Match each post-training quantization method to what it protects against

Attempt it

AWQ

GPTQ

SmoothQuant

FP8 PTQ (cast-and-go)

Pick AWQ or GPTQ when…

Quick reference

TL;DR

Each quantization method fixes a different failure mode: GPTQ minimizes second-order error, AWQ protects salient weight channels, SmoothQuant migrates activation outliers, and FP8 leans on wider dynamic range.

Memory aid

Sign in for the mnemonic that makes this stick. See Pro pricing.

Easy to grasp

Imagine compressing a detailed photo into far fewer colors. If you pick colors blindly, the important bright spots turn into ugly blotches. Different methods fix this differently. One studies which pixels matter most and rounds those carefully (GPTQ). One notices a few super-bright pixels dominate and scales them up before compressing so they survive (AWQ). One realizes the brightness problem actually lives in the lighting, not the photo, and shifts the glare onto the photo where it is easier to handle (SmoothQuant). And one just uses a color format that already spans a wider brightness range, so no special trick is needed (FP8). Same goal, lower bits, but each guards against a different way the picture would otherwise get wrecked.

Key concepts

Post-training quantization GPTQ AWQ SmoothQuant FP8 quantization

Quick scan · 2–4 min

Everything important, quickly.

4 min: split weight-only INT4 (AWQ, GPTQ) from W8A8 INT8 (SmoothQuant) and FP8 cast-and-go, then explain each failure mode and the pick rule for 70B decode.

Side-by-side

Method	What it quantizes	Failure mode it fights	Typical bit width
GPTQ	Weights only	Cumulative layer reconstruction error	INT4 / INT3
AWQ	Weights only	Destruction of salient weight channels	INT4
SmoothQuant	Weights and activations	Activation outliers blocking INT8	INT8 (W8A8)
FP8 cast-and-go	Weights (and often activations)	Narrow dynamic range of INT8	FP8 (E4M3)

Real-world examples

Real products, models, and research that use this idea.

vLLM and SGLang in 2026 ship AWQ and GPTQ INT4 kernels as a default way to serve Llama 3.1 70B on a single 80 GB GPU.
TensorRT-LLM uses SmoothQuant to enable W8A8 INT8 serving on NVIDIA H100 tensor cores for latency-sensitive deployments.
DeepSeek V4 and Mistral Large 3 publish FP8 checkpoints that serve with cast-and-go FP8 on H100 and B200 hardware with minimal calibration.
Hugging Face Transformers integrates GPTQ and AWQ through the optimum and autoawq libraries for one-line INT4 model loading.
NVIDIA TensorRT Model Optimizer applies FP8 and INT8 SmoothQuant recipes as part of the standard production quantization toolkit.

Interviewer follow-ups

What an interviewer would ask next. Try answering before peeking at the approach.

QWhy does AWQ scale activations down when it scales salient weight channels up?

Sign in or go Pro to see the approach for this follow-up.

QWhy can SmoothQuant rebalance outliers offline without changing the model output?

Sign in or go Pro to see the approach for this follow-up.

QWhy does FP8 tolerate outliers that wreck INT8 at the same bit width?

Sign in or go Pro to see the approach for this follow-up.

60-second cheat sheet

The night-before-the-interview bullets. Scan these on the way to the call.

Sign in for 5 unlocks/day or go Pro for unlimited revision sheets. See Pro pricing.

References

Primary sources. Skim if you want the original framing.