mlsanityData to 14 Oct 2026

Towards Training-free Automatic Proxy Discovery via Large Language Models for Mixed Precision Quantization

arXiv 2512.07419 ↗First posted 8 Dec 2025cs.LG cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

Mixed-Precision Quantization (MPQ) liberates Deep Neural Networks (DNNs) from the Out-Of-Memory (OOM) bottleneck and has garnered increasing research attention. However, conventional methods either rely on costly differentiable optimization search, which is neither efficient nor flexible, or learn a quantized DNN from a proxy (e.g., HAWQ) manually designed by human experts, which is labor-intensive and requires extensive expert knowledge. Can we design a proxy without involving any human experts or training? In this paper, we provide an affirmative answer by proposing a novel Large Language Model (LLM)-driven Training-free Automatic Proxy (dubbed TAP) discovery framework. It reforms the design paradigm of MPQ by utilizing LLMs and evolutionary search strategies to automatically find superior TAP tailored for MPQ. In addition, to bridge the gap between black-box LLMs and the challenging MPQ task, we introduce a lightweight Direct Preference Optimization (DPO)-based strategy controller that dynamically reweights the selection probabilities of the three prompt templates for evolutionary search strategies according to fitness signals, without fine-tuning the LLM. This forms a task-aware feedback loop that improves proxy generation across evolutions. Extensive experiments on mainstream benchmarks demonstrate that TAP achieves state-of-the-art performance. Finally, we believe that our TAP will significantly contribute to the MPQ community by providing a new perspective on LLM-driven design algorithms.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 29 of its 36 references that are in the map. First posted on arXiv 8 Dec 2025.

Builds on

29 of 36 references.

References · 29

  1. 2025Qwen3 Technical Reportcited by 8,820
  2. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 8,065
  3. 2017Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inferencecited by 1,593
  4. 2021BRECQ: Pushing the Limit of Post-Training Quantization by Block Reconstructioncited by 479
  5. 2018PACT: Parameterized Clipping Activation for Quantized Neural Networkscited by 467
  6. 2018HAQ: Hardware-Aware Automated Quantization with Mixed Precisioncited by 443
  7. 2019HAWQ: Hessian AWare Quantization of Neural Networks with Mixed-Precisioncited by 402
  8. 2021Post-Training Quantization for Vision Transformercited by 278
  9. 2018LQ-Nets: Learned Quantization for Highly Accurate and Compact Deep Neural Networkscited by 274
  10. 2020ZeroQ: A Novel Zero Shot Quantization Frameworkcited by 269
  11. 2019HAWQ-V2: Hessian Aware trace-Weighted Quantization of Neural Networkscited by 261
  12. 2019Low-bit Quantization of Neural Networks for Efficient Inferencecited by 193
17 more
  1. 2020LSQ+: Improving low-bit quantization through learnable offsets and better initializationcited by 189
  2. 2021PTQ4ViT: Post-training quantization for vision transformers with twin uniform quantizationcited by 179
  3. 2021FQ-ViT: Post-Training Quantization for Fully Quantized Vision Transformercited by 151
  4. 2018Mixed Precision Quantization of ConvNets via Differentiable Neural Architecture Searchcited by 125
  5. 2020Rethinking Differentiable Search for Mixed-Precision Neural Networkscited by 82
  6. 2023Towards Accurate Post-Training Quantization for Vision Transformercited by 71
  7. 2022Patch Similarity Aware Data-Free Quantization for Vision Transformerscited by 56
  8. 2020FracBits: Mixed Precision Quantization via Fractional Bit-Widthscited by 55
  9. 2021OMPQ: Orthogonal Mixed Precision Quantizationcited by 39
  10. 2023EMQ: Evolving Training-free Proxies for Automated Mixed Precision Quantizationcited by 22
  11. 2021Differentiable Dynamic Quantization with Mixed Precision and Adaptive Resolutioncited by 20
  12. 2021Generalizable Mixed-Precision Quantization via Attribution Rank Preservationcited by 11
  13. 2019Constructing Energy-efficient Mixed-precision Neural Networks through Principal Component Analysis for Edge Intelligencecited by 10
  14. 2024MPTQ-ViT: Mixed-Precision Post-Training Quantization for Vision Transformercited by 10
  15. 2019Mixed-Precision Quantized Neural Network with Progressively Decreasing Bitwidth For Image Classification and Object Detectioncited by 5
  16. 2025Learning from Loss Landscape: Generalizable Mixed-Precision Quantization via Adaptive Sharpness-Aware Gradient Aligningcited by 3
  17. 2025Where and How to Enhance: Discovering Bit-Width Contribution for Mixed Precision Quantizationcited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2023SEAM: Searching Transferable Mixed-Precision Quantization Policy through Large Margin Regularization
  2. 2024APTQ: Attention-aware Post-Training Mixed-Precision Quantization for Large Language Models
  3. 2025InfoQ: Mixed-Precision Quantization via Global Information Flow
  4. 2026Quantized Evolution Strategies: High-precision Fine-tuning of Quantized LLMs at Low-precision Cost
  5. 2025Tequila: Trapping-free Ternary Quantization for Large Language Models
  6. 2023AutoQNN: An End-to-End Framework for Automatically Quantizing Neural Networks
  7. 2026Q-Strata: Hierarchical Bit Allocation for Mixed-Precision Quantization of Mixture-of-Experts LLMs
  8. 2024AutoMixQ: Self-Adjusting Quantization for High Performance Memory-Efficient Fine-Tuning