Upload tokenizer.json from xlm-roberta-base
#8
by
HelgiHilmarsson
- opened
Before change, the following code,
from transformers import pipeline
classifier = pipeline("zero-shot-classification", model="joeddav/xlm-roberta-large-xnli")
gives a TypeError:
TypeError: Descriptors cannot not be created directly
Adding the tokenizer from xlm-roberta-base worked locally.
ok, downgrading protobuf to 3.20.0 also worked
HelgiHilmarsson
changed pull request status to
closed