dcl

3 ms·

1.
by dcl 6d ago
the smallest problem that cannot fit in a brain would be pretty interesting
2.
by dcl 6d ago
What do you think all the PhD's and post-doc students are doing...? It's not called graduate-descent for nothing :p
3.
by dcl 7d ago
In the early days of SC2, I remember people using genetric programming to optimize build orders. I remember a slightly unorthodox Zerg Roach Rush which was _really_ fast.
4.
by dcl 13d ago
Using AI to catch up on reinforcement learning advances, trying to 'solve' Gin Rummy. Will move on harder games after that.
5.
by dcl 23d ago
That's kind of wild. Our first PC was a an IBM PS/2 486SX 33Mhz, 4MB RAM, that was purchased in 1993.
6.
by dcl 23d ago
They overclocked well though, I think you could run the 300Mhz chips at >400Mhz. I also believe you could get motherboards that supported 2 Celeron chips. I have no idea how effective/useful it was, but it was certainly a cheap&#x2F
7.
by dcl 24d ago
thank you
8.
by dcl 24d ago
Thats not exactly 'validated'. Feels very noisy, it is not a good bar for either - does this code do what the user actually asked - is this code actually 'good' There would be so many examples of coding projects that the
9.
by dcl 24d ago
Any more info on this?
10.
by dcl 24d ago
Well thats very interesting. Thank you. Will be interesting to see how hard/easy it is to translate my Claude skills, loop design, etc to the new harness. This kind of raises another question to me regarding the coding benchmarks, how
11.
by dcl 24d ago
Very keen to try this after using Claude Code over the last few months. Should I just point Claude Code to Muse Spark endpoint (because I'm familiar with Code)? What do people think of Muse Code or other coding agent harnesses?
12.
by dcl 24d ago
Hopefully, 'validated' AI code
13.
by dcl 26d ago
I played this as a young teen in 1998 and it's still the most powerful gaming experience and memories I have.
14.
by dcl 26d ago
I believe a lot of NIC's could autodetect and adapt to the cable? I'm pretty sure I had 2 computers connected to each other using a regular network cable.
15.
by dcl 1mo ago
iPods were a very popular Apple product once upon a time...
16.
by dcl 2mo ago
Question: For things like this to be truly worthwhile, do you have to ensure the training, post-training of LLM's is filled with brilliant memelang code? The models are probably already good at 'thinking' in SQL at the moment
17.
by dcl 2mo ago
One question I have about stuff like this: How does it affect the models intelligence or thinking ability. If this modifies the output or chain of thought in any way, it may impact what the model is capable of right? Especially if it's
18.
by dcl 2mo ago
finally the benchmark for me
19.
by dcl 2mo ago
vaporwave fans rejoice
20.
by dcl 2mo ago
I have to explain this to managers, execs and stakeholders all the time. ML models work great for systems where the rules/dynamics do not change over time. With Forecasting, in a lot of domains where you want a forecast, everything is
21.
by dcl 3mo ago
That is going to be absolutely wild for whoever can access/afford it.
22.
by dcl 3mo ago
A CPU only benchmark for typical data-science and ML tasks that I can run on a few different systems that will help me figure out what the next system I should buy is. I have no idea what Apple silicon is like for these tasks.
23.
by dcl 4mo ago
Deliberately producing misaligned and deceitful AI systems now. Great.
24.
by dcl 4mo ago
Ferrari uses cars like this to test loyalty. If you want to get 'on the list' buying cars like this is one of the ways to do it, especially if you haven't spent considerable $ with them before.
25.
by dcl 4mo ago
imagine no more: https://chatgpt.com/s/m_6a14fc70630c8191bedf8e06913f4d34
26.
by dcl 4mo ago
This. If you want to get on the list to buy the new supercars, you're going to have to start here. And you better add some expensive options.
27.
by dcl 4mo ago
This is the car you will need to buy to get on the list to buy the Ferrari you kind of want - but not the Ferrari you really, really want, that will cost you a lot more.
28.
by dcl 7mo ago
Anyone else think Britain is going to eject the Royal Family within 100 years?
29.
by dcl 8mo ago
Because it's a 4gb download?
30.
by dcl 8mo ago
S&P gets you dividends though, so the interpretation of that chart is tricky. Holding the S&P, you can still do better than holding gold, even when the GOLD/S&P ratio is positive.

More ›