Papers
arxiv:2611.23388

QDA: Closed-Loop LLM Quantization Design with Agents

Published on Nov 17
· Submitted by
Xingyang Li
on Nov 19
Authors:
,
,
,

Abstract

Post-training quantization is the standard way to serve large language models at low cost. However, the best quantization recipe differs across models and precisions. As a result, experts must redesign it for every new model, which is time-consuming and often sub-optimal. Fortunately, an LLM agent can take over this design by refining the recipe over rounds of building and evaluating candidates, so that more rounds lead to better recipes. However, evaluating one quantized reasoning model takes tens of GPU-hours, and combining methods from different codebases means porting each to the new architecture by hand. We present QDA, a system in which an agent designs the recipe for each model and precision. To make evaluation fast, QDA scores each candidate in minutes by its output KL divergence from the original model on reasoning traces that the agent never sees, and checks promising ones with long generations to catch failures such as repetition loops. To make building fast, QDA provides a library of public PTQ methods that are compatible with one another and with new model architectures. Across four reasoning models, QDA improves accuracy over the strongest public recipe by 0.9, 1.4 and 2.2 percentage points on average at W4A16, W4A8 and W3A16, using 64 GPU-hours per search. On the 320B-parameter MoE model GLM-5.3-Flash at NVFP4 W4A4, it recovers 98.3% of the FP8 model's accuracy, against 96.9% for the best vendor checkpoint. Every QDA checkpoint also runs on existing serving kernels with no inference overhead. We will release the library, the evaluator and the harness.

Community

Paper submitter

QDA brings a coding agent into LLM quantization design. For each model and precision, the agent designs the post-training quantization recipe on its own, with no human in the loop. Its checkpoints beat the strongest public recipe on four reasoning models, by 0.9, 1.4 and 2.2 points on average at W4A16, W4A8 and W3A16, and run on existing kernels with no inference overhead.

The agent works the way a quantization expert does: it proposes a recipe, builds it, measures it and decides what to change next. Two things make this loop fast enough for up to 36 candidates in one eight-hour search: a KL score that takes minutes instead of dozens of GPU-hours, and a library of 21 PTQ methods that stack in any order. We find that no method wins everywhere: the best recipe differs for every model and precision, and the agent finds it by searching.

On the 320B-parameter GLM-5.3-Flash at NVFP4, one 6.5-hour, $69 search recovers 98.3% of the FP8 model's accuracy, against 96.9% for NVIDIA's release. The tools matter too: given NVIDIA ModelOpt's tools instead of ours, the same agent falls 0.5–1.3 points behind on Qwen3.6-35B. A search costs $19–75 of agent tokens.

Sign up or log in to comment

Get this paper in your agent:

hf papers read 2611.23388
Don't have the latest CLI?
curl -LsSf https://hf.co/cli/install.sh | bash

Models citing this paper 30

Browse 30 models citing this paper

Datasets citing this paper 1

Spaces citing this paper 0

No Space linking this paper

Cite arxiv.org/abs/2611.23388 in a Space README.md to link it from this page.

Collections including this paper 0

No Collection including this paper

Add this paper to a collection to link it from this page.