girvo

9 ms·

151.
by girvo 4mo ago
I don’t know what school you went to, but mine covered war crimes committed by the Allies plenty. That doesn’t mean Nazi Germany wasn’t utterly disgusting though.
152.
by girvo 4mo ago
> You don't have to agree with Neo-Nazi propaganda to acknowledge that what we are taught about WWII and Weimar Germany in school, isn't the truth either. Go on then. Say what you mean.
153.
by girvo 4mo ago
Right. Tokens/s decode isn't the most important thing to me: wall clock time for task completion is. And tracking all of that, on my GB10-based Asus box, Step 3.7 Flash at IQ4_XS beats Qwen 3.6 27B despite the latter having MTP, o
154.
by girvo 4mo ago
That beautiful orange text on a black background gives me nostalgia warm and fuzzies
155.
by girvo 4mo ago
For sure. I'd argue this cycle is operating at a different scale though
156.
by girvo 4mo ago
Depressingly in todays attention economy, you absolutely should! It'll be good for you directly. At least one upside is it can help others too I suppose, I just get sad when I think about how important attention is now
157.
by girvo 4mo ago
I wonder what the better pithy phrase would be then for "thing that is obviously useful, but is being hyped beyond it's (current) ability by those with a vested interest in doing so"
158.
by girvo 4mo ago
This is so needlessly combative and performative, which is especially hilarious as you claim the others are the ones being so. > AI is better at this than you. You just won't admit it. And it's going to get 10,000x better tha
159.
by girvo 4mo ago
Both? I mean both can be true.
160.
by girvo 4mo ago
You just described IBM's whole Hyperledger Fabric thingy. I worked with it once upon a time, with the biggest insurance companies in my country where they plus a regulator all ran nodes.
161.
by girvo 4mo ago
No, I'm not. Because, as I said, we've literally done it first for the first half of our application, and it took us eight engineers for ~5 months. Its genuinely weird to have you say that so confidently lol
162.
by girvo 4mo ago
> They can be still be applied now using coding agents, if you're willing to push back against the default setup and change your mode of thinking a little bit. Of course it doesn't help that an entire industry is dedicated to p
163.
by girvo 4mo ago
I'm currently using it to do a large migration from one Relay environment to another, but this is possible because 1. We've done it by hand for another route already, which the LLM uses as reference 2. Theres a strong validation s
164.
by girvo 4mo ago
Struggle harder than you otherwise would, sadly.
165.
by girvo 4mo ago
I have a "task router" that is a small local LLM on my mac mini (Qwen 3.5 0.8B) that I use to decide (when activated) with Pi whether to route a given task to my local LLM (Step 3.7 Flash) or to <given cloud provider>, if th
166.
by girvo 4mo ago
Funnily I think you are still right re. the specific account under discussion.
167.
by girvo 4mo ago
Open weights like this make me wish I had a bunch more DGX Sparks to cluster so I could fit it!
168.
by girvo 4mo ago
Chinese model companies are already beginning to close, instead of opening. The latest big Qwen models are not open, for example. And it doesn't look like they will be, either.
169.
by girvo 4mo ago
> The total power of all GPUs on the planet dwarf their capabilities That just isn't true. It misunderstands exactly how much silicon has gone directly to those companies, and exactly how much more powerful said silicon is compared
170.
by girvo 4mo ago
> We've been hearing about the risks of engineered viruses and homemade superweapons since GPT 3.5, so where are they? We've had abliterated open weights models much stronger than GPT 4 for over a year now. What I can't un
171.
by girvo 4mo ago
When you see the "illustrious" US government doing things like this, do you blame them? I don't.
172.
by girvo 4mo ago
Well, it's different in that at least the Chinese companies release weights unlike the American ones!
173.
by girvo 4mo ago
MiniMax M3 is surprisingly powerful, and open weight (or is about to be). There's others in this space too: MiMo v2.5, GLM 5.1. There's quite a few to pick from if you want strong models running on "your" hardware.
174.
by girvo 4mo ago
Immediately. It's always been a smokescreen, and markets have never been truly free. Thumbs on scale, at all times.
175.
by girvo 4mo ago
That lobster font we all used for our startup names was legendary
176.
by girvo 4mo ago
Work forces me to on the work laptop. But Ublock Origin Lite is good enough for that use-case. I use firefox everywhere else.
177.
by girvo 4mo ago
It really is lol
178.
by girvo 4mo ago
Same. I also found out that my old Xiaomi account is apparently considered "mainland china" and I can't put any phone number except a chinese one on it lol. I'm not trusting these people with anything that's for sur
179.
by girvo 4mo ago
> There are 2 potential outcomes: either the sky really does fall, and there's a meaningful uptick in bad things happening to iPhone users, in which Apple can easily point the finger at the EC and say "they made us do this&quot
180.
by girvo 4mo ago
> But when an LLM does it on an area we know, we notice and suddenly it's too much. Well of course. The owners of the companies building this are constantly talking about it replacing us all. Why would it be surprising that it would

More ›