Search the archive

NEW TO THE WEB PLATFORM IN FEBRUARY

The stable releases of Firefox 123 and Chrome 122 have updates including declarative shadow DOM, a new Storage Buckets API, and a feature that preloads resources with the 103 Early Hints HTTP code.

PINGORA

Pingora is a Rust framework for building fast, reliable, and programmable networked systems. Cloudflare uses it to serve more than 40 million requests per second.

AWS LAMBDA WEB ADAPTER

AWS Lambda Web Adapter is a tool for running web applications on AWS Lambda. It allows developers to build web apps that use an HTTP API with familiar frameworks like Express.js, Laravel, Next.js, Flask, and more on AWS Lambda.

HOW DO COMPANIES SHIP CODE TO PRODUCTION?

The typical code deployment process for companies starts with planning and development, where developers work in sprints while product owners refine requirements as user stories. Automated systems are then used to verify that the code passes tests. The code is deployed to dev and QA environments for testing before being deployed to production.

ON GETTING PAID FOR OPEN SOURCE

Open source can be integrated into your job, but not every company offers these roles. This author's open source work began as part of their regular job, evolving into dedicated company time and side consulting gigs. Despite various funding models, including sponsorships and partnerships, they found it difficult to make a living solely from open source outside of a company.

PREVENTING SQL INJECTION ATTACKS IN NODE.JS

SQL injection can occur when user input is used in SQL queries without validation, allowing attackers to manipulate queries for malicious purposes. To prevent SQL injection attacks, developers should use query placeholders or prepared statements to ensure user input is treated as data, not code. They should also validate and sanitize user input to enforce format and character restrictions.

ASK HN: DID YOU ENCOUNTER ANY LEAP YEAR BUGS TODAY? (HACKER NEWS THREAD)

The author of this Hacker News thread mentions that their billing team just finished patching a bug on February 29th that erroneously charged monthly subscribers for an extra day. Other commenters share their stories.

GOOGLE BRINGS STACK OVERFLOW’S KNOWLEDGE BASE TO GEMINI FOR GOOGLE CLOUD

Google Cloud and Stack Overflow have formed a strategic partnership to integrate their platforms, bringing AI-powered code assistance and trusted developer knowledge directly to both the Google Cloud Console and Gemini.

LEARN CSS LAYOUT THE PEDANTIC WAY

In-depth, technically detailed chapters about CSS layout.

WEB CHECK

An all-in-one OSINT tool for analyzing any website.

SERVING MY BLOG POSTS AS LINUX MANUAL PAGES

This author devised a system that allows readers to view their blog posts as Linux manual pages. The system generates roff formatted Linux manual pages, which users can specifically request using HTTP headers.

ELON MUSK’S BOLD CLAIM: TESLA’S ROADSTER TO HIT 0-60 MPH IN LESS THAN ONE SECOND

Elon Musk claims that the Tesla Roadster, which he plans to unveil by the end of 2024, can accelerate from 0 to 60 mph in less than a second. The current record holder for a production car's fastest 0-60 mph is the Dodge Challenger SRT Demon 170, which can do it in 1.66 seconds. If Musk's claims are true, the Tesla Roadster will be faster than the speed of sound. Tesla aims to ship the car in 2025, four years after its original 2020 target.

ADOBE’S NEW PROTOTYPE GENERATIVE AI TOOL IS THE ‘PHOTOSHOP’ OF MUSIC-MAKING AND EDITING

Adobe's Project Music GenAI Control is a prototype tool that allows users to generate and edit music using text prompts. It has controls for customizing results and remixing sections of audio. The tool can generate audio based on a reference melody and extend the length of audio clips. It is unclear whether the tool will allow uploads of any audio as reference material or how long clips can be extended for. The tool isn't available to the public yet and no release date has been announced.

MAN IN REMISSION FROM BLOOD CANCER AND HIV AFTER REMARKABLE TREATMENT

A Californian man who contracted blood cancer while living with HIV is in remission from both. The patient's treatment involved receiving a transplant of stem cells with a rare genetic mutation that provides resistance to HIV. The treatment targeted the blood cancer - HIV resistance was an extra benefit. The patient is the fifth person ever to overcome both illnesses.

VARDA CAPSULE REENTRY - FIVE MINUTES FROM LEO TO EARTH (6 MINUTE VIDEO)

Varda Space Industries' W-1 spacecraft successfully landed in Utah on February 21. This was the first time a commercial company landed a spacecraft on US soil. The spacecraft reentered the Earth's atmosphere at speeds over Mach 25 before safely deploying its parachute and landing. Varda has released a video from a camera onboard the W-1 spacecraft showing it orbiting the Earth in low-Earth orbit, smoothly separating from Rocket Lab's satellite bus, and then reentering Earth's atmosphere.

Join Next ‘24 this April in Las Vegas ↗️

*Valid through 11:59 PM PT on 02/29/2024 or until sold out.*

WHAT’S ON THE AGENDA: genAI [11] deep dives; 500+ sessions

across 10 different tracks; new AI demos [12] and trainings; [13] and expert-led builder sessions including AI coding challenges.

PINGORA

Pingora is a library for building fast, reliable, and programmable network services. It has been used to serve more than 40 million Internet requests per second at Cloudflare for more than a few years. Pingora features customizable load balancing and fail over strategies, support for a variety of observability tools, gRPC and websocket proxying, and TLS over OpenSSL or BoringSSL.

PIPELINED QUERY LANGUAGE

pql (pipelined query language) is a pipeline-based query language, inspired by Microsoft's Kusto Query Language, that compiles into SQL. The generated SQL is database agnostic. pql is designed to make developing queries simple by piping the result of each statement into the next. Examples are available.

APPLE SPENT MORE THAN $10 BILLION ON APPLE CAR BEFORE CANCELING PROJECT

Apple spent more than $10 billion working on the Apple Car since the project first launched in 2014. The money was spent on research and development, with thousands of engineers and car experts working on the project. Apple was never able to find the right leader for the project - the project had four different leads over its course of 10 years. Most of the more than 2,000 employees who worked on the car project will be redistributed, while some will be laid off. Apple will apply its learnings from the car project to other devices.

ASK HN: SLOW THINKERS, HOW DO YOU COMPENSATE FOR YOUR LACK OF QUICK-WITTEDNESS?

This Hacker News thread contains tips for slow thinkers to overcome and compensate for their lack of quick thinking ability. Many slow thinkers have potential, but they lose their capacity when forced to make quick decisions or generate arguments on the spot. Ways to get around this include taking your time to answer questions and preparing for situations in advance. Finding and acknowledging weaknesses, and then finding specific techniques or processes to overcome them, will significantly improve most outcomes.

regret it. Start with an exclusive offer for TLDR readers

BITCOIN’S SO HIGH, IT CRASHED COINBASE TODAY

Bitcoin's recent surge to $60,000, close to its all-time high last seen in 2021, caused Coinbase's app to show some of its customers a zero balance across their accounts - customer funds are safe.

NVIDIA GEAR RESEARCH GROUP CREATED TO DEVELOP AI ROBOTS AND NPCS FOR PHYSICAL AND VIRTUAL WORLDS

Nvidia's Generalist Embodied Agent Research (GEAR) group will further advance and build capable AI-based entities that can operate with skill in both the virtual and physical worlds.

STRIPE VALUATION SURGES TO $65BN AS IPO SPECULATION GROWS

Stripe has signed agreements with investors to provide liquidity to its current and former employees for a tender offer.

INTEL PUTS 1NM PROCESS (10A) ON THE ROADMAP FOR 2027 — ALSO PLANS FOR FULLY AI-AUTOMATED FACTORIES WITH 'COBOTS'

Intel's 10A process will begin development but not enter high-volume manufacturing in 2027.

WHY I USE FIREFOX

Firefox is very customizable, runs on its own browser engine, and has the best support for extensions on Android and a great picture-in-picture player, webpage translation is completely private, and Mozilla can't decrypt user data on its servers.

GOOGLE CEO TELLS EMPLOYEES GEMINI AI BLUNDER ‘UNACCEPTABLE’

Google has taken Gemini's image-generation feature offline for further testing due to problematic issues related to bias.

YOUR REFERRAL LINK - https://tldr.tech/tech?ref=2613930

META LLAMA 3 MODEL PLANNED FOR JULY

Meta's Llama team is reportedly still working on alignment to enable the model to loosen up and answer more queries that aren’t actually controversial.

MODAL ADDS WEBSOCKETS

Most serverless GPU offerings require one POST request per model inference. However, this is a challenge in cases where you want to stream model outputs. Websockets make streaming possible.

TIM COOK SAYS APPLE WILL ‘BREAK NEW GROUND’ IN GENERATIVE AI

Apple CEO Tim Cook announced at the company's annual shareholders meeting that the company plans to break new ground in generative AI in 2024, promising transformative opportunities for users.

VIDEO AS THE NEW LANGUAGE FOR REAL-WORLD DECISION MAKING

An amazing paper that frames video as a potentially better way for AI to interact with human users. It shows how video models can be used as planners, agents, compute engines, and environment simulators.

1BIT LLMS

Most language models use 16 or more bits to represent a parameter in a model. This results in powerful models that can be expensive to run. This work proposes a method that uses 1.58 bits per parameter - each parameter is in {-1, 0, 1}. This method matches performance exactly up to 3B parameters. No code or models have been released yet.

ENHANCING VISUAL PERCEPTION IN AI MODELS

This research focuses on advancing multi-modality foundation models like GPT-4V in low-level visual perception tasks. The large-scale experiment gathered feedback on 18,973 images from 58,000 humans and created the Q-Pathway dataset to analyze clarity, color, and brightness.

Save your spot for free!

DRUG DESIGN WITH GRAPH DIFFUSION POLICY OPTIMIZATION

Graph diffusion policy optimization enhances graph generation models for drug design using a unique reinforcement learning technique. This approach promises better performance in creating complex and diverse graph structures and could set new standards in the field.

LLMS HANDLE 100K TOKENS WITH DUAL CHUNK ATTENTION

Dual Chunk Attention (DCA) extends the capabilities of large language models like Llama2 70B, allowing them to handle over 100k tokens without additional training. It breaks down attention computation into chunks, enhancing the model's understanding of both short and long-range context.

DISTILABEL FRAMEWORK FOR ALIGNMENT DATA COLLECTION

STARCODER V2

The Big Code project has released another version of its flagship coding model. The model has a 16k context window and was trained on 4T tokens. Its performance is strong, but still under that of DeepSeek coder.

LEARNING FROM HETEROGENEOUS GRAPHS

HiGPT is an approach to learning across diverse heterogeneous graphs without the need for fine-tuning. Its integration with a novel graph tokenizer and a vast corpus of graph instructions allows it to excel in adapting to various data distributions.

ONLINE SHOPPING WITH MULTI-MODAL KNOWLEDGE DISTILLATION

PromptMM improves recommendation systems on platforms like Amazon and TikTok by using Multi-modal Knowledge Distillation. It addresses inaccuracies in user preferences and simplifies systems by distilling important features from various content types (visual, textual, or acoustic) to prevent overfitting.

DPO TO ENCOURAGE DESCRIPTIVENESS (GITHUB GIST)

A minimal code set up with TRL to tune a model to be more descriptive.

SHAPE SUFFIXES FOR ML CODING

A coding style at Character AI dramatically improves the readability of shapes in tensors.

ADOBE GENERATIVE AI DOCUMENT & PDF TOOL (PRODUCT)

Adobe Acrobat's new AI Assistant can query documents for answers with attribution, create document summaries, offer a list of suggested questions for deeper insights, assist in drafting content for various communication needs, and provide outlines and key points so users can swiftly navigate essential document information.

YOUR REFERRAL LINK - https://tldr.tech/ai?ref=2613930