#ot1-perplexing-regexing
1 messages · Page 75 of 1
thanks man
yeahhh that´s vim, one question do you use lsp?
tabline 👎🏼
date in statusline 👎🏼
i do
hahaha i think you´ve spent a lot of your time just setting your nvim
one could say that
can you tell im listening to a JS podcast 
its lsping time
robinerd
anyway done with the workout. time to go home. peace. ✌️
later rex
Topic?
Or is general js
me looking for who asked: 
have you looked into customizing telescope's layout
sadge
let me see your telescope
I have the default ivy layout too
I wanted to yoink in case you tried something
i'll experiment with it and let you know
i'm not a fan of the large padding between the boxes
I'm not a fan of the whole layout at all
i kinda like this
yeah
maybe this is more to your liking?
no preview though
I want a preview lol
don't know about the themes, i'm just looking online for layout inspiration
are you on windows but using wsl?
indeed
oh okay, you have a good laptop
like, maybe that, but without the file preview header, because it's obvious
true
300th anniversary episode so they picked highlights from previous 100 https://open.spotify.com/episode/0vNYqfcEanjrNDDdqv01Lw?si=hGvcmWFRSJSanCl3vHmtKQ
embed fail
anyway, one of the hosts just mentioned how she needs to update her linkedin bc her company is undergoing "restructuring" so shes back on the market again

then another host was like, i remember recently doing that and doesnt recommend cold applying since companies get like 5000 apps sometimes and the ATS system sucks
so he says the cheatcode is to just have a coffee chat with an employee and if it goes well, ask for a referral
you cut the queue with that manuever
but honestly with these JS juggernauts, the moment they mention theyre looking for work, all sorts of people will contact them directly tbh
so i wouldnt be worried
i added the coffee part, the guy just mentioned reaching out to people
you have to be wary of the rex filter
like, do you invite people for a coffee chat over LinkedIn?
obv

yeah. you never done that before?
rex does it bc hes rex
but we already established this
I don't use LinkedIn much other than the jobs section lol

at their stage, they know the types of companies they would want to work for vs. not. plus they actually have a network so they know their friends experience, etc.
they invite all these JS guests onto the show too
can you imagine 300 episodes

Sigh
When can i be like this
maybe you should start listening to this JS podcast

its actually pretty good
@quasi blaze it's your otn 🥺
lonk pls
That wasn't Keezy; it was his shadow ™️
There is no shadow anymore
He has integrated, wholly and completely
Become one with the collective. Resistance is futile.
Im not entirely sure i want to work with js yet
Im actually having a struggle figuring out what i wanna do
If i stay here longer i'd be stuck with monkey frontend work
bro finally made it? ? ?? ? ? he cooked? ? ? ?? ?
listen its just a podcast. not a relationship. doesnt mean you have to work with JS daily.
i mean look at me
you think i work with JS daily?
ha
I would definitely date JS personified

no comment except for
×2
💀
He forgot the +C when he was integrating his shadow
speaking of philosophy, i found a subreddit for egoism memes today
Not respecting your own property is such a shit cat take imho
rex's podcast snippet #874:
"The job of linking together a bunch of APIs."
"Yeah I mean I call that software development."
💀
glue code
python developers seething
🅱️ython 🅱️rogrammers
just in case:
did you know they auto set-up this for you now. previously, had to do this manually.
i get an awful taste in my mouth when i look at this
do you also want some bleach
Suggest more topics here!
It's time for the game! Let's begin.

4️⃣
6️⃣

:tada: @quasi rose won this game! :tada:

rex: is this even real?
this is despicable
You misspelled delicious. Understandable mistake.
Have you tried?
There’s no way that could be good
technically true
good is subjective
there is evil: psycopaths, murderers, villains
then there is this
my good sir are you alright
I supposed it might be less terrible with the flavorless noodles
It's chocolate, hazel nut, and noodle. It would taste exactly as it sounds.
they smell kinda funny though
But that texture though….
🥴
but spaghetti was my favorite food and i didnt notice a difference when i had to switch
the texture would be fine
have you guys seen the movie elf
yes
now, HE knows how to make noodles
whats it abt
dementia patient goes crazy
cant spoil it but theres more than dementia
its like found footage
hm
idk sounds a bit heavy for my silly little brain
especially with black friday
:incoming_envelope: :ok_hand: applied timeout to @stone belfry until <t:1699750685:f> (10 minutes) (reason: emoji spam - sent 22 emojis).
The <@&831776746206265384> have been alerted for review.
L
:(
i just wanted to post crat 
no
si, y tu?
(spanish for no)
tan maleducado smh
sisisi
its been 3 years since spanish II
yeah, I work in a factory in south Texas, I've met lots of interesting people
yo no se, es algun servidor en que estoy, pero tienes emojis buenos
epik
The Spanish learning server has these wolf emojis -- 
They used to have one called ":wolfthuglife:" which was sunglasses and a leather jacket, and that was my favoritest emoji ever, but it went away for some reason and now I am very sad
that is the weirdest emoji i have seen in a long while
wait nooo why they remove that emoji
the silliest emoji
i saw a ping here
where is the ping
he swaggin
stollen
He is your forever
bet i will join

Yes?
permission to leek city?
Sure?
dallas is not south texas
bro got the 3 hour commute
dallas to south texas is >>> 3 hours
you was at the club , bottles up when I first met you . couldn't get enough, had to get you straight to my bedroom 😏 @high verge
what
ever
(i have no game)
NO
25729 / 50000 total words
Half way there, living on a prayer 👀 Solid NaNoWriMo thus far.
Ahead of schedule! You're doing great, keep it going 
I think I just figured out what the story is about too. haha
Row houses everywhere
phone pictures of screen
in an npm build script are you supposed to have "build": "npm install && other stuff"
or are you supposed to run npm install and npm build separately
ive never seen anyone use those 2 things together
usually we have separate build and install phases in CI
i use this at my company because i need to npm i to run the unit tests :)))))
oh, this also lets me run different build steps in parallel, i use this with a monorepo at work with separate build steps for FE and BE

apparently "NaNoWriNo progress" matches syllable-wise with "living on a prayer"
@alpine tusk This should work
for (size_t i = 0; auto& x: whatever) {
// rest
i++;
}
Where whatever is an iterable of sorts
STL iterators don't expose index as far as I know
Just like iterators in Python only iterate
do for loops not need the third bit?
wait, how does this work
Like in C#
And Java and JavaScript and whatever
In JavaScript, it's of instead of :
To see what
the initializer
the if initializer is since C++17
I'm no C++ programmer
ah, init in ranged for is C++20
I mean I've done a little bit of C++ and Lua in my life
if(auto x = get_sth(); x) {
// I like it
}
explains why I haven't seen it
Not enough to discover initializers in if statements
not while
But those are if statements on repeat
there you don't need it
you can do
if(auto x = get_sth()) {
// I like it
}
Ye
since forever
if you need a more specific check the initializer form is great
Ye, I was originally doing to do a .is_present on that
scope pollution
this will also call your dtor later I would assume
you could do this
{
auto x = get_st();
if (x) {
// I like it
}
}
```but kinda eww
fair
They literally just copied and pasted a C++ class definition and changed braces to colons and only added def to instance methods
So you get class variables and C++ constructor/destructor definitions
Nice docs
I know
you get the scoping, but indenting galore
basically just some syntactic sugar for this yeah
and symmetry with for
They're probably restricted to one statement
As opposed to throwing it into a scope with (m)any statements beforehand and afterwards
it's one scope for the whole if else chain afaik
Happy Diwali 🪔
you too!
You know you are crazy when you watch a video the length of the average free code camp python tutorial
i fucking hate how undertale fans keep trying to make sans look cool in their mods
he's wearing fucking crocs and shorts
but they insist on pouring so dramatic effects on him it ends up making him look edgier than shadow the edgehog
man
I forgot how low the bar for fanfics is when a game hits tumblf
tumblr
Hi
they are 1d
They are zero dimension, you can never see them coming and they are infinite
Really creepy lines
Does anyone have knowledge of css?
I can't promise to answer, but ask your CSS question ahead
My css animation is extending beyond the specified width and height.
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<style>
* {
margin: 0;
padding: 0;
box-sizing: border-box;
}
.container {
width: 100vw;
height: 100vh;
border: 5px solid black;
background-color: rgb(199, 199, 109);
}
.box {
width: 350px;
height: 80px;
background-color: blue;
border-radius: 10px;
margin: 10px;
border: 2px solid red;
position: relative;
animation-name: rightMovement;
animation-duration: 5s;
animation-iteration-count: 2;
animation-delay: 2s;
}
@keyframes rightMovement {
from {
top: 0;
left: 0;
}
to {
top: 0;
left: 1200px;
}
}
</style>
</head>
<body>
<div class="container">
<div class="box">
</div>
</div>
</body>
</html>
dont rizz me, dont come by ohio, we are done.
When i run this my box somehow get out of the width and height which is width:100vw;
height;100vh;
idt the top and left properties care about that
what do you want to do with the animation? it seems like it's supposed to do that
I am beiggner i was just following a youtube video. I did same as he did but idk why i am facing this
what did it show in the video?
Why do you have so many tabs open 😭
Make fun of them for the phone picture of screen instead
@marsh python sorry for the ping, got a user sending out malicious stuff
you could just tell mods
@tropic lichen (sorry for the ping, seen you active in another channel)
dm @polar knoll about it
heya, i've dealt with that person now - but yeah, like ryuga said, DM'ing @polar knoll is the best way to contact us mods for stuff like this
No worries, was more as an fyi while i was sorting noms
do people actually respond to these...?
i know a guy who has like 200 tabs open at once
like rn
what type of pc wtf😭
@old gyro i found.... something
from math import pi
import math
def converter(n, decimals=0, base=pi):
# If negative, save this and convert to positive.
if n < 0:
negative = True
n = abs(n)
else:
negative = False
# Generate the values for each row (like in binary but instead of 2^x it is base^x) until
# you reach a value equal to or larger (n/2)-1
headers = []
header = 1
while header*math.floor(base) < n:
headers.append(header)
header *= base
headers.append(header)
print(headers)
# Add values for the decimals if necessary
header = 1
for i in range(decimals):
header /= base
headers = [header]+headers
# Convert the number into the appropriate base representation using the headers
number = ''
for header in headers[::-1]:
# If it is the first decimal, add the decimal point.
if header == 1/base:
number += '.'
if header <= n:
amount = min(math.floor(n/header), math.floor(base))
n -= amount*header
# If it more than 9, format as hex
if amount > 9 and base > 10:
number += 'ABCDEFGHIJKLMNOPQRSTUVWXYZ'[amount-10]
else:
number += str(int(amount))
else:
number += '0'
if negative:
number = '-'+number
return number
i think it converts to pi somehow?
it makes use of pi constant
and by looking at this
yes i know, i wrote the code
therefore making it the perfect base ||/s||
I mean a base of pi could represent a product of any given exponents of pi, just like the denary system can represent a product of any given exponents of 10
It's just that isn't particularly useful
that is in fact something
Huh
i think it just converts to base 3 or smth
i dont really understand it
seems kinda close to the needed base tho
yeah it seems kinda weird, i passed in 3.1415 and it gave me 03 ??
it should be ~10 right?
what's the decimals thing supposed to be?
oh nevermind it's for fractional numbers
i dont understand this i think i wrote it like a year ago
i think it also uses 1 as a base
2*pi gives 20 though so maybe it just can't deal with not that close approximations
i think you can specify decimals= to make it more accurate maybe?
oh right
yeah, with decimals=100 3.1415 gave me 12. It should be smaller than 10 though since py is 10 soooo idk
File "/private/tmp/a.py", line 49, in <module>
print(converter(3.1415, decimals=1000))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/private/tmp/a.py", line 36, in converter
amount = min(math.floor(n/header), math.floor(base))
~^~~~~~~
ZeroDivisionError: float division by zero``` uhhhhhh
i dont think this code is very thorough in case you couldnt tell
pick up truck? no way you're that strong
Road work ahead? Yeah, I sure hope it does
Are you telling me a shrimp fried this rice?
cargo? yeah no shit
!e made my own simple impl, and it seems to check out
import math
def to_base(x, b, prec=6):
max_pow = 0
while b**(max_pow + 1) <= x:
max_pow += 1
res = []
for i in range(-prec, max_pow+1)[::-1]:
bp = b**i
res.append(str(int(x//bp)))
x %= bp
if i == 0:
res.append('.')
return ''.join(res)
print(f'{to_base(6.283, math.pi) = }')
print(f'{to_base(6.284, math.pi) = }')
@solemn tulip :white_check_mark: Your 3.12 eval job has completed with return code 0.
001 | to_base(6.283, math.pi) = '13.011021'
002 | to_base(6.284, math.pi) = '20.000000'
and yeah my thing gives
3.011021
apparently
13.0110211100202211300010200021020021220021221211221220200201122101100012203003001211003001101201121223 is to 20 much like 0.99999999999 is to 1
there could be more representations
Damn, that's interesting. (In hindsight, it's kinda obvious tho, I was thinking it should be around 19.something, but this is base pi 🤦)
0.999... is 1
same
but on the only final exam i had so far i didnt cheat
because when 3 teachers just walk around you it is hard to
Maybe the innate inaccuracies in pythons pi make it not exactly that, I wonder if some infinite series approximation could yield some "exact" .999999? Or am I tripping
Also, wouldn't it be almost 10 instead?
not too sure what you mean by "exact" 0.99999 here
Yeah me neither, but I mean, wouldn't 3.14... be closer to 10 in base pi than 3.011...?
that's because of the multiple representations you can have with base pi
in a regular positional base system with an integer, you can't get for example 2^4 in base 2 by using digits less than that
but since we have digits 0, 1, 2, 3 with pi as the base
you can add up past that
for example
!e
from math import pi
print(pi**3)
print(3*pi**2 + 3*pi)
@crystal spruce :white_check_mark: Your 3.12 eval job has completed with return code 0.
001 | 31.006276680299816
002 | 39.03359116403745
so in base pi, 1000 < 330
so notions of size in base pi isn't as simple as comparing digits
Ah, now it makes sense, thanks for the explanation 😃
which also explains the infinite (I think at least) number of representations for each number
since if you have some number like d.ddd3
you can replace that with d.ddd20... for example, and then add digits to converge to some value
that's the intuitive formulation at least, the formal one needs more massaging
I'm not sure about infinite
assuming we restrict things to 0, 1, 2, 3 as digits
Hmm, I guess
sum(pi^-k for k in n..∞) > pi^(-n+1)
so maybe you're right
but yeah, uniqueness is always very iffy with non-integer bases
the algo I implemented greedily tries to fit the highest power of pi possible, which makes for a unique representation
Stall surveys go hard
the bar but with helium balloons
Small bar for me, last time I checked. I have had a fucked up wrist since before then. :c I could do back muscles or simple deadlift, as force expands my wrist then, but I couldn't stabilise it enough on bench
What mathematics is used in compilers, I was reading about garbage collection and a few other topics where maths notations were implemented. Is there any specific topic used more often, or can I describe a whole compiler only using maths? Could anyone give me a good book or an idea where I can read about how maths is used in low-level computer science or computer architecture? Such as calculus related to ml.
at least in the realm of compilers - type systems specifically, https://www.cis.upenn.edu/~bcpierce/tapl/ is a great book.
are you familiar with computation theory? its very important for studying compiler design
you can also try the broader realm of type systems: type theory
My wrist hurts when i supinate it
I was practicing pops on my bike and did some improperly
Hurts to do curls on a straight bar now a bit
I think i used to do 50lbs for 10 reps, doing 3 now
I think single hand DBS would be a better alternative
classic
However, moving forward, we will be investing our efforts into RustRover, which is closed source. For the existing open-source plugin, we’ll do our best to maintain compatibility with newer versions of our IDEs, but we won’t be fixing bugs or adding new features.
well, that sucks
i'll be sticking with vscode for now
Can we preserve data sustainably?
Microsoft's Project Silica can store and retrieve data in quartz glass using laser optics and Microsoft Azure, enabling more sustainable storage that can last thousands of years.
Audio described version: https://www.youtube.com/watch?v=TGsHam6z_s4
Subscribe to Microsoft on YouTube here: https://aka.ms/Subsc...
very interestign stuff but also, why would one need such data for thousands of years
what kind of data are they storing...
@young shoal Funny you mention graph. I can't render this so my HW will be a low res png from screen snip. 
@rugged owl Im double thinking about Upenn's MSE DS program
Classes seem very intermediate level
not rigorous like UT's DS
i am soon becoming a tikz guru
If your doing online, it's not going to be that rigorous either. It'll be what you put into it.
maybe another 10 years of academia will do
I mean at least UT DS has classes like Optimization, Reinforcement Learning and Algorithms
real
You vs asking for gov money for funding.
linalg is nice
I already know Linear Algebra
What's UPenn's course work look like?
Upenn's course has introductory courses
like Intro to Stats with Machine Learning
Intro to Linear Algebra
🥲
Statistics for Data Science
those are like undergrad level classes
dont you think so?
They're standard in most DS masters
none of those classes are in UT's masters tho
DS masters has no requirement of you being in STEM as undergrad
really?
UT DS has that
You have to at least pass Calc 1 to Calc 3 with B+
and Linear Algebra with B
to be able to apply for the program
Are you applying for in-person or online?
Online
Online doesn't really have those requirements last time I checked.
That's a new test they added recent years.
Since too many were failing their basic intro classes if I had to guess.
@rugged owl
these are the courses you have to take
prior to be admitted to the program

Those aren't required courses for entry. The only soft barrier is a 3.0 undergrad, which even if you don't meet, you can use your Statement of Purpose + work experience to justify. (Or show proof of higher grades at a local university w/ classes not for degree. )
Im not sure
they said you needed to have Linear Algebra
as a prereq
you can see it here too
Yea, it literally says "Preferred Qualifications"
You can reference this for hard requirements:
https://ms-datascience.utexas.edu/sites/default/files/checklist.pdf
Yeah but at least they are not having people from Business degree in MSDS
like they do in Upenn
😄
They can accept Business degrees if their gpa is good?
and like i said before, my company would hire them
you can teach technical skills
domain knowledge is important for companies
yeah but it will be extremely boring for me while those kids learn python from scratch
https://docs.google.com/spreadsheets/u/0/d/1CaP7qBUorLHnGUJK9dle6xlqLhCDQkrPQ2lbeVwSMy4/edit?pli=1
For example, this was OMSA for georgia tech. I didn't see one for UT austin when I was applying a while back.
OMSA
Username,Status,GPA1,GPA2,Year,Sem,Application,Decision,Early,Days,Exp,GMAT,GRE-V,GRE-Q,GRE-W,A-rec,P-rec,CS1301,MM,6501
california_burrito,Accepted,2.96,2022,SP,2021-06-10,2021-09-29,111,4,0,3
is_this_taken,Applied,4.00,2022,SP,2021-06-15,1,2,1
-swimmingbird-,Accepted,3.25,2022,SP,2021...
welp. that sounds like a problem youll face at any grad program
I should talk to their coordinator
Upenn's MSE DS program just started this year so
I would be one of the first students in the program
Look at UPenn's full course list is my suggestion. They'll take existing courses from other departments and mix them together is the general MO for these universities.
I did
these are the courses @rugged owl
let me know what you think
Pretty much matches up with UT austin & Gtech. It's pretty standard stuff.
yeah but its mostly coding
That's kinda what ML is, a branch of CS.
not theoratical
like UT's
😄
look at the syllabus
UT added Health Discovery and Innovation in DS program this year
you may like this course @high haven
I would suggest UPenn based on:
- Well known school, especially engineering
- Practical > Theoretical. Day to day matters more to businesses.
- Better connections
There's no major difference between what the two schools teach. **It's your decision in the end. **
okay
Im waiting for UCLA too
😄
The online MS in Data Science degree curriculum is designed to prepare students for a successful career at highly advanced technical organizations.
now they have some neat courses in UCLA @rugged owl
would you expect
divmod(z**-2 + z + 2*z**2, z**-1 + z)
to give you
(z**-1 + 2*z, -3 + z)
# or
(z**-1 + 1 + 2*z, -z**-1 - 3)
```?
basically asking if the remainder should be [) or (]
I'm leaning towards [), because for normal polynomial division something like dividing by 1+x yields a remainder of some constant, which would be [)
isn't it [) by definition?
is there a definition for laurent polynomial division?
I couldn't find any from a Google search, just a paper that talks about there being multiple possible quotient remainder pairs
and I'm looking to apply conditions so that it's unique
so either to include the upper degree of the quotient or the lower degree
hmm
I have a uniqueness proof for either case, so either can work
just the question of which is more natural I think
and what's the original question?
just that
I'm just looking to implement divmod for laurent polynomials
for fun
I also haven't come up with an algorithm for it yet
maybe that might reveal which is more natural
i'd go with that myself
that being?
trying the algorithm on a few examples to see what happens
yeah ok
i thought you were talking about something else at first. off the top of my head i don't see why one would have to exclude any powers
where can I test my app
laurent polynomial division has extra flexibility
i did a fair amount of that, but only in complex analysis. i guess it's the mod that is giving you some special behavior
like any polynomial with a single power acts like constant polynomials because they all leave no remainder after division
I'm like a novice at complex analysis so I might be missing a lot of useful knowledge
so I thought I'd ask and see
what options do you have?
you were accepted into UPenn, yeah?
generally some things you can consider are:
- Your finances. This includes the overall tuition of the school, in state vs out of state cost benefits, and the sort of financial aid/scholarships/grants/whatever you're receiving, whether you're willing to take out loans, and how much you're able to spend in general.
- The school itself. Do you like it? How much so, in comparison to the other's you've applied to? Is it perhaps more prestigious? Which one's program do you like the most?
- The location and size.
- IIRC this is for your graduates? In that case, you could look into the research facilities and such.
Although, these are all things you'll likely have looked into when choosing where to apply in the first place. Did you have a top choice?
we call this the abacadaba method
🪄
The last two answers were A so obviously that one is out this time.
it's probably C (idk i'm superstitious)
i actually had to count the alphabet ngl
someone really said V?
I wish I was joking 
as a biz major we're not that stupid (not all of us at least)
(x)
lolll
I broke my bike coming home from school
@low chasm you still around?
Options are:
UPenn - MSE DS
UT - MS DS
UCLA - MSE DS
NJIT - MS CS
!e
echo ${"Hello Mars":0:3}```
bruh
I cant choose
UPenn is where Trump went to school.
You should just go to Trump University.
does trump university still exist?
Problem is Ive been told that I wont be able to find a job
also i should go to bed, i have school and an interview tomorrow
whoms’t said that
Lol no
damn
that MS DS is a waste of money
that’s the degree i’m going for after undergrad and a few years of work
He told me that I was driving myself under a truck
That's because clowns on YT posted shit like: "Easiest 100k jobs: 1. DS"
Now DS is over saturated.
There's a perception that MS in DS programs are just riding the data science hype to generate revenue for the university
yeah thats what he said @frozen crane
There's a good amount of truth to that lol
And by the mouth of two witnesses shall every word be established.
What gave it away
what would you recommend? @frozen crane
Yo momma
I dunno, I was doing an MS in CS, but I hated it
DS just analytics in a shiny new coat of paint.
make that 3. the differences between those who landed a job after the DS program vs. my program at my grad school was/is drastic.
imo, if you didn't do CS undergrad, your going to suffer a lot in a CS MS.
Was yours CS?
not straight CS but essentially
Machine learning masters of science in computer science? What?
gatech's CS MS isn't a career pivot analytics MS. It's an additional to a strong CS foundation.
specialization
listen, lets think about your career goals after grad school
instead of optimizing for the program
Getting an MS in CS that emphasizes ML is probably the gold standard for scientific computing jobs
???
This is the specialization courses in NJIT for MS CS
Artificial Intelligence
CS 670
Artificial Intelligence
CS 675
Machine Learning
DS 669
Reinforcement Learning
CS 677
Deep Learning
CS 732
Advanced Machine Learning
CS 782
Pattern Recognition and Applications
i would agree with this statement
I wanna work in ML NLP area @civic pasture
omg I tagged mysef
I meant to tag @high haven
AI is very core CS
What's your ~~coding ~~ CS experience in undergrad?
class AnchorBox:
def __init__(self):
self.aspect_ratios = [0.5, 1.0, 2.0]
self.scales = [2 ** x for x in [0, 1 / 3, 2 / 3]]
self._num_anchors = len(self.aspect_ratios) * len(self.scales)
self._strides = [2 ** i for i in range(3, 8)]
self._areas = [x ** 2 for x in [32.0, 64.0, 128.0, 256.0, 512.0]]
self._anchor_dims = self._compute_dims()
def _compute_dims(self):
anchor_dims_all = []
for area in self._areas:
anchor_dims = []
for ratio in self.aspect_ratios:
anchor_height = tf.math.sqrt(area / ratio)
anchor_width = area / anchor_height
dims = tf.reshape(
tf.stack([anchor_width, anchor_height], axis=-1), [1, 1, 2]
)
for scale in self.scales:
anchor_dims.append(scale * dims)
anchor_dims_all.append(tf.stack(anchor_dims, axis=-2))
return anchor_dims_all
def _get_anchors(self, feature_height, feature_width, level):
rx = tf.range(feature_width, dtype=tf.float32) + 0.5
ry = tf.range(feature_height, dtype=tf.float32) + 0.5
centers = tf.stack(tf.meshgrid(rx, ry), axis=-1) * self._strides[level - 3]
centers = tf.expand_dims(centers, axis=-2)
centers = tf.tile(centers, [1, 1, self._num_anchors, 1])
dims = tf.tile(
self._anchor_dims[level - 3], [feature_height, feature_width, 1, 1]
)
anchors = tf.concat([centers, dims], axis=-1)
return tf.reshape(
anchors, [feature_height * feature_width * self._num_anchors, 4]
)
def get_anchors(self, image_height, image_width):
anchors = [
self._get_anchors(
tf.math.ceil(image_height / 2 ** i),
tf.math.ceil(image_width / 2 ** i),
i,
)
for i in range(3, 8)
]
return tf.concat(anchors, axis=0)
stuff like this
So OOP? Did you learn algorithms?
this is a diminishing area atm. you know how saturated it is rn?
more likely for you to become an "AI developer"

im technically an NLP DS
but moving more towards AI engineer type work atm
I did computer vision in Python @rugged owl
during undergrad
Ok? I'm asking about CS fundamentals.
algro ithms
Duha.
how did you do that without knowing about dictionaries or loops 
Kill me now. Gotta get through 4 hours of lecture on business strategy before like Friday so I can actually study for the exam and take it before Monday
sg do you have off next week
I'm about to take monday off next week
that took me hours to code @young shoal
Wait, isn't next week thanksgiving?
yep
I took Discrete
Maybe I should take the whole week off
but no Algorithms class but then they have that class in my masters program
We get Thur & Fri off
didn't you have a CS undergrad?
no I had a DS undergrad
i get the whole week for school ~
how did you not have algorithms 
how do you do DS without any algorithms course as prereq 
lol we had DS algorithms
I think they mean data structures
DS undergrad seems like "data science" 
yeah, so i thought they were using DS in the same way they just used DS
That sounded like a "data science version of algorithms"
"algorithms used in data science"
We love overloaded acronyms
100%
to be fair, it's supposed to be "DSA"
DS is data science now.
You know what else is supposed to be DSA?
yo mama
Dancing Snails Alliance.
lol
I was gonna say your face
Yea, if you had DS undergrad, then I would suggest CS masters
CS MS algorithms tends to be graduate level though
there are also sometimes courses that are essentially remediary
Yeah
For example MS DS and MS CS have the same algorithm class at UT @rugged owl
I gotta call UT tomorrow they said I could start as MS DS and transfer to MS CS
Yea, they share courses.
of course the base required courses are prob the same but then you have a bunch of electives probably
I get Wednesday through Friday off 😤
it's going to be so based
I meant work. I think my classes don't give a shit. Stuff still due Thursday
rip
L bozo
do it. im off tomorrow and friday
Wait, naw. I'm only going to take Monday off.
what about the busiest nights
I'm gonna go to the mall, the bigger one in my area.
ngl I feel dead and it's just Wednesday
just two more days
also rex I live on a big hill so the running will be a bit horrible
but there's this flatter place nearby
but like a 15 minute walk away
no lakes unfortunately
dont do it. youll eff up your arches in the long run
also if you run on side of roads
dont ever do that either
there's sidewalks most of the way
the more you know
sidewalks is fine
are sidewalks fine?
im talking about like on the road itself
what's wrong with the road
oh
oh
so its very bad for your arches
I have a good solution
just run in the center
assert your dominance over the motor vehicles

I gotta get up earlier tomorrow 😩
key club meetings are in the morning for some reason
I have not done a single thing for key club so far I'm kind of questioning why I joined at this point
aboo dont be one of those kids in college
where you join every org
and all you can put on your resume is "member"
lmao

I actually do stuff for the other clubs 🥺
I'm probably in the running for a leadership position next year
there weren't any open leadership spots this year since none of the captains graduated 😔
but they're seniors now and unless they get held back their seats are up for grabs
Get a dog

Idk how you can talk about one graph for 8 minutes. Apparently this prof can.

It's a dog
🥴
Breed: Medium Dog ™️
do I just walk to the flat place 😔
because it's up the hill
american bull dog
what kinda shoes do you have
is my guess
wtf is a proper running shoe
good support for your arches
Nike calls them athletic running shoes
youre good
as long as youre not like this guy #ot2-never-nester’s-nightmare message
Rex bullying be like:
Sidebar: Whoever made this slide... I want to have a word 
bro made a whole ass bipartite graph
cancelled
average business moment
maybe only in academia
if you tried this in industry youd be cancelled
well
in some places
so not converse? >.>
heh
Business
🕵️
Business class
no. not converse. never converse.
the converse is not necessarily true
ok skateboarder
vans on the road
the doggo is SOOOO happy. he absolutely loves boarding 🛹
honestly whats impressive is how you have these tiktoks saved (?) / memorized (?) / (can you tell idk how tiktok works)
this is the work you still have left?
whens the due date/exam?
wait he's actually skateboarding
sg has tiktok brain
No, I still have this + another 2 hour section
So, 11 + 12
damn that's my last super reaction

🧢
Web is even worse?
Australian web is the worst. You walk through that shit and suddenly, spiders.
@young shoal i need a zoomer to explain how sg has these tiktoks without having tiktok

who should i ask
idk
who even has tiktok here
sg bro
Cannot predict now

he's cappin
rex is 0-2 with sir lancebot today
.8ball can rex redeem
Yes definitely
😤
.8ball will it be soon or
Don't count on it

😭
you aren't ahead though
I gotchu
you can visit the web version of tiktok without having an account

thanks aboo

Yt comments is the last place I expected to see LLMs in
the vid: https://youtu.be/YMEuy0G5n34?
ALL CREDIT GOES TO POTENTIAL HISTORY.
Original: https://youtu.be/TFJ_YqnlElY
Potential History made the meme, I just cut down the footage to only the meme part. I'm willing to take it down if Potential History doesn't want me to post it. I just figured his meme was especially funny and decided to cut it down.
tbf I think it's a really good use of LLMs but do these people have compute to burn or what 
not a worse usage than discord
the worst usage of a text gen AI ive seen so far is mee6
genuinely just a waste of compute
its like character.ai but with a redditor
math
meth
A ball is dropped from a height of X meters. With each jump, it loses 20% of its previous height. Calculate the total distance and number of jumps it took until it was less than 1 meter.
seems simple
show your work
after n bounces the height will be X^0.8n, so its a case of figuring out n for X^0.8n < 1/X, right?
yes, and how does one do that?
or in other words, when X^0.8n < X^-1
if both sides are multiplied by X, we get X^(1 + 0.8n) < 1... at which point i think i've either backed myself into a corner or have reached as far as i can get
lol
X*(0.8)^n
not 1/X on the RHS
isn't it?
just 1
X*(0.8)^n < 1
ooh right
yea for whatever reason i thought it being dropped from X metres meant 1/X was 1m
do more… math? or something else
you probably predivided
drugs
which ones?
and then forgot u predivided
yes
meth
noted
yeah
so in that case yeah, like hsp said, X * (0.8^n) < 1
waiting
yeah?
it's not plus
why not
i'm focusing on just the number of jumps
why would it be
i feel like distance can be figured out after
that'd be calculating total distance the ball has traveled downwards
I thought it was min jumps
total distance and total jumps
yes, but you're trying to find what n is first
you could maybe work the other way tbf, figure out distance and then derive bounces
you can figure out total distance afterwards
I'm not saying calculate total distance first
I'm just saying what I think how the equation is forming

oh. doesn't the distance mean there's going to be factors of two because it's bouncing up
yes
that's dumb as shit
am I wrong? 
distance would include the ball going down and up
I'm assuming total distance means from dropping the ball to until it reaches the peak of the bounce that is less than a meter
yeah
which is what I've accounted for with the sum
after n bounces the height is X * (0.8^n), since it loses 20% of its height each bounce
not each half of the bounce or whatever
you're not necessarily wrong, I just have no clue where you're going with that
lol
you only accounted for going down though
I mean you can calculate distance down + distance up for total distance at the end
log
^
it's just that the pertinent information is that "bounce height less than 1", and total distance is kinda adjacent to that
no, the inverse
We've gotten X * (0.8 ^ n) < 1, and have to rearrange to get n by itself
log(X) + n*log(0.8) > log(1)
something like that?
log reverses the inequality IIRC
log is monotonic increasing













