Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
byyoung3
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
7 ms
·
61.
▲
by
byyoung3
2y ago
guys after using this model more and more, the more off it feels. It just doesnt seem to have the "understanding" like how it usually does.
62.
▲
New ChatGPT GPT-4o Model
7 points
by
byyoung3
2y ago
|
8 comments
63.
▲
by
byyoung3
2y ago
Impressive stuff nonetheless.
64.
▲
by
byyoung3
2y ago
u should maybe focus on security instead of AI haha
65.
▲
by
byyoung3
2y ago
they deserve it. definitely recommend their book
66.
▲
by
byyoung3
2y ago
you guys really didn't get the joke did you
67.
▲
by
byyoung3
2y ago
i wonder if this guy ever imagined he would be on hacker news
68.
▲
by
byyoung3
2y ago
this is not about outages
69.
▲
by
byyoung3
2y ago
theres some debate around the idea that gluten might be mildly toxic to everyone, so I could see this being really useful to everyone
70.
▲
by
byyoung3
2y ago
lots of compounds are neurotoxic when taken in large doses, but completely safe when taken in lower doses. This isn't really surprising
71.
▲
Fine tuning GPT-4o to reason like r1
(wandb.ai)
1 points
by
byyoung3
2y ago
|
0 comments
72.
▲
by
byyoung3
2y ago
lmao i think its those with smartphones are unfairly penalized
73.
▲
by
byyoung3
2y ago
Read John sarnos book and you will recover. It might not be quick but you can. Don’t give up. Also look into Harvard RSI action
74.
▲
by
byyoung3
2y ago
its infuriatingly bad
75.
▲
Converting GPT-4o into a reasoning model
(wandb.ai)
2 points
by
byyoung3
2y ago
|
0 comments
76.
▲
by
byyoung3
2y ago
need to test for heavy metals as well as water contaminants
77.
▲
by
byyoung3
2y ago
sounds like a contaminant
78.
▲
by
byyoung3
2y ago
"Cline approaches AI assistance differently from most tools in the market" - I'm not trying to be a hater but I feel like it isn't that much different from cursor?
79.
▲
by
byyoung3
2y ago
It’s more that that they aren’t innovating or doing anything novel. Eg. “Tensorcracked is hiring to build document processing with LLMs” are fairly irritating
80.
▲
by
byyoung3
2y ago
those lemmas are wild
81.
▲
Comparing o1 vs. R1 on math competition problems
(wandb.ai)
3 points
by
byyoung3
2y ago
|
0 comments
82.
▲
by
byyoung3
2y ago
u should make a better landing page
83.
▲
by
byyoung3
2y ago
haha same
84.
▲
by
byyoung3
2y ago
its increasing my curiosity because it allows me to run more experiments
85.
▲
by
byyoung3
2y ago
hackernews is peak nerdiness
86.
▲
by
byyoung3
2y ago
"the absence of a thing doesn't prove the opposite" - it definitely makes a strong case for the opposite
87.
▲
by
byyoung3
2y ago
I actually believe the DMV workers care that it does actually suck (eg. they want it to suck haha)
88.
▲
by
byyoung3
2y ago
fasting and low carb probably wouldn't hurt
89.
▲
by
byyoung3
2y ago
username checks out haha
90.
▲
by
byyoung3
2y ago
Data is oil.
More ›