That's the way to go. Anyone who publishes any of their work in the open web and who values what they are doing should do whatever possible to block the AI scrapers (which is often difficult or impossible, I know) until all that's left for them to feed on is their own garbage.
Let the garbage-spewing machines choke on their own garbage
Allowing someone to create my persona which they control sounds like a fever nightmare to me. I'm not touching this with a stick until I can host it - and control it - myself
People don't like reading articles "redacted" by AI because that feels disrespectful
Imagine approaching someone you wanna talk to and they just silently point to their assistant, suggesting you talk to them instead. I want to hear your thoughts from you, not filtered by some assistant/machine
This intrigued me a lot! I have been searching for a way to break out of the social silos, but all the other alternatives just weren't doing it for me. They usually chased thr nostalgia of the 90s for nostalgia's sake, but apart from making me nostalgic for 10 minutes I haven't found much use of them.
This seems different. It embraces the modern web and tries to fit in, while still giving the creator full control over their yard.
I might be overly optimistic, but I'm excited for this! I will try to make my myzopotamia.dev blog join the IndieWeb as soon as I find some time
I'm pretty new to this so I can't speak for IndieWeb, but I wouldn't blame them. People who seek independence from the social silos are usually engineers or other web-aware people. It is only natural that they want to solve the problem with the knowledge they possess and tools they know - and I think it's fine. If you look back at how internet evolved, it was always like this - the geeks and nerds coming up with something fun but naturally hard to get into for the regular folk. But then it would get more and more approachable until it adopted for the said regular folk.
I think the IndieWeb is at this early stage. The adoption for a broader audience will come naturally (if this survives), but for now - let them cook
First of all, techy nerdy people like things to be easy too. They're just somewhat more likely than other people (on average) to overcome not-easiness.
Second of all, how long is IndieWeb supposed to cook for before it's supposed to be ready for a broader audience? This is no shade on the IW folks if they like what they're building. But if what they're building is supposed to catch on somewhat, what's the path supposed to be? The project seems to be 15 years old already: https://indieweb.org/IndieWebCamps#
I'm trying to go the same route, but I have a 5070Ti with only 16GB VRAM (I bought it for gaming) and I'm not sure how to run anything decent on it. I have 64 GB RAM if that matters
I run it on a 12GB 4070 with 32GB system RAM. 35B A4B means only part of the model is active at a time so it takes a lot less VRAM than a dense 35B model would.
The main thing in LM studio (or whatever software you use, assuming it has fairly up to date stuff and exposes the toggles) is to offload MoE layers to the CPU, and use K/V cache quantization at Q8_0 or Q4_0.
Since you have more VRAM than I do, you could probably get away with MoE offload of like 15-20 so some remains on the GPU.
Just make sure GPU offload is turned all the way up. And I use 64k context size, although with 16GB VRAM you can probably do more.
You can find the best performance spot by playing with MoE offload until you find the number that gives the highest tok/s on your hardware.
Thanks for sharing that. I have the same card but 96gb ram. I use PI.dev to connect to LM-Studio. I may have to switch away from LM-studio if I can improve token speed. I think I range from 32-40t/s. qwen3.6-35b-a3b-genesis-v2-apex-mtp.
You can try tweaking MoE offload, I found the sweet spot after a few tries and even changing it by 1 can reduce speed by a few tok/s. I think around 45 is the average I get but sometimes it'll hit 50.
Wasn't there some scientific paper recently that proved that every operation can be represented as a logarithm? Like, the same as every logic gate can be derived from NAND gates
Any tips on which model to use and how to use them? I have 64 RAM and 16 VRAM (I know it's not a lot, it's a gaming GPU) and I'm trying to find a good model to use but it's a bit of a struggle
Let the garbage-spewing machines choke on their own garbage
reply