# Mar 15, 2024

## Spreadsheets are all you need

- **Spreadsheets-are-all-you-need.ai** offers a **low-code introduction** to AI, specifically Large Language Models (LLMs), through a real implementation of GPT-2 in Excel, targeting a broad audience including technical executives, developers, and AI policymakers.

## How Figma's databases team lived to tell the scale

- Figma's databases team successfully implemented **horizontal sharding** for their Postgres stack, achieving nearly infinite scalability after a nine-month journey.

## Why are most sofas so bad?

- **Sofas made in the last 15 years** are often constructed from low-quality materials like sawdust compressed with cheap glue, resulting in furniture that degrades quickly and is not worth the cost of reupholstering.

## Quiet-STaR: Language Models Can Teach Themselves to Think Before Speaking

- **Quiet-STaR** generalizes the Self-Taught Reasoner (STaR) approach, enabling language models (LMs) to generate **rationales for each token** to elucidate future text, thereby enhancing prediction accuracy.

## Show HN: Skyvern – Browser automation using LLMs and computer vision

- **Skyvern automates browser-based workflows** by integrating Large Language Models (LLMs) and computer vision, offering a robust alternative to traditional script-based automation methods that are prone to failure with website layout changes.

## Ollama now supports AMD graphics cards

- **Ollama now accelerates its features** using AMD graphics cards on both Windows and Linux, broadening its hardware compatibility.

## What I learned from looking at 900 most popular open source AI tools

- **Chip Huyen's analysis** of 900 popular open source AI tools reveals a dynamic shift towards application development and AI engineering, reflecting the rapid evolution of the AI ecosystem, particularly around foundation models.

## TextSnatcher: Copy text from images, for the Linux Desktop

- **TextSnatcher** enables users to **copy text from images swiftly** using OCR technology, specifically leveraging the [Tesseract OCR 4.x](https://tesseract-ocr.github.io/tessdoc/Home.html) for character recognition.

## [N] Ooops... OpenAI CTO Mira Murati on which data was used to train Sora

- **OpenAI's CTO, Mira Murati, reveals the data sources for training Sora**, sparking discussions on potential legal repercussions.

## Postgres is eating the database world

- PostgreSQL has evolved beyond a simple relational database to a comprehensive **data management framework**, challenging traditional OLAP with extensions like **ParadeDB** and **DuckDB** enhancing its analytical capabilities.

## [D] What are some well-written ML codebases to refer to get inspiration on good ML software design?

- **Scikit-learn** exemplifies **good ML software design** with its intuitive fit/predict paradigm, serving as a benchmark for newcomers in machine learning.

## Show HN: Matrix Multiplication with Half the Multiplications

- The **Free-pipeline Fast Inner Product (FFIP)** algorithm and its hardware architecture, introduced by T. E. Pogue and N. Nicolici, significantly enhance ML accelerator efficiency by reducing the need for multiplier units by nearly half without sacrificing performance.

## The Mythical Non-Roboticist

- **Benjie Holson** addresses the flawed notion that robotics programming can be simplified for "non-roboticists" through basic software frameworks, highlighting the intrinsic complexity of robotics that cannot be easily abstracted away.

## SuperPrompt: Better Text to Image Prompts in 77M Parameters

- Brian Fitzgerald developed **SuperPrompt**, a 77M parameter T5 model that enhances SDXL prompts, achieving or surpassing the performance of larger 1B+ parameter language models in prompt quality and alignment.

## Apple Buys DarwinAI Ahead of Major Generative AI Updates Coming in iOS 18

- Apple's acquisition of DarwinAI, a company known for creating **AI technology** for inspecting manufacturing components and developing smaller, more efficient AI systems, signals a strategic expansion of its AI capabilities.
