Research map / Architectures and efficiency / Quantization
EMQ: Evolving Training-free Proxies for Automated Mixed Precision Quantization
Peijie Dong, Lujun Li, Zimian Wei, Xin Niu, Zhiliang Tian, Hengyue Pan
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use searched EMQ proxy instead of Synflow as MQ proxy | MQ proxy · MQ-Bench-101 | Spearman@100%: 31.57 -> 79.21 | Improved |
| add operation sampling prioritization (OSP) (ablation) | Branched structure search space · Validity rate | validity rate (%): 26.40 -> 36.45 | Improved |
| add diversity-prompting selection (DPS) to evolution (ablation) | EMQ evolution search · proxy search | search fitness | Improved |
| add equivalent checking and early rejection (ablation) | EMQ evolution search · Branched search | evaluated proxies: ~1 x 10^4 -> ~3 x 10^2 | Improved |
| use branched structure instead of DAG-based computation graph (ablation) | EMQ search space · proxy search | search performance: validity rate 5.40 -> 26.40 | Improved |
| use branched over sequential structure (ablation) | EMQ search space · proxy search | search performance | Improved |
| use EMQ instead of OMPQ for ResNet-18 PTQ at 4.5M | ResNet-18 PTQ · ImageNet | top-1 accuracy: 68.69 -> 69.66 | Improved |
| use EMQ instead of OMPQ for ResNet-18 QAT at A=8 | ResNet-18 QAT · ImageNet | top-1 accuracy: 72.30 -> 72.31 | No clear effect |
| use EMQ instead of OMPQ for ResNet-50 QAT at A=5 | ResNet-50 QAT · ImageNet | top-1 accuracy: 76.28 -> 76.70 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 4 of 4 on most shared results.
- OMPQ — OMPQ: Orthogonal Mixed Precision Quantization (AAAI 2023) · better on 7 of 7 shared results
- BRECQ — BRECQ: Pushing the Limit of Post-Training Quantization by Block Reconstruction (ICLR 2021) · better on 1 of 1 shared result
- HAWQ-V2 — HAWQ-V2: Hessian Aware trace-Weighted Quantization of Neural Networks (NeurIPS 2020) · better on 1 of 1 shared result
- SNIP — SNIP: SINGLE-SHOT NETWORK PRUNING BASED ON CONNECTION SENSITIVITY (ICLR 2019) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Learning from Loss Landscape: Generalizable Mixed-Precision Quantization via Adaptive Sharpness-Aware Gradient Aligning (ICML 2025 · similarity 0.89)
- Retraining-Free Model Quantization via One-Shot Weight-Coupling Learning (CVPR 2024 · similarity 0.88)
- CBQ: Cross-Block Quantization for Large Language Models (ICLR 2025 · similarity 0.87)
- ParetoQ: Improving Scaling Laws in Extremely Low-bit LLM Quantization (NeurIPS 2025 · similarity 0.87)
- GSQ: Highly-Accurate Low-Precision Scalar Quantization for LLMs via Gumbel-Softmax Sampling (NeurIPS 2026 · similarity 0.87)
- SynQ: Accurate Zero-shot Quantization by Synthesis-aware Fine-tuning (ICLR 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
