harrouet

5 ms·

31.
by harrouet 2mo ago
Is it an AI self-realizing its own existence?
32.
by harrouet 2mo ago
Inspiring article, I will keep the following bits: - with AI, bad engineers (bad designs?) lead faster to catastrophe - Frontier AI companies pay their SW engineers very much --> we need highly-skilled sw engineers and the lesser ones wi
33.
by harrouet 2mo ago
As a young telecom engineer working in "Intelligent Networks" in the late 90's in the UK, I learnt that UK phones had a code that could be dialed during a call, so that the line could not hang up anymore. That was geared agai
34.
by harrouet 2mo ago
Uber is the wrong example here, as it has destroyed $27B of value since inception. Or maybe that's exactly the right example...
35.
by harrouet 2mo ago
Actually if that can bring up your morale, just replacing coal power plants with nuclear (or other 24x7 clean energy source) would keep up at the 1.5°C objective.
36.
by harrouet 2mo ago
These laws, which exist in a similar way in Europe, intend to prevent media control by closed groups to take over democracy. To be honest, scraping them on broadcast TV will probably have little influence, even though in the wrong direction
37.
by harrouet 2mo ago
Ah the old debate about function-based views and class-based views. I totally understand why the OP would use only FBV, however when writing REST APIs you will want CBV to reuse base classes such as ListView.
38.
by harrouet 2mo ago
I have the same feeling about quoting Gartner studies, tbh. The paper is actually poor in data and mostly requotes other studies.
39.
by harrouet 2mo ago
Is this guy the reason why xcancel.com exists ?
40.
by harrouet 2mo ago
I am talking about reconfigurable gates to implement an LLM in silicon, i.e. an FPGA...
41.
by harrouet 2mo ago
"we have more than 3,000 tests"... Great, but since no human has looked at them, how do you know that they are relevant and actually test undesired behavior? Also, would you be able to point at which tests should be added in the f
42.
by harrouet 2mo ago
It is very hard to make sense of these benchmarks given that at least some of these models have probably been trained on the code base of these projects.
43.
by harrouet 2mo ago
I could definitely image Apple embedding a kind of LLM-optimized FPGA: slow to load (update) an LLM, but blazing fast at computing tokens. Who needs memory when your model is set in silicon ?
44.
by harrouet 2mo ago
Agreed that SwiftUI because bloated and hacky, when it promised simplification and transparency. But is there a single UI abstraction that succeeded in affordability for newcomers while maintaining performance and solid architecture? It is
45.
by harrouet 2mo ago
Kakehashi is perfectly fine for the 2/3 of humanity that live in Asia. Open up your mind and learn other cultures.
46.
by harrouet 2mo ago
Exactly what I was looking for for my non-US keyboard :)
47.
by harrouet 2mo ago
100% with the OP: At the end it is more productive to be focused on the right target, than to gain 15% of productivity because you use whatever tools. Note that this is not mutually exclusive, but putting brain power in thinking about the p
48.
by harrouet 2mo ago
It depends on what you are looking at. Time to 1st token is faster on the M5 because of HW accelerators helping the prompt interpretation (and it is CPU-bound). Token generation after that is GPU-bound and will profit from the higher bandwi
49.
by harrouet 2mo ago
I see so many sites that pretend that they have 350 /legitimate interest/ partners. Time to crack down on abuses.
50.
by harrouet 2mo ago
Hint for Microsoft: - Display drivers don't need access to the network - Display drivers don't need to write to the window manager - Display drivers just need to display the data that needs displayed - If you sign a driver, you ta
51.
by harrouet 3mo ago
Thank you! Thank you for saying it out loud ! At the start, LinkedIn was great and you could really connect you to a vibrant community of former colleagues and weak relationships. I actually got fruitful introductions and evn job offers fro
52.
by harrouet 3mo ago
This is exactly the kind of technology that I expect Apple to ship anytime soon given the RAM prices and their HW/SW integration skills: - ship super fast SSD (tbh they are already top notch) - add a specific cache layer for tokens - k
53.
by harrouet 3mo ago
...and it can lead to parallel processing discussions for very large arrays
54.
by harrouet 3mo ago
You beat me to it, I was going to post the same. Clear red flag, only useful for whistleblowers.
55.
by harrouet 3mo ago
As an economist, you may say what, you may say when, but never both at the same time.
56.
by harrouet 3mo ago
While working for a telecom operator, I tested the idea of having people paying more for dedicated support. We did a market study. I turned out that customers are not ready to pay for support. Cognitively, paying for a service and paying on
57.
by harrouet 3mo ago
True because wars finance value destruction. But with $1.4T announced capex for the Frontier AI labs, we're not far from the 2nd (illegal) war in Irak: $1.8T of direct military spending. With that said, I don't know how Frontier A
58.
by harrouet 4mo ago
Counter theory: the M6 is so good that they want to keep some oomph for the M7 line up.
59.
by harrouet 4mo ago
Get a 15" MBA...
60.
by harrouet 4mo ago
At this point and given the cost of memory, it will probably make sense to invest in faster SSD to allow for good performance with less memory

More ›