The coolest use for the Vision Pro
431 by robbiet480 | 196 comments on Hacker News.
news
Wednesday, 29 July 2026
New best story on Hacker News: Show HN: I was tired of opening 2 tabs for every HN link, so I made a userscript
Show HN: I was tired of opening 2 tabs for every HN link, so I made a userscript
408 by twalichiewicz | 114 comments on Hacker News.
HN is great for the links people share, but a big part of the value I get comes from reading the discussion around them. I realized I was always opening the article in one tab and the comments in another, constantly switching back and forth. I figured there was probably a simpler way, so I threw together this userscript to merge the two. 1. Clicking a link from Hacker News opens the article with a side panel containing the discussion. It doesn't require your credentials, is resizable, and is easy to tweak if you want to customize it. 2. If you land on an article that has previously been shared on HN, the script finds the existing discussion and adds a button in the top-right to open the panel. Feedback welcome.
408 by twalichiewicz | 114 comments on Hacker News.
HN is great for the links people share, but a big part of the value I get comes from reading the discussion around them. I realized I was always opening the article in one tab and the comments in another, constantly switching back and forth. I figured there was probably a simpler way, so I threw together this userscript to merge the two. 1. Clicking a link from Hacker News opens the article with a side panel containing the discussion. It doesn't require your credentials, is resizable, and is easy to tweak if you want to customize it. 2. If you land on an article that has previously been shared on HN, the script finds the existing discussion and adds a button in the top-right to open the panel. Feedback welcome.
New best story on Hacker News: Show HN: Open-source engine running Gemma 4 26B in 2 GB RAM on any M-series Mac
Show HN: Open-source engine running Gemma 4 26B in 2 GB RAM on any M-series Mac
416 by gitpusher42 | 136 comments on Hacker News.
Hi HN, I built a specialized inference engine for running 4-bit Gemma 4 26B-A4B-IT on any M-series Mac using about 2 GB of RAM. It is called TurboFieldfare and is written in Swift and Metal. I have always adored on-device AI. It feels like magic that you can run a powerful NN on your Mac or iPhone. So I wanted to push the limits a bit and run a model whose weights don’t fit in memory. The model’s 4-bit quantized weights occupy roughly 14 GB, which makes running it with conventional inference tools almost impossible on an 8 GB or even 16 GB Mac once the OS, applications, and KV cache are included. The trick is to keep the shared part of the model and the KV cache in RAM, then stream only the routed experts needed for each token from SSD. An SSD is way slower than RAM, so the runtime uses a small expert cache and bounded parallel `pread`. While those reads are in flight, the GPU runs the shared part of the layer. I ran more than 100 experiments. Most didn’t work. A few got me here. The experiments are described in the GitHub repo. It currently generates 5–6 tok/s on an 8 GB M2 MacBook Air and 31–35 tok/s on an M5 MacBook Pro. I also added an experimental OpenAI-compatible local server. It supports streaming and tool calls, and reuses one prompt prefix from the KV cache. Try it! The Mac app is easy to install. On the first run, it will download 15 GB of weights from Hugging Face. The model is surprisingly capable. I would love any kind of feedback!
416 by gitpusher42 | 136 comments on Hacker News.
Hi HN, I built a specialized inference engine for running 4-bit Gemma 4 26B-A4B-IT on any M-series Mac using about 2 GB of RAM. It is called TurboFieldfare and is written in Swift and Metal. I have always adored on-device AI. It feels like magic that you can run a powerful NN on your Mac or iPhone. So I wanted to push the limits a bit and run a model whose weights don’t fit in memory. The model’s 4-bit quantized weights occupy roughly 14 GB, which makes running it with conventional inference tools almost impossible on an 8 GB or even 16 GB Mac once the OS, applications, and KV cache are included. The trick is to keep the shared part of the model and the KV cache in RAM, then stream only the routed experts needed for each token from SSD. An SSD is way slower than RAM, so the runtime uses a small expert cache and bounded parallel `pread`. While those reads are in flight, the GPU runs the shared part of the layer. I ran more than 100 experiments. Most didn’t work. A few got me here. The experiments are described in the GitHub repo. It currently generates 5–6 tok/s on an 8 GB M2 MacBook Air and 31–35 tok/s on an M5 MacBook Pro. I also added an experimental OpenAI-compatible local server. It supports streaming and tool calls, and reuses one prompt prefix from the KV cache. Try it! The Mac app is easy to install. On the first run, it will download 15 GB of weights from Hugging Face. The model is surprisingly capable. I would love any kind of feedback!
Tuesday, 28 July 2026
Subscribe to:
Posts (Atom)
-
U.S. Postal Service starts nationwide electric vehicle fleet, buying 9,250 EVs 444 by lxm | 336 comments on Hacker News.
-
Learn Postgres at the Playground – Postgres compiled to WASM running in browser 543 by samwillis | 144 comments on Hacker News.
-
NSA, NIST, and post-quantum crypto: my second lawsuit against the US government 486 by trulyrandom | 143 comments on Hacker News.