« first day (2932 days earlier)      last day (27 days later) » 
03:00 - 21:0021:00 - 00:00

6:13 AM
@Blue Huh? Did something happen with Balarka?
 
@SirCumference no, I think he just lost interest when we stopped talking about memes and started talking about physics instead.
2
 
Oh congrats oh modship btw :D
 
@SirCumference He just got busy with university life I guess. Also, he's staying in a hostel (dorm) now, so probably he doesn't get time to visit chat as much
 
I see. Was he ever a physics enthusiast, or mainly math?
 
When you're with classmates you don't feel like chatting on SE tbh :P
@SirCumference Thanks! :)
@SirCumference Mainly math
Maybe some of the mathy parts of physics...I dunno
 
6:17 AM
@Blue Ah I'm not too sure about that :P But I agree that the chat's been dying down recently. Pretty sure Bernardo, Kaumudi, Kenshin, and a lot others have been absent, not to mention Balarka and 0celo
 
Balarka used to live in the same city as me (albeit a 2 hrs drive away). It's unfortunate we couldn't ever settle on a time to meet. :P
@SirCumference Yeah, I agree. Kaumudi hasn't turned up in while (several months actually)
 
@Blue Ah, I'm pretty sure I'm the only one here who lives in my current state, or my hometown NY
 
We're better off without Kenshin though XD
 
@Blue Huh, something happen?
@Blue Most of the people on that list aren't even predominantly interested in physics. They were just fun to talk to :/
Seems in the meantime it's become either physics or politics here :P
 
@SirCumference Umm, well I sorta found him annoying. He used to claim to have solved famous open problems in math and ask us not to steal and publish his world-class physics research theories :P Now I don't know whether he mostly meant that as a joke. In any case, it didn't turn out well
 
6:22 AM
0celo was the biggest hit to the chat imo. Not gonna rebring up the issue, but he was definitely one of the most active here
 
Yeah, I do miss 0celo
Let's hope he'll be back in a few months, as a better person
 
@Blue Oh, guess I don't remember that lol. Welp, there are certainly other people who used to be active back in the day whose names I'm too tired to recall
@Blue Not even half way done :/
I don't think breaking online chat rules makes you a bad person per se :P
 
Okay, replace that with "with a better online personality" ;)
 
Fair lol.
 
We used to get several flags a day in this room, and we were attracting the attention of the community mods - not in a good way. Now we get maybe one flag a month. I consider this an improvement.
On the SE suspensions are not a punishment. They are a signal from the SE that the SE thinks it is better off without you. From the SE's point of view it is hard to disagree with the suspensions from this room.
 
6:27 AM
@JohnRennie I don't think flags are the best measure for the quality of chat conversations. We have more people following the rules, yes, but a lot of the good/interesting discussions have died down
 
The work interesting means different things to different people. Here it should mean interesting to physicsts. This is after all the official chat room of the Physics SE.
 
@JohnRennie *That's a signal that the mods think SE is better off without you. Plenty of people have vocalized their opinions that 0celo should be unbanned. But I'm not going to bring up a 6 month old topic again
 
The mods act on behalf of the SE. I've been party to conversations with the community mods and I can assure you that the suspensions here reflect the views of the SE. And ultimately it's their site.
 
@JohnRennie I'm also not going to debate what should be "interesting" here. Most everyone here has a shared passion for physics, but that does not mean we should discourage other conversations because of that.
 
There is no rule saying you cannot discuss anything but physics. But the plain fact is that this chat room is now largely controversy free, and in the past it had got to the point where the SE were seriously concerned about it.
 
6:32 AM
Yes, but I'd say it's boring as well, because many people have left. That's just my opinion though, and whether it's related to the decrease in conflicts I'm not sure
 
5 mins ago, by John Rennie
The work interesting means different things to different people. Here it should mean interesting to physicsts. This is after all the official chat room of the Physics SE.
 
Again, I don't see the logic behind that, but I don't want to debate philosophies on what people should be allowed to discuss
Chat activity's died a lot, many of the people have left, etc.
 
On the brighter side new users like enumaris and danielunderwood have joined :) But yeah, we need more mathematicians!
 
If you strongly want this chat to be only physics oriented, I guess you're getting that. But it's silly to discourage everyday chat simply because we are all interested in physics. The two aren't related
@Blue Yeah, it's nice to see new people show up and stick around :)
Wow darn, I realized even heather's gone
@Blue I think we might be the only undergrads still here (along with Eulb if he's still around). To think, a year ago half the people here were just starting college. Chat's changed a lot
 
I guess that's right. Obe (Eulb) hasn't been here for several months too
The problem solving room is doing pretty well though
Lots of high schoolers
Yashas and I had created that room just before our JEE tests. So it's nice to see the progress :P
JR and Sammy are the professors there now :)
 
6:50 AM
@Blue Well I think there's a difference between that room and here. Most high schoolers are on those rooms to do well in their physics class, not because they plan to pursue physics
 
@SirCumference Yeah, that's for sure
It doesn't involve any of the juicy aspects of physics or math ;)
 
Yep. Tbh the most interesting things that happen here are now gateprep's incidents and the occasional slereah meme. That's a bit of a low bar, not to sound harsh
Welp I ought to head to sleep. 'Night
 
Goodnight! :)
 
 
1 hour later…
8:00 AM
there seem to be no chat rooms in reality. I have never heard of a real-world chatroom.
 
8:25 AM
@0celo7 lives
Phew
 
 
2 hours later…
10:46 AM
What does "positive integer" mean in english
is it $n > 0$ or $n \geq 0$
 
@Qmechanic I'm curious about your edit here; is it our policy that software recommendations be treated as resource recommendations? I thought otherwise.
 
11:04 AM
@JohnRennie Hi
Is this correct?: "the recently proposed methods hold great promise for increasing the efficiency of ...."
 
@lılostafa morning :-)
 
or is it "hold a great promise"?
 
I would say are very promising rather than either.
But if you're going to phrase it your way use hold great promise
 
@JohnRennie It's actually "these lasers [are very promising][hold great promise] for increasing the efficiency of..."
If I use "are very promising", shouldn't it continue like "are very promising options for..."?
 
Ah, in that sentence I'd probably use hold great promise.
 
11:13 AM
Thanks :-)
 
@DavidZ : I would prefer if res. recom. includes all kinds of res. recom., e.g. books, software, etc, for simplicity. Here is a related meta discussion without a clear conclusion.
 
11:34 AM
@DavidZ I don't see any meaningful difference between recommendations for software and books as far as topicality here is concerned.
I do wonder whether it's better asked on scicomp, but that's a different story.
 
@Qmechanic Thanks, good find. As noted in that question, I would prefer the opposite, that be reserved for educational resources, and that we have a separate tag like for those software recommendation questions which are on topic.
I guess we should probably resolve this one way or another - not right this moment, but it would be nice to figure it out soon.
 
11:52 AM
Any takers for fixing this question's title?
-3
Q: What could likely be the reason?

Usha N The answer which I wrote is: "The voltmeter is connected parallel to 3 ohm resistor but not connected to 6 ohm so it showed the values of 3 ohm" Is it right answer?

 
@Chair It's off topic as is, and as long as that remains the case, I wouldn't worry too much about fixing the title. (Not that it's a bad thing to try to get it fixed, I'm just noting why I, and anyone else who feels similarly, wouldn't feel motivated to do that.)
 
@DavidZ The problem is that most homework-questions seem to have very distinctly off-topic titles. "What could likely be the reason?" could potentially be an example of a good homework-type question, since the OP's asking about how something works. This one is blatantly off-topic only once you acknowledge that the typed our question from the assignment is literally "What could the reason be?"
 
12:23 PM
> and anyone else who feels similarly
that's a nice qualifier there
you should make a club
@Chair The problem with blatantly non-descriptive titles is that they pollute lists
this one is at score below -3 so it won't be shown on the front page
it won't come up in any search that isn't opening itself to scraping the bottom of the barrel
and it will be deleted by the system within nine days
(unless it's fixed, that is)
does it really need a title fix?
 
@EmilioPisanty Now that it's been closed without an answer, it probably wouldn't benefit too much, since it's a sure candidate for the Roomba.
 
12:55 PM
Welp
Preparing to send a resume and cover letter to that Bruxelles lab
 
Let $\mathcal{C}$ be the boundary curve to the calotte surface in the polar angle interval between $\theta$ and $\pi$ in the figure
Ignore the red line
Why is the outwards normal to the surface $-\hat{r}$??
Direction of the curve is counter clockwise, seen from above
 
1:26 PM
If anyone still has an appetite for fun question titles,
 
1:42 PM
@Slereah did you sort references and leave time for them
 
I included references on my resume, yes
the thesis advisors I had
 
Are you sending more off to other places
Just remember you're supposed to know nothing but some of the basics :p
 
I'm an old man
I gotta sell myself
Jangle my sinister to-tems
Behold
Am I not credible?
I'm guessing I am much lower priority since I am 1) old 2) not their student 3) not from a fancy university
 
Nah man, none of that matters haha, it's all completely random
 
Not very reassuring
 
1:54 PM
Right, which is why you want to fire off about 10
 
Oh that I will
But trust me 10 isn't gonna cut it
I need to send as many as I possibly can
I mean the hard part isn't so much finding a thesis, it's finding a financed one
 
A good idea on top of this is maybe pick a few people and skim their papers and send a personal email saying you're interested in the topic and are wondering if they are taking people on
 
Yeah there's a few cool people it seems
 
But like everything is set up to be ready for you to say no at every step so you've got freedom
 
There's Henneaux as I said, but I can't really say "Hey i didn't read your book but I really want to"
Dammit Henneaux, put more examples!
 
1:57 PM
The thing is, you actually can
But someone said he's not taking people on right
 
Who is this someone
 
In here a few days ago
 
And has Henneaux seen my rippling musculature yet
Take me Henneaux
I will be great
 
Why not try it
 
I mean I guess I will
It's too bad the people I actually read a lot of papers from are not really anywhere nearby
or are dead
 
2:05 PM
"I brought with me a bag of physics books to read, but I did not imagine that I could start original research. You see, I had the idea that before I started research on any topic, I first had to know everything that had been done in that area, and I knew that I was far from knowing everything about anything. It wasn’t long before people at the Institute let me know that everyone there was expected to be working on some sort of research." Weinberg
 
Oh don't worry
I don't pretend or plan to know everything before applying
But I'd like to at least underline that I haven't been doing nothing these past 8 years
Also the motivation
 
Yeah
 
Gotta show some
And if I get an interview I'll brush up on some QFT before going!
Because it's been a while since I've done even master-level QFT seriously
Well, at least for the interacting case
 
yeah some look aight
Orsay has a bunch of nice labs
Like research on black holes, modified gravity, string theory, constructive QFT
That kind of stuff
Also there's other belgian universities that look nice and are nearby
Like the VUB
Also Ku Leuven
although they're like
In the flemish zone
So it's gonna be trickier
I'll have to use my fanciest english
"Een link tussen anti-de Sitter en de Sitter"
"Restricties door kwantumzwaartekracht op effectieve veldentheorieën via snaartheorie en holografie"
The horror
 
2:24 PM
man, rolling shutter is the weirdest thing
is it just me? It feels like the reflection should be a reflection of the blade's rolling-shutter pattern, instead of a continuation like if instead of a reflective surface you had a window
I mean, I can make sense of why, but it's still bizarre
actually, no, wait, I can't fully make sense of it
 
Hmm
If I am doing the flux of a solenoidal field over some surface, can I instead do flux of a surface closing the first surface but with an inwards normal vector instead?
Because the flux over the closed surface is $0$
 
3:16 PM
could someone give an example of a system that is not a thermodynamic system?
and would that also mean that in such a system, the second law of thermodynamics doesn't necessarily hold? (I would think that the first law still holds, because it's basically a statement of energy conservation?)
I'm guessing a mechanical system, though in a technical sense, even a mechanical system could be considered as a thermodynamic system, if you specify the thermal properties of the components of your system
 
@EmilioPisanty Dunno if you were tagged correctly in OP's reply, but physics.stackexchange.com/questions/440668/…
 
@JohnRennie a visible-light dupehammer, seriously?
how do you get that without getting the optics tag badge?
 
Many of those could use the optics tag but don't desperately need it.
 
Probably visible light problems are more specific than optics problems (e.g. rainbows, scattering of light in sky, colors etc. type questions)
 
39
Q: Could we make tags imply other tags?

cletusThe prime example is C# and .Net. 95% of C# questions are also tagged .Net. The reverse isn't the case so it's not a synonym. Basically, C# is a child of .Net. Would it be possible to create some sort of one-way linkage like this?

This would be a nice feature to have ^
Often some tags automatically imply another primary tag
 
3:46 PM
tag disambiguration wikipedia like feature?
 
we will not be doing trees, in any way, shape or form — Jeff Atwood Jul 6 '09 at 15:30
:/
@Secret Not exactly disambiguation; more like hierarchy (aka tree-structure)
 
i see
 
Unfortunately I think that increases complexity in the codebase
And they're not gonna implement it
 
@Blue ... until you think about what it would look like in practice and you realize that it would actually be a pretty terrible idea
 
@Blue I think that the concept of a hierarchy would be useful in a lot of cases though: like a great example here is the relationship between newtonian-mechanics and classical-mechanics, both of which are pretty active tags.
 
3:52 PM
@EmilioPisanty Umm, why exactly? (apart from the fact that it might make the response time a bit slower and increase the underlying complexity)
 
@Chair It's debatable that that is a strict inclusion
 
There's also the fact that some tags are not exactly strict subsets of another. So that's one valid point
 
but even if it were, the number of cases where you have strict parent-and-child relationships between tags is so minimal that it'd be essentially useless
@Blue not some. Almost all.
 
@EmilioPisanty I've never seen any counter examples. Lagrangian- and hamiltonian- formalism are clearly not subsets of CM, but newtonian mechanics is AFAIK.
 
@EmilioPisanty True, I agree. But in those cases where they are subsets it would save a lot of manual-retagging time. Especially on large sites like SO
 
3:54 PM
https://chat.stackexchange.com/transcript/message/47606663#47606663
On that, well, that's what happens for most people who move from high school to uni. They mostly get so busy they disappear from whatever they used to hang out in
 
@Blue which cases? You'll be hard-pressed to find more than four or five pairs per site.
 
It's probably a huge deal on SO though: like all python-library-related stuff would need the python tag; the same goes for all programming languages.
 
For instance you have Python3.x, Python3.x and Python
 
all of which would need to be community-vetted, with substantial disagreement over edge cases
 
@Secret Great, that gives me about 8 months :P
 
3:56 PM
ultimately, though - tags aren't folders, and SE is committed to the former over the latter.
You're barking at the moon on that one.
 
@EmilioPisanty Good points. I guess the ratio would be higher on SO than sites like PSE or MSE. For instance, searching for Python gives me so many related tags: stackoverflow.com/tags?tab=popular
Searching for java also returns around 30+ tags
 
@Blue SO is a bad testbed, simply because of its sheer size.
the question volume allows it to sustain a ton of tags that wouldn't be viable at any other SE site.
 
Has there ever been a meta post about splitting SO into sites for separate programming languages? I'm sure there must have been one, but a couple of searches yield nothing.
 
Unless you're proposing software features specifically with SO in mind, pulling examples from there is a bad way to go about it.
 
@EmilioPisanty Well, I was citing examples from SO because if SE is going to ever implement that feature it's going to be on SO first. And if they do that, implementing it on the other SE sites shouldn't be far-fetched. Wouldn't slow down the response/loading time too much
There are very low chances that they'll build something like that (a tree structure) specifically for the sites other than SO
 
4:02 PM
ugh. Have it your way. I'm out of this discussion.
 
Oops :P I do agree you that have several valid points
I was just putting forward mine
Anyhow, good day!
 
@EmilioPisanty expecting the tagging habits of our members to be make sense is somewhat optimistic :-)
 
Why does a burger have less energy than steak ?
 
@Blazar ::vague question alert:: :)
What type of burger?
 
@Blazar it doesn't usually. Burgers tend to contain more fat because they're made from cheaper meat, and fat has twice the energy content of meat weight for weight.
I would expect 100g of burger to contain more calories than 100g of lean meat.
 
4:13 PM
> Why does hamburger have less energy than steak?

Because hamburger is in the ground state!
8
 
Lol
@JohnRennie ✌️
 
@Blue where did you get that from? :-)
 
I am not able to understand the use of meta ? Like, why do we have something like meta when we have SE
 
@Blazar the meta is for questions about how the SE works.
So for example our main site physics.stackexchange.com is for questions about physics while physics.meta.stackexchange.com is for questions about how physics.stackexchange.com works.
 
vzn
4:18 PM
@Blue Aaronson is a real trip, check out his blog, quite colorful + high visibility scottaaronson.com/blog seems weird how little he/ its cited in here, in contrast to eg Lumo who gets a lot of attn. btw what Bruner paper are you referring to?
 
@Blue I'll add it to my list of stock jokes!
 
@JohnRennie wow, but who has the authority to modify something in SE ?
 
Yeah, I love Aaronson's blog :D
The breadth of his knowledge has always amazed me
 
vzn
@Blue also its very impressive how he gets dozens of comments per blog on occasion, sometimes over 100... reader engagement these day like that seems quite remarkable... otherwise cyberspace audience spread thin so to speak...
@Blue (looks great) would really like to see some more careful review/ analysis/ revisiting the CHSH analysis wrt Bell inequalities...
 
@Blazar The employees of Stack Overflow Inc, of course! You'll occasionally see one of them responding to a feature request or bug report on our meta site (physics.meta), but they're frequently communicating (though arguably without listening) with the community on meta.stackexchange.com, where people discuss the working of all 174 stack exchange sites.
 
4:48 PM
hmmm
I got a grey notification icon but after clicking it, I don't see a highlighted message
odd
dmckee's message is gone...o.O
 
@enumaris Yeah, @dmckee had messaged you but later deleted it.
 
@Slereah $n≥0$ is a nonnegative integer
 
I see...
 
5:35 PM
@enumaris I thought about it and realized that I couldn't find any actual evidence for what I had written. It was about the 2000 Florida re-count and the positions to the camps staked out.
 
0
A: How can I plot a heatmap on a sphere given a list of latitudes and longitudes?

E.P.There's two separate problems here: defining suitable bin sizes on a sphere so that you can build an appropriate colour function, and then plotting that colour function on a 3D sphere. I'll provide a Mathematica solution for both. 1. Making up example data To start with, here's some example dat...

here goes
 
6:03 PM
@dmckee ah I see XD
Tencent Games has contacted me...apparently they have a division here in LA, that's interesting :D
 
@enumaris Game development job?
 
6:23 PM
Game AI Engineer
but I dunno if they mean like RL type of AI or more traditional game AI stuff
I'm obviously not familiar with traditional gaming AI
 
Probably deep learning type. They made several popular modern games like PUBG after all
Sounds like a good job to me ;)
 
we'll see :D
As of March 2018, it is the largest video game company in the world.[4]
I didn't know that lol
Riot Games (100%)[7] They own Riot Games which develops LoL
minority stake in Blizzard and Activision
interesting
welp, just finished presenting to the CTO, now I'm tasked with making a presentation to the CEO lol... I do gotta say that my current job gives me a lot of access to senior leadership
 
6:55 PM
I saw a game AI position at Epic that looked pretty neat, but they seemed to be looking for more of a game dev background
 
yeah I dunno
 
I have no idea what game AI entails though. I'd guess RL-type stuff, but I'm not sure
 
this recruiter told me my profile was "shining" and matched the job perfectly lol...
doesn't hurt to connect anyways
 
@EmilioPisanty Only 706 of 3144 questions tagged are tagged
 
if it's more traditional game AI oriented I probably can't do that job
@ACuriousMind not surprising since most people don't have a catalog of tags memorized :P
 
6:57 PM
I probably wouldn't have even know of the visible light tag if it weren't for this discussion
 
sometimes recruiters will just ghost me tho...so I'm not holding my breath on this one
 
Like contact you then ghost you?
 
openai hasn't contacted me back after my last convo with them...
Yep...it's kinda weird...contact me, ask me to add them or w/e, and then ghosted
 
Huh that's weird
And yeah I haven't heard back from my code challenge and am pretty much just watching my inbox at this point
Though the position I applied for is in a new office for them, which they're evidently still searching for
 
Looking at the questions, many of them could perhaps use the tag, but many couldn't - e.g. because they are about some quantum-mechanical aspect of visible light and not about any optical principles
 
7:01 PM
@ACuriousMind build an A.I. to figure out all the tags
and auto tag
ez
 
Are there things in QM that are specifically about visible light and not the whole spectrum?
 
do you apply to a lot of positions?
 
@danielunderwood Not really, but the question can be about some phenomenon observed for visible light (and the asker doesn't necessarily know that it's more general!)
 
or are you like mostly waiting for people to contact you first?
 
@enumaris Result: Every question is tagged with
 
7:02 PM
if the A.I. says so, it must be true
I trust A.I. unconditionally
(I want this recorded so that I will get priority treatment after we are all enslaved by A.I.)
 
I've pretty much applied for everything. Most of the recruiters I've heard from are from companies that I had applied for anyway. There have been a few that weren't, but those positions were pretty bad
 
@enumaris Very well. You will be executed first.
 
@danielunderwood I see...
@ACuriousMind what!? why!?
 
I don't really have much of a profile anywhere except github though. I have a linkedin, but just have job titles on there
 
:(
@danielunderwood I find Linkedin is more quality
 
7:05 PM
@enumaris It is more merciful than having you watch all the other humans being executed.
 
indeed and monster are flooded with random recruiters that just send you a bajillion JD's that may or may not fit your background...
@ACuriousMind hmmm
I don't have a profile on github
no repository
I haven't written a library I want to open source yet
 
But the green squares!
 
wat
 
The dark green squares help you land a job apparently ;)
 
There's a chart of all your contributions from the past year on your github profiles. You get green squares on days you make contributions
 
7:09 PM
I see...
 
....mine is mostly white. No jobs for me
 
Mine are green with blank chunks every so often
 
if I ever get back into writing my RL library and finish it
I might github it
but right now my policy gradient is failing so no point lol
 
You don't need a library for Real Life, just go outside :P
 
reinforcement learning!
 
7:40 PM
@enumaris is there a standard starting place/reference for handling time-series data?
 
like...RNN's or something?
I don't know of a standard for something as general as "time-series data"
 
magic would be preferred
 
I learned RNN's through a bunch of different resources...but the Coursera Course is the most clear
Andrew Ng...good teacher
 
But yeah I thought RNN, so I guess I'll try that...when my scripts to preprocess data finish in about 10 years
 
tbf though I did learn a lot of about RNNs before I took the Coursera Course so I had some background already
RNN is just one tool for time series data though
you can also use convolutional architecture for it
for even just feed forward
or attention only
like a transformer
depends on the specific task you want to do
but "time series data" is uber broad
 
7:43 PM
Is anyone a special relativity person?
 
hmm sounds like I should throw my data into a bunch of different stuff and see what I get
 
I'm special
 
7:59 PM
If a particle decays into two photons, what is its mass given the momenta of the photons?
 
@SpaceOtter ...what particle decays into two photons?
 
I don't know, I just want to work out it's mass haha
 
There's no particle that does that
 
I know
It's a theoretical question in my problem set for this week and the maths isn't working for me
Here's the question:
By using the Lorentz invariant for energy and momentum, or otherwise. Calculate the mass
of the particle which decays into two photons with momentum $(0.76𝒊 + 0.16𝒋) 𝑀𝑒𝑉/𝑐$ and
$(1.1𝒊 + −0.23𝒋) 𝑀𝑒𝑉/𝑐$.
Momentum of the old particle is easy enough but I don't know how to proceed without working out the mass of the photons
 
Photons have no mass.
 
8:09 PM
@danielunderwood you should understand what the different architectures do so you can better know which architecture might fit your task
 
Well I have been meaning to read those attention papers
But data + bucket of models = learning!
 
hmmm
attention itself isn't difficult to figure out or implement
attention is all you need is a bit more different than what one is normally used to
the basic idea behind the generic "attention" mechanism is to use the entire time series outputs (of an RNN) rather than just the last state
so your "final output" isn't just the final state of a RNN, it's some weighted average of all the states
how that weighting occurs is learned and termed the "attention mechanism"
that's about all it is
 
Ahh that's not bad at all
oh no
> Leoforce is the very first company to launch AI / robotics in recruitment and is on a high-growth trajectory
 
o.o
 
They have some data science positions open in the area I'm looking for jobs in
hmmm
 
8:22 PM
there is only the implementational detail of how to learn weights for the attention mechanism since its inputs should not be forced to have the same length...I forgot this detail lol
try it :D
 
But then I'd be like a super-recruiter
 
probably lol
Andrew goes over Attention in his course
I'm gonna see if I can access the course through my work computer...it would be useful to my work as well
ooo it's not blocked!
nice
 
The intro one?
 
coursera courses lol
I can rewatch the attention courses right now to see the implementation detail I forgot lol
 
Do you do all the homework problems or just watch the videos?
It kind of felt like I wasn't getting much out of the homework (for the intro course at least)
 
8:36 PM
I did all the homework
helps to reinforce the concepts imo
 
8:50 PM
ahhhh got it
the weights $\alpha$ are not themselves learned, but they are generated based on a small neural net that computes them at each time step. The weights in that small neural net are shared between time steps so everything makes sense. There we go :D
 
I mean they would be learned if the small nn is being trained, no?
And are the homeworks in the other courses matlab as well?
 
the other courses are in python lol
matlab is way slow
$\alpha$ are the outputs of the neural network which is trained
 
03:00 - 21:0021:00 - 00:00

« first day (2932 days earlier)      last day (27 days later) »