Pith. sign in

REVIEW 30 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

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 · ORCID
classification cs.LGcs.CV
keywords brecqquantizationneuralerrorfirstframeworkmodelspost-training
verification ladder T0 review T1 audit T2 compute T3 formal
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). Continue with ORCID to comment.

Forward citations

Cited by 30 Pith papers

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

  1. FPTQuant: Function-Preserving Transforms for LLM Quantization

    cs.LG 2025-06 conditional novelty 7.0 of 10

    FPTQuant introduces function-preserving transforms that make transformer activations amenable to static 4-bit quantization with minimal inference overhead.

  2. Hidden Language Consistency Phenomena in Reasoning LLMs

    cs.CL 2026-08 conditional novelty 6.0 of 10

    Reasoning models often stop using the requested language as problems get harder, and this language breakdown can make accuracy look better than it is.

  3. Task-Specific Zero-shot Quantization-Aware Training for Object Detection

    cs.CV 2025-07 conditional novelty 6.0 of 10

    A zero-shot quantization-aware training method for object detectors that synthesizes task-specific images with bounding-box labels via adaptive label sampling, then distills task-specific knowledge into the quantized network.

  4. DMQ: Dissecting Outliers of Diffusion Models for Post-Training Quantization

    cs.CV 2025-07 conditional novelty 6.0 of 10

    A post-training quantization method that combines learned channel scaling and power-of-two scaling keeps diffusion image quality high at 4-bit weight, 6-bit activation precision.

  5. TruncQuant: Truncation-Ready Quantization for DNNs with Flexible Weight Bit Precision

    cs.LG 2025-06 conditional novelty 6.0 of 10

    TruncQuant uses a floor-based quantizer with 2^n scaling instead of rounding with 2^n-1, so truncating high-precision weights via bit-shift exactly matches direct low-precision quantization, recovering accuracy lost b...

  6. BASE-Q: Bias and Asymmetric Scaling Enhanced Rotational Quantization for Large Language Models

    cs.LG 2025-05 conditional novelty 6.0 of 10

    BASE-Q combines bias correction and asymmetric scaling under fixed rotations to improve 4-bit weight-activation quantization, narrowing the accuracy gap to full precision by up to 50.5% over prior rotation-based methods.

  7. Saliency-Aware Quantized Imitation Learning for Efficient Robotic Control

    cs.RO 2025-05 conditional novelty 6.0 of 10

    A saliency-weighted quantization-aware training method lets 4-bit quantized imitation-learning policies match full-precision success rates across robot manipulation, driving, and control benchmarks.

  8. Loss Landscape Analysis for Reliable Quantized ML Models for Scientific Sensing

    cs.LG 2025-02 conditional novelty 6.0 of 10

    Loss landscape flatness and connectivity visually correlate with robustness to noise and bit flips in two quantized scientific sensing models, but the correlation is not quantified and the a priori claim remains unvalidated.

  9. DGQ: Distribution-Aware Group Quantization for Text-to-Image Diffusion Models

    cs.CV 2025-01 conditional novelty 6.0 of 10

    DGQ quantizes text-to-image diffusion models to 4-8 bits without fine-tuning by preserving activation outliers and applying prompt-specific log quantization to cross-attention scores.

  10. PTQ4VM: Post-Training Quantization for Visual Mamba

    cs.CV 2024-12 conditional novelty 6.0 of 10

    PTQ4VM quantizes Visual Mamba models to 8-bit and 6-bit with small accuracy loss by adding per-token static quantization and joint learning of smoothing scales and step sizes.

  11. PTSBench: A Comprehensive Post-Training Sparsity Benchmark Towards Algorithms and Models

    cs.LG 2024-12 conditional novelty 6.0 of 10

    PTSBench benchmarks post-training sparsity techniques and model families, finding learning-based allocation and block-wise reconstruction most effective, and attention-based models most sparsity-friendly.

  12. RILQ: Rank-Insensitive LoRA-based Quantization Error Compensation for Boosting 2-bit Large Language Model Accuracy

    cs.LG 2024-12 conditional novelty 6.0 of 10

    RILQ improves 2-bit LLM accuracy by optimizing a model-wise discrepancy loss at the final layer, making LoRA error compensation rank-insensitive.

  13. PassionSR: Post-Training Quantization with Adaptive Scale in One-Step Diffusion based Image Super-Resolution

    cs.CV 2024-11 conditional novelty 6.0 of 10

    PassionSR quantizes one-step diffusion super-resolution models to 6-8 bits via learnable quantizer boundaries, equivalent transformations, and a two-stage calibration, while keeping quality close to full precision.

  14. Exploring the Robustness and Transferability of Patch-Based Adversarial Attacks in Quantized Neural Networks

    cs.CR 2024-11 conditional novelty 6.0 of 10

    Patch-based adversarial attacks remain highly effective and transferable across quantized neural networks, and a proposed quantization-aware defense training method substantially reduces their success rate.

  15. Efficient Reasoning on the Edge

    cs.LG 2026-03 accept novelty 5.5 of 10

    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.

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

    cs.SD 2026-03 conditional novelty 5.0 of 10

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

  17. Harnessing Input-Adaptive Inference for Efficient VLN

    cs.CV 2025-08 conditional novelty 5.0 of 10

    A three-part input-adaptive inference method (view masking, adaptive early exit, view caching) cuts VLN computation roughly in half on seven benchmarks with moderate success-rate loss.

  18. Compress Any Segment Anything Model (SAM)

    cs.CV 2025-07 conditional novelty 5.0 of 10

    Birkhoff compresses 18 SAM variants to about one-fifth their size with less than 1% accuracy loss, data-free, using a trajectory-based codebook and a fused GPU operator.

  19. PQCAD-DM: Progressive Quantization and Calibration-Assisted Distillation for Extremely Efficient Diffusion Model

    cs.CV 2025-06 conditional novelty 5.0 of 10

    PQCAD-DM combines progressive two-stage quantization with calibration-assisted distillation to halve the sampling steps of a quantized diffusion model while keeping FID scores close to baseline.

  20. Post-Training Quantization for Video Matting

    cs.CV 2025-06 conditional novelty 5.0 of 10

    PTQ4VM applies block-wise, globally affine-calibrated, optical-flow-guided post-training quantization to video matting, reaching near full-precision accuracy at 4-bit weights and activations.

  21. Accelerating Diffusion Transformer via Error-Optimized Cache

    cs.CV 2025-01 conditional novelty 5.0 of 10

    A precomputed trend correction injected into cached diffusion transformer blocks reduces FID by up to 28.8% at heavy caching with negligible added compute.

  22. D$^2$-DPM: Dual Denoising for Quantized Diffusion Probabilistic Models

    cs.CV 2025-01 conditional novelty 5.0 of 10

    Modeling quantization noise in compressed diffusion models as a time-step-dependent joint Gaussian, then correcting its mean and variance during sampling, improves FID over prior PTQ methods and can beat the full-prec...

  23. TTAQ: Towards Stable Post-training Quantization in Continuous Domain Adaptation

    cs.LG 2024-12 conditional novelty 5.0 of 10

    TTAQ combines weight standardization, perturbation consistency reconstruction, and adaptive balanced loss to improve post-training quantized models under continual test-time domain shift.

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

    cs.CV 2026-07 conditional novelty 4.5 of 10

    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 of 10

    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. Progressive Element-wise Gradient Estimation for Neural Network Quantization

    cs.LG 2025-08 conditional novelty 4.0 of 10

    PEGE replaces the straight-through estimator with a curriculum-driven blend of quantized and full-precision weights plus an additive discretization-error correction, reporting small accuracy gains on low-bit CNNs.

  27. Rethinking the Outlier Distribution in Large Language Models: An In-depth Study

    cs.CL 2025-05 reject novelty 4.0 of 10

    The paper claims that most massive activations in LLMs are residual-propagated 'fake' outliers that can be removed without accuracy loss.

  28. Can Post-Training Quantization Benefit from an Additional QLoRA Integration?

    cs.CL 2025-02 conditional novelty 4.0 of 10

    Adding QLoRA on top of a 4-bit post-training quantized LLM improves task accuracy over quantization alone, and can match or beat full 16-bit fine-tuning on some tasks.

  29. Efficiency Meets Fidelity: A Novel Quantization Framework for Stable Diffusion

    cs.CV 2024-12 conditional novelty 4.0 of 10

    A serial-to-parallel distillation pipeline with per-timestep activation quantization makes W4A8 quantized Stable Diffusion models generate images much closer to the full-precision model than prior PTQ methods.

  30. PQD: Post-training Quantization for Efficient Diffusion Models

    cs.CV 2024-12 reject novelty 3.0 of 10

    PQD calibrates diffusion-model quantization on time steps drawn from a tuned normal distribution, reporting competitive 8-bit FID on 64x64 ImageNet but much worse 4-bit FID and no quantitative text-to-image results.

Pith tools