Implementation of paper 《Modeling Task Relationships in Multi-task Learning with Multi-gate Mixture-of-Experts》
census data set from PaddleRec
- create
data
anddata/tfrecords
folders - download and move
train_data.csv
andtest_data.csv
todata
folder - Run with default config:
python main.py
- batch norm for census data
- try tencent video data set
- MMOE with attention
- grad norm