namisan / mt-dnn

Multi-Task Deep Neural Networks for Natural Language Understanding
MIT License
2.24k stars 412 forks source link

question about task split and pretrain model #243

Open chaser-girl opened 2 years ago

chaser-girl commented 2 years ago

Hello! Your work is very effective. I wonder if tasks must be separated from the benchmark model before they can be used as pre training language models for other tasks? Does the provided. PT file split the task?