Gradient Compression Supercharged High-Performance Data Parallel DNN Training.
Youhui BaiCheng LiQuan ZhouJun YiPing GongFeng YanRuichuan ChenYinlong XuPublished in: SOSP (2021)
Keyphrases
- database
- data sets
- test data
- data processing
- small number
- data compression
- raw data
- data collection
- data points
- data analysis
- data sources
- prior knowledge
- training data
- high quality
- training examples
- high dimensional data
- experimental data
- databases
- neural network
- data distribution
- spatial data
- synthetic data
- data structure
- input data
- data warehouse
- end users