HMC-TRAN: A Tensor-core Inspired Hierarchical Model Compression for Transformer-based DNNs on GPU.
Shaoyi HuangShiyang ChenHongwu PengDaniel ManuZhenglun KongGeng YuanLei YangShusen WangHang LiuCaiwen DingPublished in: ACM Great Lakes Symposium on VLSI (2021)
Keyphrases
- hierarchical model
- hierarchical models
- latent variables
- image compression
- high order
- higher order
- human body
- compression ratio
- real time
- compression algorithm
- data compression
- fuzzy logic
- graphics hardware
- compression scheme
- data sets
- parallel implementation
- fault diagnosis
- dimensionality reduction
- prior knowledge
- graph cuts
- graphics processing units
- bayesian networks
- graphics processors
- high voltage