prettyblocks

Which do you suggest?

I liked it too. It was full of tropes in a cute way that I enjoy from contemporary sci-fi, and it was pretty sick to go see in Dolby with the haptic seats.

Good to know!

I guess I just wanted to build my own, because my use case was really narrow and I knew it wouldn't take a lot of time. Guaranteed to not have any ads and trackers. Also needed an excuse to fully take github copilot's agentic coding for a spin.

I vibe-coded myself a flashcard app for latin phrases that I've used quite a lot. https://github.com/eliyastein/latin-learner

I'm on an M2 with 24GB ram and it feels like it flies as fast as ever.

Philip K Dick's worst nightmares are coming true.

Came to say exactly this. It's my favorite, but I still feel like I haven't read it after like 5 reads.

It's not meaningless to the folks collecting fees for securing the chain.

Blockchains might not have hundreds of millions of users, but popular L1s have volumes in the many billions of dollars.

NYC is ramping up on this as well.

Felt like counter culture to me when I went to my first one (DC11). I remember punk kids selling manuals and lineman sets they stole out of the back of telco trucks outside the entrance of Alexis Park.

Local inference is becoming completely commoditized imo. These days even docker has a local models you can launch with a single click (or command).

The license looks quite permissive though as long as you're not using n8n as the platform for your business. According to the license, if you just wrap/productize the flows you create and users don't bring their own n8n credentials or API keys, then you're good. It seems like there's a ton you can build off of it and still stay within the parameters of the license.

Given how well it seems to be going in those specific areas, it seems like it's more of a regulatory issue than a technological one.

You can build a pretty good PC with a used 3090 for that budget. It will outperform anything else in terms of speed. Otherwise, you can get something like an m4 pro mac with 48gb ram.

It's a masterpiece. Surprised it hasn't been adapted as a film or series.

Interesting. I run my local llms through ollama and it's zero trouble to get that working in openwebui as long as the ollama server is running.

I've been using openwebui and am pretty happy with it. Why do you like lm studio more?

Not a lawyer, but I think intent is a big part of legality and I hope that doing something like this for research purposes should be relatively safe.

I did lose weight, but it wasn't crazy dramatic.

I had moderate NFLD and managed to get rid of it completely by just eliminating fried food, most dairy, and sugary snacks for about 6 months, I also almost never eat red meat. It's not so much about being overweight, but what you're eating.

Royal Tenenbaums & Rushmore have always been my favorite, the way they hit every single emotional chord without being overcooked, and with characters that are relatable.

Yeah, I'm not contesting that at all. I think Telegram sold its soul a long time ago, so I kind of just treat it as a self-contained socially oriented splinter-net with its own integrated app environment... and it's actually really cool and feature rich. I also find Grok to be an LLM that performs pretty well, so I don't mind the integration, much the same way I didn't mind the AI integrations into WhatsApp, but I understand that many people just want a messenger that's a messenger so I can understand the pushback.

As a Telegram user, I don't mind having an interface to Grok that is outside of X.

I've been trying to maintain a (mostly vibe-coded) zsh/omz plugin for tab completions on your LLM cli and the rate at which you release new features makes it tough to keep up!

Fortunately this gets me 90% of the way there:

llm -f README.md -f llm.plugin.zsh -f completions/_llm -f https://simonwillison.net/2025/May/27/llm-tools/ "implement tab completions for the new tool plugins feature"

My repo is here:

https://github.com/eliyastein/llm-zsh-plugin

And again, it's a bit of a mess, because I'm trying to get as many options and their flags as I can. I wouldn't mind if anyone has any feedback for me.

I'm curious how effective these models would be at recognizing if the input video was ai generated or heavily manipulated. Also various things around face/object segmentation.

I'm very interested in working with video inputs, is it possible to do that with Qwen2.5-Omni and Ollama?

It could very well be a brazen scam.

Of course you can say that when people are obviously copy and pasting AI slop, but if someone used AI in order to find a valid security issue, and reported it effectively, how would they know that AI was even involved?