saurkt

6 ms·

1.
by saurkt 5y ago
Turing team member here. You can also try the demo at the following link which operates at pixel level for images (no meta-data which is how similar systems like search engines work). Another cool thing is that the model does inherent OCR w
2.
by saurkt 5y ago
(Team member of this project) Just a clarification, both Microsoft and Nvidia have ownership of this model. Here is the Microsoft version of same announcement. https://www.microsoft.com/en-us/research/blog/usi
3.
by saurkt 6y ago
A similar feature of soft Ctrl+F is available in the web version of Microsoft Word. https://mspoweruser.com/find-feature-in-microsoft-word-whole... Disclaimer: Microsoft employee
4.
by saurkt 7y ago
Yes, these things keep us up at night as well :-).
5.
by saurkt 7y ago
If you want access, please send an email to [turing_ AT _microsoft _DOT_ com]. Remove underscores and spaces.
6.
by saurkt 7y ago
We don't have an exact date, but, we plan to share more details in a later submission. If you want access, please send an email to [turing_ AT _microsoft _DOT_ com]. Remove underscores and spaces.
7.
by saurkt 7y ago
Thanks for your kind words. Yes, we would like to next train a language representation model. And our hunch is that probably something which is a mixture of language representation and language generation would be able to get the best of bo
8.
by saurkt 7y ago
(Similar to the response for another question.) BERT is a language representation model while Turing-NLG is a language generation model (similar to GPT). They are not directly comparable (they can potentially be massaged to mimic the other,
9.
by saurkt 7y ago
SQUAD and GLUE are tasks for language representation models -- aka BERT-like. This is a language generation model -- GPT-like. Hence, SQUAD/GLUE test sets are not really applicable. We are reporting on the wikitext and lambada sets tha
10.
by saurkt 7y ago
Not yet. We will try to run against those benchmarks soon.
11.
by saurkt 7y ago
One of the team members from Project Turing. Happy to answer any questions.