Comparing confidence metrics of the networks themselves is like comparing two athletes by asking them each how good they are and declaring athlete B the winner of the race because he thought he was better than athlete A thought about himself.
@eis | Hacker News
From 11/22/2010, 4:18:24 PM till now, @eis has achieved 3781 Karma Points with the contribution count of 298.
Recent @eis Activity
I see you are working at "dagshub". Maybe you can let the people know that it's not a good show to create fake accounts here to push the story and leave useless praising comments.
This is mentioned in the article. It was local debt, not foreign.
During Russia’s financial crisis and ruble collapse of 1998, President Boris Yeltsin’s government defaulted on $40 billion of its local debt.
The last time Russia fell into default vis-a-vis its foreign creditors was more than a century ago, when the Bolsheviks under Vladimir Lenin repudiated the nation’s staggering Czarist-era debt load in 1918.By signing a contract with TSMC? That's the fab according to the article. 5nm is not bleeding edge and as others move to N5P or N4, capacity gets freed at N5.
You should probably specifically mention PoW and not crypto mining in general. There are crypto mining mechanisms like PoS which use a negligible amount of energy.
Yup exactly. But other bets (surely they are not betting on the exact same thing with just different odds and even if they were they) don't influence the chances of this bet. These are independent events. Either you think you are likely to win or not. To refuse 1:1 odds on the grounds that you could make more money somewhere else means either A: you do it for the money and not the sport or B: don't have enough money to wager on all these bets but then he should ask for a lower amount or refuse with that reasoning. Maybe I am missing some other possible explanation? The reason of refusal is very important to understanding the motivation behind it.
If there are two people offering you a bet:
Person A offers you 5:1 odds in your favor saying that a random dice throw will yield a number small than 3.
Person B offers you 1:1 odds in your favor saying that a random number chosen between 1 and 10 will yield a number bigger than 6.
Thinking about the wager with Person B is independent of the wager with Person A. When deciding which bet to engage in the answer is both because in both cases you should be convinced that your chances of winning are >50%. Refusing the second bet would lower your overall expected winnings.
Because that is not what happened. Grimburger did not offer (from his point of view) worse odds than 1:1. And patio11 refused 1:1 because it was apparently not interesting to him.
Now one could take that as a salesmans tactic to try and extract better odds from Grimburger but at that point the monetary aspect would become the focus and not the wager itself. A wager between two people who are in it for the sport and both sure of their positions should carry 1:1 odds. One could ask for a lower amount or refuse completely on monetary grounds but not request odds in ones favor.
Asking for 5, 10 or 25 to 1 odds in your favor means you are expecting at least 20%, 10% or 4% chance that you are right.
The fact that you don't want to engage in 1:1 odds means you are less sure of your own position than he is. Just sayin' :)
Another negative story about a YC company which has risen to the front page of HN and then seems to be getting a lot of pressure to get ranked down. It has more points and is fresher then a handful of other stories on the frontpage but ranks lower and is falling quickly. And when mentioned in the past, the reason often is the invisible anti-flamewar or similar features. What is it this time?
Elon needs to buy HN and open source the algorithm! /s
Hi, good to hear that you guys care about testing. One thing apart from the Github issues that led me to believe it might not be super stable yet was the benchmark results on https://h2oai.github.io/db-benchmark/ which make it look like it couldn't handle the 50GB case due to a out of memory error. I see that the benchmark and the used versions are about a year old so maybe things changed a lot since then. Can you chime in regarding the current story of running bigger DBs like 1TB on a machine with just 32GB or so RAM? Especially regardung data mutations and DDL queries. Thanks!
It's good to hear that persistent indexes are coming soon. I saw it was on the roadmap but didn't know how far out this feature was. Do you have an idea when that release could be out?
BTW Do you have some kind of code/docs one can take a look at regarding the index structure? I'm a part-time data structure nerd :)
I was just yesterday exploring DuckDB and it looked very promising but I was very surprised to find out that indexes are not persisted (and I assume that means they must fit in RAM).
> Unique and primary key indexes are rebuilt upon startup, while user-defined indexes are discarded.
The second part with just discarding previously defined indexes is super surprising.
https://duckdb.org/docs/sql/indexes
This was an instant showstopper for me or I assume most people whose databases grow to a bigger size at which point an OLAP DB becomes interesting in the first place.
Also the numerous issues in Github regarding crashes make me hesitant.
But I really like the core idea of DuckDB being a very simple codebase with no dependencies and still providing very good performance. I guess I just would like to see more SQLite-esque stability/robustness in the future and I'll surely revisit it at some point.
Here an english article: https://www.reuters.com/world/europe/spanish-prime-ministers...
Spanish prime minister's telephone infected by Pegasus spyware
4 points • 0 comments
That paper seems to be for the obsolete Mir-BFT protocol. The library readme mentions that they have switched to a newer successor algorithm called Insanely Scalable SMR (ISS) but they have not changed the name of the library which can lead to confusion. I have found the right paper which is unfortunately not linked to:
Elon Musk to buy Twitter for $44B
9 points • 0 comments
Russian forces invade Ukraine after Putin orders attack
2278 points • 1814 comments
A tip for people who get lost in the interface:
1. Just close the tutorial
2. Scroll down to the ToS checkbox and check that
3. On top in the "input utilization" row make sure only text is checked
4. Enter your text (use only landscape terms)
5. Press the arrow to the right inside a rectangle button located below the text input.
6. Maybe zoom out a bit because the result will be the image on the right which for me was out of view by default. Had to zoom to 50% to see the whole UI.I missed the part where it said it was trained only on landscapes. So I retried it with just those and got this:
"river flowing through desert": https://i.imgur.com/QSjH5hk.png
"sunset waterfall": https://i.imgur.com/wS3EEci.png
Or how about a lovely "green shoreline": https://i.imgur.com/RkbTV99.png
It's not just a matter of showcasing the best results. It's a matter of night and day difference between normal output and the one showcased. They are not even remotely close. I actually wonder why they decided to release this to the public in its current form. I was very impressed by the noise suppression app they released so I expected something that delivers decent results.
site design / logo © 2022 Box Piper