Is ‘data-driven’ just a mantra or a reality for most businesses?
In a 451 Research survey of 679 business and IT decision-makers, most respondents claimed to make most of their decisions based on data. But as data sources become increasingly distributed, access to data can become an uphill battle. Businesses are inundated with applications which often don’t sufficiently integrate with one another.
Tesla is developing a revamped version of the Model 3. Project Highland aims to reduce the number of components and complexity in the vehicle. The vehicle will go into production in Tesla's Shanghai and Fremont plants. While the Model 3 still looks the same as when it launched, Tesla's software updates have led to improved performance over the years.
Protests and Covid lockdowns at Apple's key assembly plant in China are causing long delays on iPhone 14 Pro models. US customers can now expect to wait as long as 37 days for a device. Apple plans to catch up on production in 2023 to make up for orders lost this quarter. Wait times for iPhone 14 handsets remain at less than three days.
DLSS 3 (Deep Learning Super Sampling) combines AI-powered image upscaling and optical multiframe generation. It combines sequential frames with an optical flow field used to predict changes between frames, then it slots AI-generated frames between traditionally rendered frames. DLSS can speed up performance by over 500% in some games. A video with a side-by-side comparison of DLSS 3-rendered graphics and normal graphics is available in the article.
Scientists from CNRS and the University of Tokyo have developed an enzyme-based technique that provides clues on how to read and manipulate data written on DNA. The enzymes react to create chemical neurons that reproduce the ability for complex calculations exhibited by true neurons. These chemical neurons can execute calculations with data on DNA and express the results as fluorescent signals. The technique could result in better screening for certain diseases.
TorchScale is a PyTorch library for scaling Transformers. It can improve modeling generality, capability, stability, and efficiency. TorchScale can scale Transformers to 1,000 layers and beyond. It supports arbitrary depths and widths.
JupyterLite is a JupyterLab distribution that runs entirely in the browser. It features a Python kernel running in a Web Worker and JavaScript and P5.js kernels running in an IFrame. Users can edit, save, and download Notebooks from the browser's IndexDB.
Authentication is critical. Getting it right starts with understanding how different types of applications manage login and authentication. This technical guide by FusionAuth covers the key differences between traditional, single page, and mobile app authentication, with detailed workflows for Native Login and OAuth 2. Download the guide for free
This article contains four and a half lessons from a developer who has been laid off six times in their life. Most of the time, being fired has nothing to do with competence. Choosing the right place to work will reduce a lot of stress and worry. Don't be afraid to leave a job if it isn't working out. Try to do your best and make sure you keep life in perspective.
OpenAI recently released text-davinci-003, a GPT-3 model that improves on a number of behaviors that are important to developers. It produces higher quality writing, can handle more complex instructions, and is better at longer form content generation than previous models. This Twitter thread looks at some of the issues that are still present in the model and provides some examples of what the model does better than its predecessor. The model still has issues with written math problems, but it is better at writing poetry and job descriptions.