Search the archive
Full-text search across article titles and summaries. Refine with sector, year, content type, and reading-time filters.
Results
14,500 matchesDERIVATIVES TRADING FOR DUMMIES
TIME, SLOTS, AND THE ORDERING OF EVENTS IN ETHEREUM PROOF-OF-STAKE
FBI SEARCHES HOME OF TOP FTX EXECUTIVE
The FBI searched the home of Ryan Salame, a former FTX executive and major Republican campaign donor. Salame has recently come under scrutiny due to the $24 million in campaign contributions he made during last year’s primaries. These contributions were part of an aggregate $90 million worth of donations made to political campaigns by former FTX employees. The crux of the scrutiny is that authorities and court filings are claiming most of these funds were misappropriated from FTX exchange users. The search indicates the ongoing investigation into FTX's collapse, as federal prosecutors prepare for the trial of Sam Bankman-Fried, the cryptocurrency exchange's founder, who faces charges of fraud and illegal campaign finance schemes.
SUI PERFORMANCE UPDATE
As it gears up for mainnet launch on May 3, Sui has achieved a peak throughput between 10,871 to 297,000 transactions per second with 100 globally distributed validators as it has been running tests on the network. The time to finality is around 480 milliseconds. Sui uses Programmable Transaction Blocks (PTB), allowing complex, composable sequences of transactions with atomic execution. PTB supports up to 1024 transactions, enabling unbounded expressivity and efficiency. Sui plans to refine scalability and benchmark tooling, horizontal scalability for intra-validator scaling, and resilience to individual validator under-performance in the future.
EVERYTHING YOU NEED TO KNOW ABOUT PROTO-DANKSHARDING
GOOGLE CLOUD TO SUPPORT POLYGON TO HELP GROW ECOSYSTEM
CURVE FINANCE
ONE PERSON. 114 SCAMS.
THE INTERACTIVE GUIDE TO RENDERING IN REACT
React is generally a simple model to understand, where a view is a function of the state. However, there is one concept that is harder to understand: rendering. Exactly when and how does React update the view? This interactive guide makes understanding React rendering simple.
SCALING A MONOLITH WITH 5 DIFFERENT PATTERNS
INCONSISTENT JAVASCRIPT IMPORTS CAN LEAD TO DUPLICATE CODE
HOTSPOT PERFORMANCE ENGINEERING FAILS
HOW THE AGILE MANIFESTO CHANGED NOTHING
A SHORTCUT TO BECOME A GO-TO ENGINEER IN YOUR TEAM
LINEN
REACT HOTKEYS HOOK
React Hotkeys Hook is a React hook for using keyboard shortcuts in components in a declarative way.
NEXT ROUTE VISUALIZER
Next Route Visualizer is designed to display the routes for the ‘app’ directory of Next.js 13 in a visual tree-like fashion. This package aims to help developers better understand the routes in their Next.js app directory by providing a clear visual representation of the routes and their relationships. This package can be useful if you are new to the Next.js 13’s new ‘app’ directory and need to visualize it.
WHAT DEVELOPERS NEED TO KNOW ABOUT GENERATIVE AI
THE FUTURE OF PROGRAMMING: RESEARCH AT CHI 2023
CHI 2023 is a conference for leading human-computer interaction research. This blog post lists some interesting papers related to programming, such as visual methods to debug better.
LEARN WEBGPU
This documentation walks you through the use of the WebGPU graphics API to create native 3D applications in C++ from scratch, for Windows, Linux, and macOS.
WHAT’S NEW IN CHROME DEVTOOLS 113
Chrome 113’s DevTools come with a host of new features, but the biggest feature is the long-requested ability to override network response headers. With response header overrides, you can locally prototype fixes for various headers, such as CORS headers.
ACCESSIBLE TARGET SIZES CHEATSHEET
BRAVE SEARCH REMOVES LAST REMNANT OF BING FROM SEARCH RESULTS PAGE
Every Web search result seen in Brave Search is now served by Brave’s own index. They have removed all search API calls to Bing, which previously represented about 7% of query results.
GOOGLE CLOUD SERVICES HIT BY OUTAGE IN PARIS
OF PEOPLE’
MICROSOFT PHONE LINK BRINGS IMESSAGE TO WINDOWS TODAY
NEW NANOPARTICLES CAN PERFORM GENE EDITING IN THE LUNGS
SKY
PLATFORM ENGINEERING IN THE ERA OF LLMS
DEEPFLOYD IF
DeepFloyd IF is a modular model designed to generate high-quality images from text prompts. The base model creates 64x64 pixel images, while two super-resolution models create 256x256 px and 1024x1024 px versions of the image. DeepFloyd IF achieves a zero-shot FID score of 6.66 on the COCO dataset, outperforming current state-of-the-art models. Examples of images generated by DeepFloyd IF are available in the repository.
MEET THE PEOPLE WHO USE NOTION TO PLAN THEIR WHOLE LIVES
This article looks at the different ways that people are using Notion. A growing number of people are using Notion to organize their personal lives. Notion is software intended for work, but it can be used to track habits, make schedules, create and share lists, and more. Its high level of customizability sets it apart from other work apps and is what makes it so popular.
WHY DOES DID GOOGLE BRAIN EXIST?
RESCUE DEAL
PAYMENTS 101 FOR A DEVELOPER
AMAZON STARTS LAYOFFS IMPACTING HR AND CLOUD UNIT
TECH JOBS REMAIN A GOOD BET DESPITE MASS LAYOFFS IN SILICON VALLEY
PENTESTGPT
PentestGPT is a penetration testing tool powered by ChatGPT designed to automate the penetration testing process.
HOW BEREAL MISSED ITS MOMENT
Sponsored
ENGINEERING ROLES AT TIER-1 COMPANIES
NEW WAYS TO MANAGE YOUR DATA IN CHATGPT
ANTHROPIC PARTNERS WITH SCALE
STABILITY AI UPSCALE MODELS
LANGUAGE MODELS FOR PLANNING
Language models are great at producing plausible text and in certain cases reasonable plans. However, in general, they fail to generate long horizon plans. Without changing the model at all, even via fine-tuning, with special prompts we can dramatically improve the success of both plan generation and plan execution. Specifically, language models generate a plan language (fed a few-shot example) and that plan is executed by an actual planner then the resulting plans are translated back to natural language. The improvement is massive.
MACHINE TRANSLATION NEEDS ATTENTION
The Structured State Spaces for Sequences (S4) model, successful in various tasks, is applied to Machine Translation (MT), but lags behind the Transformer by 4 BLEU points and struggles with long sentences. This gap is attributed to S4's inability to summarize full source sentences in a single hidden state, which can be remedied by introducing an attention mechanism.
HOW TO TRAIN LLMS TO BE COMPETENT AND ETHICAL
One negative side effect of training agents to maximize reward is that they often become power seeking, meaning they try to increase their resources and access to information. This is a novel finding and somewhat hard to measure, this paper introduces a benchmark to help understand the power seeking behavior of these language models. Additionally, they give insights on how to train models to both act with high utility and according to moral guidelines. The world needs more high quality benchmarks and this is an excellent step in that direction.
DEEPFLOYD IF CODE RELEASED
The next image model from Stability AI has a code release. The model weights, blog, inference, and tooling will be coming in a few days. It is a multi-stage model with a frozen T5 text encoder and two super resolution models. One big win here is the ability to render extremely high quality text. The claim is that this model will likely be for research only at first, then open after some community feedback.
POSEVOCAB: LEARNING POSE EMBEDDINGS FOR HUMAN AVATAR MODELING
PoseVocab, a novel pose encoding method for human avatar modeling, constructs key poses and latent embeddings to effectively encode dynamic human appearances and achieve pose generalization and temporal consistency. The method outperforms state-of-the-art baselines in synthesis quality through the use of feature lines, a compact 3D representation, and a hierarchical query strategy for interpolating pose embeddings.