Closed jmtatsch closed 11 months ago
Thanks @jmtatsch !
Always happy to help. Could you please point me to the correct vocab.txt?
@jmtatsch You can obtain it by downloading the tokenizer bert-multilingual-case
from HuggingFace. This is what bark uses. I'm wondering if I shouldn't push it directly in ggml_weights
. At least for now since we only support bark. WDYT?
I always like it as easy as possible for the end user which should net the most users and thus the most contributions. Also it reduces the potential amount of issues the user will open up ;)
partially fixes #71