I think 11 GB is more than enough for 60k files unless each one of those files is gigantic! Checkout Sylvian’s comments in the issue I linked above. You should be able to resolve it!
Can we train this model to build hindi-english transliteration ?
It’s a good start, but you’ll need a training set with literal translations. Websites or books are often good. Quality assurance (and copyright) can be an issue.
For transliteration you’ll need an encoder-decoder model. For encoder, you can use the nlp-for-hindi ulmfit embeddings/encoder, but you’ll need to train your own decoder.
I’ll be happy to answer any questions you may have.
hey do you share entire model with code
hey can you pleas share the colab or github where everything works so it would more helpful as i also try the same for quotes generation in hindi