nateraw

/

llama-2-7b-english-to-hinglish

like 9

never mind. it's listed

Ah sorry just saw this. yea I grabbed it from findnitai/english-to-hinglish and split to a static train/test split that I pushed to nateraw/english-to-hinglish. So all credit on dataset goes to @findnitai :)

For sake of transparency, when I did the actual run for this model, I removed a few rows (3-4 rows) from nateraw/english-to-hinglish's train set that had obvious hinglish profanity (bhen****, etc.). Those examples are left in the version I pushed though and I don't have record of them since I did this manually in a text editor.