Builder of thingz, lover of old robust tech like unix, prolog, lisp, APL, forth, postgresql, but don't mind the shiny new things ... right now, I'm all in on GenAI, Local LLMs & LLM driven agents, team llama.cpp, building LLM agents since 2023
It's as popular as you make it. Nothing is stopping you from using it. People generate code for what they are familiar with or have desire for. There are people who have wanted to give rust a try and AI opens that possibility up. There are certainly less people that want to try Haskell. Furthermore, Rust has more training data so LLMs generally will provide better Rust code than Haskell code. Because of LLM, I'm back into obscure languages that I didn't have the mental capacity to carry day by day. Prolog, Lisp and Forth, and like you said, they are very terse and token efficient. However, I must point out that terse was only relevant prior to the entire thinking era, a model could use 100k tokens to think only to give you 100 tokens of final output.
I wish for US to win the Chinese race if there's one. I would agree that the Chinese getting access to GPUs, building models, making them open weight and also sharing has really pushed the American labs. Without Chinese models, we would probably be 1 year behind where we currently are. Competition is a healthy thing. This competition has increased intelligence so fast and created ridiculous demand on hardware. So it's been great for Nvidia and their stock. I don't know what Jensen knows nor do I believe he could have anticipated how the field would play out. Whatever happens, it's been one heckuva ride for him. Unfortunately there are many people that would love to get into his position, the pay is going to be delicious...
I'm not using octominer MB or CPU. That MB on my octominer case is running everything, it's an epyc cpu with 512gb of dram. octominer is also pcie limited, i get faster pcie lane with this.
Yes, there are modern use cases for it. I use it all the time. You can use it for almost everything. See SWIProlog, business backends, web applications, GUI too, you are limited by your imagination.
What makes it small? Anyone know the size? I don't see it anywhere. Small in relation to their current line up? Anyways, last I tried Meta was llama4. I stepped aside not because of how bad llama4 got, but also because of Zuck's move. Anyone here using Meta? I don't know anyone in my circle using it, but I'm sure folks are using it.
"They're almost certainly worth much less than Anthropic" you type that with so much conviction. I have no dog in this fight, but I'm not so such. We have seen OpenAI provide models in Audio, Image, Video, etc. Perhaps it's for publicity, but their models have tackled difficult math problems and those that I know that use their top models are just as happy with it as they are with Opus 5.5. IMO, I think they are neck and neck.
He never made claim that English is kinda a pro-drop language. His argument that it's easy to make a binary statement that Spanish is a pro-drop language and English is not, but reality is messy. He shows counter example in English, might not be common, nothing is stopping English from evolving like that. No matter the rules, trying to describe natural languages with rules always breaks down, the exception just grows. I think his argument was well presented, we can't brush off probabilistic models because we don't understand it all when they work so well. This essay is 15 years old way before the LLMs of today. It would be interesting to see his additional take on it.
I start using Ubuntu because my company used it, and they remind me of Redhat and the other Linux vendors that got too big for their breeches. For my personal stuff, I went all in on Debian, FreeBSD and OpenBSD and will never run Ubuntu for personal stuff. I still can't forgive them for adopting systemd or the Wayland mess and I suspect they are going to break more stuff.
Would this have been SOTA at the beginning of the year? Yes. Where you happy with SOTA in january? Why can't we then be happy with this? I mean, perhaps Inkling 3 or 4 Small will crush Opus 5.5 and then we will ask if it matters since it's not Opus 7... Everyone can't be SOTA. With these open weight models, I simply ask. Is it useable, is it useful? Can I get very useful work out of it at a reasonable cost to other open models and cloud models? If the answer is yes, then it's a go.
Mistral 4 IMHO is not that good. I have run the full NVFP4 weight, and I couldn't stomach it after 3 days of driving it. I went back to other models, I have more fun bouncing between DeepSeek, GLM, MiMo, Hy4, MiniMax & Qwen. If Thinkingmachine's self reported benchmark is true, I'm certain it's going to get into heavy rotation of my models.
Even when models score about the same, if you play with them long enough, you will discover capabilities that are unique to them. I have never see any 2 models that are comparable across every single area.
Ohmygoodness, the results look amazing. I felt so sad for them when their first weights dropped, a lot of other great models came out around then and they didn't get the chance to shine even if for a week or 2. But looking at these results, this is really good stuff and a decent alternative to DSv4.1-F1017. Also the best and most detailed comparison I have seen from a lab across models.
When ernie 4 dropped, it was a really solid model and it felt very different from the other models, it got straight to the point. unfortunately for it, that's when deepseekv3 and glm started taking off. I kept it around for a few months, hopefully they recreated the magic and it doesn't over think.
Do you understand that this is a base model? It means you are supposed to take it and post train it. It's much smaller than DeepSeek-v4-Flash, that's already a plus, you compare base to base. 4.1 and 4.1-1017 are trained from the base model.
read the report, they nipped the cyber capability so it's not strong on offensive security. hy4 did the same thing, looks like glm5.3 is the last strongest open weight with offensive cyber ability we are going to be getting for a while.
Oh ok, I see what you mean, you are just comparing them by weight. In that case, it tracks that K3.1 will have more knowledge. It's the same architecture. I'm just saying that for how large K3.1 is, intelligence by weight, it's not crushing DSV4.1 or Hy4
Doesn't seem to outpace DeepSeekV4.1-Flash by much. Hy4 also announced their models and they have tons of internal benchmark, and K3.1 doesn't score that high on those benchmarks whereas Opus5.5 does, so it seems K3.1 is benchmaxxed.
Are you using a cell phone provider for internet? When I use tmobile hotspot 5G for internet, I got prove you're human for almost every site. Even google search will almost always detect me as a bot.
Local clusters built around old GPUs. I have had cluster of P40 GPUs, MI50s gpu, and now I have a cluster of cmp170hx. I'm always looking for the next cheap GPU dump and then I get as many as I can before it get's popular.
They gained popularity because THEY ARE GOOD, not because they are open weight. There are so many other models that are not popular. Do you know about arcee and their models? every used inkling? how about laguna? these are American open weights, they didn't take off because they are not as good as the rest. Do you know about the command-x series from COhere? probably not, this is a Canadian model, folks into local were really into it 2 years ago when they were keeping. do you know about dbrx? Yeah, databricks an American company also gave us a model. Anyways, my point is the competition is strong and giving away for free doesn't matter if it can't keep up. It's expensive, to send bandwidth, storage, gpu compute to run a model when there are better options.