Top all-time · #6495
Jack Cook
Personal blog · jackcook.comaiprogramming
Jack Cook writes personal-site, a personal blog at jackcook.com about ai and programming. Since 2021, 12 different Hacker News readers have linked it 16 times, most often “A look at Apple’s new Transformer-powered predictive text model”. It is still writing: the last post was on 2024-11-09, and you can follow it by feed.
Too few sampled comments (10) to say how often HN recommends it.?
Why HN links it
HN readers link Jack Cook's posts on Apple's iPhone predictive text, which reportedly uses a small GPT-2-style model, and on how Mamba state space models work. Commenters mostly cite them as sources.
Summarised by Claude from a sample of 10 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
Hello! I’m Jack, and I’m currently studying neuroscience at the University of Oxford on a Rhodes scholarship. Previously, I was working on generative AI at The New York Times R&D while I studied computer science at MIT. I’ll return to MIT to start my PhD in computer science in fall 2025.
Most-linked posts?
- A look at Apple’s new Transformer-powered predictive text model7 people · 7 mentions · first linked 2023-09-08
- Mamba: The Easy Way6 people · 6 mentions · first linked 2024-02-23
What Hacker News says?
They have a GPT2 like model that's deployed right now for autocomplete https://jackcook.com/2023/09/08/predictive-text.html
Comment linking /2023/09/08/predictive-text.html
on “Gemma 3 270M: Compact model for hyper-efficient AI”Submitted When machine learning tells the wrong story as a story · 305 points
HN discussionAccording to this article[1] it has about 34 million parameters. https://jackcook.com/2023/09/08/predictive-text.html
Comment linking /2023/09/08/predictive-text.html
on “Ferret: A Multimodal Large Language Model”Submitted Mamba: The Easy Way as a story · 279 points
HN discussionThe auto correct is already backed by a smallish LLM, FYI. https://jackcook.com/2023/09/08/predictive-text.html
Comment linking /2023/09/08/predictive-text.html
on “Ferret: A Multimodal Large Language Model”Submitted Apple’s new predictive text feature appears to be powered by GPT-2 as a story · 5 points
HN discussionhttps://thegradient.pub/mamba-explained/ https://jackcook.com/2024/02/23/mamba.html https://www.kolaayonrinde.com/blog/2024/02/11/mamba.html
Comment linking /2024/02/23/mamba.html
on “Codestral Mamba”Submitted Verification on Twitter and Facebook hurts Black candidates for NYC mayor as a story · 4 points
HN discussionThere was a recent thread on explaining Mamba https://news.ycombinator.com/item?id=39501982 ( https://www.kolaayonrinde.com/blog/2024/02/11/mamba.html ) There was another one on the same thing, probably better https://news.ycombinator.com/item?id=39482428 ( https://jackcook.com/2024/02/23/mamba.html )
Comment linking /2024/02/23/mamba.html
on “Jamba: Production-grade Mamba-based AI model”Submitted A look at Apple’s new Transformer-powered predictive text model as a story · 2 points
HN discussionApparently the iPhone predictions literally do use GPT-2, or at least a model based on it: https://jackcook.com/2023/09/08/predictive-text.html
Comment linking /2023/09/08/predictive-text.html
on “Unexpected Keyboard”Submitted A look at Apple’s new Transformer-powered predictive text model as a story · 1 points
HN discussionThere is also this: https://jackcook.com/2024/02/23/mamba.html
Comment linking /2024/02/23/mamba.html
on “Mamba Explained: The State Space Model Taking On Transformers”See also: https://jackcook.com/2024/02/23/mamba.html
Comment linking /2024/02/23/mamba.html
on “Mamba Explained: The State Space Model Taking On Transformers”The MAMBA [1] model gained some traction as a potential successor. It's basically an RNN without the non linearity applied across hidden states, which makes it logarithmic time (instead of linear time) inference with a parallelizable scan [2]. It promises much faster inference with much lower compute costs, and I think up to 7B params, performs on par with transformers. I've yet to see a 40B+ model trained. The rese…
Comment linking /2024/02/23/mamba.html
on “Ask HN: Is anybody building an alternative transformer?”Craig Federighi SVP of Software at Apple “The keyboard now leverages a transformer language model, which is state of the art for word prediction, making autocorrect more accurate than ever.”[1] "New features available with macOS Sonoma" doc states "Autocorrect fixes mistakes for you more accurately than ever before by leveraging a new transformer language model in English, French, and Spanish keyboards." [2] "Introd…
Comment linking /2023/09/08/predictive-text.html
on “OpenJDK Interim Policy on Generative AI”
