Theodor Hillmann

To develop Skills or anything for claude code sucks because it cannot instantiate itself anymore – since a while – Its so damn annoying if you want it to eval a copy of it self you need kind of a complex system @bcherny

@hiaaryan So great!

a few days ago, i promised something. introducing → Sileo ✨ an opinionated physics based toast library, with SVG filters and performance optimisations. try it here: sileo.aaryan.design happy sunday ✌️

it runs on your machine - a old mac in some corner, full access to local files and network. letting an AI run autonomously on your filesystem on a timer is a little unhinged. but it's also kind of the point.  github.com/t0dorakis/murm…

a prompt in a file on a timer. that's the whole thing. i built a small daemon called murmur. you write a prompt in a markdown file, give it a cron schedule, it runs claude code on repeat. no platform, no orchestration. a file in your repo, version controlled, next to your code.

A modern day pest 🥸👇🏻

@sama Yeah @sama you really nailed the starbucks aesthetic 🎖️

i try not to think about competitors too much, but i cannot stop thinking about the aesthetic difference between openai and google

@Yampeleg ASML sold less Machines then expected and they lowered their outlook thats what dragged everything down. Also the market prices in lower interest rates for longer and people are cutting risk by selling highflyers like Nvidia

dafaq is going on with compute stocks?

@t3dotgg Well there are some similarities between Figma and Factorio when „developing“ complex prototypes – a beautiful mess

There are two types of devs: those that like Figma and those that like Factorio

@historyinmemes The beauty of analog film vs encoded. If the new lander would ever return we would get better quality (assuming the lander is a good DoP.

Prompt engineering means exploring textspace until you find an input token sequence that is (1) well represented in the training set and (2) precedes the type of output you’d like to see. Two conditions seem necessary for a good prompt: 1. Low perplexity 2. Nearby (in embedding-space) to emblematic training data Surprising no one has made a prompt eng tool that does embeddings search over the training set of a given model. This would greatly reduce the time it takes to find a good prompt and could be a huge advantage for a particular model.

TL;DR: We use SDS from text-to-video models to animate vector-graphics sketches! Please check our project page for more details, and more penguins: livesketch.github.io

An app that takes in your LinkedIn profile and interests to show you a map of possible future life paths 🗺️ I like to think of it as verse-jumping (from EEAAO) but irl

my friend at openai took the day off to do acid in the park. he’s about to turn his phone back on

Anon Reddit account created today shared this about Sam/@OpenAI situation. Plausible? Posted 5 minutes ago with 0 upvotes. So boring it’s almost believable.

@vandalezza Hahahahaha

blur trick for the optical fit

Business BDSM

Today, Generative AI takes its next big step forward. Introducing Gen-1: a new AI model that uses language and images to generate new videos out of existing ones. Sign up for early research access: bit.ly/3RxvBZr

Google Calendar, but it shows the cost of the meeting