Research map / Architectures and efficiency / Quantization
FP=XINT: Representing Neural Networks via Low-Bit Series Basis Functions
Boyang Zhang, Daning Cheng, Yunquan Zhang, Jiake Tian, Jing Li, Fangming Liu
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 |
|---|---|---|---|
| expand full-precision model into low-bit series basis models (no calibration or fine-tuning) vs best PTQ baseline | ResNet-50 · ImageNet W4A4 | top-1 accuracy: 75.09 -> 77.03 | Improved |
| series expansion vs CL-Calib at 2-bit weights, 2-bit activations | ResNet-50 · ImageNet W2A2 | top-1 accuracy: 58.30 -> 62.13 | Improved |
| series expansion vs PTQ baselines on BERT W4A4 | BERT · SQuAD1.1 | F1: 75.97 -> 79.30 | Improved |
| series expansion vs GPTQ on 4-bit LLM | LLaMA3-8B · MMLU average | accuracy: 63.0 -> 64.7 | Improved |
| increase number of expansion terms (ablation) | ResNet-50 · ImageNet INT4 | accuracy | Mixed |
| disable clipping (Wsa = 0) in saturation basis (ablation) | series expansion · ResNet-18/50 | accuracy | No clear effect |
| expand weights beyond 2-3 terms (ablation) | series expansion · various | accuracy | No clear effect |
| use 4-bit basis instead of 8-bit basis | ResNet-50 · efficiency | latency (ms): 0.74 (down 30%) | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- PD-Quant — PD-Quant: Post-Training Quantization Based on Prediction Difference Metric (CVPR 2023) · better on 3 of 3 shared results
- QDrop — QDrop: Randomly Dropping Quantization for Extremely Low-bit Post-Training Quantization (ICLR 2022) · better on 3 of 3 shared results
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
- HOPE: High-Order Polynomial Expansion of Black-Box Neural Networks (TPAMI 2024 · similarity 0.86)
- Approximation to Smooth Functions by Low-Rank Swish Networks (ICML 2025 · similarity 0.85)
- QuadEnhancer: Leveraging Quadratic Transformations to Enhance Deep Neural Networks (NeurIPS 2025 · similarity 0.85)
- QBB: Quantization with Binary Bases for LLMs (NeurIPS 2024 · similarity 0.85)
- Function-space Parameterization of Neural Networks for Sequential Learning (ICLR 2024 · similarity 0.85)
- An Efficient Matrix Multiplication Algorithm for Accelerating Inference in Binary and Ternary Neural Networks (ICML 2025 · similarity 0.84)
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.
