Closed rokrokss closed 5 years ago
Hi, thanks for your interest! I didn't keep prior files in previous experiments. In my experience, training from scratch should be fine for most of the sequence labeling datasets.
I think I understood how it works now! By the way, the implementation is very concrete and clever. I'm planning to study with your work here. Thanks for your kind answer.
I'm interested in your work, how can I get the prior files?