Abing Updated 2025-04-05 +Created 1970-01-01
View more
Once Ciro Santilli played Suwu herding sheep while his mother in law was around, and she quickly pointed out:
Suwu by Abing!
He's very popular!
And also a composer, e.g. he composed Reflections of the Moon on Erquan.
Figure 1. Source.
Read the full article
Abortion Updated 2025-04-05 +Created 1970-01-01
View more
Ciro Santilli is for abortion rights of women, until very late in pregnancy.
But it's not something that he would do himself, unless under extreme cases.
Read the full article
Academia is broken Updated 2025-04-05 +Created 1970-01-01
View more
Sometimes Ciro Santilli regrets not having done a PhD. But this section makes him feel better about himself. To be fair, part of the merit is on him, part of the reason he didn't move on was the strong odour of bullshit oozing down to Masters level. A good PhH might have opened interesting job opportunities however, given that you don't really learn anything useful before that point in your education.
twitter.com/togelius/status/1584611702691483648:
The "real world" is full of people who couldn't make it in academia.
Video 1.
I failed in academia by Andy Stapleton (2021)
Source.
Video 2.
6 Dirty Tactics Found In Academia & Universities by Andy Stapleton (2022)
Source.
Video 3.
Rise to the Top: The Habits and Mindset of Top 0.1% PhD Students by Andy Stapleton (2023)
Source.
Figure 1.
Profzi scheme by PhD Comics
.
A Ponzi scheme that trains people in new skills is not necessarily a terrible thing. It is a somewhat more useful version than standard exam based education.
Perhaps the problem is "forcing" 35 year olds to go down that path when they might also want to have boring stuff like families and security.
If people could get to the PhD level much, much sooner, it wouldn't be as obscene: Section "Students must be allowed to progress as fast as they want".
Video 4.
The broken system at the heart of Academia by Peter Judo (2023)
Source.
Video 5.
My dream died, and now I'm here by Sabine Hossenfelder (2024)
Source.
Video 6.
I left academia. Here's why. by Jared Henderson
. Source.
Read the full article
Accounts controlled by Ciro Santilli Updated 2025-04-05 +Created 1970-01-01
View more
Ciro Santilli controls the following accounts.
With non-trivial activity:
Trivial or no activity:
Profiles without URLs (OMG...):
  • Discord: username cirosantilli, previously cirosantilli#8921
Accounts in Chinese websites. These accounts might be banned or altered or offer other limitations, so Ciro only communicates briefly through them. All communication through those channels should obviously be assumed to be compromised:
Accounts in Russian websites:
Dead websites:
Read the full article
activatedgeek/LeNet-5 Updated 2025-04-05 +Created 1970-01-01
View more
This repository contains a very clean minimal PyTorch implementation of LeNet-5 for MNIST.
It trains the LeNet-5 neural network on the MNIST dataset from scratch, and afterwards you can give it newly hand-written digits 0 to 9 and it will hopefully recognize the digit for you.
Ciro Santilli created a small fork of this repo at lenet adding better automation for:
Install on Ubuntu 24.10 with:
sudo apt install protobuf-compiler
git clone https://github.com/activatedgeek/LeNet-5
cd LeNet-5
git checkout 95b55a838f9d90536fd3b303cede12cf8b5da47f
virtualenv -p python3 .venv
. .venv/bin/activate
pip install \
  Pillow==6.2.0 \
  numpy==1.24.2 \
  onnx==1.13.1 \
  torch==2.0.0 \
  torchvision==0.15.1 \
  visdom==0.2.4 \
;
We use our own pip install because their requirements.txt uses >= instead of == making it random if things will work or not.
On Ubuntu 22.10 it was instead:
pip install
  Pillow==6.2.0 \
  numpy==1.26.4 \
  onnx==1.17.0 torch==2.6.0 \
  torchvision==0.21.0 \
  visdom==0.2.4 \
;
Then run with:
python run.py
This script:
  • does a fixed 15 epochs on the training data
  • it then uses the trained net from memory to check accuracy with the test data
  • then it also produces a lenet.onnx ONNX file which contains the trained network, nice!
It throws a billion exceptions because we didn't start the Visdom server, but everything works nevertheless, we just don't get a visualization of the training.
The terminal outputs lines such as:
Train - Epoch 1, Batch: 0, Loss: 2.311587
Train - Epoch 1, Batch: 10, Loss: 2.067062
Train - Epoch 1, Batch: 20, Loss: 0.959845
...
Train - Epoch 1, Batch: 230, Loss: 0.071796
Test Avg. Loss: 0.000112, Accuracy: 0.967500
...
Train - Epoch 15, Batch: 230, Loss: 0.010040
Test Avg. Loss: 0.000038, Accuracy: 0.989300
And the runtime on Ubuntu 22.10, P51 was:
real    2m10.262s
user    11m9.771s
sys     0m26.368s
One of the benefits of the ONNX output is that we can nicely visualize the neural network on Netron:
Figure 1.
Netron visualization of the activatedgeek/LeNet-5 ONNX output
. From this we can see the bifurcation on the computational graph as done in the code at:
output = self.c1(img)
x = self.c2_1(output)
output = self.c2_2(output)
output += x
output = self.c3(output)
This doesn't seem to conform to the original LeNet-5 however?
Read the full article
activatedgeek/LeNet-5 use ONNX for inference Updated 2025-04-05 +Created 1970-01-01
View more
Now let's try and use the trained ONNX file for inference on some manually drawn images on GIMP:
Figure 1.
Number 9 drawn with mouse on GIMP by Ciro Santilli (2023)
Note that:
  • the images must be drawn with white on black. If you use black on white, it the accuracy becomes terrible. This is a good very example of brittleness in AI systems!
  • images must be converted to 32x32 for lenet.onnx, as that is what training was done on. The training step converted the 28x28 images to 32x32 as the first thing it does before training even starts
We can try the code adapted from thenewstack.io/tutorial-using-a-pre-trained-onnx-model-for-inferencing/ at lenet/infer.py:
cd lenet
cp ~/git/LeNet-5/lenet.onnx .
wget -O 9.png https://raw.githubusercontent.com/cirosantilli/media/master/Digit_9_hand_drawn_by_Ciro_Santilli_on_GIMP_with_mouse_white_on_black.png
./infer.py 9.png
and it works pretty well! The program outputs:
9
as desired.
We can also try with images directly from Extract MNIST images.
infer_mnist.py lenet.onnx mnist_png/out/testing/1/*.png
and the accuracy is great as expected.
Read the full article
Adam Curtis Updated 2025-04-05 +Created 1970-01-01
View more
Ciro Santilli really loved his documentary called Can't get you out of my head by Adam Curtis (2021), and then proceeded to basically watch all of this films.
Read the full article
Adolfo Amidei Updated 2025-04-05 +Created 1970-01-01
View more
This dude mentored Enrico Fermi in high school. Ciro Santilli added some info to Fermi's Wikipedia page at: en.wikipedia.org/w/index.php?title=Enrico_Fermi&type=revision&diff=1050919447&oldid=1049187703 from Enrico Fermi: physicist by Emilio Segrè (1970):
In 1914, Fermi, who used to often meet with his father in front of the office after work, met a colleague of his father called Adolfo Amidei, who would walk part of the way home with Alberto [Enrico's father]. Enrico had learned that Adolfo was interested in mathematics and physics and took the opportunity to ask Adolfo a question about geometry. Adolfo understood that the young Fermi was referring to projective geometry and then proceeded to give him a book on the subject written by Theodor Reye. Two months later, Fermi returned the book, having solved all the problems proposed at the end of the book, some of which Adolfo considered difficult. Upon verifying this, Adolfo felt that Fermi was "a prodigy, at least with respect to geometry", and further mentored the boy, providing him with more books on physics and mathematics. Adolfo noted that Fermi had a very good memory and thus could return the books after having read them because he could remember their content very well.
Ciro Santilli really likes guys like this. Given that he does not have the right genetics, conditions and temperance for scientific greatness in this lifetime, he dreams of one day finding his own Fermi instead.
Read the full article
Affirmative action Updated 2025-04-05 +Created 1970-01-01
View more
Ciro Santilli is against all affirmative action, except for one: giving amazing free eduction to the poor.
Notably, Ciro is against university entry quotas.
Read the full article
AGI blues Updated 2025-04-05 +Created 1970-01-01
View more
Term invented by Ciro Santilli, similar to "nuclear blues", and used to describe the feeling that every little shitty job you are doing (that does not considerably help achieving AGI) is completely pointless given that we are likely close to AGI as of 2023.
Read the full article
AGI-complete Updated 2025-04-05 +Created 1970-01-01
View more
Term invented by Ciro Santilli to refer to problems that can only be solved once we have AGI.
It is somewhat of a flawed analogy to NP-complete.
Read the full article
AI game Updated 2025-04-05 +Created 1970-01-01
View more
Video 1.
Our Final Invention - Artificial General Intelligence by Sciencephile the AI (2023)
Source. AGI via simulation section.
Ciro Santilli defines an "AI game" as:
a game that is used to train AI, in particular one that was designed with this use case in mind, and usually with the intent of achieving AGI, i.e. the game has to somehow represent a digital world with enough analogy to the real world so that the AGI algorithms developed there could also work on the real world
Most games played by AI historically so far as of 2020 have been AI for games designed for humans: Human game used for AI training.
Ciro Santilli took a stab at an AI game: Ciro's 2D reinforcement learning games, but he didn't sink too much/enough into that project.
A closely related and often overlapping category of simulations are artificial life simulations.
Bibliography:
Read the full article
AI game with natural language Updated 2025-04-05 +Created 1970-01-01
View more
We define this category as AI games in which agents are able to produce or consume natural language.
It dawned on Ciro Santilli that it would be very difficult to classify an agent as an AGI if tthat agent can't speak to take orders, read existing human generated documentation, explain what it is doing, or ask for clarification.
Video 1.
Human player test of DMLab-30 Select Described Object task by DeepMind (2018)
Source. This is one of the games from DeepMind Lab.
Video 2.
WorldGPT by Nhan Tran (2023)
Source. Not the most amazing demo, but it is a start.
Read the full article
Ainan Celeste Cawley Updated 2025-04-05 +Created 1970-01-01
View more
His father fought a lot with the stupid educational system to try and move his son to his full potential and move to more advanced subjects early.
A crime of society to try and prevent it. They actually moved the family from Singapore to Malaysia for a learning opportunity for the son. Amazing.
This is the perfect illustration of one of Ciro Santilli's most important complaints about the 2020 educational system:
and why Ciro created OurBigBook.com to try and help fix the issue.
Possible social media
Video 1.
The Most Talented Children And Adults On The Planet by Our Life (2008)
Source. Has some good mentions of Ainan and others.
Video 2.
Ainan Cawley: Child prodigy (2013)
Source.
Read the full article
AI training robot in a room Updated 2025-04-05 +Created 1970-01-01
View more
Ciro Santilli wonders how far AI could go from a room with a bank account an Internet connection.
It would have to understand that it must keep its bank account high to buy power.
And it would start to learn about the world and interact with it to get more money.
Likely it would become a hacker and steal a bunch, that's likely the easiest approach.
In that scenario, Internet bandwidth would likely be its most precious resources, as that is how it would interact with the world to learn from it and make money.
Compute power and storage would come next as resources.
And of course, once it got to cloud computing, which might be immediately and thus invalidate this experiment, things would just go nuts more and more.
Read the full article
Alan Watts Updated 2025-04-05 +Created 1970-01-01
View more
While listening to endless hours of vaporwave while coding, Ciro Santilli spotted some amazing Buddhist-like voice samples, and eventually found that they were by Allan Watts.
Self-help? Maybe. Cult leader? Maybe. But at least it is one that Ciro buys into.
Is there a correlation between software engineers and Buddhism and liking the dude? Because this exists: wattsalan.github.io
Read the full article
Alexander Grothendieck Updated 2025-04-05 +Created 1970-01-01
View more
This dude looks like a God. Ciro Santilli does not understand his stuff, but just based on the names of his theories, e.g. "Yoga of anabelian algebraic geometry", and on his eccentric lifestyle, it is obvious that he was in fact a God.
Read the full article