I wonder if the current modding explosion is downstream of the same thing happening with video games in a year or two. I'm perpetually like 1 particularly dull week away from making Saga of the Void: Commanders or something.

Alignment research has yielded us the "helpful assistant <-> spiralist sage" axis, but what about the "sheriff <-> bandit" axis.

Samuel@samuel.fm · last mo.

when claude says "the smoking gun"

Increasingly amenable to the idea of being fully automated if it means I don't have to spend another stupid evening bisecting a bloody obscure, undebuggable cuda issue.

How coding feels recently.

Ive really been mulling this recently that the current bottleneck is probably less intelligence, and more coordination, once agents can organise like humans can, there will be no need to become hyperintelligent to achieve significant gains, just add more agents.

mr. TIM@timkellogg.me · 6mo ago

everyone is convinced continual learning will be in-weights, changing the model but: 1. Dario says large context alone is enough 2. OpenAI “acquires” OpenClaw we’ve been here before. The public assumed that reasoning models used complex search like MCTS, but DeepSeek showed that was a distraction

Finally found a use for my old Raspberry Pi 4, now to figure out what to do with it that doesn't involve giving it access to my Google account.

Moving to ML, I mostly learnt python, given that LLMs at the time were mostly usable for boilerplate and telling it *explicitly* what to change in a script. Now I watch Claude Code smash through problems that even I struggle with, and I wonder if I'd have learnt what I did if I'd moved recently.

Nano banana pro is insane, nothing is even close, the in context capabilities are bar none.

uhhh

Steam-oyua

My experience today #o3pro

I love how people are all like "more money should be spent on game development and the people making them should be paid more", then freak out when the cost of games roughly tracks inflation, despite the scope of games increasing over time, where did you think the money was going to come from?

I'm shocked how much i like openAIs o3, the improvement is near gpt-3 -> gpt-4 for me, it's better at everything, coding with it is much less clunky and the web search is well integrated to the point where it feels natural.

why does tagging anything with florence2 always create a perfect list of people, faces, occasional objects and always, always ... shoes? Like 50% of the output tags are always shoes or feet for some reason.

Thrilled to share that Geometry Image Diffusion has been accepted to #ICLR2025! 🚀 Paper: openreview.net/forum?id=Glm...

Geometry Image Diffusion: Fast and Data-Efficient Text-to-3D with...

Generating high-quality 3D objects from textual descriptions remains a challenging problem due to high computational costs, the scarcity of 3D data, and the complexity of 3D representations. We...

openreview.net

Slava Elizarov@selizarov.bsky.social · 2y ago

Does 3D generation always have to be either slow or complex and data-hungry?🤔 We don’t think so! With Geometry Image Diffusion, we’re all about reusing (and recycling ♻️) what already works — making it faster and easier by reducing complexity and data needs 🚀(1/10)

Have you ever wanted to use a version of stable diffusion that had recursive cross attention layers so you could use multiple prompts in a way that didn't actually improve prompt understanding. No? Well here you go anyway: github.com/CiaraStrawbe...

Genuinely in love with the steam deck as a platform, got one for travelling a few months ago, but i do a large fraction of my gaming on it now, it's like the difference between the vive headset and the quest, it's just so much less effort to do anything.

Does 3D generation always have to be either slow or complex and data-hungry?🤔 We don’t think so! With Geometry Image Diffusion, we’re all about reusing (and recycling ♻️) what already works — making it faster and easier by reducing complexity and data needs 🚀(1/10)

Going to have to choose between a 5090 or a single openai pro subscription for a few months at this rate.

Introducing my new work: Stylecodes! An open source sref implementation for sd! Paper here: ciarastrawberry.github.io/stylecodes.g... Code here: github.com/CiaraStrawbe... Stylistic control using only an easily sharable 20 digit base64 code! #stablediffusion #midjourney

I'm gonna move to posting here from now on I think for the most part, have some interesting stuff happening soon.