — 1× in this paper · explained in 12 episodes
• Quantization-Aware Training — 1× in this paper · explained in 3 episodes

Sources

• Estimating or Propagating Gradients Through Stochastic Neurons for Conditional Computation
• A Survey of Quantization Methods for Efficient Neural Network Inference
• Optimizing Large Language Models through Quantization: A Comparative Analysis of PTQ and QAT Techniques
• Scaling Laws for Neural Language Models
• Unifying Block-wise PTQ and Distillation-based QAT for Progressive Quantization toward 2-bit Instruction-Tuned LLMs
• PV-Tuning: Beyond Straight-Through Estimation for Extreme LLM Compression
• A White Paper on Neural Network Quantization
• Overcoming Oscillations in Quantization-Aware Training
• Energy and Policy Considerations for Deep Learning in NLP

Related papers

• Lions and Muons: Optimization via Stochastic Frank-Wolfe under Heavy-Tailed Noise
• Adam-HNAG: A Convergent Reformulation of Adam with Accelerated Rate
• Incremental Learning in Mirror Flows
• Second-Order Optimization with Asynchronous Curvature
• Online Control via Counterfactual Tracking
• Asynchronous Replanning in Two Population Linear Quadratic Mean Field Games: Information Requirements and Stability