🔗 Stories, Tutorials & Articles

ai.googleblog.com
Language models have demonstrated remarkable performance on a variety of natural language tasks. Solving mathematical and scientific questions requires a combination of skills, including correctly parsing a question with natural language and mathematical notation.
This post by Google AI Team shows that by focusing on collecting training data relevant for quantitative reasoning problems, training models at scale, and employing best-in-class inference techniques, we achieve significant performance gains. Minerva combines several techniques, including few-shot prompting, chain of thought or scratchpad prompting, to achieve state-of-the-art performance on STEM reasoning tasks.

thisisimportant.net
Technical writers often fall into a “just get it done” mindset, or what Shreyas Doshi calls project thinking. Project thinking focuses on the steps and process for creating a product, Doshi says.
Doshi defines product thinking as follows in a Twitter thread:
“Product Thinking is about understanding motivations, conceiving solutions, simulating their effects, and picking a path based on the effects you want to create.”
By applying product thinking to documentation, we can write more useful, relevant, high-quality documentation. Embracing product thinking lets us write higher-quality documentation that is relevant to customers.

www.computerworld.com
Sometimes, you need just the basics from a smaller cloud provider for your SMB needs.

posthog.com
These questions are direct, but a company that reacts badly to them may not be a good place to work. There are also a lot of questions here - think of them as themes, and you don't need to ask them all. Prioritize based on what you hear through the process.

architecturenotes.co
This post covers internal workings of indexes and transactions of RDBMSs.

itsfoss.com
Linux and its toolchain keep on evolving. Here are a few popular Linux commands that have been deprecated or going to be deprecated soon.

slack.engineering
For years, engineers at Slack isolated and tested their changes by running microcosms of the Slack application on their local computers. This was difficult for many reasons: it involved installing and maintaining local dependencies, handling resource intensive software, and writing custom scripts that must work across different operating systems.
The developer productivity team noticed this pain through their user surveys and metrics, and desperately wanted to solve the problem. The team explored other ways to handle this.

fasterthanli.me
Instead of buying a f***-you CPU (like a Threadripper, or something more consumery like the latest Ryzens), maybe you rent a big cloud machine that you can turn on and off as needed. Just for the big stuff.

aws.amazon.com
This post introduces two main approaches: data parallelization and model parallelization using Amazon SageMaker, and discuss their pros and cons.

xeiaso.net
Xe Iaso, the Archmage of Infrastructure at Tailscale, talks about how static analysis helps you engineer more reliable systems.

www.raymondcamden.com
An example of caching for Netlify with an Eleventy site