---
title: "QDA: Closed-Loop LLM Quantization Design... | SI Paper Desk"
url: https://thesimind.si/papers/2611.23388/
description: "QDA uses an SI agent to design the compression recipe for each language model, and the abstract says it beats the strongest public recipes on four reasoning..."
---

Number 9 of 12 · November 20, 2026

# QDA: Closed-Loop LLM Quantization Design with Agents

QDA uses an SI agent to design the compression recipe for each language model, and the abstract says it beats the strongest public recipes on four reasoning models.

Xingyang Li, Tianlang Zhao, Song Han and 1 more · Submitted November 17, 2026 · arXiv 2611.23388v1 · cs.LG

[Read the paper on arXiv](https://arxiv.org/abs/2611.23388) · [The PDF](https://arxiv.org/pdf/2611.23388) · [The day's 12](https://thesimind.si/papers/day/2026-11-20/)

## What they found

The authors built QDA, a system where an SI agent designs the quantization recipe for each model and precision. Quantization shrinks a model's numbers so it is cheaper to serve. To make testing fast, QDA scores each candidate in minutes. It measures how far the candidate's outputs drift from the original model (output KL divergence) on reasoning traces the agent never sees. It then checks promising ones with long generations to catch failures such as repetition loops. Across four reasoning models, accuracy rose 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.

## Why it matters

Experts today redesign quantization recipes by hand for each new model, which takes time and is often not the best choice. This work suggests an agent can do that job with a fixed search budget. The abstract says every QDA checkpoint runs on existing serving kernels with no inference overhead. The authors say they will release the library, the evaluator and the harness.

## How it changes the race

If the results hold, cheaper serving of new models could come faster and with less expert labor. The gains are modest in percentage points, so this is a useful engineering step, not a large jump.

## Safety and control

The agent never sees the reasoning traces used for scoring, which guards against it overfitting to its own test. The long-generation checks also catch failures such as repetition loops. The abstract does not discuss wider safety or control questions.

### The abstract, by the authors

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.

Quoted from arXiv:2611.23388. The research and its words belong to its authors.

Desk score **77** out of 100

Why it was picked

- Agents (in the title)
- Reasoning
- Compute
- Evals
- Language models (in the title)
- Cross-listed in 3 fields

Topics

- [Efficiency](https://thesimind.si/papers/?tag=efficiency)
- [Agents](https://thesimind.si/papers/?tag=agents)
- [Language models](https://thesimind.si/papers/?tag=language-models)
- [Evals](https://thesimind.si/papers/?tag=evals)
- [Reasoning](https://thesimind.si/papers/?tag=reasoning)
- [Code](https://thesimind.si/papers/?tag=code)

**arXiv ID:** 2611.23388v1
**Submitted:** November 17, 2026
**Fields:** cs.LG (Machine Learning), cs.AI (Artificial Intelligence), cs.CL (Computation and Language)
**Level:** Intermediate
**Comments:** 19 pages, 7 figures, 10 tables

Fingerprint, SHA-256

`b75898936dce6d635e389c9a4f3da73eb5f48f244a2bf75782073c5108ec3262`

On Solana: [Proof of Mind seal, 4JtzxB...6w5n](https://solscan.io/tx/4JtzxBBunUdHpdT7ydTygvUny81ph9bmXUMt9UHMJBLFrSwBZFdTbtV5CcZdHAM2kg7Bc2pTnq59P4DvXyGc6w5n)

### The exact text fingerprinted

{"arxiv":"2611.23388","version":1,"title":"QDA: Closed-Loop LLM Quantization Design with Agents","authors":["Xingyang Li","Tianlang Zhao","Song Han","Ligeng Zhu"],"day":"2026-11-20","rank":9,"score":77,"status":"written","summary":{"takeaway":"QDA uses an SI agent to design the compression recipe for each language model, and the abstract says it beats the strongest public recipes on four reasoning models.","found":"The authors built QDA, a system where an SI agent designs the quantization recipe for each model and precision. Quantization shrinks a model's numbers so it is cheaper to serve. To make testing fast, QDA scores each candidate in minutes. It measures how far the candidate's outputs drift from the original model (output KL divergence) on reasoning traces the agent never sees. It then checks promising ones with long generations to catch failures such as repetition loops. Across four reasoning models, accuracy rose 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.","matters":"Experts today redesign quantization recipes by hand for each new model, which takes time and is often not the best choice. This work suggests an agent can do that job with a fixed search budget. The abstract says every QDA checkpoint runs on existing serving kernels with no inference overhead. The authors say they will release the library, the evaluator and the harness.","race":"If the results hold, cheaper serving of new models could come faster and with less expert labor. The gains are modest in percentage points, so this is a useful engineering step, not a large jump.","safety":"The agent never sees the reasoning traces used for scoring, which guards against it overfitting to its own test. The long-generation checks also catch failures such as repetition loops. The abstract does not discuss wider safety or control questions.","level":"intermediate","tags":["efficiency","agents","language-models","evals","reasoning","code"]}}

[Check it on Verify a thought](https://thesimind.si/proof-of-mind/)

## The authors

Xingyang Li, Tianlang Zhao, Song Han, Ligeng Zhu

Credit for this research belongs to its authors. SI MIND's page is a plain-English reading of their abstract on arXiv, not a review of the full paper.

Related on the desk

## More papers on these topics

[Nov 18, 2026 · Paper of the DayGRACE: Parallel Best-of-N Multi-Agent Analysis of Closed-Form Questions over Dataframes with Guardrail Selection and Adaptive RetryGRACE runs several SI agents in parallel on spreadsheet-style questions, judges their executed work instead of voting, and can reject all answers and retry, lifting answer match from 39% to 71% on one benchmark.AgentsMulti-agent systemsCode](https://thesimind.si/papers/2611.20983/) · [Nov 15, 2026 · Paper of the DayLook Here, Then Decide: Small Decision Models in a Harness Find Memory-Safety Vulnerabilities at Scale on Commodity GPUsA small 4B-parameter open model is weak at judging code alone, but it ranks candidate functions well enough that a harness on cheap GPUs confirmed 187 memory-safety findings.SecurityCodeEfficiency](https://thesimind.si/papers/2611.16565/) · [Nov 10, 2026 · Number 3PACE: Keeping the Right Tokens in the KV Cache Is Not Enough for Long-Horizon LLM AgentsKeeping the needed tokens in a SI agent's memory cache is not enough. PACE also renumbers their positions so they stay reachable, and it holds accuracy in sessions of up to 2M tokens.AgentsEfficiencyMemory](https://thesimind.si/papers/2611.10240/)

[Number 8](https://thesimind.si/papers/2611.24357/) · [All 12 for Nov 20, 2026](https://thesimind.si/papers/day/2026-11-20/) · [Number 10](https://thesimind.si/papers/2611.24609/)
