basically how does code work? is it like for eg in maths if we know the formula we can solve the question...?
#voice-chat-text-1
void cedar
i got it; I will come back here and will let you guys know how i did it
you guys are really awesome
i really appreciate the support!
what are your thoughts on current market?
yeah
job
misty sinew
where are you from?
void cedar
wait, you're not employed yet?
misty sinew
bro, you sounds like 30
misty sinew
bruhhhh i thought youre some senior soft dev (by ur voice)
bruhh, fr
proper junco
def factorial(number):
for x in range(1,number):
number = number * x
return number
void cedar
i am a science dude:)
misty sinew
i am a science dude:)
going to AI/ML?
void cedar
going to AI/ML?
i dont have any idea about that currently i am just trying to figure out how i can learn "a programming language"
misty sinew
i dont have any idea about that currently i am just trying to figure out how i c...
I see, have you ever tried before?
misty sinew
from what did you start?
misty sinew
trust me youre a great person!
fr he is
void cedar
i am learning basic py from YT
misty sinew
tried what?
programming
void cedar
okay there's one stupid question:
how is this going to work, let's say i know a language now, how do I implement it?
i want to start with simple and then go on
yeah i get
i think I eventually will get it, will i?
misty sinew
i really want to do it, but then i dont know anything:(
what do you mean by, "then I don't know anything"
void cedar
what do you mean by, "then I don't know anything"
i wanted to be a programmer i wish i work but considering the current downfall and recession i am afraid and that's the reason i am kinda delusional
i have started writing a code in vscode from last week
that's python
misty sinew
i wanted to be a programmer i wish i work but considering the current downfall a...
bro, see this type of situation will come and go, if you're distracted by these things then that obvious you won't be able to concentrate on learning.
so first of all clear these assumptions, and concentrate on learning ik its harder, I've faced this too but focusing on learning first its faar better than first focusing on how much I can earn it with that.
start from basics, ig a course from freecodecamp maybe helpful, where the dr chuck teaches the python
let me see if I can find it
void cedar
are you a dev?
misty sinew
are you a dev?
just joined several months ago
Here it is, https://www.youtube.com/watch?v=8DvywoWv6fI
Ik this one is old, but believe me this will be helpful
misty sinew
are you from asia?
yup
void cedar
I am learning python from a national dev
misty sinew
aah that's good
void cedar
and I feel like he also does explain better
misty sinew
then at what point you're feeling that you're lost?
void cedar
im just confused with my uni and my own learning
I want to learn python first
and then I will focus on java
i feel like i will be left behind if i learn java as they say
misty sinew
Aah this syllabus sucks, fr me it was same, ig Java and (CPP or C#) both were in one semester
misty sinew
i feel like i will be left behind if i learn java as they say
you won't, if programming concepts are clear.
the languages are just medium to connect with it. each and every language has its own unique value you've to find it what will be useful what will be not and at what point.
misty sinew
in simple words, have you played valorant?
void cedar
yeah
misty sinew
counter strike?
void cedar
:)))
misty sinew
call of duty?
misty sinew
see if you know shooting on head, then you can do it in every game, but each and every game has different layout and different sensitivity you've to know and learn to control it.
void cedar
are you employed?
proper ridge
!paste
coarse hearthBOT
lunar salmon
class Admin:
def __init__(self):
self.name = "Shayan"
self.age = "23"
self.__access = "root"
def get_access(self):
return self.__access
def set_access(self, new):
self.__access = new
admin = Admin()
print(admin.name)
print(admin.age)
print(admin.get_access())
print("-------------------------")
age = 25
name = "Borzu"
Admin().set_access("user")
print(admin.name)
print(admin.age)
print(admin.get_access())
Likewise, have a nice day!
lunar salmon
class Name:
def __init__(self, first_name, last_name):
self.first_name = first_name
self.last_name = last_name
def answer(self):
return f"{self.first_name} {self.last_name}"
@classmethod
def splitter(cls, strng):
first_name, last_name = strng.split()
return cls(first_name, last_name)
name = Name.spliter("Shin Ryuko")
print(name.answer())
hoary citrus
class = cls
lunar salmon
class = cls
tnq
proper ridge
def _non_parameterized_single_qubit_gate(self,
gate: Literal["I", "X", "Y", "Z", "H", "S", "T"],
qubit_indices: int | Collection[int]) -> None:
# Define the gate mapping for the non-parameterized single qubit gates
gate_mapping = {
"I": I,
"X": X,
"Y": Y,
"Z": Z,
"H": H,
"S": S,
"T": T
}
# Apply the gate to the specified qubit(s)
if isinstance(qubit_indices, Collection):
for index in qubit_indices:
self.circuit.append(gate_mapping[gate](self.qr[index]))
else:
self.circuit.append(gate_mapping[gate](self.qr[qubit_indices]))
def _parameterized_single_qubit_gate(self,
gate: Literal["RX", "RY", "RZ"],
angle: float,
qubit_index: int) -> None:
# Define the gate mapping for the parameterized single qubit gates
gate_mapping = {
"RX": Rx(rads=angle),
"RY": Ry(rads=angle),
"RZ": Rz(rads=angle),
}
# Apply the gate to the specified qubit
self.circuit.append(gate_mapping[gate](self.qr[qubit_index]))
def _single_qubit_gate(self,
gate: Literal["I", "X", "Y", "Z", "H", "S", "T", "RX", "RY", "RZ"],
qubit_indices: int | Collection[int],
angle: float=0) -> None:
# Define the gate mapping for the non-parameterized single qubit gates
gate_mapping = {
"I": I,
"X": X,
"Y": Y,
"Z": Z,
"H": H,
"S": S,
"T": T,
"RX": Rx(rads=angle),
"RY": Ry(rads=angle),
"RZ": Rz(rads=angle)
}
# Apply the gate to the specified qubit(s)
if isinstance(qubit_indices, Collection):
for index in qubit_indices:
self.circuit.append(gate_mapping[gate](self.qr[index]))
else:
self.circuit.append(gate_mapping[gate](self.qr[qubit_indices]))
@delicate wren alisa, is the new abstraction a good change or did I overdo it?
It works, and passes mypy, but as you mentioned before it's not an indicator of good code.
delicate wren
```py
def _non_parameterized_single_qubit_gate(self,
...
why doees the second one not have the if?
proper ridge
why doees the second one not have the `if`?
Why would it have an if?
hoary citrus
Why would it have an `if`?
You know German right?
thin lintel
proper ridge
Does it come with the flashy nose?
thin lintel
Dried reindeer meat. It is a traditional food and a delicacy of Northern Finland, prepared at springtime. Like jerky, of which kuivaliha is a variant of, its origins lie in the need for food preservation. Kuivaliha is a very useful snack when camping, etc., for its light weight and good nutrition values.
umbral rose
All I want for Christmas is one of Santa's reindeer π
umbral rose
Back in a bit. Gotta bathe these grubby kids.
stuck bluff
Metal detector use here requires a mining license.
Use.
I may be misinformed.
Premium Burials.
sly pond
stuck bluff
It would have to be a bit of an indictment against the competency of doctors of the time.
Well don't they have that hammer thing with the pope? Or is it a spoon? i can't remember.
thin lintel
The corpse is placed in a cypress wood coffin, meant to signify humility and refer to the fact that the pope was an ordinary man first. Papal items, including palliums, or lambswool stoles, coins and medals minted during the individual's pontificate, are also placed in the casket
proper ridge
why doees the second one not have the `if`?
Why would it have an if?
stuck bluff
We have a number of night noise animals here.
Foxes, possums and alpacas.
Bats, occasional birds.
A species of owl.
Which I've never seen, but heard.
proper ridge
I'm gonna take a break, brb.
thin lintel
very noisy
stuck bluff
Didn't hear it, felt the air. The air felt soft.
thin lintel
also noisy
stuck bluff
You see some images of these birds on logs and stuff, and unless you know they're there, they can be very difficult to spot.
stuck bluff
Looks fast.
Dusks and dawns often are graced by a chorus of kookaburras, where I am.
Also Australian magpies.
...and all the others, of course.
Sometimes the magpies get confused at the moon.
Bye bye
proper ridge
@delicate wren I put the help ticket here, if you could have a look I'd deeply appreciate it.
https://discord.com/channels/267624335836053506/1261658498471559168
proper ridge
Dusks and dawns often are graced by a chorus of kookaburras, where I am.
I am so stupid that I half-read your message and for a second thought dusks and dawns are actual bird species and patted myself on the back for thinking Dusk till Dawn could be a cute wordplay.
misty sinew
it's so silence
proper ridge
qubit
versed elk
qubits
sly pond
morning
us
it's early
take out the z's
a lot easier
then think about hte movie frozen
π
it's early
I'm gonna go make coffee, I do have an earbud in, so I am listening
it is life
so cybersec, what drew you in?
sly pond
Yea, coffee maker is downstairs
Yea, I'm pretty quiet in the morning, it's why I'll often just type
gotta have the coffe and wake up
sly pond
Ooh, permanent besties!
Geopolitical dance rivals
Did you see break is in the olympics?
I wonder if India and Pakistan entered, and if they did, it would be an amazing final
no
rarely
Have you changed your diet?
new medicines?
stopped anything?
well then...
I think that's the cause, you're not comfortable in your sleeping environment
hoary citrus
Yes it is a fact
What he said?
same
Is your profile pic a solar eclipse?
Nice
hoary citrus
It looks like the cloud is smiling
hoary citrus
It was for much time?
hoary citrus
That cloud didnt go
sly pond
The cloud came in at like noon
and got thicker and thicker
and ruined things, but in 2017 it was a sunny sky
though totality was nearly a minute shorter
Went to Charlotte for that one, and that's where I took my profile pic
sly pond
editing
then completed the edit
though maybe I was typing, I do sometimes hit the key and delete
hoary citrus
everyone do this
sly pond
I missed that
hoary citrus
Are you proggamming or not that is what he said
hoary citrus
mostly nothing
yeah
hoary citrus
there are elections in us?
sly pond
there will be later this year
but my US news tab is very empty
only one story in there
Apparently nothign else has happened in the US this weekend
stuck bluff
Evening.
Already gone? I imagine not.
Is a fire over when someone stops lighting it?
sly pond
stuck bluff
What actually happened almost doesn't matter. What we need to look at is what effect it will have.
Hm. I don't know what happened. I have thoughts.
Tbf, they make conspiracy theories very easy.
Sorry, the shooter was what?
proper ridge
You're right, you don't even need a sniper to see the guy. I'm sort of wondering how he missed the shot from that close.
stuck bluff
Also, what was with the secret service just letting Trump just stand around pumping his fist? Like, your guy just got shot and you're not forcing him off the stage?
stuck bluff
Did they know it was the only one?
In that moment.
Did they know?
"Yeah, it's cool, just let your face stick out."
How was his reaction normal?
Like, if I get shot at, I'm pissing my pants and freezing.
stuck bluff
Like, if I get shot at, I'm pissing my pants and freezing.
I imagine. I've never been shot at.
Or maybe not. I might just blank. I don't hope to ever find out.
sly pond
β€οΈ bingbot
thin lintel
herschel walker
thin lintel
sly pond
gonna run
fair heron
hoary citrus
@fast path hi
versed lava
@sly pond i know where you're from as long as your nickname is your name
are you from
P-O-L-A-N-D
did I guess?
sly pond
kinda
versed lava
maybe he comes from Poland, but he lives in the US
hoary citrus
maybe
sly pond
Live in the US, most my life
primal mason
stuck bluff
https://tenor.com/view/baldur%27s-gate-3-baldur%27s-gate-bg3-emperor-empereur-gi...
Just vibing?
tranquil drift
yessir
stuck bluff
delicate wren
huh
Russian SCP fork uses Rust+Python for the backend
fork of this
https://github.com/scpwiki/ftml
stuck bluff
delicate wren
Russian SCP fork uses Rust+Python for the backend
if anyone is interested for whatever reason
https://github.com/SCPru/RuFoundation
proper ridge
gusty ether
π
wide cargo
hola
stuck bluff
I'm not useful, there.
grand iris
can i screen share
mild flume
!stream 1115939192141266954
coarse hearthBOT
β @grand iris can now stream until <t:1721060823:f>.
buoyant dagger
!e
import random
random_choice = random.randint(0, 1)
if random_choice == 1:
print("Heads")
elif random_choice == 0:
print("Tails")
:white_check_mark: Your 3.12 eval job has completed with return code 0.
Tails
thanks ppl
buoyant dagger
!e
names = ["Angela", "Ben", "Jenny", "Michael", "Chloe"]
import random
# Get the total numbers of items in list.
num_items = len(names)
# Generate random numbers between 0 and the last index.
random_choice = random.randint(0, num_items - 1)
# Choose and print a random name.
print(f"{names[random_choice]} is going to buy the meal today!")
:white_check_mark: Your 3.12 eval job has completed with return code 0.
Chloe is going to buy the meal today!
wary fable
@raven orbit
delicate wren
!e
```py
names = ["Angela", "Ben", "Jenny", "Michael", "Chloe"]
import random
...
!d random.choice
coarse hearthBOT
buoyant dagger
!d random.choice
I was not allowed to use random.choice in my course π
coarse hearthBOT
buoyant dagger
!e
import random
names = ["Angela", "Ben", "Jenny", "Michael", "Chloe"]
# Get the total number of items in the list.
num_items = len(names)
# Generate a random number.
random_choice = random.randrange(num_items)
# Choose and print a random name.
print(f"{names[random_choice]} is going to buy the meal today!")
:white_check_mark: Your 3.12 eval job has completed with return code 0.
Chloe is going to buy the meal today!
!e
import random
names = ["Angela", "Ben", "Jenny", "Michael", "Chloe"]
random_choice = random.randrange(len(names))
print(f"{names[random_choice]} is going to buy the meal today!")
:white_check_mark: Your 3.12 eval job has completed with return code 0.
Angela is going to buy the meal today!
hey!
yea, I do...
oh god, I forgot your username... I can't ping you...
I'm new to the server so can't speak in the VC right now...
come back...π
hello @stuck bluff
I can't speak ;-;
stuck bluff
misty sinew
Yea, I gotta spend more time here ig...
bruh, I can't even type in the VC room 1 chat lol!
should I share my screen? You guys can help me correct my mistakes... @stuck bluff @proper ridge
stuck bluff
@misty sinew π
misty sinew
ah! I see... So how do I "interact" with the returned value?
oooh... I wrote a code which does not require the return keyword but I can interact with the defined functions regardless.... Can you explain ti me why does that work?
wait, I'll copy and paste the code
def happy_bday():
print("HAPPY BDAY!")
print(f"enjoy being X years old")
print() # <=====to leave a line... makes the code look tidy, nothing else.
happy_bday()
happy_bday()
#So. What did we do here?
#Well, we "defined" a functin called happy bday since we had to compile it multiple times(Don't ask why, I don't fucking know.)
#How to define a function: def function():
# print("whatever u want it to say")
def happy_bday(name, age): #here, we've defined a perameter
print(f"HAPPY BDAY, {name}")
print(f"enjoy being {age} years old")
print()
happy_bday("Pranjal",19) #here, we've assinged the name parameter a "value"
happy_bday("Samarth",19)
happy_bday("some random guy",1000000)
#----------------------------------------------------------------------------------------------------------------------------------
#Let's fuck it up a little, shall we? π
def happy_bday(age, name):
print(f"HAPPY BDAY, {name}")
print(f"enjoy being {age} years old")
print()
.
I make my notes in VS code so it's elaborated...
yep
I mean, I can call and change the defined functions by doinf this too... wait.., what do you mean by that? "it is meant to print somethin and not get info out of it" I didn't get that.... @stuck bluff
I'm so dumb... ;-;
ah... I see
hmmm
ok... I'll need some time to figure that out... Bruh, you guys are so smart! me scared :3
misty sinew
ok... I'll need some time to figure that out... Bruh, you guys are so smart! me ...
πββοΈ yup right...me scared too
I shall apply myself.
Damn, you sound so wise! It's nice to find ppl who can guide me!
Thank you so much! β€οΈ
haha!
you read philosophy?
Are you a professor at some uni? You sound like one... Most of the ppl I've asked help from just ignore me or laugh at my face.... or, my screen... ok, that was a bad joke
I'm sure that I won't be the last lol
abyway, gtg... let's hope that the next time I'm here, my voice wouldn't be supressed by the evils of this world....
[Lightning strike to make it look more dramatic]
k, bye!
hearty heath
No worries π
hearty heath
Let me take a quick look at the code, one sec.
hearty heath
So the wrapping function just checks some stuff before calling the wrapped method?
Why do you use Sequence? Just for generality?
Do the methods mutate the data?
proper ridge
def _single_qubit_gate(self,
gate: Literal["I", "X", "Y", "Z", "H", "S", "T", "RX", "RY", "RZ"],
qubit_indices: int | Sequence[int],
angle: float=0) -> None:
# Define the gate mapping for the non-parameterized single qubit gates
gate_mapping = {
"I": I,
"X": X,
"Y": Y,
"Z": Z,
"H": H,
"S": S,
"T": T,
"RX": Rx(rads=angle),
"RY": Ry(rads=angle),
"RZ": Rz(rads=angle)
}
# Apply the gate to the specified qubit(s)
if isinstance(qubit_indices, Sequence):
for index in qubit_indices:
self.circuit.append(gate_mapping[gate](self.qr[index]))
else:
self.circuit.append(gate_mapping[gate](self.qr[qubit_indices]))
__len__, __getitem__, and __iter__.
wary fable
No mic today, in a quiet zone
proper ridge
Final goal is to have the qubit_indices of the decorated methods to be int | Sequence[int].
The gate methods themselves don't mutate the qubit_indices. I only need mutability for self.circuit_log because I mutate that in self.vertical_reverse() for instance.
hearty heath
Hey Opal
wary fable
Hey Opal
hearty heath
I am confused sorry π
hearty heath
The wrapper returned by the decorator doesn't have a different signature from the function it's wrapping right?
proper ridge
def X(self,
qubit_indices: int | Sequence[int]):
...
@gatemethod
def X(self,
qubit_indices: int | list[int]):
...
Vc0 is full of streamer brainrot lol
Some of these books aren't as useful as others lol
And some I kinda want a copy for myself lol
It usually gets taught alongside sed
Oh boy
Some of these books are OLD
misty sinew
Hello
stuck bluff
@gusty ether @misty sinew π
wary fable
Pretty sure this is pre-json
gusty ether
π
misty sinew
<@870716429568184360> <@456226577798135808> π
π
wary fable
Trying to see if they have a copy of "Hating on Languages You Don't Use"
Settled on these three for this week. The library card is such an unsung hero when you are on a budget.
hearty heath
So as I understand it, your intention with the decorator is to return a wrapper function (with the same signature as the wrapped function) that basically checks some stuff and does some logging, and then calls the wrapped function? It shouldn't matter what the wrapping function does to values it's passed, as long as it's not trying to do things that can't be done to values of their type π€ Have you learned about using ParamSpec to type hint decorators? https://www.youtube.com/watch?v=fwZoxWyMGM8
But I agree with Opal. Keep it simple π
proper ridge
!zen
coarse hearthBOT
uneven python
```
def X(self,
qubit_indices: int | Sequence[int]):
...
@gatemethod
...
if the decorator is the same as we talked about earlier, you'd typehint qubit_indices as list[int]. the @gatemethod would then return something expecting a int | Sequence[int]
(to be fair, i think just having it inside the method definition could be easier, because then you dont need decorators for each signature. but its an interesting problem anyways)
gusty ether
is there a situation where you can say typing is additional work and slows you down in development when you need to move fast?
hearty heath
Type hinting should be secondary when writing Python code in my opinion.
Although it can help guide the design of your code
uneven python
is there a situation where you can say typing is additional work and slows you d...
i am used to statically typed languages so not having a typehint somewhere (where it cant be inferred) makes me confused. type stubs also allow to chart out stuff
proper ridge
if the decorator is the same as we talked about earlier, you'd typehint ``qubit_...
Yes, it's that. I was trying to see if there were other ways around it. You're absolutely right about it being a list.
This is because as you saw if I type hint the decorator (with the callable), it would only be applicable to one signature as opposed to how I have it now which works with any signature.
hearty heath
Type hinting can be fun though. It's like solving little puzzles π
proper ridge
You know, since it's just a checker for values, and just logs the params, what if I just put it inside instead of using it as a decorator?
uneven python
if the decorator is the same as we talked about earlier, you'd typehint ``qubit_...
^ yeah, i suggested that right now. i think thats the least complex approach. maybe will look a bit repetitive and copy-pastish but whatever
proper ridge
^ yeah, i suggested that right now. i think thats the least complex approach. ma...
Yes, thank you so much. Still trying to wrap my head around what wouldn't make people barf.
I honestly want to write this in a clean manner.
hearty heath
Yeah sorry @proper ridge I'm not really familiar enough with the code, so I don't really understand what's going on. I recommend watching that video I linked if you're generally usure about how to go about type-hinting decorators with wrapping functions.
uneven python
use haskell for 10 minutes will learn more than from any video about typehinting
proper ridge
Yeah sorry <@664667836605661217> I'm not really familiar enough with the code, s...
Please don't apologize. I have to apologize for wasting your time (and Lambda's and others) with my question.
uneven python
nah solving type problems is my hobby
hearty heath
Sorry, just going to check in on vc0 for a sec
proper ridge
nah solving type problems is my hobby
I truly wish I could ask you to have a look at the full module (basically just the ABC and the 4 subclasses) and let me know what you think, but I don't dare ask.
proper ridge
Sorry, just going to check in on vc0 for a sec
Thank you so much LX!
gusty ether
typing vs comments: maybe we use typing if we don't want to add comments on small chunks of functions? hehe
buoyant dagger
!e
target = 100
for number in range(1, target + 1):
if number % 3 == 0 and number % 5 == 0:
print("FizzBuzz")
elif number % 3 == 0:
print("Fizz")
elif number % 5 == 0:
print("Buzz")
else:
print(number)
:white_check_mark: Your 3.12 eval job has completed with return code 0.
001 | 1
002 | 2
003 | Fizz
004 | 4
005 | Buzz
006 | Fizz
007 | 7
008 | 8
009 | Fizz
010 | Buzz
... (truncated - too many lines)
Full output: https://paste.pythondiscord.com/MDIWC4A2VWBTH6BYCOJU2QSYOY
proper ridge
!zen
coarse hearthBOT
misty sinew
The code I wrote 2-3 years ago, now whenever I read that code. I'm like when did I write this code huh, it can be done in this in the easiest and shortest way.
it's true we need practice
keen orbit
i cannot speaketh
stuck bluff
buoyant dagger
!e
#Password Generator Project
import random
letters = ['a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', 'x', 'y', 'z', 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z']
numbers = ['0', '1', '2', '3', '4', '5', '6', '7', '8', '9']
symbols = ['!', '#', '$', '%', '&', '(', ')', '*', '+']
print("Welcome to the PyPassword Generator!")
# nr_letters = int(input("How many letters would you like in your password?\n"))
nr_letters = 12
# nr_symbols = int(input(f"How many symbols would you like?\n"))
nr_symbols = 8
# nr_numbers = int(input(f"How many numbers would you like?\n"))
nr_numbers = 4
password_list = [""]
for char in range(1, nr_letters + 1):
password_list.append(random.choice(letters))
for char in range(1, nr_symbols + 1):
password_list.append(random.choice(symbols))
for char in range(1, nr_numbers + 1):
password_list.append(random.choice(numbers))
random.shuffle(password_list)
password = ""
for char in password_list:
password += char
print(f"Your password is: {password}")
:white_check_mark: Your 3.12 eval job has completed with return code 0.
001 | Welcome to the PyPassword Generator!
002 | Your password is: n+s5V(gtU++ez+U8D*lT($08
!pypi qoin
coarse hearthBOT
white oyster
<@492357889336279071> π
HELLUUUUU
scenic echo
<@312590170438369280> π
π₯Ή Hi! I just wanna listen
obtuse valve
guess whos baccc
distant nova
cute lady bug?
hoary citrus
@obtuse valve Everyone say pytorch is best
obtuse valve
depends on your compute limits i guess
eg you can have a "jetson nano" and you can claim that it's a "tensor processing unit"
alternately you can use distillation i guess
but then again, what is your compute requirements
"edge compute"
reinforcemen tklearning whoohohohohohohohohohoohjohohhoho
hoary citrus
See ya guys
obtuse valve
seeya
hoary citrus
GN
obtuse valve
gn
lol the wifi in the attic sucks
iirc mr hemlock told me that it was default in rotterdam
obtuse valve
lol the wifi in the attic sucks
.
stuck bluff
@south flame π
stuck bluff
south flame
okey
stuck bluff
@rough pasture π
rough pasture
Hi @stuck bluff
south flame
i leave 30min
rough pasture
I don't have a permission to speak! How can get that?
south flame
<#764802555427029012>
there @rough pasture
rough pasture
Thanks man @south flame
south flame
Thanks man <@309443001711722498>
your welcome
proper junco
stuck bluff
@fiery sage π
fiery sage
π
stuck bluff
@paper raptor π
paper raptor
<@1237779660931076258> π
@stuck bluff π
proper junco
text[:,1] = [x-min(text[:,1]) for x in text[:,1]]
text[:,1] = [x/max(text[:,1]) for x in text[:,1]]```
dry pendant
import numpy as np
create an array
data = np.array([[12, 29], [33, 50],
[5, 15], [0, 10]])
normalized_data = (data-np.min(data))/(np.max(data)-np.min(data))
normalized data using min max value
print(normalized_data)
stuck bluff
!e py import numpy as np arr = np.array([50, 2, 12]) arr = arr - arr.min() arr = arr / arr.max() print(arr)
coarse hearthBOT
!e ```py
import numpy as np
arr = np.array([50, 2, 12])
arr = arr - arr.min()
ar...
:white_check_mark: Your 3.12 eval job has completed with return code 0.
[1. 0. 0.20833333]
vocal stream
yo
vocal stream
@proper junco If your low-income in the Us it helps 10000%
college affordability eu vs us
I joined late mb
I dont have vc unlocked yet xd
@dry pendant whats the startup concept?
you can talk i can hear you if you want or you can type
dry pendant
break up courses into individual concepts. https://conceptual.fyi/ (eventually)
it's what I'm working on tonight and for the last 3 years.
vocal stream
break up courses into individual concepts. https://conceptual.fyi/ (eventuall...
ah I see interesting
Ignore my non-sense messages trying to get up to 50 without spamming chats
yes
dry pendant
I'm trying to find out what people want and how much they would pay for on-demand concept based learning.
delicate wren
@dry pendant keyboard sounds are quite loud; do you have Krisp enabled?
misty sinew
hi @delicate wren
dry pendant
didn't know that was a thing.
vocal stream
We love the asmr keyboard
dry pendant
I'll see
dry pendant
<@776004461659160577> keyboard sounds are quite loud; do you have Krisp enabled?
got a headset instead of the laptop mic.
vocal stream
@dry pendant keep the same settings you had before and just lower ur sensitivity
your mic got worse
you can go into voice settings and just lower the bar to where you keyboard isnt picked up
delicate wren
<@776004461659160577> keep the same settings you had before and just lower ur se...
I don't think sensitivity would help there
vocal stream
the mic is muffled somewhat
vocal stream
I don't think sensitivity would help there
would stop the keyboard no?
delicate wren
not if it's that loud
vocal stream
maybe
i see what youre saying
my keyboard is super loud but discord normally can tell since its farther from the mic
@proper junco live coding may be able to?
or no
delicate wren
<@471093405246554114> live coding may be able to?
it's restricted similarly everywhere
vocal stream
ah rip
misty sinew
I can't stream in any channel..
vocal stream
I can't stream in any channel..
its just global disabled
delicate wren
!source video
coarse hearthBOT
delicate wren
hmm
proper junco
delicate wren
!d numpy.fft.fft
coarse hearthBOT
vocal stream
quite simple yes π€
misty sinew
I gotta get back to work. Take care guys :p
vocal stream
byebye
delicate wren
FFT sounds like a dumb and straightforward way to estimate the offset
(though ig that's not the problem being solved)
vocal stream
is there a reason screenshot isnt enabled anywhere? @delicate wren
delicate wren
there's also a problem that it's different scale, right?
as in count of points
they don't align, right?
interpolation is needed anyway
dry pendant
vocal stream
Oh im not trying to screenshare
I was just wondering why its disabled
feel like it could be useful to help etc
delicate wren
I was just wondering why its disabled
prior incidents
vocal stream
prior incidents
assumed so yeah
delicate wren
you can make a function that does that
!e
def try_(f, except_):
try:
return f_()
except Exception:
return except_
print(try_(lambda: 1/0, "error"))
:white_check_mark: Your 3.12 eval job has completed with return code 0.
error
!e
import numpy as np
import matplotlib.pyplot as plt
x = np.linspace(-np.pi, np.pi, 100)
xp = np.linspace(-np.pi, np.pi, 10)
fp = np.sin(xp)
f = np.interp(x, xp, fp)
plt.plot(x, f)
plt.savefig('out.png')
:white_check_mark: Your 3.12 eval job has completed with return code 0.
delicate wren
!e
import numpy as np
import matplotlib.pyplot as plt
x = np.linspace(-4, 4, 100)
xp = np.linspace(-np.pi, np.pi, 10)
fp = np.sin(xp)
f = np.interp(x, xp, fp)
print(f)
:white_check_mark: Your 3.12 eval job has completed with return code 0.
001 | [-1.22464680e-16 -1.22464680e-16 -1.22464680e-16 -1.22464680e-16
002 | -1.22464680e-16 -1.22464680e-16 -1.22464680e-16 -1.22464680e-16
003 | -1.22464680e-16 -1.22464680e-16 -1.22464680e-16 -2.80651313e-02
004 | -1.02467186e-01 -1.76869241e-01 -2.51271296e-01 -3.25673351e-01
005 | -4.00075406e-01 -4.74477460e-01 -5.48879515e-01 -6.23281570e-01
006 | -6.71997169e-01 -7.11585676e-01 -7.51174182e-01 -7.90762689e-01
007 | -8.30351195e-01 -8.69939702e-01 -9.09528208e-01 -9.49116715e-01
008 | -9.83454176e-01 -9.69705232e-01 -9.55956288e-01 -9.42207344e-01
009 | -9.28458400e-01 -9.14709456e-01 -9.00960512e-01 -8.87211568e-01
010 | -8.73462624e-01 -8.38181399e-01 -7.77528288e-01 -7.16875177e-01
... (truncated - too many lines)
Full output: https://paste.pythondiscord.com/6EMNALDTJYK5USYSRREDMPZCEQ
delicate wren
hmm
ah, it just copied
!e
import numpy as np
import matplotlib.pyplot as plt
x = np.linspace(-4, 4, 100)
xp = np.linspace(-np.pi, np.pi, 10)
fp = np.sin(xp)
f = np.interp(x, xp, fp)
plt.plot(x, f)
plt.savefig('out.png')
:white_check_mark: Your 3.12 eval job has completed with return code 0.
dry pendant
How much are you paying for education materials in the last year? (learning python, university etc..)
from pathlib import Path
for line in Path("myfile.txt").read_text().splitlines():
if line.startswith('#'):
print(line)
if you already have the file read in then:
[line for line in line_array]
vocal stream
password_list = []
for char in range(1, nr_letters + 1):
password_list.append(random.choice(letters))
for char in range(1, nr_symbols + 1):
password_list.append(random.choice(symbols))
for char in range(1, nr_numbers + 1):
password_list.append(random.choice(numbers))
password = ""
random.shuffle(password_list)
password = ''.join(password_list)
dry pendant
password_list = []
for char in range(1, nr_letters + 1):
password_list.appen...
check out the video: https://www.youtube.com/watch?v=EnSu9hHGq5o
vocal stream
vocal stream
sly pond
There will be coding in the future
and people code today, they also coded yesterday
So I guess I don't understand the question
the future is the future
buoyant dagger
!pypi qoin
Here
proper ridge
!pypi qoin
coarse hearthBOT
distant nova
love that
amber arrow
hey
midnight dagger
@still warren hello
still warren
Hello
I am muted by default
what is the voice chat about
Now i meant even if i tried to talk the app doesnt let me
Where ?
Discords gui is confusing to me
midnight dagger
still warren
Too much elements in the same pace
You are not currently eligible to use voice inside Python Discord for the following reasons:
You have sent less than 50 messages.
I must send 50 messages guess
Does that count ?
coarse hearthBOT
:incoming_envelope: :ok_hand: applied timeout to @still warren until <t:1721313093:f> (10 minutes) (reason: burst spam - sent 8 messages).
The <@&831776746206265384> have been alerted for review.
midnight dagger
high prairie
hello
high prairie
All Good! wbu
I'm good what do you do
hoary citrus
I'm good what do you do
Learning python
high prairie
Learning python
ok
high prairie
u?
I also started to learn for some work purpose
prior to this was a js developer
now switching
hoary citrus
Okay
high prairie
yes
hoary citrus
are you a student ?
yes
hoary citrus
which course you are pursuing
Doing projects
high prairie
Doing projects
okay got it
hoary citrus
okay got it
yes
high prairie
Doing projects
what you are developing
hoary citrus
what you are developing
For now a chat bot
high prairie
For now a chat bot
nice
high prairie
π
misty sinew
When I drive I don't break signal, but what I do is turn left then turn right then again I turn left
high prairie
When I drive I don't break signal, but what I do is turn left then turn right th...
same
π
whenever its possible
misty sinew
same
I follow rules!! ahhahahahahah
what???
80%-20%??!!!
oh I see, i get that
bro achieved everything ig
bro how many years of experience do you have?
bruhhh
you sound older
tell me how old I'm from my voice
elfin breach
tell me how old I'm from my voice
you are muted.
midnight dagger
i'll go have dinner guys, ttyl
rough pasture
Hey guy's
misty sinew
misty sinew
Amrika may work ni krna ye
?
misty sinew
send the global one
But why?
rough pasture
Hi @misty sinew
misty sinew
Hi <@456226577798135808>
Hi
rough pasture
Are you a developer?
misty sinew
Are you a developer?
I guess so,
rough pasture
Which language are you using? Python!?
misty sinew
Yes I am
misty sinew
So where can I start?
Have you ever done programming before?
rough pasture
Yeah! I'm a front-end web developer...
misty sinew
Yeah! I'm a front-end web developer...
Oh so you just want to enter the backend, right?
rough pasture
Yeah that's right
rough pasture
I use JavaScript for the frontend
misty sinew
cz majority of the public is
north american / amriki
The person with I was in VC, we were discussing the internet plans of countries, that's why I shared an India specific link.
misty sinew
I use JavaScript for the frontend
I see, if you're just planning to create APIs and this stuff.
Then start with the basics of python, once you're done then go for the database, learn the basics of SQL and database. Now when you're done picking up a web framework, meanwhile learning it learn how things work internally. You don't have to learn the whole framework at once but when you're sure you're able to build APIs for CRUD with full confidence, then go for advanced part learn async and more database concepts, meanwhile you should also explore how to deploy the backend, going through the flow you won't need a roadmap you'll learn concepts one by one by learning and exploring it, also however if your goal is still building APIs you should also build web apps that aren't consuming APIs at first but from using template engine to doing CRUD.
rough pasture
Okay @misty sinew thanks for this information... Appreciate it man...
hoary citrus
@rotund bough yeah it is working but how?
rotund bough
idk
distant nova
idk
hoary citrus
@distant nova ?
distant nova
had to go srry
calm tusk
hexed sentinel
..
coarse hearthBOT
:incoming_envelope: :ok_hand: applied timeout to @hexed sentinel until <t:1721411256:f> (10 minutes) (reason: duplicates spam - sent 4 duplicate messages).
The <@&831776746206265384> have been alerted for review.
split hare
@misty sinew I dont have ability to use voice chat yet
do you think you could help me with a quick python related question?
sly pond
@lapis matrix
lapis matrix
Have we considered the footwear that women typically wear in slippery winter climates? There is often less surface friction, and the quality of these shoes is usually not designed for winter activities.
I think it's just the curse of dimensionality it's going to take a longer Time to find the consistencies between different data points or understand which data points contribute the most towards finding a good prediction on A model That's backs up the data
sly pond
@crude void Hello
thin lintel
Β¨'
proper ridge
!paste
coarse hearthBOT
proper ridge
@proud chasm
stuck bluff
Sue Vlaki.
sly pond
oof
thin lintel
pseudo raven
pretty regis @stuck bluff
thin lintel
proper ridge
Opal, battleship?
https://battleship-game.org/en/id42890869
coarse hearthBOT
pseudo raven
@midnight dagger check those keys exist in the db
midnight dagger
all_results = []
for myid in unemp_df.index:
results = fred.get_series(myid)
results = results.to_frame(name=myid)
all_results.append(results)
unemp_df.tail(2)
pd.concat(all_results, axis=1).drop(['LNS14000150','LRUN25TTUSM156S'])
midnight dagger
<@1127458312938598460> check those keys exist in the db
i dont understand what u r saying
midnight dagger
all_results = []
for myid in unemp_df.index:
results = fred.get_series(myid...
@proper ridge
pseudo raven
the error is telling you these dont exist 'LNS14000150','LRUN25TTUSM156S' or you're referencing them incorrectly
proper ridge
i dont understand what u r saying
Send your full code.
And any dataset you're importing.
torn canopy
What should I do to turn on the microphone?
torn canopy
thx
proper ridge
Pleasure!
I believe the issue is you're trying to drop a row but using drop column command.
drop usually removes columns.
You're trying to remove a row.
torn canopy
<#764802555427029012>
Isnβt the specification for opening a microphone too strict?
midnight dagger
how to convert it into a code thing
proper ridge
!code
coarse hearthBOT
pseudo raven
PICO CTF
proper ridge
Isnβt the specification for opening a microphone too strict?
No, many people join and scream or cuss and leave. This is to ensure quality of the vc.
pseudo raven
proper ridge
If you want to be part of the community, the requirement is not being strict by any means.
torn canopy
No, many people join and scream or cuss and leave. This is to ensure quality of ...
Ok, I understand
midnight dagger
!pip install fredapi > /dev/null
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import plotly.express as px
plt.style.use('fivethirtyeight')
pd.set_option('display.max_columns', 500)
color_pal = plt.rcParams["axes.prop_cycle"].by_key()["color"]
from fredapi import Fred
fred_key = '551bda501958d4184e31f14ded34986d'
from kaggle_secrets import UserSecretsClient
user_secrets = UserSecretsClient()
secret_value_0 = user_secrets.get_secret("fred-api")
fred = Fred(api_key=fred_key)
sp_search = fred.search('S&P', order_by='popularity')
sp500 = fred.get_series('SP500')
sp500.plot(figsize=(10, 5), title='S&P 500', lw=1) #lw means line width
unemp_df = fred.search('Unemployment rate state', filter=('frequency','Monthly'))
unemp_df = unemp_df.query('seasonal_adjustment == "Seasonally Adjusted" and units == "Percent"')
unemp_df = unemp_df.loc[unemp_df['title'].str.contains('Unemployment Rate')]
all_results = []
for myid in unemp_df.index:
results = fred.get_series(myid)
results = results.to_frame(name=myid)
all_results.append(results)
unemp_df.tail(2)
pd.concat(all_results, axis=1).drop(['LNS14000150','LRUN25TTUSM156S'])
@proper ridge
proper ridge
Bubye
unborn marlin
@tacit jetty Many companies fail, but there are tons of employed people in the industry
proper ridge
<@664667836605661217>
As far as I can see, you're trying to drop a column instead of a row.
Check out the method for dropping rows.
midnight dagger
k
unborn marlin
@tacit jetty Break the cycle.
proper ridge
k
It will help if you carefully read the documentation for pandas. It gives alot of examples.
fluid cedar
proper ridge
midnight dagger
It will help if you carefully read the documentation for pandas. It gives alot o...
yeah
stuck bluff
@eternal beacon π
fast path
hey guys, any chance I can be granted permission to share my screen?
I'd love to pass my knowledge to everyone who's interesting in learning python and other technologies
stuck bluff
Screen sharing permissions can be requested of voice-regular admin-and-above level users, ideally when they're already in the voice chat.
fast path
a few of us joined the code / help channel eager to learn but I'm unable to screen sharing
stuck bluff
Screen sharing permissions can be requested of voice-regular admin-and-above lev...
In practice, this usually means Mr. Hemlock, but Mindful Dev, LX and sometimes some others may be available.
stuck bluff
Discouraged, but nobody is able to really stop you.
fast path
sure, I won't do it then
I'll keep an eye on the server until I spot one of the admins are available
thanks for your help my G
stuck bluff
It's the weekend period, during which Mr. Hemlock is usually unavailable, and Mindful Dev, who sometimes is in on weekends has been drawn away to other duties of late.
fast path
got you! that makes sense
I'll wait
no big deals
just want point out to the fact that you look like a really good person
thanks ma man
can't thank you enough for your kindness in using your time to help me
thin night
Cool
terse sierra
hoary citrus
What happened?
midnight dagger
<@1127458312938598460> yo
hi
midnight dagger
ur netherlands
no
midnight dagger
no
hearty heath
Yaw
Yeah sorry
Is it Python code? @royal seal
Whoops, sorry for the ping
I meant to ping @obtuse valve
Oh ok
Yeah
So the beauty of Q-learning is that you don't need to look ahead. You update the Q-value of a transition each time you make that transition.
In practice, you learn a model for the Q-value that hopefully generalises to transitions you havent seen.
Online, I think. If I remember what that term means
Yeah, so as an RL agent, you get a reward (possibly zero) each time you make a transition.
Right π€
So, you're not training your agent in a simulation?
obtuse valve
s_0 -> a_0 -> state transition -> s_1
but idk what s_1 is, unless i make action a_0
hearty heath
That's true, but that's how you learn right?
I guess you start off with random values, and take kind of random actions.
Then tend towards taking actions that more-and-more exploit what you've learned.
obtuse valve
1000000 random simulations
- label all "winning" ones 1.0
- label all "losing" ones 0.0
like all states within the simulation
hearty heath
Yeah that's a different kind of thing, that's search.
But when you do Q-learning, you're kind of implicitly searching.
obtuse valve
s_0 -> a_0 -> state transition -> s_1
I DONT UNDERSTAND THIS :(
hearty heath
It doesn't have to be a table.
Yeah
So, I think the term you're looking for is temporal difference learning possible?
I'm not current on ML techniques though (like, I literally did this stuff ten years ago before the current AI boom 
).
Errrr let me do some reading
Do you have a university library? I recommend getting your hands on a copy of AIMA if you can. The chapters on MDPs and RL are pretty good and clearly written.
hearty heath
Oh right ok
hearty heath
It's alright π
obtuse valve
your pfp π
it's smth from aardman studios
i can recognise it
machiavellian penguin
i forogt his name
massive baddie
hearty heath
Yeah, it's Feathers McGraw π
from the wrong trousers
They're releasing a sequel later this year I think
obtuse valve
hmm
hearty heath
hearty heath
Oh ok
obtuse valve
bc of shit wifi in the attic
so how does it work in q learning (dont mind the td learning)
i will get shit sorted hopefully
interesting shit
i heard of dqn before
PERFECT
another mujoco sim
and tutorial from pyotrch???????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????/////////////
OMG
yk dqn? it was "novel" about 10 years ago
obtuse valve
yeah, just RL stuff
hearty heath
yk dqn? it was "novel" about 10 years ago
Haven't heard of it
obtuse valve
hm
hearty heath
Brb while I do some reading
hearty heath
Deep q-learning I assume is just q-learning using deep neural nets
obtuse valve
yes
q learning sounds a lil fucked imho, i STILL think that i need to sort of "simulate" the next state
or somehow have the agent to "learn" the environemtn
and the state transition function
pine jacinth
Hey <@706205884161720404> π
Hi @hearty heath ! Yeah I can't unmute unfortunately but we chat here at least
obtuse valve
yk any RL algs that learn the state transition function itself?
that could be VERY powerful
or even better, learn the env
muzero is what im thinking of
iirc DQN was a big deal in 2012
like it played atari games from what ive read
hearty heath
yk any RL algs that learn the state transition function itself?
That would be "model-based" reinforcement learning. Q-learning is "model-free", in the sense that the agent doesn't know or learn the transition model explicitly; what it knows about state transitions is implicitly part of the q-values.
Nice quote from AIMA:
[...] Reinforcement learning differs from "just solving an MDP" because the agent is not given the MDP as a problem to solve; the agent is in the MDP. It may not know the transition model or the reward function, and it has to act in order to learn more. Imagine playing a new game whose rules you don't know; after a hundred or so moves, the referee tells you "You lose". That is reinforcement learning in a nutshell. [...]
hearty heath
With monopoly, you could go with a model-based approach, because you already know the rules of the game.
obtuse valve
iirc some ai has learnt driving on gta
like it was wild
like the nn was outputting like frames of driving etc
fucked up
i couldnt find it
hazy minnow
v
obtuse valve
was about to say
was it him or was it yannic kilcher
but yeah ik sentdex made a series on it
goated rich boi that gets sponsored all the time
but he's quite knowledgeable
hearty heath
Ok, so there's "passive reinforcement learning" where the policy of the agent is fixed and you're learning the utilities of the states under that policy. And there's "active reinforcement learning" where the agent is simultaneously trying to figure out a policy.
obtuse valve
perfect timing
i still dont't understand utility
like u value
like it's obv in alphazero
(ish)
obtuse valve
i _still_ dont't understand utility
sounds so similar to q value
(sorry im presently distracted by my alphazero run)
alright back
hearty heath
So this would be the utilities you get by just "solving the MDP"; it's the utilities of the states under the optimal policy:
hearty heath
ive seen this fucking diagram fo so many times
Oh really? π It's an example from that book.
obtuse valve
it tells me the expected return, which is the cumulative rewards, of an action in a specific state?
obtuse valve
Oh really? π It's an example from that book.
i guess youtube likes ripping off from textbooks
not this particular example but yeah
royal rivet
hello
hearty heath
it tells me the expected return, which is the cumulative rewards, of an action i...
Yeah exactly. It's how much reward you expect to get starting in a state and following the optimal policy (with rewards being exponentially discounted as they go further into the future).
obtuse valve
Yeah exactly. It's how much reward you expect to get starting in a state and fol...
that's the q value right
the u value sounds sus
like what else can it tell me that i would find useful that isnt alr described by
- policy
- value
- action value (q)
fuck all it seems
hearty heath
that's the q value right
Q-value is the same thing, but for a state and a tentative action, whereas the u-value/v-value/utility is the maximum over all possible actions you could take from that state.
hearty heath
U(s) = max(Q(s, a) for a in actions)
obtuse valve
U(s) = max(Q(s, a) for a in actions)
damn so im right
ok so it's "value" but for q
very fucking interesting
since "value" is technically the expected return of a state following pi
royal rivet
Do you know python language?
obtuse valve
as in, the best actions taken from pi directly from the state, as opposed to q where its goal is to measure the "action value" specifically
obtuse valve
Do you know python language?
sure
obtuse valve
as in, the best actions taken from pi directly from the state, as opposed to q w...
like "how good is this action if i were to follow the policy from this action onwards"
like that sounds so fucking complete now
rl is now full circle
i get a function/cool abstraction for every little thing
royal rivet
where are you from?
obtuse valve
instead of max(Q(s,a) for a in actions) β:
U(s) β
genius
we RL bros are so excessive and i like it :D
ty for clarifying though :p
im happy
hearty heath
but isn't that a*?
Yes, I'm talking about the value of the state under the optimum policy. I think under a particular policy Ο(s): U(s) = Q(s, Ο(s)) (assuming your policy is deterministic and not random)
It's a bit harder going the other way (getting the Q-value from the U-value) because you need a transition model.
brb
obtuse valve
ok so in theory, with perfect V*:
a* = argmax(pi*(s))
Q(s,a*) = U(s)
im trying to process this
shit backticks no work
obtuse valve
It's a bit harder going the other way (getting the Q-value from the U-value) bec...
well hm, you can use the "advantage function"
which is, iirc:
A(s,a) = Q(s,a) - V(s)
royal rivet
see you soon
obtuse valve
you can do:
u_val = U(s)
a_star = None
q_vals = {}
for a in actions:
if u_val == Q(s,a):
a_star = a
advantage = A(s,a)
q_vals[(s,a)] = advantage
idk im thinking on top of my head
hearty heath
im trying to process this
Well it's because you'd have to derive the Q-values by looking one step forward to the U-values of the successor states.
Btw, u-value and v-value and "utility" are the same thing. Different books use different terminology. Just in case that wasn't clear.
obtuse valve
well they are a bit nuanced
like the exact thing that they are measuring arent exactly the same
is it
eg they use distinctly different formulae
hearty heath
No I think they're the same thing generally:
In this chapter we use U for the utility function [...] but many works about MDPs use V (for value) instead.
hearty heath
which is, iirc:
A(s,a) = Q(s,a) - V(s)
Oh, I hadn't heard of that π
obtuse valve
Oh, I hadn't heard of that π
interesting
they describe the nn as a "q network"
i wonder if nns output values or q values
since in alphazero they output values
and more speficially, values that are clipped from -1 to 1
using softmax
@hearty heath this is what i think i dont get
i see some subtraction in a ()^2, ig somehting to do with mse???
hearty heath
Yeah
obtuse valve
hmm expected value of some mse? sussy baka
idk how that works, like i thought expected values = p_1 * w_1 + p_2*w_2 ...
hearty heath
<@476543547210858506> this is what i think i dont get
Err, I think theta_i represents the parameters of your model in iteration i, and that's the loss function. It would help if I could see the preceding text tbh π
hearty heath
So like Q(s, a; theta) is your neural net or whatever, with parameters theta, and its meant to approximate Q*(s, a).
obtuse valve
loss = MSELoss()
for i in dataset:
q_i = model(i)
loss(preds_i - q_i)
loss.backward()
model.zero_grad()
optim.step()
or something
restarted stream
and iirc the optim is sgd
hearty heath
In the reinforcement learning community this is typically a linear function approximator...
I guess this is not the case anymore π
obtuse valve
obtuse valve
> In the reinforcement learning community this is typically a linear function ap...
i was thinking of this, like were people using like excel regression shits to play games???????????????????
like hmm what's the best y=mx+c for this game π€
hearty heath
i was thinking of this, like were people using like excel regression shits to pl...
They probably used like a weighted sum of features model.
obtuse valve
like hmm what's the best y=mx+c for this game π€
you can argue that this is still true today
hearty heath
Like where you hand-craft the features.
obtuse valve
yeah ik handcrafted stuff
still very much alive in the chess community
even though there are nnues
which are basically cpu nn
but it can incrementally update the outputs
which makes it viable on cpu
anyways
obtuse valve
<@476543547210858506> this is what i think i dont get
it seems like this formula uses Q values as targets
so this nn outputs "state action values" and not "state values"
i shouldve seen it coming
"Q-Network"
π€¦
hearty heath
Β―_(γ)_/Β―
obtuse valve
i can probs write this out in ~3 lines of pytorch code
THIS SHIT LOOKS SCARY
like the one that says "differentiating..."
is this just regular backprop
ive read it a few times
like dL/dy right
or more precisely, the gradients of each weight i in nn theta
hearty heath
Yeah I think so. I'm still wrapping my head around the notation actually. I don't understand what a subscript on an expectation is supposed to denote.
obtuse valve
i think it's a shit way of saying "given", ik it's weird
im still guessing
like E[...|s=s, a=a...]
that's the only logical shit
ugh dinner is late tonight
so i might have dinner in ~10 mins
fuck
right in the middle of a focus sesh
hearty heath
that's the only logical shit
Oh right ok.