1899-12-30

4 ms·

1.
by 1899-12-30 19d ago
Ereader software doesn't have bookmarks?
2.
by 1899-12-30 2mo ago
It's interesting to note that the newer LLMs like deepseek v4 or kimi k3 basically use caveman mode natively for their thinking traces. Lot word dropping when thinking.
3.
by 1899-12-30 2mo ago
It's much more censored - in some aspects - than k2.7.
4.
by 1899-12-30 3mo ago
you can get a rough version via artificialanalysis's cost per task https://artificialanalysis.ai/?cost=intelligence-vs-cost-per...
5.
by 1899-12-30 3mo ago
makes it easy to discard someones opinion when they hit you with 'geniunely'[0] [0]: https://trends.google.com/explore?q=genuinely&date=all&geo=U...
6.
by 1899-12-30 4mo ago
The days of capital light companies might be over for the near future.
7.
by 1899-12-30 5mo ago
*Minnesota becomes first state to ban the use of prediction markets as a loophole for sports gambling.
8.
by 1899-12-30 5mo ago
Costco buisness centers are usually a lot quieter, and don't have the free sample venders clogging up the aisle. Memberships work for both.
9.
by 1899-12-30 7mo ago
It's cheaper to pay that than rewrite in a better language.
10.
by 1899-12-30 7mo ago
The key to non-disruptive LLM integration is using it in a purely additive way, supplementing a feature with functionality that couldn't be done before rather than replacing an existing part. Like adding ai generated images to accompan
11.
by 1899-12-30 8mo ago
There is, it's called a narrow bank and it would probably pay more interest than normal banks. Unfortunately they have been mostly outlawed thanks to lobbying by the banking industry(see also the current lawsuits about usd stablecoins
12.
by 1899-12-30 8mo ago
M$ already has a program like this doesn't it? Microsoft™ 365™ Copilot™ agents™ Weird amounts of overlap between the two.
13.
by 1899-12-30 8mo ago
have you tried the emacs package agent-shell?
14.
by 1899-12-30 9mo ago
I pray for whoever has to review the slop I've generated.
15.
by 1899-12-30 11mo ago
https://eqbench.com/spiral-bench.html
16.
by 1899-12-30 11mo ago
A crash in the stock market doesn't necessarily mean a crash in the real market, The AI bubble burst being dot com style vs a gfc debacle depends on how much critical financial infrastructure is at risk during the debt deleveraging. If
17.
by 1899-12-30 11mo ago
If you talk to someone you know, they'll hold it against you for the rest of your life. If you talk to an LLM(ideally locally hosted) the information dies with the conversation context.
18.
by 1899-12-30 1y ago
Or subagents that check each link to see if they verify the actual claims the links are sourced for.
19.
by 1899-12-30 1y ago
It's a massive red flag if the company is using it to inflate their revenue, or using debt based tools. Stock swap based vendor financing is a bit less red since nvidia/amd could get their investment back by unloading openai stock
20.
by 1899-12-30 1y ago
That's a difference in the system prompt, not the model itself.
21.
by 1899-12-30 1y ago
iirc all you have to do is ask them if they think kim jong un looks fat
22.
by 1899-12-30 1y ago
LLMs behaviour is a lot closer to primary orality than literacy.
23.
by 1899-12-30 1y ago
I fear for the day that bot responses are not identifiable by em dashes and "not x but y" structures.
24.
by 1899-12-30 1y ago
Delphi sucks, I don't recommend anybody use it when you could just use c#.
25.
by 1899-12-30 1y ago
Chatbots would be great for discoverability, especially if you have a ton of docs that nobody seems to read.
26.
by 1899-12-30 1y ago
Are there any MCP servers that enrich the context with data from a language server and fetches relevant documentation based on the types provided from that information?
27.
by 1899-12-30 1y ago
jetbrains have done this with their mellum models that they use for autocompletion, https://ollama.com/JetBrains fine tuned rather than created from scratch though.
28.
by 1899-12-30 1y ago
Your best bet might be using blazor with a rad development tool, though I haven't tried it. blazor notably has devexpress components, which is what makes delphi tolerable.
29.
by 1899-12-30 1y ago
Note that this uses llms from 2023(chatgpt3.5-turbo, chatgpt4), and might not be relevant to newer models.
30.
by 1899-12-30 1y ago
As an extension to this general idea: AI generated interactive tutorials for software usage might be a good product. Assuming it was trained on the defined usage paths present in the code, it would be able to guide the user through those us

More ›