🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

If in the latest two years I developed enough clues and experience about LLMs, then you can assume Claude 4 (even Opus) is not as good as Gemini 2.5 PRO. I tested it for a few hours now, and that's my conclusion.

If you read about O3 finding a SMB bug in the Linux Kernel, I did a few tests and I what I suspected looks true: Gemini 2.5 PRO can more easily identify the vulnerability. My success rate is so high that running the following prompt a few times is enough: gist.github.com/antirez/8b76cd…

It's a long story but trust me: Airbnb is a pretty distinctional company in the way it handles the relationship with the hosts. Also one of the poorest software experience across the industry.

I understand that in Germany / Austria those waters are very popular. I tried to understand what's the point but then realized it and I'm providing with an improved version here. From a discussion with my brother Dario Sanfilippo, living in Vienna.

I'm seeing that many of the best use cases Redis users are applying vector sets to, are not related to learned embeddings. For instance, I have a friend that is doing a huge scale molecular search application, and they encode the molecule representation directly.

Every time you use BambuStudio, PrusaSlicer, OrcaSlicer, and the other evolutions of Slic3r, remember that all started with the italian Alessandro Ranellucci doing all the initial work needed: teamdigitale.governo.it/it/people/ales…

antirez bsky social Regular reminder for the corporate folks: if your fancy slicer is built on open source, maybe toss a coin (or some dev hours) back to the community that made it all possible 😏 Thanks Alessandro - we see you!

Human coders are still better than LLMs: antirez.com/news/153

Inspired by antirez bsky social 's C course, I have recreated the iconic ASCII c-donut using #threejs and #pmndrs postprocessing ASCII effect. YT tutorial: youtu.be/MIZbAO_rNWU #glsl #postprocessing #ascii #asciiart #donut #cdonut #c

Vector Math and Vector visualizations with Redis Vector Sets @redisinc antirez bsky social youtu.be/UdaBWTW4BjE?si…

Prompt caching is affecting LLM user experience in a severe way. A few weeks of this issue with Gemini regurgitating instantaneously a reply slightly related to my question.

antirez bsky social When GPT 4.1 was added to OpenAI's Web interface, I got all sorts of oddities that first day with it giving totally irrelevant responses (possibly based on other people's prompts). I suspect it was caching related. It def feels like a vector for potential problems.

One of the most embarrassing things in today software: I have the same folder with thousands of .crt games in my Commodore 64 Kung-Fu II card and in my MacOS M3 Max computer. The former opens the directory faster than the latter.

Are you from Malta? My uncle moved there years ago and now he is reinvented himself restoring wood doors, windows, and so forth. If you need his services please drop me a DM or an email and I'll reply with his phone number.

I'm sorry to inform you and your optimism on coding agents that our technology is based on things like the Linux kernel, or a modern browser: completely outside of today's possibilities. When you analyze AI, think at the Linux kernel.

About the recente Apple paper: many of us believe CoT is a form of sampling/search in the representations the model possess: this allows to "ground" the reply with a large context of related ideas/info. No surprise that the Tower of Hanoi is a problem where sampling doesn't help.