Meta unveils a new large language model that can run on a single GPU (3 minute read)
Meta has announced a new large language model called LLaMA-13B that is 10x smaller than OpenAI's GPT-3 model but with reportedly better performance. LLaMA models range from 7 billion to 65 billion parameters in size, while GPT-3 has 175 billion parameters. Training of the LLaMA models was done using publicly available datasets. Meta hopes that LLaMA can be used for natural language research and to power applications such as question answering, natural language understanding, and reading comprehension. LLaMA-13B could pave the way for access to ChatGPT-like performance on consumer-level hardware. |
Apple's Secret 'XDG' Team Is Working on More Than Just a Glucose Monitor (20 minute read)
Apple is renowned for its secrecy, but the company's Exploratory Design Group (XDG) is even more mysterious. The XDG is a startup-like operation within Apple made up of only a few hundred engineers and academics. It is well-funded so that staff are able to explore countless ideas. The group is currently researching glucose monitoring, new display technology, artificial intelligence, and AR/VR headset features to help people with eye diseases. This article explores XDG's structure and operations and takes a look at the technologies it is developing. |
Science & Futuristic Technology |
Planning for AGI and beyond (7 minute read)
OpenAI's mission is to ensure that artificial general intelligence (AGI) benefits all of humanity. AGI could increase abundance, turbocharge the global economy, and aid in the discovery of new scientific knowledge. It has the potential to give everyone incredible new capabilities and help with almost any cognitive task. However, AGI also has serious risks of misuse, mistakes, and societal disruption. OpenAI believes it is not possible or desirable to stop AGI development and the company is exploring how to get it right as we transition into a world with superintelligence. |
SpaceX unveils next-gen Starlink V2 Mini satellites ahead of Monday launch (4 minute read)
SpaceX has announced that it will launch its next-generation Starlink V2 Mini satellites as early as February 27th. The mission, named Starlink 6-1, will carry 21 Starlink V2 satellites into low Earth orbit. The satellites will operate under SpaceX's Starlink Gen2 FCC license, which allows SpaceX to launch up to 7,500 of a nominal 29,998 satellites. Starlink V2 satellites reportedly offer almost 10 times more bandwidth than V1.5 satellites. |
Programming, Design & Data Science |
The Missing Semester of Your CS Education (Website)
Computer Science classes often cover topics such as operating systems and machine learning, but they rarely teach proficiency with tools. This course gives students the opportunity to become proficient in the command line, text editor, version control systems, and more. The mastery of tools not only saves time but also opens up possibilities to solve complex problems. By investing in their tool proficiency, students can make the most of their education and career. |
Konsta UI (GitHub Repo)
Konsta UI is a library of pixel-perfect mobile UI components built with Tailwind CSS. It includes iOS and Material Design components for React, Vue, and Svelte. All components are created using official design guidelines to ensure a native-like look and feel. Konsta UI is designed to be used with parent frameworks such as Ionic and Framework7. |
Meet the $10,000 Nvidia chip powering the race for A.I. (7 minute read)
The Nvidia A100 is a roughly $10,000 chip that has become a critical tool in the artificial intelligence industry. It is capable of performing many simple calculations simultaneously, making it ideal for training and using neural network models. Originally designed for 3D graphics in gaming, the A100 is now configured and targeted at machine learning tasks. Nvidia, which has 95% of the market share for graphics processors that can be used for machine learning, has seen its shares rise 65% so far in 2023. The company introduced the H100, an optimized successor of the A100, in 2022. The H100 is the first data center GPU to be optimized for transformers. |
Report: More Twitter drama after Slack shutdown; employees play hooky (3 minute read)
Twitter's internal Slack channels and Jira tracking software were suddenly shut down on Wednesday. A Twitter employee had speculated that the company may have stopped paying their Slack bills. Twitter told employees that the Slack channel was down for 'routine maintenance'. A Slack spokesperson confirmed that Slack did not deactivate Twitter's workspace or user accounts. |
LLM security (GitHub Repo)
A repository that demonstrates the consequences of giving large language models interfaces to other applications. |
All in Bits - Sr. Backend Engineer (Fully Remote, $130,000 - $250,000)
All in Bits is hiring a Senior Backend Engineer to define requirements, iterate on software solutions, and deliver useful developer APIs for a company that empowers builders to launch sovereign blockchains and decentralized applications that are secure, scalable, and interoperable. |
If your company is interested in reaching an audience of tech executives, decision-makers and engineers, you may want to advertise with us.
If you have any comments or feedback, just respond to this email!
Thanks for reading,
Dan
|
|
| |
No comments:
Post a Comment