Apr 11, 2024

• Anyone got a contact at OpenAI. They have a spider problem

• OpenAI's GPTBot fetched over 3 million pages from a single content farm, indicating a potential oversight in its web crawling algorithm.

• Aider: AI pair programming in your terminal

• Aider is a command line tool that enables pair programming with GPT-3.5/GPT-4, directly editing and committing changes to code in local git repositories, supporting a wide range of programming languages and offering a unique capability to work with large codebases.

• Show HN: Sonauto – A more controllable AI music creator

• Sonauto-platform appears to be inaccessible without JavaScript enabled, suggesting a reliance on dynamic content for user interaction.

• Humane AI Pin review: not even close

• The Humane AI Pin, a wearable device designed to replace smartphones by using voice commands and AI, fails to deliver on its promises due to numerous functional shortcomings.

• Embed Crystal code directly in Ruby

• crystalruby is a Ruby gem enabling Ruby code to inline and execute Crystal code, significantly boosting performance for certain operations, as demonstrated by benchmark comparisons between Ruby and Crystal implementations of the Fibonacci sequence.

• Transformers.js – Run Transformers directly in the browser

• Transformers.js brings the power of Hugging Face's transformers library to the web, enabling state-of-the-art machine learning directly in the browser without a server, covering tasks in natural language processing, computer vision, audio, and multimodal.

• Udio: Generate music in your favorite styles with a text prompt

• Udio is an innovative app for music creation and sharing, leveraging text-prompting to enable users to generate music in their preferred styles effortlessly.

• Implementation of Google's Griffin Architecture – RNN LLM

• RecurrentGemma, developed by Google DeepMind, introduces a novel language model architecture named Griffin, which enhances long sequence generation through a blend of local attention and linear recurrences, detailed in the Griffin paper.

• McCarthy's Ambiguous Operator (2005)

• John McCarthy's amb operator, introduced in 1961, elegantly solves computational problems by selecting values that prevent future errors, a concept that predates but anticipates modern backtracking algorithms.

• Lens sort, a masked pixel sort glitch effect

• Lenssort introduces a novel approach to image manipulation by combining facial recognition with pixelsorting, creating unique, glitched images reminiscent of Snapchat lenses, utilizing the pixelsort project.

• Nimble: A new columnar file format by Meta [video]

• Meta introduces Nimble, a new columnar file format designed to optimize machine learning workloads by making data reading up to 2x faster, significantly reducing the time spent on reading data before processing.

• Quantum Algorithms for Lattice Problems

• Yilei Chen introduces a quantum algorithm for the Learning With Errors (LWE) problem, leveraging Gaussian functions with complex variances and windowed quantum Fourier transform techniques, aiming to solve LWE in polynomial time.

• Show HN: Next-token prediction in JavaScript – build fast LLMs from scratch

• The next-token-prediction library enables the creation of language models for high-quality predictions such as next word or phrase, offering an alternative to decoder-only models like GPT and Mistral.

• Building an AI Coach to Tame My Monkey Mind

• The author combats chronic imposter syndrome and the hedonic treadmill by using ChatGPT as a therapeutic tool, finding it offers insightful external perspective and emotional clarity.

• Storm: LLM system that researches a topic and generates full-length wiki article

• STORM, a Large Language Model (LLM) system, is designed to automate the creation of Wikipedia-like articles by conducting Internet-based research to generate outlines and full articles with citations, as detailed in their NAACL 2024 paper.