jbaudanza

Yeah certainly, the SVG animation should be playable at any speed.

If you're learning to write, I'd recommend looking up the official stroke order. You can just google "Hangul stroke order" and find plenty of guides. And then, just practice writing in a notebook a lot.

Thanks! I added this to the training data.

That's an interesting idea. Velocity is something that can be derived from the x, y inputs that the model already has. So, if the velocity contained useful information, I would expect the model to be able to learn this during the training phase. The pressure, on the other hand, was new information that the model didn't have before.

The text embedding structure is very specific to Hangul. Someone that knows more about Chinese than me could probably design a suitable text embedding structure that would work well. I’m happy to collaborate if you want to fork my project and try.

Thank you for the feedback. I’ll incorporate this into the new training run.

I’m sorry about your father. Reproducing his handwriting would be a beautiful project. What kind of samples do you have from him?

Oh thank you, good catch. "쀍" is definitely not a common character in my training set. I'll add it to the next run!

Ah yes, I'm well aware! My company (hiworker.com) makes AI products for ship building and construction companies that have a large foreign workforce. We have some customers in Ulsan. It's a beautiful place, I love visiting.

Thank you, you're right! It definitely seems to have a problem when the end of the string ends with 다. It seems to work if you end the sentence with "다." instead of "다". I'll look into this more. There should be plenty of "다" data in the training set, maybe there's something deeper going on.

PS: I saw your feedback report. 화이팅! :)

I think this would work well for Chinese, but you’d probably need to modify the text embedding system. Let me know if you’d like to try, happy to collaborate.

> Nice! Did you find any weird/funny case while training the model? [I remember autotranlation/readers inventing stuff a few years ago.]

Mostly, it was me trying things the naive way, running into problems, and then solving them, usually by copying what Alex Graves did in 2013 (but not always!)

For example, to make things simple I first built the model to simply predict the next x, y direction of the pen. This worked for simple pen strokes, but I noticed the model had a difficult time turning corners.

To fix this, I changed the model to instead predict an Mixture Density Network (Same as Graves). This is explained better in his paper, but essentially, instead of predicting one x, y direction, you predict 10, and then randomly sample one of those predictions. Also instead of predicting scalar x, y values, you predict a parameters for a gaussian distribution, and then sample from that.

It always amazes me how much randomness is involved in intelligence.

> If you get someone to make a few samples and photos, it would be very nice.

I'm in Seoul, if you know anyone who would be interested in collaborating, please send me a note! [email protected]

I had the same problem. I think a lot of people reading this will have weak hamstrings. I think it's a natural consequence of desk work.

Adding in some Romanian Deadlifts really forced me to focus on my hamstrings when doing lifts. It helped me breakthrough some plateaus with the conventional deadlift.

3) Whisper really needs to be paired with Silero VAD, otherwise the hallucination problem makes it almost unusable.

I have a question about Mistral OCR. If I give the model a PDF that is 90% text, is it actually performing OCR on an image representation of the text? Or is it smart enough to extract the text directly and only use OCR on images?

I really like this concept, and am working on a similar product for my own self-study.

I would like to see more language support. For me, I'd like to study Korean

The UX around card navigation needs some work. It took me a long time to figure out that after I chose a smiley face, that I was moving to a new card. This was especially true if the next card is similar to a previous card.

Interesting choice of a 24kHz sample rate for PCM audio. I wonder if the model was trained on 24kHz audio, rather than the usual 8/16kHz for ML models.

Righty tighty, lefty loosey

This looks interesting.

Most devices have noise suppression built in. Do you use RNNoise in tandem with the hardware noise suppression, or do you disable it?

Also, are there any plans to implement this with react-native on mobile? And if so, how would you implement this, since audio worklets and WebAudio aren't available.

Yes! I’ve had the same experience with spaced-repetition. I’ve used Memrise.com to learn a large set of Korean vocab words.

I’ve found it’s improved my general recall ability, not just for things I’ve scheduled into a spaced-repetition program.

It’s like the program trains my brain on how to recall things, and then I can use that skill on anything.

I’ve also found that if I slack off my training schedule, my general recall ability gets worse too.

Also, alcohol for sure impacts my memory. I find even one drink will make it harder to recall things

I know this is theoretically true, but I've never found this to be the case in practice.

Coinbase, in fact, has made it a rule that all things must be put in useMemo. https://attardi.org/why-we-memo-all-the-things/

Facebook is working on an experimental compiler called React Forget that will transparently memoize everything for you.

https://www.reddit.com/r/reactjs/comments/rcn5ks/react_forge...

I've been using react native heavily for the past 18 months.

Because there's no JIT, Javascript is slow on mobile, especially on Android. I spend a lot of my creative energy trying to optimize the render function. I often times wonder if I should have chosen Flutter instead.

Each framework has its bullshit you don't discover until it's too late. So, I'm sure if I had chosen Flutter I'd be complaining about something else.. but boy am I tired of stuffing things into useMemo, useCallback, and stressing about the identity of things in dependency arrays.

There's some exciting stuff coming up this year with React/React Native: Hermes, React Native "New architecture", react concurrency mode (although I still don't 100% understand what this is). I hope one of these things improves the current situation.

Yeah. And with the soft-launch system, they can be hidden away in any house or enclosed space.

From the useMemo docs: https://reactjs.org/docs/hooks-reference.html#usememo

You may rely on useMemo as a performance optimization, not as a semantic guarantee. In the future, React may choose to “forget” some previously memoized values and recalculate them on next render, e.g. to free memory for offscreen components. Write your code so that it still works without useMemo — and then add it to optimize performance.

I think about this section a lot. If they actually changed useMemo to sometimes "forget" values, it would break so many useEffect dependency arrays (including my own).

The situation is much more severe with react-native, which doesn’t have a JIT compiler available.