Different variations of diacritics
I would like to get multiple variations of diacritics for sentence
For instance with 'Shalom Olam'
ืฉืืื ืขืืื
The diacritics are 'Shlom Olam'
ืฉืึฐืืึนื ืขืึนืึธื
I tried to implement beam search but couldn't get different variations
Thanks
Seems the vocalization is for peace of world not Hello world! :)
Indeed, the current architecture does not allow retrieving multiple variations of diacritics for each word/the sentence. We are looking into training a model with a different architecture, but that is currently only in research.
I noticed some differences in the nikud from Dicta website in terms of modernity
For instance when I hit ืฉืืื ืขืืื in Dicta website it's Shalom Olam but in the model it's Shlom Olam, it's like the model nikud is a bit 'less modern' than Dicta website. That's why I asked for a way to get variations.
If you plan another one, I wish that you can include more modern nikud, and Shva Na and Atama'a! ; )
Thank you very much for the model. very appreciated.