diff options
Diffstat (limited to 'configdefinitions/src/vespa/hugging-face-tokenizer.def')
-rw-r--r-- | configdefinitions/src/vespa/hugging-face-tokenizer.def | 13 |
1 files changed, 10 insertions, 3 deletions
diff --git a/configdefinitions/src/vespa/hugging-face-tokenizer.def b/configdefinitions/src/vespa/hugging-face-tokenizer.def index bc0d5300de5..896a7b03234 100644 --- a/configdefinitions/src/vespa/hugging-face-tokenizer.def +++ b/configdefinitions/src/vespa/hugging-face-tokenizer.def @@ -8,7 +8,14 @@ model[].language string # The path to the model relative to the application package root model[].path model +# Include special tokens in output addSpecialTokens bool default=true -maxLength int default=512 -truncation bool default=true -padding bool default=false + +# Used for truncation/padding. Use -1 for model default. +maxLength int default=-1 + +# Truncation strategy. Use NOTSET for model default. +truncation enum { ON, OFF, NOTSET } default=NOTSET + +# Padding strategy. Use NOTSET for model default. +padding enum { ON, OFF, NOTSET } default=NOTSET |