Hello again.
Links of the week
Consider using Kubernetes ephemeral debug container
I'll admit I am late to learning these exist. About once a month, maybe more, I need to spin up some sort of debug container inside of Kubernetes. It's usually for something trivial like checking networking or making sure DNS isn't being weird and it normally happens inside of our test stack. Up until this point I've been using a conventional deployment, but it turns out there is a better option.
kubectl debug
.
It feels like we have more tools than ever to throw at monitoring but we're not making progress. Instead the focus seems to be on increasing the output of applications to increase the revenue of the companies doing the monitoring. Very little seems to be happening around the idea of transmitting fewer logs and metrics over the wire from the client. I'm running more complicated stacks to capture massive amounts of data in order to use it less and less.
Emerging Architectures for LLM Applications
There are many different ways to build with LLMs, including training models from scratch, fine-tuning open-source models, or using hosted APIs. The stack we’re showing here is based on in-context learning, which is the design pattern we’ve seen the majority of developers start with (and is only possible now with foundation models).
FTC Request, Answered: How Cloud Providers Do Business
The RFI has 20 questions addressing security risks, market power, and business practices. I’m only going to address the ones that I have the most knowledge and experience with, and also are the most interesting to me; that’s what you get when I’m volunteering!
Hugging Face Transformers: Fine-tuning DistilBERT for Binary Classification Tasks
And yes, I could have used the Hugging Face API to select a more powerful model such as BERT, RoBERTa, ELECTRA, MPNET, or ALBERT as my starting point. But I chose DistilBERT for this project due to its lighter memory footprint and its faster inference speed. Compared to its older cousin, DistilBERT’s 66 million parameters make it 40% smaller and 60% faster than BERT-base, all while retaining more than 95% of BERT’s performance.
How to make the Mac better for developers
If I need to join a Zoom call, I don't want to think about whether Zoom is going to work. If someone calls me on Slack, I don't want to deal with quitting and opening it four times to get sound AND video to work. When I need to use third-party commercial software its often non-optional (a customer requires that I use it) and I don't have a ton of time to debug it. With the Mac, commercial apps are just higher quality than Linux. They get a lot more attention internally from companies and they just have a much higher success rate of working.
Book of the Week
Do you have any more links our community should read? Feel free to post them on the comments.
Have a nice week. 😉
Have you read last week's post? Check the archive.