What's with ChatGPT4 using Python scripts that take several seconds to evaluate simple calculations with generated code snippets? Has anyone else noticed that?

Open Source in one tweet.

Keyboard for scale 😅

Team OpenFaaS Ltd is getting ready for a long Easter weekend 🐇 and doing some final testing of the 🦊 GitLab for actuated support. Reach out if you'd like easier maintenance and stronger security when using Docker.

actuated.dev/blog/secure-mi…

Ivan's nailed the user experience for labs launched from the CLI. Here's an OpenFaaS CE session deployed into the multi-node K3s lab.

Interesting that my GitHub Sponsors Portal running on faasd has become a target for scammers and bots. Not a huge amount of traffic, but interesting to see what they're trying to find.

github.com/openfaas/faasd

Ran an SSH debug session with actuated and asked llama2 why GPUs might be needed in CI, and what gaps we're leaving when we omit testing

The answer from 13b was much better than I've seen with self-hosted models in the past

What other models should we be testing?

Run - explore and create AI models with ollama and GitHub Actions - accelerated with a GPU with actuated

#ai #ml #llama2 #cicd #e2etesting
actuated.dev/blog/ollama-in…

Guess where all those Arm builds are running? actuated 🤗

'E2E tests are flaky'

Yep, often true.

But, E2E tests aren't inherently flaky. They're flaky for a *reason*:

🚩 The UI is buggy.
🚩 The backend is often down.
🚩 The DB is constantly changing.
🚩 The network is slow or unreliable.

E2E tests aren’t flaky. The environment is.…

We could use some of that sunshine in the UK Craig Box 😅

Congrats on ambient mode !

This does check out. Very high energy efficiency on the Ampere chips.