Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
DantesKite
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
11 ms
·
61.
▲
by
DantesKite
3y ago
If you’re curious, you can check the progress of many open source LLM’s and how they perform on various evals here: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderb...
62.
▲
by
DantesKite
3y ago
This is really cool. Just being able to take blood tests in general to get a better sense of what might be going on can be so informative.
63.
▲
by
DantesKite
3y ago
> Maybe JavaScript is actually what saved your PHD, although cookie clicker was what pointed you in that direction. Yes the article strongly implies this with some added nuance: > I’m extremely grateful to Cookie Clicker for the journ
64.
▲
by
DantesKite
3y ago
Humans are like that too.
65.
▲
A Bayesian view of Amazon resellers (2011)
(johndcook.com)
114 points
by
DantesKite
3y ago
|
36 comments
66.
▲
Open-Source AI Camera
(github.com)
2 points
by
DantesKite
3y ago
|
0 comments
67.
▲
by
DantesKite
3y ago
This reminds me of an observation I had: I think a lot of the best dating apps aren't strictly dating apps. For example, you'd be surprised at the number of people who have met on Twitter. It's quite a lot. Even John Carmack
68.
▲
Lessons from the Years of Lyndon Johnson by Robert Caro
(dwarkeshpatel.com)
1 points
by
DantesKite
3y ago
|
0 comments
69.
▲
by
DantesKite
3y ago
It's not clear to me the Reddit users will be permanently gone either, unless there's some place like Reddit taking in a lot more users right now. But I haven't heard of any place like that so far. No real competitor for user
70.
▲
Contra Marc Andreessen on AI
(dwarkeshpatel.com)
41 points
by
DantesKite
3y ago
|
29 comments
71.
▲
Ask HN: What do you think of actors spreading malicious packages with ChatGPT?
5 points
by
DantesKite
3y ago
|
0 comments
72.
▲
by
DantesKite
3y ago
This would work better if there was a viable competitor to Reddit. As it stands, there doesn't seem to be anywhere for these people to go, so they'll likely come back after a few days. Habits are hard to break. Habits are a moat t
73.
▲
Can you trust ChatGPT’s package recommendations?
(vulcan.io)
29 points
by
DantesKite
3y ago
|
28 comments
74.
▲
Transformative AGI by 2043 is <1% likely
(twitter.com)
2 points
by
DantesKite
3y ago
|
0 comments
75.
▲
The US is getting its first new nuclear reactor in 40 years
(grist.org)
190 points
by
DantesKite
3y ago
|
358 comments
76.
▲
Improving GPT-4’s codebase understanding with ctags
(aider.chat)
1 points
by
DantesKite
3y ago
|
0 comments
77.
▲
by
DantesKite
3y ago
You should try OpenAI embeddings. They're fairly cheap to run over a large amount of text (should cost <$10 if you have thousands of documents I believe, but correct me if I wrong). Then you can run searches for content even if the
78.
▲
by
DantesKite
3y ago
Side projects can be invigorating if you can find something you've always wanted to build. I know GPT-4 in particular has made exploring new, small projects fairly easy. You could do something similar. Mix up the tediousness of doing t
79.
▲
by
DantesKite
3y ago
One of the things that isn’t talked about enough in a reasonable and calm way is the impact COVID can have on the immune system long term. It’s talked about in terms of Long COVID and whether or not it’s psychosomatic, the policies made dur
80.
▲
by
DantesKite
3y ago
Tempted to train a small language model on this text and see what happens lol.
81.
▲
by
DantesKite
3y ago
I like that it seems like he optimized the game for the Steam Deck.
82.
▲
by
DantesKite
3y ago
It's given me a place to understand problems I might not have otherwise solved or been aware of. It's a bit like being given a piece of paper to solve math problems, instead of trying to do it all in your head. Makes a huge differ
83.
▲
Open LLM Leaderboard
(huggingface.co)
6 points
by
DantesKite
3y ago
|
0 comments
84.
▲
by
DantesKite
3y ago
I ask it sometimes whether or not it can automate a common workflow I do on the computer and it'll occasionally surprise me with something I didn't know existed. Like being able to do rule-based sentiment detection analysis and in
85.
▲
by
DantesKite
3y ago
Makes sense. Metabolic processes and immune pathways are closely interlinked together (such as through the mTOR pathway). Activate mTOR and you get the signal for T cells to proliferate as well as muscle growth, especially in the context of
86.
▲
by
DantesKite
3y ago
I've always wanted lab-grown meat. Not the kind that looks like sludge, but the kind that would be indistinguishable from a beef patty you buy at the store, not least because it would be a more efficient way of producing food at scale.
87.
▲
by
DantesKite
3y ago
Another take I’ve heard is that lighter eyes allow you to see the dark ring around the iris (limbic rings), which apparently fades as you age, becoming progressively lighter, forming a sort of proxy for age.
88.
▲
by
DantesKite
3y ago
I wonder if they’ll use this to train large language models.
89.
▲
by
DantesKite
3y ago
> The UN says that the death penalty has not proven to be an effective deterrent globally and is incompatible with international human rights law, which only permits capital punishment for the most serious crimes. I don’t know it seems t
90.
▲
by
DantesKite
4y ago
I tried it and it doesn't seem to recommend as many books as other book recommendation sites I use. I prefer Readow.AI for instance because it recommends a lot more books, it's quicker, and the books it recommends are occasionally
More ›