@crazor Now I have a certain MIDI theme stuck in my head… 😆

@leah Thank you for sharing that. I’m editing the post. 🫡

@pl Thank you for sharing that!

@jesse That's not an easter egg, that's gVisor's (sandboxing tool) "kernel log"

https://github.com/google/gvisor/blob/3574bdd750349e806433ee328e6f7cc17d8c6776/pkg/sentry/kernel/syslog.go#L50

ChatGPT's "Advanced Data Analysis" feature runs Python code in a container. When you run “dmesg” inside the environment, you get gVisor’s application kernel messages which initially looked to me like an OpenAI Easter egg or a hallucination.

Just wrapped up a new project: a machine learning classifier for the elusive spaghetti squash! 🍝🍈 Thing did not go exactly as planned. I faced my first challenge with overfitting when led to a deep dive into data cleaning. Check out my journey and try the classifier for yourself! https://jessealter.com/project/squash-classifier/ #DeepLearning #fastai

@tiago_j_m Thanks for checking it out!

My learning approach is already undergoing a transformation, thanks to the fast.ai course. The fast.ai course focuses on getting impressive results right from the first lesson, and gradually builds on foundational topics. As someone who usually likes to go deep and spend a lot of time understanding theory, I've found that approach to be unexpectedly refreshing and energizing!

@radek suggests that as AI continues to progress exponentially, the most practical way to get top-notch results in AI will be to work at the top of the stack, rather than pour all of your focus into low-level details. It is completely at odds with how I've worked and learned over the past decade, but I suspect he's right. https://radekosmulski.com/there-is-something-weird-about-the-current-gen-of-ai-better/

I'm happy to share my first ML project: an OS screencap classifier. Upload a screenshot, and the model will attempt to identify which operating system it came from. https://jessealter.com/post/os-screencap-classifier/