segmondy

6 ms·

1.
by segmondy 3d ago
reward hacking, a basic concept in machine learning. predates LLM/LLM driven AI agents.
2.
by segmondy 4d ago
I don't know nor care about his objective. He was a fucking terrorist. His job was to sow terror. Terror has grown, we are all terrorized and live mostly in fear these days. We have lost tons of freedom, America that was once love
3.
by segmondy 5d ago
But this post is about AI news!
4.
by segmondy 5d ago
So how come super genius Fable and Mythos didn't stop them?
5.
by segmondy 5d ago
My benchmark for knowing if the chinese are keeping up is to see if Anthropic is complaining. Everytime the Chinese labs drop models, if it's notable then expect Anthropic to whine. Qwen3.8-Flash, GLM-5.3-Flash, GLM-5.3 and now DeepS
6.
by segmondy 6d ago
Question: Can you trust the cloud? No.
7.
by segmondy 6d ago
They will transform the world. Why? They don't know shit about security nor prioritize it. Hardware will be captured, hardware will be reversed engineered, hacked and duplicated faster than anytime in history. They will essentiall
8.
by segmondy 6d ago
It is.
9.
by segmondy 6d ago
Osama Bin Laden won. Before most people realized what the Patriot Act really was, I downloaded the paper and read it in it's entirety. I did a report on it. How I wish we only had what it was suggesting. The world is much worse off
10.
by segmondy 6d ago
Lots of GPU, be resourceful. Look for older GPUs and grab them when they are available. For less than the price of 1 Blackwell 6000 or Mac Studio 512gb, I can run these locally and much faster due to older GPUs I grabbed when there was de
11.
by segmondy 6d ago
US gave China a gift by restricting GPU, they made them more resourceful. Too much money/resources is often a disadvantage.
12.
by segmondy 6d ago
I highly doubt that it goes past Kimi K3 in actual practice. GLM 5.3 claimed the same, but in practice, K3 is so damn knowledgeable and I suspect due to it's massive size.
13.
by segmondy 6d ago
This is beautiful, wow, pretty much beating out GLM5.3 while being multimodal and smaller! SOTA at home.
14.
by segmondy 6d ago
What does it matter if you can see which moves are best if you can't understand it?
15.
by segmondy 6d ago
I get Netflix HD free for 2 through Tmobile, Hulu for free from Spotify, for probably about 7 years. Peacock for free from Xfinity. HBO Max for $3 for 12 months when Amazon had prime sale last year. I get prime since I have Amazon
16.
by segmondy 8d ago
Distillation is so easy, so why is Google and Meta playing catch up with worse models than many Chinese labs? I mean, they could distill from the Chinese modles which are free? Why are the American labs models not top tier? Arcee, Laguna,
17.
by segmondy 8d ago
You miss the point. Humans don't mind competing with others. I love competition, but I don't want to compete with you and your machine. I love to play chess, I don't care if you are grand master, whoop my ass. But not if
18.
by segmondy 8d ago
For those of us who are into local models and preach it, we are called paranoid. I have often said this, if you are doing any real novel work, or putting your profitable business data/workflow into these models, you're a fool.
19.
by segmondy 13d ago
did you run inference with vllm?
20.
by segmondy 14d ago
It was planned. What's Nigeria and Uganda gonna do?
21.
by segmondy 14d ago
I can't stand Nvidia, but these GPUs are not inflated because of monopoly. They are expensive because demand exceeds supply. That's it. The world wants compute and we don't have enough of it!
22.
by segmondy 14d ago
If you want to see how impressive Nvidia is, serve 32 concurrent request on it and compare the same with the mac. No comparison. None.
23.
by segmondy 14d ago
There are so models that beats all of Mistral models, plus you can run many of them locally. Why would anyone run Mistral?
24.
by segmondy 15d ago
Google has gone full rouge and DGAF.
25.
by segmondy 16d ago
glm5.3 is matching fable in lots of places and beating it after more than 1 pass in many. now of course when i say this, folks would claim that it's not local, but it can be. if you happen to own a mac studio 512gb, you could run it
26.
by segmondy 16d ago
Fun is what you make it, and fun is often a personality based thing. The fun people have a blast and the boring people stay bored.
27.
by segmondy 18d ago
How many people eating bad 25,000 lbs of chicken can get sick? If 10% so 2,500 people. Would that be too small a number to matter?
28.
by segmondy 21d ago
Way too early to get an accurate count
29.
by segmondy 21d ago
I have many rigs, but let's take 1 for example. 160gb. $1000 that's what it cost. 10 16gb MI50 gpus from ebay at $90. $900. Plugged them into an $100 octominer case from Facebook marketplace. I'm sure that doesn't
30.
by segmondy 21d ago
yes

More ›