PulseAugur
EN
LIVE 02:56:19

LLM agents automate and optimize large language model quantization

Researchers have developed QDA, a novel system that utilizes LLM agents to automate and optimize the process of post-training quantization for large language models. This closed-loop design refines quantization recipes by iteratively building and evaluating candidates, significantly reducing the time and expertise required compared to manual methods. QDA achieves improved accuracy over existing techniques, notably recovering 98.3% of FP8 model accuracy for the GLM-5.3-Flash model, and will release its library, evaluator, and harness. AI

IMPACT Automates and optimizes LLM quantization, potentially reducing serving costs and improving model performance.

RANK_REASON The cluster describes a new research paper detailing a novel method for LLM quantization.

Read on Hugging Face Daily Papers →

AI-generated summary · Google Gemini · from 5 sources. How we write summaries →

LLM agents automate and optimize large language model quantization

How we ranked this

Signal score
0 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Research
The cluster describes a new research paper detailing a novel method for LLM quantization.
Source corroboration
5 independent sources
Strong cross-source corroboration — multiple independent publishers covered this within the clustering window.
Topics
model release, infra
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
5 days old
Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.
Coverage growth since scoring
+2 source(s) since last score
New sources have picked up this story since our last re-score. Score will update on the next scoring pass.

Full methodology in our editorial standards.

COVERAGE [5]

  1. arXiv cs.LG TIER_1 English(EN) · Pier-Jean Malandrino ·

    Tetra: Serving Leech-Lattice Quantized LLMs at 2.7 Bits per Parameter

    arXiv:2609.35465v2 Announce Type: replace Abstract: Leech-lattice quantization gives good quality at two bits per weight, but its codebooks hold more than 10^14 points, too many for a lookup table. Our earlier kernel expanded the codes at load time and read 4.804 bits per weight …

  2. Hugging Face Daily Papers TIER_1 English(EN) ·

    Calibrate in Context: Efficient Data-Free Quantization from Small Synthetic Pools

    We enable accurate low-bit recovery from small synthetic pools by improving both image synthesis and student preparation. A grid with lower local response error can incur greater error in the student's block context. Functional Grid Calibration (FGC) selects activation scales and…

  3. arXiv cs.LG TIER_1 English(EN) · Xingyang Li, Tianlang Zhao, Song Han, Ligeng Zhu ·

    QDA: Closed-Loop LLM Quantization Design with Agents

    arXiv:2611.23388v1 Announce Type: new 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 …

  4. arXiv cs.LG TIER_1 English(EN) · Seohyeon Cha, Huancheng Chen, Dongjun Kim, Haoran Zhang, Kevin Chan, Gustavo de Veciana, Haris Vikalo ·

    CoreQ: Mismatch-Corrected Successive Rounding for LLM Quantization

    arXiv:2602.05902v3 Announce Type: replace Abstract: Post-training quantization (PTQ) enables efficient deployment of large language models by mapping pretrained weights to low-bit formats without retraining, typically using a small calibration set to minimize a layer-wise calibra…

  5. Hugging Face Daily Papers TIER_1 English(EN) ·

    QDA: Closed-Loop LLM Quantization Design with Agents

    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. Fortuna…