David Chanin

The hardcoded 52 makes sense as 26x2. So that hardcoded value would need to be changed to use this for a real font, as real fonts have more than 52...

yes that makes sense. Sorry when I said removing from the generator I meant removing the fc layer and `content_logits`, as well as the `criterion_ce` and `loss_char` stuff entirely. It...

There's no Gemfile in this repo, how can I bundle install the requirements needed at the right versions to get this to run locally? Is there a Gemfile with the...

I struggled to get sesame working due to this issue, so instead trained a T5 transformer model using the open-sesame data/splits. It gets pretty comparable results in trigger ID and...

Another option is to use a lazy-loader for SQLAlchemy that is more intelligent about bulk loading. We wrote a custom loader that will inspect the DB session and bulk-load all...

Highlighting differences between strokes would require another data source since the makemeahanzi data doesn't provide an obvious way to map the strokes in 2 different characters to each other as...

`onStartUserStroke` sounds like a reasonable addition. If you want to add it, definitely feel free to open up a PR. It should be pretty straightforward to base it on the...

What you wrote should work @yxxhero, but it's awkward because `animationWriter._character` might be `null` until the character data can load successfully. I think it makes sense to add a new...

It might be that `_character` is null at first because the writer hasn't finished loading the character data yet. I'm hoping adding a proper method to get the character will...

v3.1.0 adds a `writer.getCharacterData()` method that you can use to get the contents of `writer._character` after the data has loaded. The method returns a promise which will contain this data....