Full-text search across article titles and summaries. Refine with sector, year, content type, and reading-time filters.
Sponsored
While the OWASP Top 10 provides valuable insights into open source security, it’s equally critical to manage risks within your first-party code. Check out Snyk’s Top 10 Code Vulnerabilities report
The Vercel Platforms Starter Kit is a full-stack Next.js template for building multi-tenant applications with custom domains. It features programmatically assigned unlimited custom domains, subdomains, and SSL certificates, fast performance, an AI editor, image uploads, custom styles, and more. The kit creates apps that can serve multiple customers across different subdomains/custom domains with a single unified code base.
LineSelect is a shell utility that allows users to interactively select lines from stdin and output them to stdout. It effectively pauses the pipeline for the selection. LineSelect makes it easier to select files with ls, manage docker containers, grep log files, and more.
Twitter's recent volatility provided Meta with an opening to compete. Meta had been planning to release Threads for a while, but the recent backlash over Twitter's limit on how many tweets people can see per day prompted the company to push it out this week. Meta expects tens of millions of people to try the service within the next few days. Threads is now available in 100 countries, but not in the EU due to complexities complying with some of the laws coming into effect next year.
AI licensing isn't as simple as applying proprietary/open-source software licenses. It has many layers with multiple components and additional licensing considerations. This article presents a set of AI licensing categories to standardize discussions on AI licensing. It names different categories of license types based on the function they serve and which AI component they serve. Using the right terms will help the industry move forward in developing standards around each of the categories.
Sponsored
Has AI changed software development? Chime in to shape the future of dev tools, win prizes, and uncover cutting edge advancements shaping software today. Take the Developer Nation Survey
Firefox 115 introduced a new 'feature' that only allows some extensions monitored by Mozilla to run on specific websites.
Some people are frustrated with React Server Components, others are frustrated with the way the React team communicates, and the React team is frustrated with the pile-on they are facing.
Large language models do both of the things that their promoters and detractors say they do, at the same time, on the same prompts, and while it is difficult to tell which they're doing, both of them are useful.
This tool takes English instructions and compiles them into PySpark objects like DataFrames.
This page contains a collection of interviews, podcasts, posts, books, and other resources from Keith Rabois, a general partner at Founders Fund.
This article presents three alternative AI-generated endings for humanity in an attempt to mimic a physicist's agnostic approach to predicting how the universe will end.
How do you align a model much smarter than humans? OpenAI has been working on this question across teams for some time, but now they're bringing 20% of the entire org (compute-wise) to bear on this challenge. They're hiring extensively and many top scientists internally are working on this problem.
The popular AI powered image editing suite raised money to continue pushing forward text driven creativity expressed via pixels.
It seems like dilated attention now finally works! This is an exciting advancement if it holds at scale, because it means that we can essentially treat entire corpi of documents as context. All of the Python code on GitHub is only a few hundred billion tokens. LongNet is a linear attention variant, but they seem to have stabilized training, and it doesn’t sacrifice performance on shorter sequences.
This study presents a technique that improves AI's understanding of image-text tasks by focusing on fine details and keywords, enhancing performance, especially when faced with unseen scenarios.
One of the top multimodal models today was developed by the UniLM team at Microsoft. They have now released demo code, weight checkpoints, and more for you to play with and integrate into your systems.
There's a lot of excitement around using vector databases to augment the memory of different language models. This vector database has a simple interface and is primarily used for ChatGPT plugins. It runs many times faster than competitors while maintaining simplicity. It doesn't have the bells and whistles of many other offerings, but it performs the task well for plugins.
This research introduces SAM-DA, a new system that improves real-time tracking of unmanned aerial vehicles (UAVs) at night. By using the Segment Anything Model (SAM) to generate detailed training samples from limited nighttime images, it significantly enhances the tracking ability while using fewer resources.
WebGLM is a new online tool powered by artificial intelligence designed to answer questions by searching the web and providing the most relevant information. It improves upon previous models by offering increased accuracy and efficiency, and studies show that it performs as well as, if not better than, more complex systems.
If you’re doing anything with generative AI and playing internally with your own LLMs, you should check out this post about Streamlit.
Celebrities are partnering with brands to create AI-generated duplicates of themselves that can be used in marketing campaigns. This gives celebrities more control over their own images and allows brands to use the stars in ways they never could before.
We’ve covered this in previous newsletters, but it just got an upgrade. Inspired by Andrej’s great NanoGPT repo, the NanoT5 repo can train a model to equivalent T5 paper performance 150x faster. It’s definitely worth checking out if you care about training models from scratch.
Ryan Moulton discusses the various methods through which digital minds can acquire knowledge, such as direct programming, machine learning, simulation, and training.
The article explores the issue of AI model weights and their availability as open-source. It argues that while the source code of AI models may be open, the weights—which contain the actual learned knowledge—are typically not shared openly due to various reasons such as intellectual property concerns, privacy, and commercial interests.
BlackRock's CEO, Larry Fink, stated yesterday that Bitcoin is now viewed as an "international asset". The statement comes after the asset management firm made moves to offer a spot Bitcoin ETF. Fink also compared the role of Bitcoin to digital gold, offering an alternative to physical gold investments as an inflation hedge. Despite his initial skepticism, Fink asserted that BlackRock's aim is to democratize crypto and provide cheaper options for investors.
In Q2, crypto M&A numbers fell, with intra-sector deals down 50%. According to Architect Partners, there were only 32 transactions during this period, a steep decline compared to the previous quarter's high of 54. Total deal value also dropped to $370 million from approximately $400 million last quarter. The analysis attributed this dip to crypto companies becoming more selective in their business choices. Laser Digital’s acquisition of Elysium Technology Group and Ripple's purchase of Metaco were some of the big deals of the month.
Blur v2 launched with two major feature rollouts: trades now consume 50% less gas due to optimizations and users can now place bids on specific traits. As part of this update, Blur v1 bids are no longer functional, meaning users will have to make the same bids through the updated contract. Trait bidding is available for all collections and specific ones can earn extra points. Other features previously launched this season include lending, BNPL, mobile support, notifications, and an optimized v2 marketplace.
Axiom, a solution giving developers enhanced access to on-chain data on Ethereum, is now live. With this, developers can now request larger amounts of data and compute power for their smart contracts. Using zk proofs, Axiom aims to ensure data accuracy and trustworthiness, enabling the creation of data-rich applications with the same level of interaction as traditional web apps, but with the security of Ethereum. Currently, developers can access block headers, account details, and historic storage values through Axiom’s platform, and the company plans on providing access to transaction, receipt, and consensus data in the near future.
This write-up explains how Ethereum's scaling future may depend on rollups, a shared sequencer architecture designed to improve atomic cross-rollup interoperability. The current rollup solution faces issues with centralized sequencers and a lack of effective solutions for interoperability between optimistic rollups. Shared Validity Sequencing is a decentralized model of sequencing, intended to allow atomic bridging of tokens and enable cross-chain actions. In this design, the technology can serve entire rollup ecosystems with unified layers of native assets. Among the advantages, this solution promises enhanced sovereignty for applications, flexible pricing for locks, and immediate sharding possibilities.
This tweet discusses setting up a 2 of 2 multisig setup for increased security using BIP85 on a Coldcard for bitcoin transactions. It provides step-by-step instructions for creating a secure system that incorporates the need for just one seed plate and pass phrase, two hardware wallets, and no wallet files. The process incorporates a geographic key distribution tactic to ward off potential offline theft and offers suggestions for extending the setup for a 3 of 3 spread over three locations. The tweet goes on to explain how you could potentially use this setup for an inheritance.
Brett Harrison, former FTX US CEO, founded a venture called Architect earlier this year, aiming to simplify AI-based crypto trading. It has closed a $5 million seed round from investors like Coinbase Ventures and Anthony Scarmucci. Architect uses algorithms and AI interfaces like OpenAI's API to help users create crypto trading strategies. The platform, currently in beta mode for specific clients, allows traders to establish these trading rules using natural language. Despite the use of AI tools, Harrison emphasizes that the effectiveness of the trading strategies still ultimately relies on the guiding team's expertise.
This blog explores the concept of blockchains and bridges in crypto networks. It defines a blockchain as a data structure for categorizing transactions, achieved via a consensus protocol for data validation. The blog likens blockchain protocols to databases, with ‘closed’, ‘public’, or ‘open’ databases based on their access permissions. It delves into the role of bridges in facilitating communication between such databases, with overviews of 'Human-operated' and 'Light client' bridges.
NFT royalty payments hit a two-year low in June, according to Nansen. The rise of royalty-optional marketplaces like Blur and OpenSea with its new policies coincided with creators' earnings decreasing significantly.
The IRS is focusing on crypto tax evasion by attempting to provide clearer guidelines and tracking crypto transactions more closely.
Coinbase Product Director Conor Grogan found over $320,000 in untouched Ethereum Classic funds for an individual unaware of their existence. This find was part of Grogan's broader effort to locate dormant ETC wallets stemming from the 2016 Ethereum blockchain fork.
In the wake of a precipitous decline in NFT prices, investors are blaming Blur's incentivized trading model. Critics argue that Blur's practices encourage rapid flipping and wash trading for rewards, essentially turning NFTs into nothing more than "altcoins with pictures."
Sponsored
BrightData [https://get.brightdata.com/i3b4q467ltcx] is the world's top web data platform with over 72 million residential IP addresses and 7 million mobile IP addresses for web scraping. Use our web scraper IDE, scraping browser, or API to unblock even the toughest sites, or just use one of our pre-crawled datasets!
A new React Handbook page that discusses two key aspects of React performance: Loading and Runtime Performance. Loading Performance focuses on how quickly a webpage's content loads. The handbook suggests optimization techniques like GZip compression, image optimization, and code splitting to improve loading performance. Runtime Performance, on the other hand, measures the application's smoothness post-load. The handbook recommends strategies such as fixing rendering issues, offloading work to Web Workers, and using memoization and virtualization.
Node.js code can contain some computationally complex tasks that will typically block the main thread. This is where multithreading comes in handy because it allows us to simply run these tasks in a worker thread. This article explores the ins and outs of multithreading in Node, including some potential pitfalls.
Darcy Clarke, a former npm CLI team manager, discusses a critical issue in the npm ecosystem known as "manifest confusion." This issue arises from the independent publishing of an npm package's manifest and its tarball, without full validation of their contents against each other. This loophole allows potential exploitation by bad actors, leading to hidden malware, unknown dependencies, and potential downgrade attacks.
The blog post on Deno's website argues that CommonJS, the module system used in Node.js, is hindering JavaScript's progress due to its synchronous loading and lack of native browser support. It suggests that ECMAScript modules (ESM), which support asynchronous loading and are browser-compatible, are a superior alternative. The post advocates for a complete transition to ESM to better align with modern JavaScript development and the shift towards cloud primitives.
Evan Hamilton and Tyler Swartz, former Reddit employees, share a four-step framework for building products for opinionated user bases. They suggest measuring and building trust with users, identifying key voices to heed, creating a private advisory council for discussions, and acting responsibly based on feedback. Hamilton and Swartz emphasize the importance of managing passionate users productively to ensure successful product development.
For developers in today's job market, finding a role that helps you thrive involves more than just technical skills and compensation. As a developer, thriving means you’re truly supported, motivated, and able to do your best work–and it significantly predicts how productive you can be at work. This article offers 4 tips that will help you find your next role that will allow you to thrive as a developer.
The Vercel Platforms Starter Kit is a full-stack template for building multi-tenant applications with custom domains using Next.js App Router, Vercel Postgres, and the Vercel Domains API. It comes with powerful features like an AI editor, image uploads, fast performance, dark mode, and more.
Prettier 3.0 is here and one major change includes the migration to using ECMAScript Modules for all source code. This change has significantly improved the development experience for the Prettier team. Read on to find out what else has changed and improved.
Remotion v4 has been released. Remotion is a way to make videos programmatically using React. It introduces Remotion Studio, a way to interactively edit props of your React app and render assets quickly. It also features Rust-powered architecture, PDF and WebP generation support, and more.
This is a hands-on article that explores the Function Calling capability OpenAI recently released. The article uses an example to use Large Language Models (LLMs) to extract structured data from unstructured Kaggle competition write-ups. The process involves defining a function for structured output, designing a prompt for extraction, and handling API rate limits. After obtaining a list of methods, the author uses an LLM to clean up and structure the data, which is then analyzed further.
The article explores how human translators are adapting to AI translation tools. Despite AI's advancement, human translators remain vital, particularly in specialized fields like law and medicine, and in hybrid services where AI drafts are human-reviewed. However, the rise of AI hasn't significantly improved translators' pay rates, underscoring that automation is a gradual process with varied impacts on different professions.
CSS has a whole slew of different color formats: hex codes, rgb(), hsl(), and lch()... the list goes on! It can be hard to figure out which one to use. This tutorial will take you on a tour of the different options and show you how they work, how to decipher them, and how to use them to your advantage.