segmondy

5 ms·

31.
by segmondy 2mo 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
32.
by segmondy 2mo 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.
33.
by segmondy 2mo 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?
34.
by segmondy 2mo ago
Way too early to get an accurate count
35.
by segmondy 2mo 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
36.
by segmondy 2mo ago
yes
37.
by segmondy 2mo ago
yes. the real question is, when?
38.
by segmondy 2mo ago
nothing, if it's that important it will bubble up on another part of the internet.
39.
by segmondy 2mo ago
There's no brand confusion, you are just unfamiliar with them.
40.
by segmondy 2mo ago
The price is free for some of us, we can run it at home.
41.
by segmondy 2mo ago
Folks need to learn to start putting such projects outside of the reach of US. Put it in an "unfriendly" country where the C&D means nothing.
42.
by segmondy 2mo ago
I think the Chinese are going to be building their own chips aided with AI. DeepSeek, z.AI, MiniMax, Moonshot, etc, it's a race. The take off has really started.
43.
by segmondy 2mo ago
Give me a break, Anthropic is expensive and their CEO is not the nicest guy around, and the games they play with other people's money/their API is not fair. I stopped using Anthropic and OpenAI once they started calling for reg
44.
by segmondy 2mo ago
Yup, OP has domain knowledge in software/security so they knew how to steer. it's like knowing how a rudder can control an aircraft doesn't make you capable of flying one.
45.
by segmondy 2mo ago
I read the entire article, I love it. Nothing boring about it.
46.
by segmondy 2mo ago
I wonder if you started with GLM5.3, how fast and cheap would it have been? Or did it really need the combo of K3 and GLM5.3
47.
by segmondy 2mo ago
Thanks for sharing, pretty cool. Whenever I read these, I want to see your prompts. Not necessarily the output from the model since that would be verbose, perhaps summarized if too much. But seeing your prompt and how you steer the mode
48.
by segmondy 2mo ago
I have thought that the thing for folks to do instead of destroying these cameras is to just put a black trash bag on it.
49.
by segmondy 2mo ago
You must not have been paying attention to development with robots, there are many videos of robots moving really fast in "non controlled environments"
50.
by segmondy 2mo ago
most likely a preview. they often release the preview via API, get more training data, post train some more then release the weight. i would expect to see it perform better in a few weeks or a month.
51.
by segmondy 2mo ago
It doesn't beat k3, it doesn't even beat deepseekv4Flash-0731 let alone glm5.2
52.
by segmondy 2mo ago
Thieves, but does it matter. But what does it matter? Thievery runs to the highest post of the land.
53.
by segmondy 2mo ago
So what? The world is more trustworthy than you can imagine. I have bought over 30 GPUs over the internet, sight unseen. From around the world and every single one has arrived good. I have also gotten over 20 from marketplace and the sa
54.
by segmondy 2mo ago
Well, some folks are going to keep trying to exclude it, but the reality is we are almost at the point even those folks would concede that they are not smarter than LLM. I suspect we will see then see more discourse about consciousness,
55.
by segmondy 2mo ago
I can run this at home. No guardrails, this is not shy of Sol and Fable, this crushes them in my book. It's not just about evals, but what I can do with the damn model.
56.
by segmondy 2mo ago
You don't need $10k to run DeepSeek, I run it on a $1000 system.
57.
by segmondy 2mo ago
I've 0 issues with gemma4 and I downloaded it early.
58.
by segmondy 2mo ago
I suppose you are not at the end of the hateful content. It's easy when it's a group, but how small does the group have to be before they should deplatform it? 100k, 10k? 1000? 10, 1? Is one person okay? If I and a bunch of fel
59.
by segmondy 2mo ago
... because they are often the first to quant it. sometimes the actually model providers will release wrong chat templates or values in the model config which leads to bad quants. how would you know a quant is good if you don't make
60.
by segmondy 2mo ago
the single shell command is the terminal bench.

More ›