Fix dgs types - Githubissues

sign-language-processing / datasets

TFDS data loaders for sign language datasets.

https://sign-language-processing.github.io/#existing-datasets

83 stars 27 forks source link

Fix dgs types #18

Closed bricksdont closed 2 years ago

bricksdont commented 2 years ago

also catch "Token" strings without "s"
add frequency feature
add dgs_types to load.ipynb

the number of dataset elements increases from 5551 to 6506 (lower than 2647 because supertypes and subtypes are grouped)

AmitMY commented 2 years ago

looks good to me.

just verifying, the difference in the groups if match.group(3) != '' -> if match.group(3) is not None is because the 3rd group is optional, while the 4th group (used to be the 3rd) was not optional.