Eris

Eris

@erisaurus.bsky.social

I’m into Data, Machine Learning, TTRPGs, Magic the Gathering, and art! 🚫NFTs and AI "art" - I did machine learning before it was cool 😭

535 followers1.5K following22 posts

Pinned

My biggest pet-peeve in the current tech climate is how the term “AI” has taken over and now everything is simultaneously both AI and also just another garbage LLM

Eris reposted

I wanted a way to explore Unicode by visual similarity, not just by name or codepoint, so I built Charcutrie. It lets you browse characters that look alike, search across scripts and symbols, and even sketch a shape to find matching glyphs. (pretty badly for now :D) charcuterie.elastiq.ch#U+221E

Eris reposted

Today I announce Cantrip: On summoning entities from language in circles. In this book I unify the paradigm behind base models, chatbots, coding agents, RLMs, and RL agents, through the metaphor of magic. Code is provided. deepfates.com/cantrip

deepfates.comCantrip: On summoning entities from language in circles | deepfatesWith Cantrip, deepfates reimagines the fundamentals of language model agents. Available as a ghost library with generative test specification.

Eris reposted

One day the industry will recognize the drawbacks of AI agents and nondeterministic automation, and rediscover the UNIX philosophy of chaining together small purpose built tools in a low cost and predictable way, otherwise known as shell scripts.

Eris reposted

MUST READ thread before Monday night's forecasted G4 / SEVERE geomagnetic storm. Auroras could be seen as far south as the U.S.-Mexico border at times. Widespread mid-latitude displays are possible. Here's what you need to know to have a successful aurora chase tonight! 🧵 #heliophysics #aurora

Eris reposted

Agreed on what the actual ****. This is reflected of a pretty fundamental failure to understand what historians do, because LLMs do basically none of my job and are completely incapable of most of it. But they have the ears of the people that write the checks, so this is real bad news. 1/

Brendan Davey@brendandavey.bsky.social

Yo, historians, what the actual ****? I've seen how LLMs handle "history." This is... not good. 🗃️

Eris reposted

Even if the technocrats were right about AI replacing all jobs (they’re not), notice they never pitch any vision for what happens to humans. “No one has to work anymore” sounds great, but the only version of it where we’re not sleeping under bridges is one where these guys pay their fucking taxes

Eris reposted

I have been trying to formulate some rules for giving good talks and fundamentally it’s this: tell a story. People like stories, and they will listen to you if they trust you are telling them.

You'd think the DeepSeek-replacing-OpenAI situation was kinda obvious - I don't think there's ever been a company who's "bottled lightning" and thought to themselves "oh yeah no one could ever replicate that" and had it work out well. #ai #machine-learning

Eris reposted

LLMs aren't interesting to me as a logician because there's hardly anything to mathematically prove. "This seems to work well enough, when trained on this corpus, but sometimes it doesn't". Okay.

I love how the deepseek model is out and open source instead of this garbage closed-source proprietary approach that OpenAI has taken but it doesn’t change how fundamentally limited LLMs are at solving real problems

My biggest pet-peeve in the current tech climate is how the term “AI” has taken over and now everything is simultaneously both AI and also just another garbage LLM