Machine Learning
分组量化
Group QuantizationA method for compressing large language models (LLMs) by quantizing weights in groups rather than uniformly across the entire model. / 一种通过按组而非在整个模型中均匀量化权重来压缩大型语言模型（LLM）的方法。
- Knowledge type
- 工艺技术
- First proposed or discovered
- Proposer
- Turkey Farm member
- Creator
- 农场主
- Last updated
- 2026-11-01 18:06:39
Sub-knowledge points
No connected node yet
Higher-order knowledge points
Large Language Models / 数学理论Hardware–software co-design
Main parameters and performance indicators
group size: The number of weights quantized togetheroutlier-aware: A strategy to handle extreme values in the model weights
Open problems
No unresolved problem recorded yet
Related literature
ranked by community starsOFGQ: Hardware–software co-design for group quantization of LLMs with outlier-aware adaptive formatsOpen source
Related products and links
No item recorded yet
Creation and modification trail
1 recordsRecord title: Group Quantization.
Node update feedback
review required