I’m building a Python job queue.
I’ve been obsessed with queues & offline processing since 2011.
In fact, I released an open-source library called Schedulr back then & ran a hosted version of it for a couple of years.
It was basically "Google Cloud Run before Google Cloud Run"
Call me a masochist, but I wanted IND to lose the 2nd Test match as well after the 1st loss - because obviously there were glaring issues. Only when you hit rock bottom do people contemplate. Hope now heads will turn and will find solutions to the problems that the team has atm.
I just love Hugging Face.
Their new 200+ page Training Playbook covers everything: training frameworks, model architecture, data curation, pre/mid/post-training, eval, how GPUs work, latest research, and ablations.
Packed with practical wisdom. I read it like a novel.
Every professional developer should be (and most are) anti-AI pilled:
Non-AI, deterministic solutions are always better than AI-based, non-deterministic solutions.