jmount

15 ms·

241.
Tools for Jupyter in (and Near) Production(win-vector.com)
2 points by jmount 3y ago|0 comments
242.
by jmount 3y ago
I thought He is mostly alpha-radiation capturing a few electrons. I.e. that it is produced by radio isotope reserves and escapes no matter what one does. And not that it was in big pockets of pre-existing gas that one is exhausting. So it m
243.
What You Should Know About Linear Markov Chains(win-vector.com)
1 points by jmount 3y ago|0 comments
244.
by jmount 3y ago
So the original "14000" leak is 0.1% of the customers, but now 6.9M is claimed. 14000/6900000 ~ 0.2%, so now half the customers are indirectly leaked?
245.
The Biased Drunkard's Walk(github.com)
1 points by jmount 3y ago|0 comments
246.
A Deep Dive on the Drunkard's Walk(github.com)
1 points by jmount 3y ago|0 comments
247.
Consider Evaluating LLMs as Search Engines(win-vector.com)
1 points by jmount 3y ago|0 comments
248.
by jmount 3y ago
This is really dreary.
249.
Yet Another ChatGPT Winge(win-vector.com)
2 points by jmount 3y ago|0 comments
250.
A Slightly Unfair Game(win-vector.com)
1 points by jmount 3y ago|0 comments
251.
Unrolling the ROC Curve (With Animation)(win-vector.com)
2 points by jmount 3y ago|0 comments
252.
A/B Tests for Engineers(win-vector.com)
1 points by jmount 3y ago|0 comments
253.
by jmount 3y ago
"There is another system." (worst demo ever)
254.
by jmount 3y ago
And lead is still allowed in aviation fuel. The planes (only piston?) are spraying it all over you every day.
255.
by jmount 3y ago
Small proviso: this is only true for a reasonable number of categories (or you run into repeated experiment problems).
256.
Schemas for Python Data Frames(win-vector.com)
2 points by jmount 3y ago|0 comments
257.
by jmount 3y ago
Very cool rule. I think you can justify it by approximating each category as an independent Poisson distribution. Then for each such processes the variance equals the mean. So once you have 10 successes in a bin, you have evidence of a prob
258.
by jmount 3y ago
So those are the subjects, not co-authors.
259.
by jmount 3y ago
It isn't a cynical opinion if that turns out to be what is cynically being done.
260.
Solving for Hidden Data(win-vector.com)
1 points by jmount 3y ago|0 comments
261.
by jmount 3y ago
I thought all the Hacker News "smarter than everyone else, self prescribers" were still endorsing Adderall, mild hallucinogens, and poop-swapping. But, by all means, shine on you crazy diamonds.
262.
Detecting Data Differences Using the Sphering Transform(win-vector.com)
2 points by jmount 3y ago|0 comments
263.
Omitted Variable Effects in Logistic Regression(win-vector.com)
2 points by jmount 3y ago|0 comments
264.
by jmount 3y ago
This is the usual if a hazard isn't labeled, it isn't a hazard fallacy.
265.
by jmount 3y ago
I rode a train in Spain where the train changed gauge at a station!
266.
by jmount 3y ago
Even gamers hate gamification. External rewards often decrease activity, as the convert anything into a chore.
267.
by jmount 3y ago
Couldn't get Triffids cuttings.
268.
by jmount 3y ago
Why would you even want a no-GIL Python? Java and C showed how much more effort it takes to maintain slower thread safe code for no real benefit. Parallelize at the fork level or at the isolated numeric library level.
269.
by jmount 3y ago
The "missing 1" is a waste-category that is implicitly re-scaled. The explicit 1 formulation is used in binary softmax, and the implicit (not seen 1) is used in multinomial softmax. I suspect this is the old "notation B looks
270.
by jmount 3y ago
I think one of the flaws is: nobody seems to want to agree that editing is far more expensive than writing (the to point where editing is routinely skipped). I think it is well established in software engineering debugging is far more expen

More ›