Full-text search across article titles and summaries. Refine with sector, year, content type, and reading-time filters.
Learn how to develop, document, and test React Server Component applications in isolation using Storybook, using Mock Service Worker to mock network requests.
Tesla has told suppliers it wants to start production of a new electric vehicle in mid-2025. Codenamed 'Redwood', the next-generation vehicle is expected to drive a new wave of growth. Tesla has forecasted a weekly production volume of 10,000 vehicles. The new vehicle may be the entry-level $25,000 car that Elon Musk first promised to build in 2020. Tesla plans to make an inexpensive robotaxi based on the same vehicle architecture as the $25,000 electric car.
Lumiere is an AI video generator that uses a space-time diffusion model for realistic video generation. The Space-Time U-Net architecture that it uses generates the entire temporal duration of a video at once with a single pass. It is designed to handle both where things are and how things move and change throughout a video. Lumiere can perform text-to-video generation, convert still images into videos, generate videos in specific styles using a reference image, apply consistent video editing using text-based prompts, create cinemagraphs, and inpaint videos. It excels in creating videos of cute animals in ridiculous scenarios - examples are available in the article.
The world's deepest and largest underground laboratory is around 2,400 meters below the Jinping Mountains in southwest China. The enormous space, with a sprawling capacity of 330,000 cubic meters, is home to scientists studying dark matter, a hypothetical substance thought to make up more than 80% of the mass in the Universe. It opened in 2010 but only became operational in December 2023. The layers of rock above the laboratory shield it from background noise such as cosmic rays.
Five children with a form of genetic deafness called DFNB9 regained their hearing thanks to a new type of gene therapy. The therapy involves the use of an inactive virus carrying a functioning version of the gene associated with the disease. The children, who were previously totally deaf, recovered their hearing and were able to conduct normal conversations. Further studies will enroll more patients and follow them for a longer period of time to ensure that the treatment works.
Cargo by default now uses `strip = 'debuginfo'` for the `release` profile unless debuginfo is explicitly requested for some dependency. This new default will be used for any profile that does not enable debuginfo anywhere in its dependency chain. This article looks at how a developer noticed that Rust binaries were unnecessarily large and then developed and submitted a fix for it.
Prompt engineering is just a subset of software engineering. Prompts are tiny programs written in natural language. Guessing the right 'function calls' with the clever use of vocabulary is a huge part of the game as the API isn't specified and varies between models. Thinking of prompts as code allows prompt engineers to generate model-aware syntax highlighters for favored keywords.
A leaked memo from Google's CEO Sundar Pichai revealed the company's seven goals for the year. This year, Google aims to deliver the world's most advanced, safe, and responsible AI; improve knowledge, learning, creativity, and productivity; build the most helpful personal computing platforms and devices; enable organizations and developers to innovate on Google Cloud; provide the world's most trusted products and platforms; build an extraordinary Google; improve company velocity, efficiency, and productivity; and deliver durable cost savings. The leak comes amid a wave of layoffs at the company, with more job cuts on the horizon.
Artificial General Intelligence (AGI) is commonly defined as an autonomous system that surpasses human capabilities in the majority of economically valuable work. Recent developments in autonomous driving present a good early case study of the societal dynamics of increasing automation. Driving is a difficult problem to automate and its automation will impact a large human workforce. The autonomous driving industry has created many new jobs that didn't exist before. Instead of destroying jobs, automation caused the industry to adapt and change.
How much Apple plans to charge is still unknown, but the company currently charges a 27% commission on in-app purchases made through alternative payment systems in the Netherlands.
Netflix plans to eliminate the cheapest ad-free plan available to users.
A guide on breaking the DRM controls on a Xiaomi 4 Pro air purifier.
How Apple decides to comply with the EU's Digital Markets Act will decide whether Spotify implements in-app payment functionality that will allow users to upgrade their subscriptions or buy audiobooks with a tap.
Nvidia's RTX Video HDR feature uses AI to convert SDR color space video to HDR - it requires an HDR10-compatible monitor with HDR enabled in Windows.
The exponential improvements in tech in the 21st century can be explained by improvements in semiconductor fabrication.
Sponsored
For LegalTech Companies: Elevate your platform instantly with CaseMark’s Workflow API [7]. Seamlessly integrate our solution to offer AI-driven deposition, case, trial, and hearing summaries.
Embeddings serve as the foundation for many AI use cases. They can be expensive and slow due to rate limits. This blog walks through using Modal to scale up the embedding process efficiently and correctly.
Google, the Weizmann Institute of Science, and Tel Aviv University have unveiled Lumiere, a space-time diffusion model for realistic video generation that promises more coherent motion and better quality compared to existing AI video models. It's not yet available for public testing.
Neel Nanda is one of the pioneers of Mechanistic Interpretability (MI). This is his guide to getting into the field. It features 200 concrete open questions. MI is the study of the quantitative values of language models (e.g., actually looking at neurons). It is accessible as it doesn’t require significant computation, and is extremely promising as a research direction, although there has been minimal success to date.
Adept scaled up its Fuyu architecture and trained a larger version that is currently the 3rd best model on MMMU. The model also performs well on other tasks (e.g., achieving 40+ on HumanEval coding benchmark). There were fewer challenges with scaling this architecture than alternatives, in part due to its simplicity. However, there were still challenges with data i/o, memory bandwidth, and custom kernel development.
HEDNet is a new encoder-decoder network designed to enhance 3D object detection in autonomous vehicles, specifically addressing the challenge of sparse point distribution in 3D scenes.
The Song Describer dataset features over a thousand human-written descriptions of music recordings. This new dataset aids in evaluating music-and-language models through tasks like music captioning and text-to-music generation.
This project presents a novel prompt pool method for dialog state tracking that adapts to new user requirements without needing task identities during testing.
DiffMoog is a new audio technology that combines the versatility of a modular synthesizer with the power of neural networks. It allows for automated sound matching and custom sound creation, making it a valuable tool for audio synthesis.
A great little utility that allows you to use dictionaries with tensors in PyTorch. It improves readability and usability of code while maintaining performance.
The Turing Test and Searle's Chinese Room Thought Experiment are often cited in the discussion of AI consciousness, with the former testing if an AI's behavior is indistinguishable from a human's and the latter arguing that external behavior alone cannot prove consciousness. This debate highlights the difficulty in defining and recognizing consciousness in AI, as our understanding is primarily based on human experiences and functionalist approaches.
This article discusses evaluation metrics that don't rely on ground truth data. It explores general-purpose as well as RAG, chatbot, and summarization-specific evaluations. All metrics come with a Python implementation.
Apple is intensifying its focus on AI for the iPhone, acquiring numerous AI companies, and increasing AI job postings.
BreezeML is an AI governance platform that enables compliance teams to specify and continually monitor governance policies over every AI workflow in its organization without relying on manual and tedious coordination with data science teams.
Researchers have explored asynchronous Local-SGD (a variant of federated averaging) for training language models.
Crypto investment firm HashKey Capital has partnered with London Stock Exchange subsidiary FTSE Russell to introduce three indices that track the performance of crypto assets. These are meant to guide investors seeking to gain exposure to cryptocurrency. The indices include the top 20 cryptocurrencies by market capitalization, infrastructure-focused tokens, and tokens for crypto apps.
Telefónica is partnering with Nova Labs to extend mobile coverage in Mexico City and Oaxaca using Helium Mobile Hotspots. The technology, based on the OpenWifi standard, allows select Movistar customers access to the Helium Mobile Network using their existing SIM cards. The solution is expected to reduce infrastructure costs and environmental impact while maintaining high-quality service and expanding coverage.
Solana has introduced a concept called token extensions, a development allowing for specialized functions at the token level. Created in partnership with regulated institutions, these plug-and-play tools are designed to reduce engineering efforts in tokenizing assets on Solana. Extensions already used by Paxos and GMO Trust include permanent delegation and metadata pointers for their stablecoin launches. Tools like these enable streamlined KYC verification, transferring hooks, transaction fees, and more, boosting Solana's appeal to developers and businesses alike.
Blockchain project Noble has joined forces with asset manager Hashnote to roll out a tokenized real-world asset product on Cosmos. The initiative's first product is the USYC token, a short-duration yield fund product from Hashnote that provides exposure to short-duration securities, including US Treasury bills. This marks Noble's move into the real-world asset niche on Cosmos.
This thread provides some tips for staying ahead in the crypto market. The tips include regularly checking crypto price charts to identify trends, journaling trade rationale and outcomes, understanding market trends and structures, tracking smart money, looking into pre-launch projects, monitoring relevant data changes, and staying up-to-date through trusted curators. The author also recommends going through governance proposals daily to keep up-to-speed with potential catalysts for coins.
This post explains how web3 offers an alternative to Big Tech data dominance, allowing users to own their data. Platforms like Zora let users mint digital artwork, paying a small fee to creators. The collector owns a copy of the work in their portfolio, which in a sense creates a dataset that they have full control over. This could open up the potential for those in the space to leverage AI/ML on top of these open datasets to monetize them further or allow for better discovery of art and artists.
This post breaks down what the most popular narratives were on Crypto Twitter last year. The dominant narrative of 2023 was AI, with almost a third of the ‘mindshare’ throughout the year. Ordinals, GameFi, and Layer 2s were all also notable themes last year, with mindshares in the high single digits. New narratives like ETFs and DePIN gained momentum in late 2023, while SocialFi had a short surge in August. As of this month, ETFs remain a focal point, with AI and DePIN also gaining momentum. On the other hand, GameFi and SocialFi have lost some mindshare.
The Bitwise Bitcoin ETF (BITB) is the first US ETF to publicly reveal its Bitcoin addresses, aiming to enhance transparency and align with Bitcoin's ethos.
This post explains how for most of crypto’s history, builders have been infrastructure-focused, but there's been a shift towards app development as of late, with value increasingly captured at this layer.
Brink's, a 164-year-old security and cash handling firm, has invested in BitGo, one of the largest cryptocurrency custody specialist firms in the space.
This post explains how parallel transaction processing, a type of transaction processing that has gained popularity with increased user adoption, enhances scalability and user experience, overcoming the limitations of sequential execution models.
These experimental AI features enable users to organize their tabs, create custom themes, and get help writing on the web. Chrome users can enable these experimental features in settings, but because these features are early public experiments, they’ll be disabled for enterprise and educational accounts for now.
Lumiere is a new text-to-video diffusion model from Google AI that creates realistic, diverse, and coherent motion in videos. Instead of spatially distant keyframes followed by temporal super-resolution, Lumiere's architecture generates the entire video duration in a single pass, enhancing motion fluidity and coherence. Examples of early animations show remarkably smooth camera movements and complex object animations that span multiple seconds.
Kittl raised another $36 million just a year after raising $11.6 million during a Series B round led by IVP and including some existing investors. The startup intends to become the go-to tool for creating logos through its built-in illustration library and collection of copyright-free photos, t-shirt designs, cards, social media posts, and other media. It recently added AI-based features like AI background removal and image generation using Stability AI's and OpenAI's APIs as well as its own in-house technology.
Lessons learned during auditing a bilingual application for a public sector organization in Wales. This article contains sixteen specific accessibility insights, ranging from making context explicit for assistive technology users to considerations in design, such as using color effectively for interactivity and ensuring compliance with guidelines like GDPR in web development.