Great job. Quick inquiries.
#2
by
sapwavino
- opened
I was just about to embark on the same journey when i thought to check 🤗 first and fortunately i came across your project. Great job so far. I just have a few questions. What does your training data look like? I have a .tsv of translations of reviews and random questions from freebase_qa and amazon_reviews_multi and i'd love to go into training/finetuning. I was thinking of generating more datasets with this model translating the questions/reviews and i was wondering if that's the approach you've taken to diversify your training datasets. Also i can't seem to access your tokenizer. Is that by design?
Thank you for such amazing work.