Post-Training Statistical Calibration for Higher Activation Sparsity
Chua, Vui Seng, Pan, Yujie, Jain, Nilesh
–arXiv.org Artificial Intelligence
We present Statistical Calibrated Activation Pruning (SCAP), a post-training activation pruning framework that (1) generalizes sparsification by input activations of Fully-Connected layers for generic and flexible application across Transformers, and (2) features a simple Mode-Centering technique to pre-calibrate activation distributions for maximizing post-training sparsity. Our results demonstrate robust Pareto efficiency compared to prior methods, translating to a 1.5x additional LLM decoding speedup against CATS at iso model quality. SCAP effectiveness is empirically verified across a wide range of models, including recent Transformer Decoders, MoE, Mamba2, Encoding Transformer, and pre-quantized models, highlighting its practicality and scalability. The code is available at: https://github.com/IntelLabs/SCAP.
arXiv.org Artificial Intelligence
Dec-9-2024
- Country:
- North America > United States
- Hawaii > Honolulu County
- Honolulu (0.04)
- California > San Diego County
- Carlsbad (0.04)
- Hawaii > Honolulu County
- Asia > Middle East
- North America > United States
- Genre:
- Research Report > New Finding (0.86)
- Technology: