Closed ghost closed 3 years ago
Hi, same question here, could you please help in including "WikiAnn-NER" to facilitate testing mt5 model on this task? thank you for your help.
I also have the same question. Dose it directly decodes the tag sequences in a seq2seq style without any constraints? It seems to be time-consuming for long sentences.
Thanks @craffel !
The WikiAnn NER task has been added here: https://github.com/google-research/multilingual-t5/blob/master/multilingual_t5/tasks.py#L446
Hi I'd like to test T5 on the WikiAnn NER tasks in the paper, I checked the tasks and to me NER ones are not added, could you assist me with sharing with me how I can add them and test mT5 on NER task? thanks