Good old ReLU. I've heard that CNN's perform better when using ReLU as the activation func, and that's probably why; ReLU acts as a filter on the image's features.
Nah. Without a nonlinearity you just get a linear combination of inputs instead of the output of a deep neural network. ReLU or Ramp is just the simplest possible non linearity. Using a simple function can enable using deeper networks yielding even better performance.
It's actually somewhat of a headache, numerically. Works well enough tho.
Yes, it is inherently incorrect, I just think that increases the fun factor. It reminds me of that guy who wrote a prime-checker for his thesis and it was just a massive list of if-statement
Had one guy asking if AI is conscious. I said the whole thing is actually one very big excel sheet, so does he think that excel can be conscious? Anyway, I think the guy is now in a relationship with office.
I mean yeah, the mechanism itself is straightforward, but isn't a big thing that we don't really know what it chooses for its... basis vectors? Something like that, I don't do programming and I'm out of practice with matrices
I would be heavily more inclined to use LLMs (and also probably use it in a more correct manner) if it came in the physical form of a magic 8-ball that I would violently shake to get the answer.
You're forgetting the activation function to break linearity, otherwise you can only model linear trends. Many activation functions have "if" hidden in them (e.g. ReLU with the "max").
Oh, just wait till you hit that sweet sweet python "Everything can have an else block"
if some_shit:
try:
for stuff in thing:
if stuff is bad:
raise ShitHitTheFanError
# Massive block comment
do = the_thing()
if do is good:
break
else:
do = the_other_thing()
if not a_thing:
help_me() # May raise NoHelpFoundError
except ShitHitTheFanError:
my_life = recover()
if my_life is bad:
for fuck in my_life:
try:
fix(fuck)
except NoHelpFoundError:
break
else:
fuck_this_shit()
else:
wat_is_going_on()
else:
theres_another_one("?")
44 replies
More accurately:
>looks inside
>matrix multiplication
Max(input*weights, 0) is an if in a sense, I guess.
Good old ReLU. I've heard that CNN's perform better when using ReLU as the activation func, and that's probably why; ReLU acts as a filter on the image's features.
Nah. Without a nonlinearity you just get a linear combination of inputs instead of the output of a deep neural network. ReLU or Ramp is just the simplest possible non linearity. Using a simple function can enable using deeper networks yielding even better performance.
It's actually somewhat of a headache, numerically. Works well enough tho.
Yes, it is inherently incorrect, I just think that increases the fun factor. It reminds me of that guy who wrote a prime-checker for his thesis and it was just a massive list of if-statement
98% accurate prime checker, super fast
isPrime() return False;
A transistor is on or off. It's all if-then.
No, it's more elemental: it's all ands , ors and nots
Boolean math let's you convert it all to nand gates. For ease of manufacturing this is what is done.
Had one guy asking if AI is conscious. I said the whole thing is actually one very big excel sheet, so does he think that excel can be conscious? Anyway, I think the guy is now in a relationship with office.
Matrix multiplication, activation functions and softmax. Also some matrix addition sometimes. That's literally it
Everything is quantum field fluctuations.
That's literally it.
I mean yeah, the mechanism itself is straightforward, but isn't a big thing that we don't really know what it chooses for its... basis vectors? Something like that, I don't do programming and I'm out of practice with matrices
The human brain is far too complex to understand
The brain isn't meat, it's fat, and is known to turn to soap under the right conditions. Which coincidentally also explains the soapy tiddy memes.
🤚sapient
🫰saponified
The brain can turn into soap? Did it still work after? Can soap think?
Singing meat!?
That was old AI.
New AI uses a complex, multi-layer sequence of Magic 8-balls.
That's a common misconception. It is a complex multi-layer sequence of heavily biased Magic 8-balls.
With a post-it note saying "be polite, don't hallucinate"
"Mirror mirror on the wall please fix my awful code, make no mistakes"
I wouldn't be surprised if some stuff (as in architecture) was solved by querying a different model to solve a problem
I would be heavily more inclined to use LLMs (and also probably use it in a more correct manner) if it came in the physical form of a magic 8-ball that I would violently shake to get the answer.
You don't have to be an AI developer to not understand your code. I not understand my code every day.
-Agent
It’s just
y = k*x + ma few trillion times.Yes it's not even for loop and if. Just multiplication and addition.
You're forgetting the activation function to break linearity, otherwise you can only model linear trends. Many activation functions have "if" hidden in them (e.g. ReLU with the "max").
come on, it’s not that simple!
there’s also linear algebra
Have you ever tried to follow five hundred nested for loop and if statements? Shit is nuts...
Oh, just wait till you hit that sweet sweet python "Everything can have an
elseblock"I'm a network engineer lol python is all I know... and yup, the else is a curve ball from hell
But this is absolutely fucking hilarious
Conway's Game of Life is Turing Complete and that fact has been fucking with my mind for like 25 years.
If statements are some of the largest sources of bugs in software.
If this is true then everything is fine.
Edit: Damn, I'm part of the problem, right?
I mean, without if statements and loops theres no logic? It would only be flat code, so i mean... Yeah?
All you need for Turing completeness is finite precision floating-point operations and an infinite loop.
ml is not a hard coded set of if-else pairs. it's a glorified fit with a lot of parameters.
repeat that for like one hundred billion times and now THAT's innovation
Indeed! See: induction circuits