menaerus

32 ms·

691.
by menaerus 2y ago
I didn't find that feature to be that handy tbh. Mostly because it necessitates me to search for the pattern manually. What I usually want in my workflow is to highlight the word under the cursor automatically. This simplifies the code
692.
by menaerus 2y ago
I wasn't, you're confusing me with someone else from this thread. It's literally my first comment here.
693.
by menaerus 2y ago
Databases are a perfect example of an open-ended complexity space. SQL is a Turing-complete language and your users are programming their workloads against your database kernel. You (as a developer) know nothing about those workloads nor do
694.
by menaerus 2y ago
If you're open to a bit different and perhaps thought provoking perspective let me say that for your hypothesis to be true we would need to assume that the distribution of memory-safety bugs is uniform across the codebase. From my expe
695.
by menaerus 2y ago
This is not universally true although I see this phrase being repeated here too often. And it is especially not true with the small models. Small models are compute-bound.
696.
by menaerus 2y ago
You're (luckily) wrong on this. There's nothing that is stopping you to plug the server into your home power outlet. It will work just fine - ~0.5kW is almost nothing. Case in point - Dell also builds the workstations with the sam
697.
by menaerus 2y ago
So we get a new model that is one of the ~two best performing models on the market, and yet we are not discussing its technical capabilities but rather its inclination towards the history events. Sorry, I don't get this obsession.
698.
by menaerus 2y ago
I know it's an unpopular opinion but tbh those Google security blogs always read more as a marketing material than engineering effort. Why I think so it's because statistics always has two side of the coins - if you want to make s
699.
by menaerus 2y ago
I think you want both. To scale the model, e.g. train it with more and more data, you also need to scale your inference step. Otherwise, it just takes too long and it's too costly, no?
700.
by menaerus 2y ago
Is it still safe even if it depends on the dependencies which are not safe?
701.
by menaerus 2y ago
I appreciate the details, I am not familiar with this at all. This is pretty shitty considering that as a contractor you would need to work ~3.5 months to earn 50k for much much less work and stress. That makes me wonder what is the profile
702.
by menaerus 2y ago
No, that's only about the leverage - a perfect instrument to create even more pressure. Everything else I call a BS, sorry. Aren't founders already taking a huge leap by admitting themselves to work probably as much as ~10x more,
703.
by menaerus 2y ago
> Sure, but in some cases it's also a great way to become very very wealthy. In very very very few cases, sure. Also, the wealth isn't the only or at least not the biggest motivator for many people trying to build their startup
704.
by menaerus 2y ago
Sounds like a nightmare and a good way to go into personal bankruptcy. Unless, I guess, you had a job where you accumulated ~M of savings, which is very very few people.
705.
by menaerus 2y ago
Are there investors who wouldn't agree to pay the founders a wage? I understand the part where you invest your time and money to build the PoC but I can't imagine running a startup with VC-backed investment without paying out myse
706.
by menaerus 2y ago
Interesting story but what I don't get is this > Founders might get a wage to avoid personal financial stress How do you make a living as a founder if you don't get a wage? Is that really a norm? > but then also need skin in
707.
by menaerus 2y ago
That's good news for us I suppose.
708.
by menaerus 2y ago
It is vague because the implementation details you are asking me for are closed source, for obvious reasons. I can only guess what it does but that's besides the point. The point is rather that Devon or 1M window context qwen model mig
709.
by menaerus 2y ago
As I already said in my first comment, I haven't used those models and any of them would have been forbidden at my work. My point was rather that you might be observing suboptimal results only because you haven't used the models w
710.
by menaerus 2y ago
Several millions lines of code. Can't remember any project that I was involved with and that was less than 5MLoC. C++ system level programming. Overhyped or not Devon is using something else under the hood since it is pre-processing yo
711.
by menaerus 2y ago
You never know - one moment arguing on HN and the second moment you know, drinking at the bar lamenting on how AI is gonna replace us :)
712.
by menaerus 2y ago
The largest context that I am aware that an open-source model (e.g. qwen) can manage is 1M tokens. This should translate to ~30kLoC. I'd envision that this could in theory work even on large codebases. It certainly depends on the chang
713.
by menaerus 2y ago
Ok, sorry about that.
714.
by menaerus 2y ago
> That's actually not the case I don't know where do you get the data but the latencies are surely not decreasing and 2 cycles for L1 is totally wrong. I should have perhaps been more specific that under latency I actually mean
715.
by menaerus 2y ago
I am not sure I follow your argument. You said that icecream cannot work in multi-client multi-server case. I said that it certainly can, and it can. Perhaps you're arguing that in that multi-case scenario ccache is not usable? That I
716.
by menaerus 2y ago
I don't think my comment read as snarky but I was surprised to see the immediate downvote which presumably came from you so I only added the last sentence. This is a stupid way of disagreeing and attempting to shut down the discussion
717.
by menaerus 2y ago
Thanks but IIUC you're describing a situation where you're simply using a model without giving it a chance to learn from the whole codebase? If so, then I was asking for the opposite where you would ingest the whole codebase and t
718.
by menaerus 2y ago
I see that you deleted your previous response which was unnecessarily snarky while my question was genuine and simple I suppose. > And frankly, if you can't automate context, How about ingesting the whole codebase into the model? I
719.
by menaerus 2y ago
Did you have to do any preparation steps before you asked from a model to do the large scale change or there were no steps involved? For example, did you simply ask for the change or did you give a model a chance to learn about the codebase
720.
by menaerus 2y ago
If all of the latencies to the main memory and L1/L2/L3 caches have been only increasing throughout the years, is picking the right data memory layout more or less important than before?

More ›