Darren Tarbard

I've been making AI models and one of my latest is specially trained to think it's a real person rather than being a "helpful assistant". It's small enough to run on your own computer. See my other models and consider supporting/hiring me at gurubotai.com

@krunkosaurus nice, that's a good speed you are getting, similar to a discrete GPU but you have way more RAM than most GPUs.

@tarbard 4bit which is 90-95% as accurate as full quant

Apple M5 Max Macbook Pro with 128GB ram arrived. First unoptimized chat with Qwen3.5 is 108 TPS! Local AI is here.

Someone made a game in which you have to program a drone using a simple python-like language to fully automate various farming tasks

@TheBobPony "vibe code purple" is an interesting choice of colour

Firefox removed the fox. 😢

@SkunkWorks0x @mattshumer_ is openclaw stateless though? I would have thought it has a "memory" system

The agent doesn’t know it messed up. It has no memory of what the last run cost so it just kept doing the same thing. Norm isn’t scared. Norm has no idea anything happened. That’s the real problem with stateless agents. They can’t learn from their own mistakes because they don’t remember making them.

My OpenClaw knows he fucked up and is too scared to respond :(

To try a pre-configured version of self: After Dark on your computer install ollama from ollama.com/download and then run the command "ollama run gurubot/girl"

I've been making AI models and one of my latest is specially trained to think it's a real person rather than being a "helpful assistant". It's small enough to run on your own computer. See my other models and consider supporting/hiring me at gurubotai.com

This gives natural fibers a whole new meaning!!! Textile made from plant roots, that are truly natural. Zena Holloway is spearheading this idea and I think she is on to something. As you can see for now she is using wheat to germinate and root in very particular pattern.

@mmitchell_ai sometimes I give it instructions in pseudo code

I've been trying out "vibecoding", but, Q: Does anyone else get tired of having to *talk* in order to code? I like to program to program, not to have to say stuff.

This is not a real company

Current RAM situation

Instead of port numbers we should have port names. http://myserver:jellyfin http://myserver:immich

@johnrushx I've been doing a lot around this recently and it is constantly making me think about how humans cope with memory issues too, there are some parallels that can guide the LLM solutions but it's still tricky.

I’m feeling so pessimistic about AI today. After building dozens of AI agents for many years, here is what I got to say: The biggest challenge in building AI agents is agent’s long term memory. LLMs with huge context windows are pretty much scams, cuz they compress the text under the hood making it completely unreliable for any serious task (the error accumulation and forward propagation when run in an agentic loop ). Existing LLMs would be capable of replacing humans if only the memory problem was really solved. But..nobody has come even close to solving this yet, and it may take a decade until this is solved. I won’t even be surprised if it turns out that our general intelligence comes from our insanely amazing memory engine in our brains coupled with the logical apparatus. Right now the LLM can only apply logic, but it can’t memorize things as our brain does, therefore we’re very far from AGi or even a basic 100% autonomous ai agent. Unfortunately :( But, I have an idea to solve this at least for business ai agents. I’ll present my solution soon in my next ai agents release.

oh you're a full stack engineer? What's this then ?

@maximelabonne Nice. for anyone wondering these work fine in Ollama.

We're releasing a collection of tiny task-specific models 🥳 Want to do data extraction, translation, RAG, tool use, or math on a Raspberry Pi? We got you covered! ✅ Here are a few examples ↓

It's time to replace the t-shirt slogan "Go away or i'll replace you with a very small shell script" with "Go away or i'll replace you with a prompt"

The "Thinnest iPhone Ever" be like:

@ing_bue @fishPointer What do you have the big buttons mapped to?

@fishPointer do you need it? probaly not will it make CAD / 3D work more enjoyable? absolutely yes I’ve used this one for more than a deade, I prefer the small standalone model, to maintain close keyboard proximity, added the extra mechanical because I could

should i get one or are these a meme

@FrameworkPuter @Neolker @Prusa3D yep I have more confidence that my bambu x1c will start printing when something is sent to it than my HP printer.

@Neolker Our @Prusa3D printers are much, much more reliable than our 2D printer!

One of you must have sabotaged our office printer to force us to make one. There is no way HP is actually shipping printers that are this broken.

@RedmondAI @arrakis_ai well, he did say "insane"

@arrakis_ai I'm confused, didn't everything fail at the end?

Google AI Studio is absolutely insane… I uploaded a video and asked it to implement the features, and within just 5 minutes it built an AI startup for me