tmoertel

27 ms·

361.
by tmoertel 3y ago
Thanks for sharing this fascinating tale!
362.
by tmoertel 3y ago
Thanks for the recommendation!
363.
by tmoertel 3y ago
Does anyone know of an off-the-shelf mini PC that supports ECC memory? As memory capacities increase, I don't know why ECC isn't becoming more common.
364.
by tmoertel 3y ago
The problem with to-do lists is that they get to be huge, way more than you can do in a reasonable day, and seeing a long and ever-growing list of tasks that you can never finish creates stress and inhibits your motivation. The mind hack to
365.
Large Language Models Are Effective Text Rankers with Pairwise Ranking Prompting(arxiv.org)
3 points by tmoertel 3y ago|0 comments
366.
by tmoertel 3y ago
Agreed! When writing about challenging technical subjects, we ought to realize that a reader may not fully grasp one explanation, even if it's written clearly. Buttressing it from another angle can reconnect the reader with the intende
367.
by tmoertel 3y ago
Readers, if you want a great into to probability co-authored by Joe Blitzstein, see https://stat110.net/ .
368.
by tmoertel 3y ago
Draw at random just six people from a population of millions. Take the min and max wealth of those six people. That procedure produces a confidence interval that contains the true median wealth of the population more than 95% of the time.
369.
by tmoertel 3y ago
> > But if a sample is random and representative, you could estimate population statistics of millions (or even infinite) with a sample of just 30 or so. > While that statement is technically correct, it is useless in practice. I d
370.
Making Middle Earth maps with R(andrewheiss.com)
1 points by tmoertel 3y ago|0 comments
371.
by tmoertel 3y ago
For anyone who is curious, the weighted sampling logic is the SQL translation of Algorithm A from Efraimidis and Spirakis's 2006 paper [1]: Algorithm A Input: A population V of weighted items Output: A WRS of size n 1
372.
by tmoertel 3y ago
The `ORDER BY RANDOM() LIMIT n` idiom is fairly well established for taking a random sample of uniformly distributed rows. It is also easily augmented to take weighted samples: `WHERE weight > 0 ORDER BY -LN(RANDOM())/weight LIMIT n
373.
by tmoertel 3y ago
Note that `FROM table TABLESAMPLE SYSTEM (n/N)` is a safe substituion for `ORDER BY random() LIMIT n` only when you don't need a sample that's guaranteed to be from a uniform distribution of rows. That's because TABLESAM
374.
by tmoertel 3y ago
Note that TABLESAMPLE SYSTEM is not guaranteed to give you a uniform sample, as it samples blocks, whereas ORDER BY RANDOM() LIMIT samples uniformly distributed rows. Some database systems support TABLESAMPLE RESERVOIR to guarantee uniform
375.
by tmoertel 3y ago
The general idea is that you develop the algorithm and the proof of correctness for the algorithm at the same time. The two views are duals in a certain sense, and each gives you insight about the other.
376.
by tmoertel 3y ago
This seems very similar to the "second-order diff" trick: https://blog.moertel.com/posts/2013-02-18-git-second-order-d...
377.
by tmoertel 3y ago
But the choice isn't between doing a startup and doing nothing. It's between doing a startup and taking an established tech job at, say, a FAANG. For most definitions of success, the second route leads to success with pretty high
378.
by tmoertel 3y ago
The CO2 sensor in the View Plus id advertised as follows: NDIR Sensor (Non-Dispersive Infra-Red): Measurement range 400–5000 ppm Optimum Accuracy ±50ppm ±3% within 10 – 35°C / 50 - 95°F and 0 – 80%RH after initial calibration time Can
379.
by tmoertel 4y ago
The article you linked to suggests otherwise: > However, food economists are skeptical an inquiry would uncover wrongdoing. > “I don’t think we’ve seen anything that makes us think that there’s something there other than normal econom
380.
by tmoertel 4y ago
Does cooking not kill the parasites?
381.
by tmoertel 4y ago
"Set in Garamond with Scene as display" Does anyone know which Garamond?
382.
by tmoertel 4y ago
The problem is that "impact" is easy to establish for shiny new things, whereas for preventing problems – you know, maintaining things – the impact is how much better you have made the universe compared to what the universe woul
383.
The Computer Chess Club(archive.org)
1 points by tmoertel 4y ago|0 comments
384.
by tmoertel 4y ago
One easy way to avoid this problem is to give each feature its own independent space of “dice rolls” by hashing the user ids with feature-specific constants before interpreting them as dice rolls: feature1_enabled = hash(user_id + &qu
385.
by tmoertel 4y ago
Ah, it looks like there's a typo in your earlier post. It currently reads: "My solution numbered the balls in base 3, then did four weighings with no if/thens..." (Emphasis mine.)
386.
by tmoertel 4y ago
Related question: If the alignment of the rotating discs goes out, can these chains turn into undercarriage destroyers?
387.
by tmoertel 4y ago
One big gap is causal learning. A true general intelligence will have to learn how to intervene in the real world to cause wanted outcomes in novel scenarios. Most current ML models capture only stastical knowledge. They can tell you what i
388.
by tmoertel 4y ago
In this case the actor isn't the bee but the bee's genetic programming, which only propagates via the queen and hive. If the programming allows a bee to cary a fatal disease back to the hive, the program ceases to propagate and ma
389.
by tmoertel 4y ago
If you want to get better at writing, be deliberate : practice with the goal of learning and becoming better. (1) read good works and study their authors' choices and techniques, (2) write often to strengthen your writing muscles and
390.
by tmoertel 4y ago
Consider the monoid abstraction in the context of batch processing. Anywhere you have a monoid, you have a corresponding “banker’s law” that says you can find the generalized sum of a collection of items by partitioning the items into group

More ›