Research map / Architectures and efficiency / Quantization
MSQ: Memory-Efficient Bit Sparsification Quantization
Seokho Han, Seoyeon Yoon, Jinhee Kim, Dongwei Wang, Kang Eun Jeon, Huanrui Yang, Jong Hwan Ko
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 MSQ mixed-precision instead of uniform 4-bit DoReFa quantization | ViT-Base-Patch16-224 · CIFAR-100 | accuracy (%) and compression ratio: 90.20 -> 91.45 accuracy; 8.00 -> 9.14 compression | Improved |
| increase the L1 regularization strength lambda on LSBs (ablation) | MSQ · LSB sparsity (ResNet) | LSB non-zero rate / sparsity: lambda 5e-5 -> 1e-4 gives a smaller LSB non-zero rate | Improved |
| set lambda too high (ablation) | MSQ · general | accuracy | Hurt |
Closest papers by meaning
- ParetoQ: Improving Scaling Laws in Extremely Low-bit LLM Quantization (NeurIPS 2025 · similarity 0.84)
- AQ-DETR: Low-Bit Quantized Detection Transformer with Auxiliary Queries (AAAI 2024 · similarity 0.83)
- Finite Scalar Quantization: VQ-VAE Made Simple (ICLR 2024 · similarity 0.83)
- QTIP: Quantization with Trellises and Incoherence Processing (NeurIPS 2024 · similarity 0.82)
- Binary Quadratic Quantization: Beyond First-Order Quantization for Real-Valued Matrix Compression (NeurIPS 2025 · similarity 0.82)
- QuIP$\#$: Even Better LLM Quantization with Hadamard Incoherence and Lattice Codebooks (ICML 2024 · similarity 0.82)
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.
