Gregory's Server
Joining Overture Places with county health restaurant grades using DuckDB, Ollama, and more. https://www.dbreunig.com/2024/09/27/conflating-overture-points-of-interests-with-duckdb-ollama-and-more.html
|
I really enjoyed this interview with John Oliver about the process involved in producing Last Week Tonight https://www.youtube.com/watch?v=Q9kNMJ8SguQ "Those jokes have no structural integrity if the facts underneath them are bullshit"
Show previous comments
 That was a great conversation. Glad Oliver is still loving making his show, and hasn't fallen into the exhaustion that Jon Stewart fell into w/ Daily Show. @simon Having watched a couple of episodes of LWT off the back of that interview... Does anyone put together supercuts without the jokes? Or a #sponsorblock feed to skip them? Eg, the piece on the history of Boeing was powerful. It doesn't need the asides about the CEO's marriage, or some guy in a corporate video being too close to the camera. This is the time of the year when you can contribute to the rapid development of the Django framework and get a professional IDE for your Django development. Until October 6, buy PyCharm at a 30% discount, and an amount equal to your purchase will be donated to the Django Software Foundation. Link: https://jb.gg/e742yn I put together a whole bunch of notes on the new "Audio Overview" feature of Google's NotebookLM, where you can dump a bunch of source documents in and get a ~10m long utterly weird (and surprisingly convincing) artificial podcast episode generated where two AI characters perform a "deep dive" conversation about those sources https://simonwillison.net/2024/Sep/29/notebooklm-audio-overview/
Show previous comments
@simon man this is so believable; I tried it with my own article and they even pronounced my name wrong like every self-respecting podcaster ;) @simon The host sounds like it was trained on Leo LaPorte TWIT, not sure who the female voice is. Here's a recipe for running the Qwen2-VL vision LLM models on Apple Silicon using Python and the mlx-vlm library, via a uv shell one-liner Full details on my blog: https://simonwillison.net/2024/Sep/29/mlx-vlm/ - and here's the full output from that example prompt https://gist.github.com/simonw/9e02d425cacb902260ec1307e0671e17 A friend of mine suggested that we stop using the sparkle ✨ emoji as the symbol for AI and replace it with a magic eight ball 🎱 as it more realistically conveys how the technology works. I've been thinking about that a lot lately. 😂   The key thing here is not that I built a lot of features, but that they all come with imo solid documentation, examples, tests, tooling. And it’s absolutely minimal software, built to do exactly what I want with barely an external dependency. This is way better than what I used to do for my open source, and I did it without eating up my weekend. It’s an incredible improvement to my creativity and love of software. I get to have a life and build the things I want to build that inspire me. 3/ OpenFreeMap looks really neat - I poked around a bit and shared some notes here https://simonwillison.net/2024/Sep/28/openfreemap/ 🎉 Announcing DjangoTV. I'm ~10 minutes away from boarding my flight which means, I technically still launched this project while at #DjangoCon #Django #Python https://micro.webology.dev/2024/09/27/announcing-djangotv.html
Show previous comments
@webology What an awesome idea, I would love to see this done for PyCons too (there are just so any it's hard to keep track of them all when they upload VODs!) @webology when you're rested and have some time, would love to know how to help out while I have some free time. Themes from DjangoCon US 2024 - my notes from #djangoconUS  My first update to Niche Museums in a while... the Vincent and Ethel Simonetti Historic Tuba Collection in Durham, North Carolina! I went with a group of five during this year's DjangoCon US 2024 - it was utterly delightful https://www.niche-museums.com/112  ... just spotted they have a photo of us on their Instagram! https://www.instagram.com/p/DAZRx-QSVZK/ @simon did you ever make it to this one? I must admit that after years in and around Evanston, I never have. Sent out he latest issue of my newsletter - everything from my blog for the last two weeks turns out to be quite a lot of stuff this time round https://simonw.substack.com/p/llama-32-and-plugins-for-django Llama 3.2 is out, and it's a much more substantial release than the 3.1 to 3.2 version bump might indicate Four new models, including Meta's first two vision models (11B and 90B) and two new text-only small models (1B and 3B) My notes so far: https://simonwillison.net/2024/Sep/25/llama-32/ You can try out the vision models on https://lmarena.ai/ by navigating to "Direct Chat" and then selecting those models from the dropdown menu here's some home-brewed software for ya; personal image search using @simon's llm-clip in a little wee webpage flask thingy. And you could run it on your own customized version of CLIP too, if you really dug into things. https://github.com/shawngraham/personal-image-search-engine/ Gergely Orosz has a new podcast, and I was the guest for the first episode: "AI tools for software engineers, but without the hype" https://newsletter.pragmaticengineer.com/p/ai-tools-for-software-engineers-simon-willison We covered a bunch of ground, but my favorite topic was an exploration of the (very legitimate) reasons that many engineers are resistant to taking advantage of AI-assisted programming tools. Here's a detailed write-up of my new project: DJP: A plugin system for Django - which I introduced yesterday at #djangoconUS https://simonwillison.net/2024/Sep/25/djp-a-plugin-system-for-django/ New in DJP 0.2: @carlton suggested it would be useful if plugins could say “this middleware needs to be inserted directly before/after this other middleware” - that’s now supported: djp.Position("my_plugin.middleware.MyPluginMiddleware", before="django.middleware.common.CommonMiddleware") I'm really impressed by nanodjango (by @radiac, presented in a lightning talk at #djangoconUS just now) - it's the latest in a long line of attempts to have Django work for Flask-style single file apps but it's got WAY more features than previous attempts I've seen, like model and admin support and even async My notes here: https://simonwillison.net/2024/Sep/24/nanodjango/ nanodjango docs and tutorial: https://nanodjango.readthedocs.io/ Pretty cool that the Django Software Foundation is about 1/3 funded by individual donors (from a @jacob talk at DjangoCon US) Jacob’s talk is exploring what the DSF could do if revenue quadrupled from $250,000/year to $1,000,000/year It's the ten year anniversary of XKCD 1425, one of the all-time greats and one that resonates slightly differently today. Wrote up some notes on that here: https://simonwillison.net/2024/Sep/24/xkcd-1425-turns-ten-years-old-today/ @simon I do find it incredible that I don't find it incredible that I can take a crappy photo of a bird and very quickly get the bird identified (using pre-AI-hype google image search for example) New TIL: How streaming LLM APIs work I put together some notes after poking around with the OpenAI, Anthropic and Google Gemini streaming APIs
Show previous comments
 @simon Nice. Little note: on a recent curl, you can POST JSON with `curl --json <string>, saving the header setting. Update: `--no-buffer` always `fflush()`es the output in curl. So it might still be beneficiary. Updated my TIL with example JavaScript code for streaming events from a fetch() POST API (using an async iterator function) https://til.simonwillison.net/llms/streaming-llm-apis#user-content-bonus--2-processing-streaming-events-in-javascript-with-fetch |
