附录A中提到：
For the multi-task setting, we combine the training set of the three datasets for pre-training. We use different linear classifiers for each dataset while sharing the continuous prompts.
这意味着不同任务的训练是分开进行的？针对不同的任务，需要手动更换linear classifier？
如有理解不到位之处烦请指教。
附录A中提到： For the multi-task setting, we combine the training set of the three datasets for pre-training. We use different linear classifiers for each dataset while sharing the continuous prompts. 这意味着不同任务的训练是分开进行的？针对不同的任务，需要手动更换linear classifier？ 如有理解不到位之处烦请指教。