Full-text search across article titles and summaries. Refine with sector, year, content type, and reading-time filters.
Chardet is a character detection module developed entirely in TypeScript designed to identify encoding based on occurrence analysis. It is versatile and compatible across various environments (Node, Browser, Native) and platforms (Linux, Mac, Windows), without any dependencies or native code/bindings.
This GitHub Action makes it easy to quickly write a script in a GitHub workflow that uses the GitHub API and the workflow’s context. To use this action, a developer needs to provide an input named `script` that contains the body of an asynchronous function call.
Effective crisis management and learning from failures in tech leadership roles are essential for successful software companies. Blameless post-mortems are crucial learning tools for improving future practices. Mistakes are inevitable, but the response to these mistakes, including clear communication, owning up to errors, and implementing corrective measures, defines effective leadership.
OpenAI, which initially opted for a non-profit model when it was founded in 2015, later adopted a for-profit model. The company recently experienced a major crisis, with its CEO Sam Altman being fired, leading to employee revolt and threats of mass resignation. The crisis was resolved with Altman's return and board restructuring. This article not only summarizes the recent OpenAI saga, but also touches on OpenAI's fast-paced success with ChatGPT, the contrasting directions it has pulled the company in, and Microsoft's interest in maintaining an independent OpenAI team.
Organizational structures have evolved constantly in response to technological advancements, beginning with the New York and Erie Railroad in 1855 and its innovative use of the telegraph and organizational charts. This article draws parallels between historical shifts in organizational practices due to technological changes, like Henry Ford's assembly lines, and the current potential impact of AI on organizational processes. Using the example of Wharton Interactive, a software startup, the article illustrates how AI, specifically large language models like GPT-4, can be integrated into various organizational roles.
This article discusses 6 trends that might become commonplace in 2024 in the API ecosystem, such as APIs from multiple sources and the evolution of API gateways.
This article covers 10 HTML tricks from the past and the present, such as using HTML tables for layout.
A tutorial on building a web page layout with container queries.
Tesla has started releasing its Full Self-Driving v12 update to employees. The biggest difference with the update is how vehicle controls will be taken over by neural nets rather than being hardcoded by engineers. Tesla may be able to push the update to customers by the end of the year. Elon Musk has previously said that he believes that Tesla will achieve true self-driving capability by the end of the year.
Google will start deleting inactive Google accounts from December 1. Accounts are deemed inactive if the user hasn't logged on for at least two years. Gmail addresses for deleted accounts cannot be used again when creating a new Google account. Users just need to sign into their accounts to preserve them.
SpaceX plans to test Starship for the third time in December. There are now three ships in final production. SpaceX launched the second Starship test on November 18. The ship successfully performed a hot-stage separation, but it exploded after it completed a flap and boostback burn.
Researchers used an algorithm to sort through millions of genomes to find new rare types of CRISPR systems that could eventually be adopted into genome-editing tools. Single-celled bacteria and archaea use CRISPR systems to defend themselves against bacteriophages. The algorithm helped scientists identify the code for an entirely unknown CRISPR system that targets RNA. Researchers will be able to use the algorithm to look for other types of protein across species.
Developers should always use a Cryptographically Secure Pseudorandom Number Generator (CSPRNG) when choosing a Pseudorandom Number Generator. While people typically say to only use CSPRNGs when doing cryptography and need security against adversarial attackers, they are really the only thing that makes sense to use. This article talks about why developers should only use CSPRNGs by comparing them with other random number generation methods. CSPRNGs are better by definition as given the first n bits, there is no efficient algorithm that will predict the next bit with a success rate non-negligibly better than 50%.
RAGs is a Streamlit app that lets users create RAG pipelines from data sources using natural language. Users just have to describe the tasks and parameters they want from their RAG systems. The RAG can be queried and it will answer questions over the data.
It's difficult to commodify technical work. Despite the crazy infrastructure that allows us to mass produce things, in many cases, individual people make the machine run. Society runs on commodification, but anyone who thinks that it can run on pure commodification, without any understanding of the human complexities of the people around them, is wrong.
Binance and its CEO recently pleaded guilty to operating the world's largest criminal conspiracy to launder money and paid more than $4 billion in fines. This article looks at the history of Binance to see what led to this outcome. The ruling settles a long-running investigation involving the DOJ, CFTC, FinCEN, and assorted other parts of the US regulatory state. However, it importantly doesn't resolve the SEC's parallel action, which includes charges of operating an unregistered exchange, misrepresenting trading controls, and the unregistered offer and sale of securities.
The US government and Intel have started training a one trillion parameter model called ScienceGPT.
The traditional strategy of mastering a single tool may no longer be optimal - in the world of AI, it is better to embrace new tools and discard old ones and focus on things that are stable around the rapidly shifting software foundation.
AI helps developers skip steps in the workflow.
Users should be able to build projects without installing any extra tools or content - if it's something the build needs, then it belongs in version control.
There is less competition and teams are more motivated.
This thread contains a lot of information and resources for older tech workers who want more comfortable jobs.
A recent leak that stated the OpenAI leadership debacle was caused by an internal breakthrough that combined search and RL. This post outlines one theory explaining what might actually be happening with this new method.
An all-encompassing recap of the OpenAI fiasco. This will be studied for generations. If you haven’t followed the entire saga, this summary is a must read including all of the interesting angles and fall-out.
Anthropic included a price cut for tokens generated with the release of the newest version of Claude. This comes from both open and closed model pressure.
LoRAs are low rank learned matrices that change model behavior but are cheaper than regular fine-tuning. This work proposes an effective way to merge LoRAs together while maintaining the novel information from each.
The introduction of SegVol, a universal model for medical image segmentation, marks a new era in clinical analysis. Trained on a comprehensive collection of CT scans, SegVol excels in segmenting an extensive variety of anatomical categories.
This innovative approach accommodates a range of prompts and contexts, greatly boosting performance in segmentation tasks and showcasing impressive results in open-set challenges.
Collecting image, audio, and keystroke data from your computer is actually somewhat challenging. This library makes that process easy with the purpose of training digital Agents.
Direct preference optimization is a stable alternative to RLHF using very similar data. This repository contains an implementation for study to learn about the technique.
Kandinsky Video involves a dual-stage text-to-video process. It crafts keyframes to outline the video's story and then smoothens transitions and movements, substantially enhancing video quality and reducing computational demands.
It’s been five incredibly turbulent days at OpenAI, with the exit and then return of CEO Sam Altman. This article succinctly unpacks all of that dynamism and then goes on to explore whether OpenAI is the not-for-profit it declared itself to be in 2015 or something else entirely.
GenSLMs, an AI model for genomic data, accurately generated gene sequences resembling recent COVID-19 variants despite being trained only on early pandemic data, showcasing its potential for predicting virus evolution and aiding research.
Anthropic has released Claude 2.1, which can digest 200,000 tokens, hallucinates half as often and can do things like search the web.
Many top researchers have founded a new multimodal research lab that aims to publish great work and hopefully maintain its non-profit status.
Sponsored
Stablecoin Studio is an open-source SDK [5] that simplifies the process of issuing stablecoins using Hedera network services. “Stablecoin Studio was a game-changer for our successful stablecoin remittance pilot [5] with SCB TechX and other financial institutions. The toolkit has made it incredibly convenient to send low-cost cross-border payments.” — Kim Bying Hee, Chief of Blockchain Division, Shinhan Bank
Investment funds can now officially tokenize their assets in the UK, which officials hope will increase efficiency, transparency, and competitiveness in the investment management sector. This move, facilitated by the UK Treasury, Financial Conduct Authority, and various investment managers, requires that funds meet specific criteria for tokenization.
Blur’s Blast L2 launched last week and has attracted approximately $390 million into its contract - about $340 million in ether and $50 million in stablecoins. While it was marketed as an L2, for now, it’s simply a bridge to a contract created by the team and secured with a 3 of 5 multisig. To add to the uncertainty many in the space have about it, it’s unclear if each of the five keys was created by individual entities or people, and the deposited funds cannot be withdrawn until February 2024.
Circle and Japanese financial conglomerate SBI Holdings announced a partnership to bring USDC to Japan. The collaboration aims to expand the use of Circle's USDC stablecoin and promote banking and web3 technologies. It leverages the regulatory clarity from Japan's new stablecoin laws implemented in June. The strategic partnership will also focus on fostering the growth of web3 apps, including tokens and smart contracts in Japan's consumer market.
Rain.fi, a DeFi platform, has announced custom Peer-to-Peer loans on Solana, offering a variety of user-set parameters such as Custom LTV, APYs, and durations. Users can create loans in multiple tokens with one base currency and protect against token volatility. The platform also has an incentive for early repayment and plans to launch the service with select tokens like USDC and Solana.
This article provides a comprehensive explanation of Verkle trees, an improvement over Merkle tree structures. Verkle trees, which employ vector commitments, are more efficient regarding proofs and bandwidth costs, making them key in scaling blockchain communication. The article also discusses the construction of Verkle trees and how to commit changes to the tree and explores Verkle tree membership proofs and batch proofs.
This post discusses tribalism in the crypto industry and how builders can navigate and leverage tribalism when developing their applications. Some of the strategies mentioned in the article are diversifying one's understanding of different crypto tribes, challenging the norms within your community, and thinking like a "multi-armed bandit," balancing exploration and exploitation.
This blog post explores the concept of "intents" in distributed systems, particularly as they relate to blockchains. The blog highlights potential issues such as order flow, trust, and opacity and emphasizes the ability of intent-based solutions to flip supply-side blockchain economics on its head. It suggests that the decentralization of protocols, security models, histories, and communities enabled by intents is the future of blockchain systems.
AI's rise, specifically LLMs, offers advantages across various fields but raises concerns around privacy and security as they interact with extensive personal and sensitive data. Integrating blockchain technology into AI could create a secure framework that preserves data ownership and privacy. This article explores multi-party computation in transformers, a method that combines AI and privacy.
This thread provides a detailed breakdown of the complex Kyber smart contract exploit, detailing its unique manipulation of concentrated liquidity and an ‘infinite money glitch’ due to a technical oversight.
A Bitcoin sender accidentally paid $3 million in fees for a transaction worth $2 million, marking the highest fee in Bitcoin history.
This detailed analysis walks through how the unique nature of the crypto industry being both nowhere and everywhere played a part in Binance’s involvement with money laundering and other alleged unlawful activities.