We just launched our new eBook: Practical MLOps - How to Get Ready For Production Models! This eBook gives an overview of why MLOps matters and how you should think about implementing it as a standard practice! Including chapters from SigOpt & Tecton hubs.ly/H0z3wkH0

Fridays are for building ✨ cool s**t ✨ at Valohai and in honor of the 300+ MLOps tools out there, we chose some of the most Pokémon-sounding ones and built it into a game. Can you catch them all in Pokémon or #MLOps? 👉 valohai.com/mlops-or-pokem…

Super excited to have Mike Del Balso, the CEO & co-founder of Tecton and former product lead of Uber's Michelangelo MLOps platform to join the Valohai advisory board! - Welcome Mike! valohai.com/blog/valohai-m… #mlops #machinelearning #datascience

#GenAI is reshaping creative industries like no other be it graphics or text, less of a "the artist's tool" and more of a full-fledged collaborator 💯 we are _so_ ready to co-create with machines! 👨‍🎤🤖 #AI #aiartist #aiart #ML #flux2 #SD #digitalart #Cursor #LLMs

AMD is stepping up for ML! With streamlined support in libraries like vLLM , AMD hardware now rivals (and often outperforms) NVIDIA in LLM inference. Tried the #MI300X yet? Stay tuned for our inference benchmarks tomorrow!

do llms suffer from analysis paralysis? if a chain-of-thought model reflects on each step of the way, in which situations will at actually end up in worse conclusions or how many steps is too much self-reflection🤔 #LLMs #GenAI #ML #AI

when do we get our first organic consulting companies that vouch no to use AI to produce code or content 🤔🌱

a good writeup on both agentic RAGs and RAGs in general "vanilla RAG is like going to the library to get more information, agentic RAG is like having a smartphone to do it" 📱

A Comprehensive Survey of Small Language Models Nice survey on small language models (SLMs) and discussion on issues related to definitions, applications, enhancements, reliability, and more. arxiv.org/abs/2411.03350

all this effort on LLM evals will contribute to making neural interfaces a thing why prompt engineer anything if you can plug a conversational model to read YOUR chain-of-thought 😀

attempting to unlock your phone with a thumb coated in stubborn superglue 😥 ...hardships of the miniature hobbyist

each time I see this guy happily trekking forward, I get sus and refresh the page a couple of times to make sure some part of GitHub isn't down

TIL (after debugging a CI pipeline for hours): tracemalloc is apparently _significantly_ more performance intensive on #Python 3.11+ than on earlier versions, so don't enable it unless you need to.

heads up, Pythonistas!