AI Paper Atlas

Research map / Architectures and efficiency / Quantization

XTC: Extreme Compression for Pre-trained Transformers Made Simple and Efficient

Xiaoxia Wu, Zhewei Yao, Minjia Zhang, Conglong Li, Yuxiong He

NeurIPS 2022 · Architectures and efficiency › Quantization · approach: Ultra-low-bit LLM quantization · 0 citations (OpenAlex)

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 changedWhereMeasuredEffect
Increase the training budget from Budget-A to Budget-C (ablation)BERT-base with 1-bit quantization · GLUEAvg. all: 82.38 -> 83.44Improved
Use one-stage KD with learning-rate search instead of multi-stage KD (ablation)BERT-base with 1-bit quantization, Budget-A · GLUEAvg all: 81.59 -> 82.38Improved
Remove data augmentation (ablation)BERT-base with 1-bit quantization, Budget-A · GLUEAvg. all: 82.38 -> 81.61; -0.77Hurt
Remove data augmentation (ablation)BERT-base with 1-bit quantization, Budget-C · GLUEAvg. all: 83.44 -> 82.78; -0.66Hurt
Use lightweight layer reduction initialized from the fine-tuned teacher instead of pre-training distillation (ablation)Task-specific BERT student · GLUEAvg. all: 83.11 -> 83.50Mixed
Reduce the Skip-BERT student from 6 layers to 5 (ablation)Skip-BERT · GLUEAvg. all: 83.50 -> 83.38Hurt
Reduce the Skip-BERT student from 6 layers to 4 (ablation)Skip-BERT · GLUEAvg. all: 83.50 -> 82.18Hurt
Reduce quantized student from 6 layers to 5 (ablation)XTC with 2-bit quantization · GLUEAvg.: 82.89 -> 82.46Hurt
Reduce quantized student from 6 layers to 4 (ablation)XTC with 2-bit quantization · GLUEAvg.: 82.89 -> 81.22Hurt
Reduce quantized student from 6 layers to 5 (ablation)XTC with 1-bit quantization · GLUEAvg.: 81.71 -> 81.34Hurt
Reduce quantized student from 6 layers to 4 (ablation)XTC with 1-bit quantization · GLUEAvg.: 81.71 -> 79.96Hurt
Add low-rank full-precision weight matrices (ablation)XTC 1-bit layer-reduced models · GLUEaverage GLUE scoreHurt

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy