This article explains how quantization—originating from signal processing—reduces precision to save resources, details its application to neural network weights and activations via PTQ, QAT, GPTQ, AWQ, and SmoothQuant, and shows how vector quantization enables fast, memory‑efficient retrieval in large‑scale RAG systems.
This article explains the principles of model quantization, compares post‑training quantization (PTQ) and quantization‑aware training (QAT), details the QAT workflow in PyTorch—including fake quantization, gradient handling, and code examples—and offers practical tips for achieving high‑accuracy int8/int4 models.
The article surveys modern post‑training quantization approaches for large language models, detailing weight‑only and activation‑aware methods such as GPTQ, AWQ, HQQ, SmoothQuant, QuIP, QuaRot, SpinQuant, QQQ, QoQ, and FP8, and compares their precision levels, algorithmic steps, accuracy‑throughput trade‑offs, and implementation considerations for efficient inference.
This article explains why large language models need quantization, describes the core concepts, classification schemes, symmetric and asymmetric methods, handling of outliers, and compares post‑training quantization (PTQ) with quantization‑aware training (QAT), while detailing popular techniques such as GPTQ, GGUF, and BitNet.
Apple Intelligence introduces on‑device large language models and new AI features across iOS, macOS and iPad, but its limited memory forces aggressive quantization and a private‑cloud fallback, raising questions about whether Apple can truly safeguard user data while matching competitors like Google, Microsoft and OpenAI.
SAMP is an adaptive mixed-precision inference toolkit that automatically controls floating-point and integer operations to accelerate model inference while maintaining computational accuracy.
