For the labels containing multiple words, How to take the mean-pooling?

[TianlinZhang668]

For the labels containing multiple words, How to take the mean-pooling? Is there this code in files?

[hiyouga]

There are two approaches to tackling this problem. We can take the mean-pooling on the BERT outputs to obtain the label representations. Alternatively, we can also take the first word in the label if it does not conflict with other labels.