Weeknotes #345 — Bath full of bricks
andrewdoran.uk·2d
🎯Content Recommendation
Regression to the Mean
blog.engora.com·1d·
Discuss: Hacker News
🎯Arithmetic Coding Theory
Intelligent Search in Rails with Typesense
avohq.io·3d·
Discuss: Hacker News
🎯Gradual Typing
Replacing the Type Checker for the Swift Compiler
forums.swift.org·2d·
Discuss: Hacker News
💧Liquid Types
New 7-Zip high-severity vulnerabilities expose systems to remote attackers — users should update to version 25 ASAP
tomshardware.com·11h
💣ZIP Vulnerabilities
Building Actual Peer-to-peer Applications: Outside Misconceptions
hackernoon.com·3d
🌐Distributed Hash
Postgres Trip Report and Photos from PGConf NYC 2025
techcommunity.microsoft.com·2d·
Discuss: Hacker News
🧬PostgreSQL Forensics
Mastering Zod Validation Mastering Data Validation and Error Handling in TypeScript with Zod
blog.codeminer42.com·4d
✅Format Validation
AI Bubble: Circular Funding And Intense Competitive Rivalry
seekingalpha.com·13h
🏠Homelab Federation
The Dwarves Beneath the Kernel: Forging BTF for eBPF
bootlin.com·3d
📺VT100
We had a staycation with our adult children and their families. It was so much fun, we want to do it annually.
businessinsider.com·11h
⬜Minimalism
Crucial reveals fastest laptop memory ever - LPCAMM2 RAM reaches a staggering 8533 MT/s, but at just under $500 for 64GB, it won't be for everyone
techradar.com·4d
💾Persistent Memory
Cluster Paths: Navigating Interpretability in Neural Networks
arxiv.org·2d
🧠Machine Learning
TALENT: Table VQA via Augmented Language-Enhanced Natural-text Transcription
arxiv.org·2d
🎙️Whisper
Making Machines Sound Sarcastic: LLM-Enhanced and Retrieval-Guided Sarcastic Speech Synthesis
arxiv.org·2d
🎙️Whisper
Attention Sinks and Compression Valleys in LLMs are Two Sides of the Same Coin
arxiv.org·2d
🧠Neural Compression
EVALUESTEER: Measuring Reward Model Steerability Towards Values and Preference
arxiv.org·2d
👁️Perceptual Coding
A small number of samples can poison LLMs of any size
anthropic.com·2d·
Discuss: Lobsters, Hacker News, r/50501
🔍Vector Forensics
Ambitious Chinese GPU vendor debuts liquid-cooled, single-slot, dual GPU card with 48GB RAM and the ability to get four of them in a system - sadly, it's the In...
techradar.com·1d
🖥️Terminal Renaissance
I Tried 500+ New AI Tools, and Honestly, These Will Blow Your Mind
dev.to·2d·
Discuss: DEV
⚡Proof Automation