TL;DR summary of stories on the internet
The author delves into the history of emoji, exploring vintage Japanese drawing devices with built-in memo functions. The Sharp PI-4000 from 1994 surprisingly features recognizable emoji, predating the more well-known sets from SoftBank in 1997 and NTT DoCoMo in 1999. Further research uncovers emoji on the NEC PI-ET1 in 1991 and the Sharp WD-A521 word […]
Read more »
Flatcar Container Linux, a continuation of the CoreOS project, focuses on backward compatibility with a fixed set of software, relying on containers for customization. However, scenarios requiring extensions beyond the original design have emerged. Luckily, Flatcar has evolved with systemd-sysext to facilitate reliable infrastructure automation. This innovative solution allows for user-provided software integration at the […]
DKIM keys are vulnerable to the 2008 Debian OpenSSL bug discovered via scanning with the tool badkeys. A surprising number of hosts, including big names like Cisco, Oracle, and GitHub, were found to be vulnerable, allowing forged DKIM signatures. Most vulnerable keys belonged to a company called Cakemail. Attempts to notify them failed. Many organizations, […]
Retrieval-Augmented Generation with Postgres, pgvector, ollama, Llama3, and Go allows the creation of an AI assistant that can answer questions based on existing knowledge bases using PostgreSQL, pgvector, ollama, and less than 200 lines of Go code. By generating embeddings of documents from a story such as Sherlock Holmes using Llama3, storing them in a […]
The author starts off in a bit of a mood and needs a good martini and Laplace approximation to fix it but is out of martinis. The post, though started in February 2023, got delayed due to visas and jobs distraction. It delves into the Laplace approximation, a simple Gaussian distribution that approximates target distribution […]
es6_maps simplifies Elixir map creation by introducing ES6-like shorthand syntax, reducing repetitive code patterns. This feature, familiar to JavaScript and Rust developers, enhances language ergonomics without runtime overhead. Installation involves adding es6_maps to dependencies and compilers in mix.exs. Usage examples demonstrate creating, destructuring, and updating maps and structs. Additionally, a formatting plugin converts existing code […]
In this study submitted on 8 May 2024, the authors address the problem of “glitch tokens” in language models that can cause unwanted behavior due to discrepancies between tokenizer creation and model training. They focus on identifying these tokens through a detailed analysis of Large Language Model (LLM) tokenizers, utilizing various techniques such as tokenizer […]
The Noisy Sockets CLI provides a way to configure and manage userspace WireGuard connections. It includes apps like the Noisy Sockets Shell, a safe remote shell using WireGuard for authentication and encryption, accessible via terminal or web browser. Operations like generating private keys, adding peers, starting servers, and connecting to the shell are done through […]
to the frustration we feel at seeing the first results make nonsense, and the realization that at this rate it will be many months before we arrive at something useful, it is not surprising that many people give up the whole idea. This is a pity, because if they persevered a little longer, they would […]
In this paper, the authors commit to the ICLR Code of Ethics and present a solution to artifacts in feature maps of ViT networks. They introduce new tokens (“registers”) to address high-norm tokens in low-informative areas, improving model performance and feature map quality. This solution sets a new state of the art for self-supervised visual […]