Hey Cadia. What are the results?

[johanub]

I was wondering if you have the data over which network was the best at punctuating text. The bidirectional RNN or BERT based one?
I am interested in starting a project of my own and I would like to know which direction is the right one to look in.

Best regards Johan

[ingarun]

Hi Johan I see that we need to add a results file. The BERT model was better. But we also tested simulating ASR errors and saw that Punctuator2 was more robust against distorted sentence structures, so that is something to consider if you are going to use this with ASR results. Best, Inga

…

On Tue, Jan 19, 2021 at 11:16 PM johanub ***@***.***> wrote: I was wondering if you have the data over which network was the best at punctuating text. The bidirectional RNN or BERT based one? I am interested in starting a project of my own and I would like to know which direction is the right one to look in. Best regards Johan — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#8>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AHHJ6GC4U3HVEAODFUI5OALS2YHFRANCNFSM4WJWEYJQ> .

[johanub]

Thanks for the answer. Looking forward to the results file!