Hello, I’m using pytorch to teach NLP, and I would like to avoid giving bad habits to my students.
An exercice I give involves the usual manipulations of GloVe vectors (computing nearest neighbourg and so on). After the change of API for torchtext, I’m a bit nervous about what I should use.
The current documentation lists very few methods and variables ; calling help() on Vectors or on GloVe is a bit more informative, but, for instance, fields vectors or itos are not documented any more (or I miss something). It might come from my strongly typed languages background, but I like to know what’s ok for use and what could change - basically, the recommended way to use the classes.
Should I relax, and given that my code works, teach the use of itos to my students, or do I miss something ?
Best Regards,
Serge