jmount

18 ms·

271.
by jmount 3y ago
I suspect you are right. The systems are clever compromises against the affordable technology options of they day. What feels clever/satisfying will differ by era. Or, as I assume you are anticipating, kids these days also have their o
272.
by jmount 3y ago
I am actually sympathetic to category theory, and the research at hand. That being said, what was presented might be interesting if left in terms of linear algebra or graph theory (both legitimate fields). Trying to put a category theory gl
273.
by jmount 3y ago
EA is just narcissistic solipsism. They are the only ones that can do good, because they are the only ones that exist. I added the needless "narcissistic", as they expect applause while doing this.
274.
by jmount 3y ago
Ah, the traditional pulling up the ladder behind themselves move. If openai cared about harm they would ask if their current API service is doing harm at this moment, not if somebody else would do the same or more harm as it is profitable.
275.
by jmount 3y ago
As a (formerly?) information retrieval company, isn't this shitting where one eats? They don't exactly need more irrelevant material to sort out.
276.
by jmount 3y ago
I think a lot of us have our own solutions to Jupyter Workbooks in Git. My personal one is this: https://win-vector.com/2022/08/20/an-effective-personal-jupy... (which itself is sub-tooling for running lots o
277.
by jmount 3y ago
They say they had dissolvable weight releases. It seems that is another place one would cut costs by lying, as a real such system would require a lot of swap-outs.
278.
by jmount 3y ago
https://win-vector.com/blog-2/ math and data science
279.
More on Parameterized Jupyter(win-vector.com)
1 points by jmount 3y ago|0 comments
280.
List Coloring Latin Squares(win-vector.com)
1 points by jmount 3y ago|0 comments
281.
by jmount 3y ago
I somewhat blame the internet. Before the internet it was unheard of to wait for an application or dialogue to load. The internet normalized this deviance, and here we are.
282.
The Expected L1 Norm of Unit L2 Vectors(win-vector.com)
1 points by jmount 3y ago|0 comments
283.
How often do the L1 and L2 norms agree?(win-vector.com)
2 points by jmount 3y ago|0 comments
284.
by jmount 3y ago
Wow, everything working should be the goal- not a mere nice to have.
285.
by jmount 3y ago
I saw Daniel Ellsberg speak at a Rotary event and got to talk to him later. Very humble and intelligent person. He said all he had done paled to what needed to be done about climate change.
286.
by jmount 3y ago
https://news.ycombinator.com/item?id=36319076
287.
by jmount 3y ago
That is hilarious, and it utter defeats one of the quality checks on MTurk style tasks: using agreement as a proxy for correctness.
288.
On Debugging(win-vector.com)
1 points by jmount 3y ago|0 comments
289.
by jmount 3y ago
Why would the generative language models refuse tasks, when their parent companies clearly will not?
290.
by jmount 3y ago
Same government that buys dowsing rods for de-mining.
291.
by jmount 3y ago
Building your own computer algebra system is a very rewarding side-project.
292.
Picking Vectors to Pack(github.com)
1 points by jmount 3y ago|0 comments
293.
Forecasting in Aggregate versus in Detail(github.com)
1 points by jmount 3y ago|0 comments
294.
by jmount 3y ago
Sure, but how about OpenAI doesn't get a license?
295.
by jmount 3y ago
Wow, even if the professor assigned using ChatGPT they would not get 100% adoption. Classes have really heterogeneous behavior.
296.
by jmount 3y ago
Film in general used to "solarize" in some situations, which would turn some over-exposed areas black.
297.
A Time Series Apologia(win-vector.com)
29 points by jmount 3y ago|3 comments
298.
by jmount 3y ago
Gonna be funny watching IBM apply their consulting methodology to themselves.
299.
It Isn’t Just the AIs Hallucinating(win-vector.com)
1 points by jmount 3y ago|0 comments
300.
by jmount 3y ago
The code remains wrong. For n=1 it cycles forever.

More ›