Track Surveillance (Flock Cameras) Tech in Local Government Meetings
Enter your email below and we’ll send you a login link. After logging in, you can set your notification preferences. Your Email Sign Me Up ✓ Check your email for a login link! ✗ Please enter a valid email address. You’re logged in! Update your notification settings to receive alerts. Update Settings
Prompt caching: 10x cheaper LLM tokens, but how?
As I write this post, cached input tokens are 10x cheaper in dollars per token than regular input tokens for both OpenAI and Anthropic’s APIs. Anthropic even claim that prompt caching can reduce latency ”by up to 85% for long prompts” and in my own testing I found that for a long enough prompt, this […]
AIsbom – open-source CLI to detect ”Pickle Bombs” in PyTorch models
AIsbom is a specialized security and compliance scanner for Machine Learning artifacts. Unlike generic SBOM tools that only parse requirements.txt, AIsbom performs Deep Binary Introspection on model files (.pt, .pkl, .safetensors) to detect malware risks and legal license violations hidden inside the serialized weights. ⚡ Quick Start 1. Installation Install directly from PyPI. No cloning […]
Gh-actions-lockfile: generate and verify lockfiles for GitHub Actions
No Native Lockfile GitHub Actions has no built-in mechanism to lock dependency versions. Mutable Tags Version tags like @v4 can be silently retagged to point to different code. Hidden Dependencies Composite actions pull in transitive dependencies you can’t see or audit.
Writing a blatant Telegram clone using Qt, QML and Rust. And C++
This was a fun project for a couple of days, but I will probably shelve it for now so I can continue what I was already working on before. Read on to follow along with my journey. Spoilers: I didn’t get very far. Get ready for an Opinion Dump of an intro: I have fond […]
Biscuit is a specialized PostgreSQL index for fast pattern matching LIKE queries
Biscuit is a specialized PostgreSQL index access method (IAM) designed for blazing-fast pattern matching on LIKE and ILIKE queries, with native support for multi-column searches. It eliminates the recheck overhead of trigram indexes while delivering significant performance improvements on wildcard-heavy queries. It stands for Bitmap Indexed Searching with Comprehensive Union and Intersection Techniques. Version 2.1.4 […]
40 percent of fMRI signals do not correspond to actual brain activity
For almost three decades, functional magnetic resonance imaging (fMRI) has been one of the main tools in brain research. Yet a new study published in the renowned journal Nature Neuroscience fundamentally challenges the way fMRI data have so far been interpreted with regard to neuronal activity. According to the findings, there is no generally valid […]
This is not the future
I thought about this when reading a mastodon post which commented on a news where a project adopted a ”use Generative AI but disclose it” policy, because it is ”the future” and ”people are going to use it anyway”. I find the ”this is the future, like it or not” framing particularly disgusting, and it […]
I don’t think Lindley’s paradox supports p-circling
As usual I’d like to preface all this that I write these blogposts as attempts to make sense of a subject for my own sake. I am not an expert here and it is likely I am confused about some details. On the other hand, I think “confused” discourse can also be productive to read […]
Rust GCC back end: Why and how
Whenever you compile using Rust, the compiler goes through different passes and in the end, generated binary code for the target processor. By default, it uses LLVM as backend to generate the binary code, but more backends exist like cranelift and GCC. This post is about how it’s possible for one compiler to use different […]