Closed humzaiqbal closed 5 months ago
The CLIPA tokenizer description as it exists right now isn't exactly correct. In order to use any of the masking you need to add
"tokenizer_kwargs": { "reduction_mask": "syntax" }
as a configuration within text_cfg. This PR fixes the documentation.
text_cfg
The CLIPA tokenizer description as it exists right now isn't exactly correct. In order to use any of the masking you need to add
as a configuration within
text_cfg
. This PR fixes the documentation.