Now as of today, I am pleased to announce that Maral is here. The 7 billion parameters bilingual model which can respond to Persian and English prompts, and can produce GPT-3.5 level of answers based on the dataset we fed to it!
Great work and I'm glad to see coverage of the Persian language!
Upon inspecting the training dataset, I noticed almost all of the text is in "formal" Persian, which is fine if users want to talk to the model to write letters, website articles, etc., but what about "informal" Persian? Do you think the model could respond to users casual questions phrased like
به نظرت کدومو بخرم بهتره: این گوشیه که اندازه ی صفحه نمایشش ۶ اینچه یا اون یکی که یه کوچولو کوچیک تره ولی ۲ تومن ارزون تره؟
3 comments
[ 4.8 ms ] story [ 25.5 ms ] threadUpon inspecting the training dataset, I noticed almost all of the text is in "formal" Persian, which is fine if users want to talk to the model to write letters, website articles, etc., but what about "informal" Persian? Do you think the model could respond to users casual questions phrased like
به نظرت کدومو بخرم بهتره: این گوشیه که اندازه ی صفحه نمایشش ۶ اینچه یا اون یکی که یه کوچولو کوچیک تره ولی ۲ تومن ارزون تره؟