Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
sigmoid10
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
7 ms
·
61.
▲
by
sigmoid10
2mo ago
Exactly that. Except that ultrafast delivers this level of intelligence an order of magnitude faster. So if your competitors automatically react to news articles or financial statements with a certain level of comprehension within minutes,
62.
▲
by
sigmoid10
2mo ago
Moving from either frontier intelligence or frontier latency to a single model that does both at the same time is potentially a game changer in certain industries. I can easily see e.g. hedge funds dropping tons of money on this, becaus
63.
▲
by
sigmoid10
2mo ago
That's essentially the state of consulting outside of the big firms.
64.
▲
by
sigmoid10
2mo ago
What if your car is not brought to the wash by you (common for companies or rich people). If someone asks you how to get it and the car wash is close by, of course you will tell them to go on foot. Human life experience is not universal and
65.
▲
by
sigmoid10
2mo ago
It's not that hard though. Until we integrate extraterrestrial code into our package managers, you can call earth our closed environment. And if your company has strict offline and library requirements, you could even define it for you
66.
▲
by
sigmoid10
2mo ago
You still have to find these things though. Even if you didn't do it, someone, somewhere did spend all that mental energy. There are no shortcuts to saving effort in a closed environment. Same way you can't reduce total energy in
67.
▲
by
sigmoid10
2mo ago
Counter example: A person who always has their car brought to the car wash by someone else (could be super wealthy, could be company cars, could be a million other reasons...) is asked whether to go or drive to the car wash. Of course they&
68.
▲
by
sigmoid10
2mo ago
I call it the conservation of effort. If you are able to save mental energy in one place, it necessarily has been spent in another.
69.
▲
by
sigmoid10
2mo ago
All these studies are highly limited. In addition to general self-reporting issues, you can't say anything about natural day and night cycles unless you carefully control all light exposure (and ofc other stimulants like caffeine, exer
70.
▲
by
sigmoid10
2mo ago
If it is missing context, it will be indistinguishable from mistakes that humans could make as well.
71.
▲
by
sigmoid10
2mo ago
You are conflating several aspects here. The 6am hustle culture originated during the industrial revolution in Britain. And the reason why people struggle with melatonin cycles is because it is almost impossible to escape artificial light.
72.
▲
by
sigmoid10
2mo ago
These are usually from a missing context (i.e. specifying where the car is right now). Tokenizer issues (how many Xs are in Y) are a better giveaway, but models can solve them using tools nowadays. I think the best way to uncover an LLM is
73.
▲
by
sigmoid10
2mo ago
Eliminate all screens after sundown. It's literally the nr 1 reason why people struggle. I used to as well (I could easily work in front of a monitor until 4am without getting tired with no coffee or anything). After that, avoid any br
74.
▲
by
sigmoid10
2mo ago
I've let it research and implement pretty complex stuff for my 3d engine. Like a realtime volumetric cloud rendering solution. The kind of stuff you can do yourself, but it would probably take weeks, if not months. I spent quite a bit
75.
▲
by
sigmoid10
2mo ago
I find speed alone would be a game changer for current models. I hardly find any task anymore that the current frontier models can't do with max reasoning after several rounds of feedback (provided sufficient instruction and the right
76.
▲
by
sigmoid10
2mo ago
That's a nice anecdote, but I didn't just make up this fact. According to their paper [1], Flora has 4851 supported species, while PlantNet covered 30,261 at the time. Flora also tends to mix up cultivated/ornamental plants w
77.
▲
by
sigmoid10
2mo ago
PlantNet covers everything, Flora is more central Europe focused and has a hard time with non-native plants.
78.
▲
by
sigmoid10
2mo ago
Google was always at the frontier of real research, but has been abysmal at shipping good products (at least since Sundar). The core company is run for margins and interest rates by the business people nowadays.
79.
▲
by
sigmoid10
2mo ago
The last chief communications officer of OpenAI used to work at Apple for 7 years according to her linkedin.
80.
▲
by
sigmoid10
2mo ago
Tbh, the amount of time I spend reading documentation has fallen off a cliff. I barely even bother looking for it, much less reading it. Every good agent system with a frontier LLM as the brain and connection to the web can get you this inf
81.
▲
by
sigmoid10
2mo ago
Then it's for sure a guidance or missing context issue. When given full access to everything they need, these models deliver results on par or above the best coders I've ever met and they are lightyears ahead of the lower 90% of d
82.
▲
by
sigmoid10
2mo ago
Which model and harness did you try?
83.
▲
by
sigmoid10
2mo ago
Which model/agent/harness tool did you use? I've found what you describe was my exact experience some ~6-8 months ago, but since about a month or so the game has completely changed. Using 5.6 Sol with highest reasoning settin
84.
▲
by
sigmoid10
2mo ago
If your perception of a moat is worth more than the GDP of most countries, publishing anything must seem like a bad idea. Much less helping others build upon your work.
85.
▲
by
sigmoid10
2mo ago
Wdym? According to their FAQ, you can opt-out for Kimi memberships too.
86.
▲
by
sigmoid10
3mo ago
I feel like it is on you to understand these classes first though. In game engine programming, operator overloading is often just vector algebra. If you have no idea how matrix-vector multiplication works, then you have no business doing th
87.
▲
by
sigmoid10
3mo ago
>given that there are in fact many humans who will outright refuse machine driven cars Are there though? I feel a lot of this hate is concentrated on Tesla and most of that is probably just a projection of hate on Elon. These may be the
88.
▲
by
sigmoid10
3mo ago
I wonder how long software projects will even be able to ban LLM use in an age where exploits are found by LLMs. Like, if you have two forks of Debian and one uses LLMs to fix exploitable bugs and the other one doesn't, the level of se
89.
▲
by
sigmoid10
3mo ago
Maybe. The RCE was on the Telegram Desktop client though.
90.
▲
by
sigmoid10
3mo ago
Actually, the EU should very much crack down on any foreign company that abuses its market position to stifle local business competition. There's no good reason to allow them to break local laws and diminish the local economy just beca
More ›