Show HN: FastGraphRAG – Better RAG using good old PageRank

show-hn:-fastgraphrag-–-better-rag-using-good-old-pagerank

Streamlined and promptable Fast GraphRAG framework designed for interpretable, high-precision, agent-driven retrieval workflows. Looking for a Managed Service? » Note Using The Wizard of Oz, fast-graphrag costs $0.08 vs. graphrag $0.48 — a 6x costs saving that further improves with data size and number of insertions. Stay tuned for the official benchmarks, and join us […]

MailCatcher runs a super simple SMTP server

mailcatcher-runs-a-super-simple-smtp-server

MailCatcher Catches mail and serves it through a dream. MailCatcher runs a super simple SMTP server which catches any message sent to it to display in a web interface. Run mailcatcher, set your favourite app to deliver to smtp://127.0.0.1:1025 instead of your default SMTP server, then check out http://127.0.0.1:1080 to see the mail that’s arrived […]

Launch HN: Regatta Storage (YC F24) – Turn S3 into a local-like, POSIX cloud FS

Hey HN, I’m Hunter the founder of Regatta Storage (https://regattastorage.com). Regatta Storage is a new cloud file system that provides unlimited pay-as-you-go capacity, local-like performance, and automatic synchronization to S3-compatible storage. For example, you can use Regatta to instantly access massive data sets in S3 with Spark, Pytorch, or pandas without paying for large, local […]

Bayesian Neural Networks

Bayesian inference allows us to learn a probability distribution over possible neural networks. We can approximately solve inference with a simple modification to standard neural network tools. The resulting algorithm mitigates overfitting, enables learning from small datasets, and tells us how uncertain our predictions are. Train What’s wrong with neural networks? You may have heard […]

A joke in approximating numbers raised to irrational powers

a-joke-in-approximating-numbers-raised-to-irrational-powers

How it started The story behind this post surprisingly starts on Facebook, on a group dedicated to mathematics. A math teacher asked audience to find the (lfloor 3^{sqrt{3}}rfloor) without using a calculator. Of course, it was a matter of proving (lfloor 3^{sqrt{3}}rfloor=6), or that (lfloor 3^{sqrt{3}}rfloor=7) (my intuition said it’s 7). After a few back […]

Bit-twiddling optimizations in Zed’s Rope

bit-twiddling-optimizations-in-zed’s-rope

A couple of weeks ago I came across one of Antonio’s PRs, titled “Speed up point translation in the Rope” — now who doesn’t stop to take a closer look at a PR with that title? The description already lives up to the title. It contains benchmark results telling me that a method on our […]

Qwen2.5 Turbo extends context length to 1M tokens

qwen2.5-turbo-extends-context-length-to-1m-tokens

API Documentation (Chinese) HuggingFace Demo ModelScope Demo After the release of Qwen2.5, we heard the community’s demand for processing longer contexts. In recent months, we have made many optimizations for the model capabilities and inference performance of extremely long context. Today, we are proud to introduce the new Qwen2.5-Turbo version, which features: Longer Context Support: […]

Scientists Clone Two Black-Footed Ferrets from Frozen Tissues

scientists-clone-two-black-footed-ferrets-from-frozen-tissues

Antonia, one of the new black-footed ferret clones. The first black-footed ferret clone was born in 2020. The two new ferrets are the second and third successful clones. Roshan Patel / Smithsonian Conservation Biology Institute Researchers have cloned two black-footed ferrets from preserved tissue samples in an effort to conserve the creatures, the United States […]

A BBC navigation bar component broke depending on the external monitor

a-bbc-navigation-bar-component-broke-depending-on-the-external-monitor

Recently, my team and I fixed an absolutely bizarre bug that only one person in the team could reproduce. And to make it even weirder, she was only able to reproduce the issue when using her work laptop at home; it worked fine with the same laptop in the office. So what was the bug? […]

The Skyline algorithm for packing 2D rectangles

the-skyline-algorithm-for-packing-2d-rectangles

Publication: 2024-11-17 Packing 2D rectangles into bigger fixed-size rectangles is a need for most multimedia projects. In GPU programming, changing textures (“binding”) is expensive. Thus, when rendering text, you don’t want to have one texture per glyph. Instead, it is desirable to pack the glyphs in a single texture, called the “atlas”. In 2D-based games, […]