Research map / Architectures and efficiency / Quantization
First-Order Error Matters: Accurate Compensation for Quantized Large Language Models
Xingyu Zheng, Haotong Qin, Yuye Li, Haoran Chu, Jiakai Wang, Jinyang Guo, Michele Magno, Xianglong 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 |
|---|---|---|---|
| add first-order error term to GPTQ compensation | GPTQ · Llama3-8B WikiText2 W3 | perplexity: 9.86 -> 8.32 | Improved |
| FOEM instead of GPTAQ | GPTAQ · Llama3-8B MMLU 5-shot W3 | accuracy: 53.8 -> 56.1 | Improved |
| add first-order term to GPTQ | GPTQ · Llama2-7B MMLU W3 | accuracy: 40.3 -> 42.0 | Improved |
| add first-order term at 4-bit | GPTQ · Llama3-8B WikiText2 W4 | perplexity: 6.54 -> 6.54 | No clear effect |
| add first-order term at 4-bit | GPTAQ · Llama3.2-3B WikiText2 W4 | perplexity: 8.66 -> 8.67 | No clear effect |
| FOEM vs GPTQ on Llama3-70B | GPTQ · Llama3-70B WikiText2 W3 | perplexity: 5.37 -> 5.36 | No clear effect |
| FOEM vs GPTQ on 3-bit zero-shot BoolQ | GPTQ · Llama3-8B BoolQ W3 | accuracy: 71.8 -> 69.1 | Hurt |
| combine FOEM with SpinQuant at W4A4KV4 | GPTQ · Llama3-8B WikiText2 | perplexity: 8.55 -> 8.35 | Improved |
| FOEM on Qwen3-8B | GPTQ · Qwen3-8B WikiText2 W3 | perplexity: 8.32 -> 8.17 | Improved |
| FOEM on Mamba-1.4B | GPTAQ · Mamba-1.4B WikiText2 W3A16 | perplexity: 14.10 -> 13.91 | Improved |
| raise stabilization factor beta from 0.1 to 1.0 (ablation) | FOEM · Llama3-8B WikiText2 W3 | perplexity: 8.32 -> 10.37 | Hurt |
| raise beta from 0.1 to 0.3 (ablation) | FOEM · Llama3-8B 0-shot avg W3 | accuracy: 65.5 -> 65.8 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- GPTAQ — GPTAQ: Efficient Finetuning-Free Quantization for Asymmetric Calibration (ICML 2025) · better on 11 of 11 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
- Exploring Post-training Quantization in LLMs from Comprehensive Study to Low Rank Compensation (AAAI 2024 · similarity 0.93)
- QLLM: Accurate and Efficient Low-Bitwidth Quantization for Large Language Models (ICLR 2024 · similarity 0.91)
- Norm Tweaking: High-Performance Low-Bit Quantization of Large Language Models (AAAI 2024 · similarity 0.91)
- BiLLM: Pushing the Limit of Post-Training Quantization for LLMs (ICML 2024 · similarity 0.91)
- ABQ-LLM: Arbitrary-Bit Quantized Inference Acceleration for Large Language Models (AAAI 2025 · similarity 0.90)
- ResQ: Mixed-Precision Quantization of Large Language Models with Low-Rank Residuals (ICML 2025 · similarity 0.90)
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.
