709 karma · joined December 5, 2014
From what I understood, tree-based models can usually outperform transformers when given sufficient parameter tuning. However, models like TimeGPT offer decent performance without extensive tuning, making them an attractive option for quicker implementations.
If I'm reading this right (and I did double check this with ChatGPT [0]) if you have an app with two million unique installs annually, you owe Apple 500 000 euros. That seems to include free apps as well.
[0]: https://chat.openai.com/share/905c5c45-657b-477c-a746-0468dd...
Here are tweets from famous HN users Simon Willison[0] and Patrick Mckenzie[1] endorsing the viewpoint as well.
That said, at some point it starts to make sense to read open source repositories to pick up useful patterns.
I've recently been playing around with Midjourney for the cover images and that's been a excercise as well.
[0]: https://www.semafor.com/article/01/27/2023/openai-has-hired-...
* Startups for the rest of us
* Art of product
* Out of Beta
* Indiehackers
* Software Social
* Build your SaaS
The guy doing Startups for the rest of us, Rob Walling, also runs a bootstrapper conference called Microcof that I recommend checking out.
> Q: Is there a JIT compiler?
> A: No. We’re still exploring other optimizations.
[0]: https://docs.python.org/3.11/whatsnew/3.11.html#faster-cpyth...
[0]: https://yew.rs/
Apologies for the self-promotion, but in case any tech minded finns are reading this, I write a daily news letter about tech in finnish called Transistori [3].
[1]: https://www.bloomberg.com/opinion/authors/ARbTQlRLRjE/matthe...