-
-
-
Typically, copyright and licenses of data these kinds of models are trained on do not apply to the trained models (or at least everyone behaves that way) since the data itself is not shared. As far as I know, you can use BERT and BERTje any way you like (including commercially).
- Još 2 druga odgovora
Novi razgovor -
-
-
just sayin', a talk on this would be well received: https://pydata.org/amsterdam2020/cfp/ …
@pydataamsterdam cfp is open -
Thanks for the suggestion! May think about it.
Kraj razgovora
Novi razgovor -
-
-
Nice work! What did the 4.4GB books data in the training set consist of? Dbnl texts?
-
Thanks! The books do contain dbnl texts but the collection also contains a lot of other e-books that may not be freely available.
- Još 1 odgovor
Novi razgovor -
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.