Pith. sign in

REVIEW 26 cited by

BRECQ: Pushing the Limit of Post-Training Quantization by Block Reconstruction

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2102.05426 v2 pith:QUHPPJQI submitted 2021-02-10 cs.LG cs.CV

BRECQ: Pushing the Limit of Post-Training Quantization by Block Reconstruction

Yuhang Li , Ruihao Gong , Xu Tan , Yang Yang , Peng Hu , Qi Zhang , Fengwei Yu , Wei Wang
show 1 more author
Shi Gu
This is my paper
classification cs.LG cs.CV
keywords brecqquantizationneuralerrorfirstframeworkmodelspost-training
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

We study the challenging task of neural network quantization without end-to-end retraining, called Post-training Quantization (PTQ). PTQ usually requires a small subset of training data but produces less powerful quantized models than Quantization-Aware Training (QAT). In this work, we propose a novel PTQ framework, dubbed BRECQ, which pushes the limits of bitwidth in PTQ down to INT2 for the first time. BRECQ leverages the basic building blocks in neural networks and reconstructs them one-by-one. In a comprehensive theoretical study of the second-order error, we show that BRECQ achieves a good balance between cross-layer dependency and generalization error. To further employ the power of quantization, the mixed precision technique is incorporated in our framework by approximating the inter-layer and intra-layer sensitivity. Extensive experiments on various handcrafted and searched neural architectures are conducted for both image classification and object detection tasks. And for the first time we prove that, without bells and whistles, PTQ can attain 4-bit ResNet and MobileNetV2 comparable with QAT and enjoy 240 times faster production of quantized models. Codes are available at https://github.com/yhhhli/BRECQ.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 26 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. GPTQ-intrinsic LoRA: A Near-optimal Algorithm for Low-precision Quantization with Low-rank Adaptation

    cs.LG 2026-05 unverdicted novelty 8.0

    GPTQ-intrinsic LoRA augments GPTQ with intrinsic low-rank compensation via Hessian modification to achieve layer-wise reconstruction bounds that match information-theoretic lower bounds under structural assumptions.

  2. Reclaiming Residual Knowledge: A Novel Paradigm to Low-Bit Quantization

    cs.CV 2024-08 unverdicted novelty 7.0

    CoRa reclaims quantization residuals in pre-trained ConvNets by searching low-rank adapter architectures instead of weights, matching SOTA accuracy on ImageNet in 3-4 bit settings with under 250 iterations on 1600 images.

  3. STaR-Quant: State-Time Consistent Post-Training Quantization for Diffusion Large Language Models

    cs.LG 2026-06 unverdicted novelty 6.0

    STaR-Quant provides a state-time consistent PTQ framework for DLLMs using SGAT and TAC to improve low-bit weight-activation quantization.

  4. Selective Coupling of Decoupled Informative Regions: Masked Attention Alignment for Data-Free Quantization of Vision Transformers

    cs.CV 2026-06 unverdicted novelty 6.0

    MaskAQ generates synthetic samples for ViT quantization by identifying and selectively aligning sparse informative regions in self-attention maps, combined with periodic sample refreshing.

  5. Motion-Compensated Weight Compression

    cs.CV 2026-05 unverdicted novelty 6.0

    MCWC aligns permutation-symmetric blocks across layers to enable sequential prediction and residual entropy coding, improving rate-accuracy tradeoffs versus quantization and prior codecs on language and vision models.

  6. GEMQ: Global Expert-Level Mixed-Precision Quantization for MoE LLMs

    cs.LG 2026-05 unverdicted novelty 6.0

    GEMQ applies global LP-based expert importance estimation and router fine-tuning within progressive quantization to cut memory and speed inference in MoE LLMs with little accuracy loss.

  7. SAFE-SVD: Sensitivity-Aware Fidelity-Enforcing SVD for Physics Foundation Models

    cs.LG 2026-05 unverdicted novelty 6.0

    SAFE-SVD introduces a sensitivity-aware fidelity-enforcing SVD framework for compressing physics foundation models that maintains higher accuracy than standard methods at greater compression ratios.

  8. Nonlinear Bipolar Compensation: Handling Outliers in Post-Training Quantization

    cs.CV 2026-05 unverdicted novelty 6.0

    Nonlinear Bipolar Compensation with Bipolar Logarithmic Transformation reduces outlier effects in post-training quantization by performing compensation in a compressed transformed space.

  9. High-Rate Quantized Matrix Multiplication II

    cs.LG 2026-05 unverdicted novelty 6.0

    Waterfilling rate allocation makes quantized matrix multiplication for LLMs near information-theoretically optimal, with WaterSIC being basis-free and within 0.25 bits per entry of the limit.

  10. OSAQ: Outlier Self-Absorption for Accurate Low-bit LLM Quantization

    cs.LG 2026-05 unverdicted novelty 6.0

    OSAQ uses the low-rank structure of the Hessian to construct a closed-form additive weight transformation that suppresses outliers without changing task loss, enabling better low-bit LLM quantization.

  11. OSAQ: Outlier Self-Absorption for Accurate Low-bit LLM Quantization

    cs.LG 2026-05 unverdicted novelty 6.0

    OSAQ suppresses weight outliers in LLMs via a closed-form additive transformation from the Hessian's stable null space, improving 2-bit quantization perplexity by over 40% versus vanilla GPTQ with no inference overhead.

  12. CoreQ: Learning-Free Mismatch Correction and Successive Rounding for Quantization

    cs.LG 2026-02 unverdicted novelty 6.0

    CoreQ delivers adaptive mismatch correction via closed-form geometric coefficient and successive rounding to improve PTQ accuracy for large language models.

  13. Rethinking Output Alignment For 1-bit Post-Training Quantization of Large Language Models

    cs.LG 2025-12 unverdicted novelty 6.0

    A post-training 1-bit quantization method for LLMs that fixes error accumulation and anisotropic representation distortion to outperform prior weight-driven and naive output-driven baselines.

  14. Sampling-Aware Quantization for Diffusion Models

    cs.CV 2025-05 unverdicted novelty 6.0

    A quantization technique for diffusion models that aligns sampling trajectories to preserve high-order sampler performance under quantization noise.

  15. AHCQ-SAM: Toward Accurate and Hardware-Compatible Post-Training Segment Anything Model Quantization

    cs.CV 2025-03 unverdicted novelty 6.0

    AHCQ-SAM introduces ACNR, HLUQ, CAG, and LNQ quantization techniques that deliver 15.2% mAP gain on 4-bit SAM-B and 14.01% J&F gain on 4-bit SAM2-Tiny versus prior PTQ methods.

  16. Efficient Reasoning on the Edge

    cs.LG 2026-03 accept novelty 5.5

    LoRA adapters, budget-forced GRPO, dynamic switching, parallel verification and FPTQuant enable practical chain-of-thought reasoning on quantized Qwen2.5-7B for edge devices.

  17. Minimizing the Hidden Cost of Scales: Graph-Guided Ultra-Low-Bit Quantization for Large Language Models

    cs.AI 2026-06 unverdicted novelty 5.0

    SAGE-PTQ is a graph-guided ultra-low-bit PTQ framework that achieves 1.03 average weight bits and 0.004 scaling bits per matrix on LLMs while reporting lower perplexity and memory use than BiLLM and PB-LLM.

  18. MARR: Module-Adaptive Residual Reconstruction for Low-Bit Post-Training Quantization

    cs.LG 2026-05 unverdicted novelty 5.0

    MARR uses per-module adaptive residual scaling updated by PID feedback to balance error correction against Hessian-approximation bias in low-bit PTQ.

  19. CAR-SAM: Cross-Attention Reconstruction for Post-Training Quantization of the Segment Anything Model

    cs.CV 2026-05 unverdicted novelty 5.0

    CAR-SAM introduces MatMul-Aware Compensation and Joint Cross-Attention Reconstruction to enable stable 4-bit post-training quantization of SAM, outperforming prior PTQ methods by 14.6% mAP on SAM-B and 6.6% on SAM-L.

  20. High-Rate Quantized Matrix Multiplication II

    cs.LG 2026-05 unverdicted novelty 5.0

    With known covariance, waterfilling improves GPTQ and WaterSIC reaches within 0.25 bit/entry of the rate-distortion limit while being basis-independent.

  21. DA-PTQ: Drift-Aware Post-Training Quantization for Efficient Vision-Language-Action Models

    cs.RO 2026-04 unverdicted novelty 5.0

    DA-PTQ quantizes VLAs by compensating cross-space distortions and allocating mixed precision to minimize motion errors and kinematic drift in trajectories.

  22. Evolution Strategy-Based Calibration for Low-Bit Quantization of Speech Models

    cs.SD 2026-03 conditional novelty 5.0

    Using CMA-ES to jointly optimize activation quantization scales keeps speech-model accuracy near full precision under full INT8 and INT4 quantization.

  23. Rethinking Output Alignment For 1-bit Post-Training Quantization of Large Language Models

    cs.LG 2025-12 unverdicted novelty 5.0

    A post-training quantization technique for 1-bit LLMs that corrects layer-wise error accumulation and anisotropic representation distortion to preserve output behavior more effectively than existing methods.

  24. MixFrag: Fragility-Guided Mixed-Precision Post-Training Quantization for Vision Transformers

    cs.CV 2026-07 conditional novelty 4.5

    KL-isolation fragility plus MCKP bit allocation yields mixed-precision ViT PTQ that lags recent ImageNet PTQ but reports large COCO AP gains at MP3/MP3.

  25. Cross-Layer Error Compensation and Finite-Sample Feature-Statistics Matching for Extreme Low-Bit Quantization of Large Language Models

    cs.NE 2026-07 conditional novelty 4.0

    A whole-network final-feature and statistics matching objective improves 1.125-bit and 4.125-bit LLM quantization over layer-local and distillation baselines, but its cross-layer mechanism reduces exactly to final-fea...

  26. GoQuant: Geometric Orthogonal Residual Projection for Multiplier-Free Power-of-Two Transformer Quantization

    cs.LG 2026-05 unverdicted novelty 4.0

    GoQuant formulates quantization as dual-basis geometric projection to create higher-resolution residual lattices for 3-bit PoT transformer quantization using only shift-and-add, reporting 6.10 perplexity on LLaMA-2-7B.