Pattarawat Chormai
Pattarawat Chormai
The simplest thing I could think of is to use the training code for the transliteration mode that we have with inverse source and target data. It probably gives us...
can we be more specific here? It would be good to list requirements for this new IPA stuff.
> Well, the problem is predictability - can we find what the least surprising keys would be for "Feeling lucky?" and "Report"? imho, it is probably quite challenging to find...
> I wouldn't mark that in the letters of the footer buttons, since it gets a bit harsh on the eyes. It may be better to pop-up a little modal...
Regarding `similarity_func`, I'm not sure to what extent the current API allows, but we could show how to compose a number of `similarity_func`'s together. For example, in randomization tests, I...
I have another suggestion. Right now, it would be good if each section also includes evaluation metrics. Maybe we can start with the work tokenisation section?
IMHO, it would be good if we could have this section in Thai first. https://github.com/codeforthailand/election-live#how-we-handled-100000-active-users-on-a-single-10mo-server We can announce the time machine feature and link to this section, saying why we...
Thanks! @rapee @kristw could you please review this PR?
Also not working properly on an iPad simulator 