Replies: 1 comment 1 reply
-
< is out-of-vocabulary for the T5 tokenizer, so it gets tokenized/detokenized to ??. |
Beta Was this translation helpful? Give feedback.
1 reply
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
I train
t5
transformerHere is a sample (input, output):
input:
output:
However, for the prediction I get:
which replaces
<
with??
, what should I do to resolve this problem?Beta Was this translation helpful? Give feedback.
All reactions