How to perform data augmentation in model pipeline texcat spacy?

[blizaga]

I'm trying to follow the tutorial for data augmentation, but it only explains for text and then what about the textcat label on each text. How to do the augmentation?

I created a dataframe for text columns and labels as input to the spacy training process using the CLI. Here I use Indonesian language

[KennethEnevoldsen]

Hei, sounds like you can just apply the augmentation on the each document independently.

I am not at a computer at the moment, but I can send a snippet when I am.

[blizaga]

However, I would like to ask again, how to implement multi augmentation using the function above?

[KennethEnevoldsen]

By multiaugmentation I assume you mean creating multiple augmentations from the same text? I.e. for the text:

Hope you are well
You might get:

Hope you are ell
Hope you well are
Hope you well
hope you well

One option is to use multiple augmenters. Another is to use the repeat augmenter using combined augmenters.

[blizaga]

Can you provide code examples especially for combination augmentation?

[KennethEnevoldsen]

I have added an example in the documentation for using combine, which is as follows:

char_swap_augmenter = augmenty.load("char_swap_v1", level=.02)
synonym_augmenter = augmenty.load("wordnet_synonym_v1", level=1, lang="en")
combined_aug = augmenty.combine([char_swap_augmenter, synonym_augmenter])
# combine doc using two augmenters
augmented_docs = augmenty.docs(docs, augmenter=combined_aug, nlp=nlp)

[blizaga]

thank you for the insight, it's very useful