timdellinger

5 ms·

31.
by timdellinger 3y ago
A couple other observations on the experimental side: They define success as being a sample with >50% of the target material. I guess that's success, but wow you can't test any actual properties (hardness, electrical conductiv
32.
by timdellinger 3y ago
For all the automation effort, there's always something that has to be done by hand... From the experimental paper: "The XRD sample holders must be cleaned manually when the lab has depleted its stock"
33.
by timdellinger 3y ago
As someone who probably puts way too much time into literature reviews, my big hope is that literature reviews at the beginning if a research project will be revolutionized. Things are re-discovered across adjacent fields of study all the t
34.
by timdellinger 3y ago
A side-by-side before-and-after with the "previously best" image to compare against the new image would be great to see.
35.
by timdellinger 3y ago
As an aside, deep sea oil drilling and production platforms are incredible put-a-man-on-the-moon engineering feats. If oil weren't politically unpopular, the engineering behind these would be celebrated and admired.
36.
by timdellinger 4y ago
An argument can be made that model weights are a "creative work", in the same way that a list of pixel intensities and RGB values is a creative work. Who knows whether the argument will hold up or not. As with all IP, it's o
37.
by timdellinger 4y ago
I second the suggestion at the bottom of the (very enjoyable) post, pointing readers to the Crime Pays But Botany Doesn't content on YouTube - their passion and their approach are fantastic.
38.
Why your Fortune 500 company should start its own GPT team(timdellinger.substack.com)
2 points by timdellinger 4y ago|0 comments
39.
by timdellinger 4y ago
Serious question: what's the half-life of institutional learnings at large tech companies? It's been approximately 20 years since Clippy (~1997-2004). Has MicroSoft literally forgotten all about that, due to employee turnover?
40.
by timdellinger 4y ago
This seems to be getting the Hug of Death, but this looks like the content: https://betterprogramming.pub/pandas-illustrated-the-definit...
41.
by timdellinger 4y ago
my perspective, which might not be universal: the teen at the construction site was acquiring knowledge that wasn't otherwise available to him. the teen in front of youtube only sees previously available knowledge. we're wired to
42.
by timdellinger 4y ago
Not until they start paying 5x the market rate for my salary.
43.
by timdellinger 4y ago
The list of different ways to tune a guitar is monstrously long: https://en.wikipedia.org/wiki/List_of_guitar_tunings Same for banjo: https://zeppmusic.com/banjo/aktuning.htm
44.
by timdellinger 4y ago
In 2005, you could trust Amazon ratings and reviews. The early phase of - well, basically everything - tends to have authenticity. Then the opportunists step in. This is true of any trend, be it cultural or technology.
45.
by timdellinger 4y ago
this is a good place to start: https://scholar.google.com/scholar?hl=en&as_sdt=0%2C47&q=aca... e.g. https://eric.ed.gov/?id=EJ1178415 which describes cultural differences in cheaters
46.
by timdellinger 4y ago
Back in the 90s, I would occasionally get a phone call on my land line from the phone company, trying to upsell me on things like call waiting that they could charge me more for. My reply was that I only want two things: the ability to make
47.
by timdellinger 4y ago
Personally, I don't find this disrespectful at all. It's good for stakeholders to understand what you're doing, why you're doing it that way, and what the alternatives are.
48.
by timdellinger 4y ago
do you "get" BattleBots? Teams build a robot, and then they compete in a combat arena against other teams' robots. The design and build and strategy are as important as the person controlling the robot.
49.
by timdellinger 4y ago
?? Perhaps you're thinking of Formula 2, where teams use the same engine, chassis, and tires. Winning is more about driver skill, and about execution on the part of the team. Formula 1 is widely regarded as a team sport where designi
50.
by timdellinger 4y ago
I'm not sure if these are "book shaped" ideas, or perhaps more blog/webpage/Instagram shaped.
51.
by timdellinger 4y ago
I'll also take this opportunity to plug Semantle, which is an infuriating game that I can't resist. It's based on word2vec. You get infinite guesses, and you're given feedback on how "semantically similar" ea
52.
by timdellinger 4y ago
CHANT LIMED YOURS and I almost always get it on the fourth guess, sometimes in three guesses
53.
by timdellinger 4y ago
I assume that GPT-3 is just exhibiting rote memory. For small numbers, it has accurate answers memorized from the training set, but for larger numbers, it just "remembers" whatever is close... hence the ability to estimate. My ta
54.
by timdellinger 4y ago
my recollection is that the original journal article announcing GPT-3 included some data on how it performed against SAT-style questions
55.
by timdellinger 4y ago
if you read the thread: PMC = professional managerial class
56.
by timdellinger 5y ago
Agreed! You as a user are the product , you are not the customer . Google is a marketing company. Charge money, and their supply of product drops like a rock.
57.
by timdellinger 5y ago
The belief among GME enthusiasts is that the shorting wasn't through puts (options), but rather through straight up short selling. And that yes, this has been going on a long time and is in fact costing the hedge funds a lot of money.
58.
by timdellinger 5y ago
The alternative is to have these business processes managed using internally developed software that may or may not be supported as time passes... excel via email might (unfortunately) very well be the more robust and reliable way to go her
59.
by timdellinger 5y ago
Does anyone have a good lead on where I should direct my outrage, where decision-makers might be listening? (I'm a Toyota owner, but not effected by this.)
60.
by timdellinger 5y ago
I'll also add that "holy grail" type discoveries tend to be published in higher tier journals than this one.

More ›