Another @DeepDishSwift podcast episode is out! 🍕
This time, @airkai and I chat with @allisonmcentire about her path from teaching academic writing, through developing websites, to working as an iOS engineer 👩💻
https://t.co/PeCfQnUrBM
#DeepDishSwift
🍕 Here we go again…
Deep Dish Swift 2024 is ready to be served!
📅 May 5th to May 7th in Chicago, IL
- 2 days of Swift and iOS talks
- 1/2 of indie dev talks
- Live recording of @LaunchedFM
👇 All we needs is you...
https://t.co/Tc0Myh2qLe
If you'd like to learn how to build SwiftUI components, check out this hands-on interactive tutorial I built for you!
You will learn:
- Different techniques for composing views
- How to make your views customisable
- Making your views interactive
- And how to style your views
@RobRWAPP @natpanferova @thea_lena_36 @AlexKudelka Allison talked about how to build iOS apps that allow users to manipulate and enhance images with Stable Diffusion and CoreML. Yes another awesome topic 👏
@IOSDEVUK @thea_lena_36 After the lunch break, it’s @allisonmcentire with “Stable Diffusion, CoreML, and the future of AI for iOS” - including an interactive demo!
How to code efficiently with ChatGPT?
Pro edition 🤌🏻
1. Best practices
I use phrase like:
“Apply best practices to this code, make it maintainable and increase readability”. + paste my code
🧐Why: easy way to learn about improving code and clean up initial gpt response.
2. Specify tools
I give common info in the 1st message:
eg: “I am building nextjs app using tailwind
I want to add support for A”.
🧐Why: to save msg limit with gpt4, as no need to guess language/project structure later.
3. Break down long code
If code is too long, I would say that I will send it in 2 chunks
eg: “I will first say to you first half, don’t reply, I will send the rest in the next message”
🧐Why: there is a text limit for message, this way you can upload long code
Anything else? 😎
📢 Announcing Swift Transformers 📢
A package to run language models in native apps, on-device.
It's part of a growing set of tools to help Swift developers work with Core ML models. Read on for details, or check our post: https://t.co/W5qp33klZE
4 new tools released today!
Stable Diffusion XL running on Mac using Core ML and advanced quantization techniques!
Open-sourced today:
- SDXL support in Apple's conversion & inference package.
- SDXL support in our demo app.
- New mixed-bit quantization for Core ML.
- Core ML models ready for use.
Phew 😅
Stable Diffusion on iPhone is much faster now!
Same model, same phone (iPhone 13 Pro), same settings 🤯
The trick?
* 6-bit quantization in Core ML. Announced last week in WWDC
* Additional optimizations to the attention blocks
Check our post for details https://t.co/87Pvlrocrb
@kvlly Straight hire for me. The next question I would ask them is to explain the code line by line using diagrams and drawings to test comprehension. If they understood it themselves, that’s great.
And to make this a broader subject (which will surely get lost because everyone will just see the first tweet) if you’re evaluating a candidate on a single problem that can be solved by AI in 5 minutes you’re capturing only a tiny subset of the candidate.