girvo

14 ms·

361.
by girvo 7mo ago
All the time now… it’s wild how little usage you get with Opus on the Pro sub now haha
362.
by girvo 7mo ago
Exactly. 3.6 plus in the exact same coding agent harness is notably worse in all of my testing compared to 3.5 plus. The former gets stuck in ridiculous thought loops on the exact same tasks I’m testing. Fascinating really, I expected more
363.
by girvo 7mo ago
> I also expect that there's a lot of feeling busy while not actually moving much faster. Hey don’t say that too loudly, you’ll spook people. With less snark, this is absolutely true for a lot of the use I’m seeing. It’s notably fas
364.
by girvo 7mo ago
> I still want to _code_ not just vibe my way through tickets. You and I want this. My EMs and HoEs and execs do not. I weep for the future of our industry.
365.
by girvo 7mo ago
> I'm not sure why more people aren't jumping on it Simple: most of the people you’re talking to aren’t setting these things up. They’re running off the shelf software and setups and calling it a day. They’re not working with c
366.
by girvo 7mo ago
The value prop for the Nvidia one is simple: playing with CUDA with wide enough RAM at okay enough speeds, then running your actual workload on a server someone running the same (not really, lol Blackwell does not mean Blackwell…) architect
367.
by girvo 7mo ago
They really are. Benchmaxxing is real… but also the Qwen 3.5 series of models are still very impressive. I’m looking forward to trying out Gemma
368.
by girvo 7mo ago
It’s not a fact in the slightest.
369.
by girvo 7mo ago
What a ridiculous, reductive and dangerously accelerationist viewpoint.
370.
by girvo 7mo ago
Poor babies. Again, I don’t care about them.
371.
by girvo 7mo ago
I have been!
372.
by girvo 7mo ago
> then code quality just doesn’t really matter so much in the age of AI Except at scale it really does, because garbage in garbage out. The crappier the code you feed the current models, the worse and more confusing the broken leaky abst
373.
by girvo 7mo ago
How closely it replicates microtonic's, would be my assumption
374.
by girvo 7mo ago
I'd really recommend putting a modicum of work into cleaning up obvious AI generated output. It's rude, otherwise, to the humans you're expecting to read this.
375.
by girvo 7mo ago
I cannot bring myself to care about distillation, when these companies have built their empires on top of everyone else's stolen data, while at the same time telling the world they're out to replace us all.
376.
by girvo 7mo ago
I'd recommend it too, because the knowledge cutoff of all the open weight Chinese models (M2.7, Qwen3.5, GLM-5 etc) is earlier than you'd think, so giving it web search (I use `ddgr` with a skill) helps a surprising amount
377.
by girvo 7mo ago
“Make no mistakes”
378.
by girvo 7mo ago
GLM-5 (and 5.1) is surprisingly impressive too I’m finding.
379.
by girvo 7mo ago
Rarely! But it takes me a few minutes per component, sometimes more because I'm quite bad at electronics and always afraid of damaging things haha
380.
by girvo 7mo ago
This is very very cool :) And is _way_ better than when I'm forced to do this by hand, I'll say that much haha
381.
by girvo 7mo ago
Hopefully low, it's a static blog. (I know this isn't the actual point of your comment, apologies!)
382.
by girvo 7mo ago
> If you charted performance over time, it'd look like a sawtooth People have, though, and it doesn't show that. I think it's more people getting hit by the placebo effect, the novelty effect, followed by the models by-def
383.
by girvo 7mo ago
A bunch of the points in this AI generated blog post were like that. Makes me feel dirty when I'm 1/3rd of the way through and I realise how off it is.
384.
by girvo 7mo ago
Surprising really, because I'm a Firefox + Ublock Origin die hard and I never get Cloudflare captchas. Wonder what the difference is? I have CGNAT turned off, if that matters at all (probably not).
385.
by girvo 7mo ago
If it’s similar to what I read: various stretching exercises, couple with creams to soften the tissue
386.
by girvo 7mo ago
The 777s I’m talking about are only international flights though? Edit: oh man I needed to scroll down, about a million people told you that hah.
387.
by girvo 7mo ago
> The most valuable thing management has taught me is how to communicate with precision when someone else’s work depends on it. Ignoring the AI tells throughout the article for the moment, but I'm very surprised that someone who i
388.
by girvo 7mo ago
Yes. In that I'm saying their AI generated slop README is misleading enough to be lying. When you prominently say "all without uploading anything to the cloud." but the default does exactly that, a single half-sentence ment
389.
by girvo 7mo ago
They are, dang et al deal with it pretty strictly. Which I thank them for.
390.
by girvo 7mo ago
I don’t get it. It says nothing leaves your computer, but it’s sending things to OpenRouter, not running models locally. Perhaps I am dumb (and I always feel dumb after reading an AI generated README for yet another AI tool Tbf)

More ›