nurettin

8 ms·

151.
by nurettin 8mo ago
They probably meant goal hacking. (I just made that up)
152.
by nurettin 8mo ago
No this is the stop playing with your dolls argument. It is rock solid. As for torturing animals, you do you I guess.
153.
by nurettin 8mo ago
Right, good that you've got that going, congratulations to you and your team. Design time does make human development much easier, since you are able to see the working prototype run as you design it, not sure what the point is if AI i
154.
by nurettin 8mo ago
Yeah, people get shot/stabbed/"fall off a building by accident" every day and we should be considerate of the feelings of a petri dish.
155.
by nurettin 8mo ago
> waiting 30days, farming karma If "farming karma" is a thing, maybe that forum deserves what is coming. Either the karma mechanic is inappropriate given the demographic, or it is too hard for the users to avoid upvoting bots.
156.
by nurettin 8mo ago
For Lazarus (an IDE with visual components similar to Delphi) I switched to code-first components and did away with the form files. You can probably do this with all of these frameworks.
157.
by nurettin 8mo ago
Of course I know that bulk operations aren't the same as clock speed and they sit in a weird place. My point was that their existence is evidence of needing performance on a single core. There is also an argument to be made for re-ar
158.
by nurettin 8mo ago
I think it will enhance everyone. PMs can't develop, since llm development (adding code to whatever the llm initially spat out) still consumes time and effort, but they can now write a PoC without devs and quickly get it up and running
159.
by nurettin 8mo ago
Not sure what you mean. It lets you do 64 operations with one instruction. Where's the diminishing returns?
160.
by nurettin 8mo ago
Single core speed is absolutely a thing that is needed and preferred to multicore. That's why we have avx, amx, etc.
161.
by nurettin 8mo ago
Coming from pascal to C as a highschooler, my biggest wtf moment happened when I forgot a ; after a struct in a header. The compiler kept complaining about the code below the include and for the life of me I couldn't figure it out. Too
162.
by nurettin 8mo ago
See this is why we can't have cheap tinfoil.
163.
by nurettin 8mo ago
Yes it feels like I got it from ANC. Might not "be a thing", just coincided with my ANC use. It is my data point.
164.
by nurettin 8mo ago
Got it similarly. 7-8 years ago. Probably from ANC. It used to feel loud, now I have to remind myself to hear it. There is light at the end of the tunnel.
165.
by nurettin 8mo ago
800gb database is still a single file, if it only threads on file level, you shouldn't see a difference.
166.
by nurettin 8mo ago
I haven't opened a text editor to code in months and probably won't need to anymore. Goodbye vim and intellij, nice knowing you. It was a good while it lasted. Glad I haven't invested decades into emacs like some of my collea
167.
by nurettin 8mo ago
Completely the opposite experience here! I am a tech lead with decades of experience with various programming languages. When it comes to producing code with an llm, most noobs get stuck producing spaghetti and rolling over. It is so bad th
168.
by nurettin 8mo ago
7k for 70,000 books sounds insane.
169.
by nurettin 8mo ago
It is a bottleneck for multiple files, but will it speed up with a single file?This is how we sent files for decades. Archive, transfer, unarchive. So I'm wondering what the point is.
170.
by nurettin 8mo ago
Why not tar.gz and send as a single stream?
171.
by nurettin 8mo ago
This is a balancing act between collectors and explorers. There is probably some optimized number. Likely targeted at beginners.
172.
by nurettin 8mo ago
In my case it definitely helps, but keeping 64k context window would probably require more than 64 GB.
173.
by nurettin 8mo ago
I'm running it on a pure cpu 2020 model ryzen server with 2x128 GB RAM with llama.cpp, it seems as intelligent as gpt4. I optimized a little bit by forcing it to run on a single RAM stick and tuning llama.cpp build parameters, going fr
174.
by nurettin 8mo ago
Can you tell it to write to /tmp/s.py instead of trying to execute it inline?
175.
by nurettin 8mo ago
I am singularly impressed by 35B/A3, hope that is not the reason he had to leave.
176.
by nurettin 8mo ago
No that's new. This was the previous "accelerating science" piece. https://openai.com/index/accelerating-science-gpt-5/?hl=en-G...
177.
by nurettin 8mo ago
> he doesn't pay attention to any of the contributions to math or science that these models are making Ok but why report PR pieces as evidence for LLMs being useful? These are tools that can possibly provide output that is eventuall
178.
by nurettin 8mo ago
You need to be a special kind of troll to use LLMs to respond to someone whose entire online persona is built around "AI bubble".
179.
by nurettin 8mo ago
Ironically, the editors probably haven't opened a text editor for months.
180.
by nurettin 8mo ago
I just start llama.cpp serve with the gguf which creates an openai compatible endpoint. The session so far is stored in a file like /tmp/s.json messages array. Claude reads that file, appends its response/query, sends it to t

More ›