catalog.ngc.nvidia.com/orgs/nvidia/resources/resnet_50_v1_5_for_pytorch explains:
The difference between v1 and v1.5 is that, in the bottleneck blocks which requires downsampling, v1 has stride = 2 in the first 1x1 convolution, whereas v1.5 has stride = 2 in the 3x3 convolution.This difference makes ResNet50 v1.5 slightly more accurate (~0.5% top1) than v1, but comes with a small performance drawback (~5% imgs/sec).
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:and it works pretty well! The program outputs:as desired.
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.png9We can also try with images directly from Extract MNIST images.and the accuracy is great as expected.
infer_mnist.py lenet.onnx mnist_png/out/testing/1/*.pngThe opposite of spacelike-separated events.
Sponsor Ciro Santilli's work on OurBigBook.com Financial crime preventive measures by 
Ciro Santilli 40 Updated 2025-07-16
Ciro Santilli 40 Updated 2025-07-16Crazy shady crypto people seem to like Ciro Santilli, so this is in order.
Giving to Ciro Santilli is the worst possible way to launder your money, as donations amounts are clearly publicly disclosed (though not donor identities if they with to remain anonymous), and clear records kept of every donation made (including private note of donor identities if known). Also suspicious donations are promptly reported to the authorities.
Donation refunds upon donor's requests are only made at our discretion, and may be declined, unless required by law of course. This is to reduce the risks of us unknowingly serving as money mules or aiding money laundering.
Ciro Santilli believes that he is not require to report large donations to either:But note that Ciro will preventively report if there are any further suspicious aspects to any donations received.
- charities have to report anonymous donations of £25,000 or more as "serious incident", but Ciro Santilli does not have a registered charity: www.gov.uk/guidance/how-to-report-a-serious-incident-in-your-charity, related: docs.ourbigbook.com/project-governance
- dealers selling goods over 10,000 euros for cash must make a Suspicious Activity Report (SAR) www.gov.uk/guidance/money-laundering-regulations-high-value-dealer-registration. However Ciro Santilli does not sell any goods, only provides services, and services are excluded from the report requirements
mlcommons.org/en/ Their homepage is not amazingly organized, but it does the job.
Benchmark focused on deep learning. It has two parts:Furthermore, a specific network model is specified for each benchmark in the closed category: so it goes beyond just specifying the dataset.
Results can be seen e.g. at:
Those URLs broke as of 2025 of course, now you have to click on their Tableau down to the 2.1 round and there's no fixed URL for it:
And there are also separate repositories for each:
Some usage examples under: Section "Sequelize example".
As of 2021, this library is extremely painful to use. It does feel semi-mature, but there are just too much horrible things going on;
- the documentation is a bit messy and misses a lot of stuff. The examples are often too short, and it is hard to understand what specific options they are talking about do because they lack clear input/expected output pairs. Examples:
- the implementation has several inelegant/unintuitive annoyances/requirements of code repetition that drive you mad.The association API feels notably bad, it took a few days for Ciro Santilli to learn to do what he considers "basic" association operations, knowledge which he dumped to: stackoverflow.com/questions/22958683/how-to-implement-many-to-many-association-in-sequelize/67973948#67973948See also: how to decide if an ORM is good?.
- bugs are piling up. It appears that many key devs left, and current maintainers are just not being able to keep up.
Some glaring issues are listed at the horrors of Sequelize.
Ciro Santilli is very happy to meet people with related interests, he really loves his like-minded online friends. Even if you don't have something a specific goal in mind for the contact, please just say hi.
To contact Ciro publicly about any general subject that is not covered in a more specific GitHub repository, including saying hi or suggestions about his website either:
- create a GitHub issue at: github.com/cirosantilli/cirosantilli.github.io/issues/new or Giscus thread
- at mention Ciro's main Twitter account
For comments about China, first read:
and then create a GitHub issue at: github.com/cirosantilli/china-dictatorship/issues/new
Publicly viewable contact is preferred if possible to more effectively share Ciro's wisdom with the world.
But if you feel more comfortable with private contact, no problem, either:
- email: username
cirosantilliwith provider ProtonMail which has domain nameproton.me. Note that Ciro also controls the Gmail address with that same username, and keep in mind that dots are ignored in Gmail addresses. But ProtonMail preferred because why should we give our private minds to the CIA by default? Push notifications disabled. - Signal: username
cirosantilli.89signal.me/#eu/BuJjV0enXYTOnFFc1FZV5LFcWodL1a2Oy9VZ5uyQU7xrYxqw+npIcaHBqghWudrI. 89 is a reference to Ciro Santilli's birth year 1989. Push notifications enabled, but treat like email unless we are actively chatting back and forth.
For other less good methods that will also likely work, use direct messages of the following profiles from under Section "Accounts controlled by Ciro Santilli":Ciro's Twitter DMs are also open, but note that Ciro receives endless Chinese language SPAM there which Twitter is doing nothing to combat, so it's not as reliable.
- LinkedIn: www.linkedin.com/in/cirosantilli
- Reddit: www.reddit.com/user/cirosantilli
- Facebook: www.facebook.com/cirosantilli
- Telegram: telegram.me/cirosantilli. Note that end-to-end encryption is present on secret chats only, which don't have device sync. Ridiculous.
If you are a privacy freak or are going to tell Ciro state secrets Ciro has this GNU Privacy Guard public key: pubkey.gpg, but it's not something that he has ever really used.
Disqus comments were removed from his website in 2019-05-04, a manual dump is available here, removal rationale at: why Ciro Santilli removed Disqus comments from his website in 2019-05-04.
70,000 28x28 grayscale (1 byte per pixel) images of hand-written digits 0-9, i.e. 10 categories. 60k are considered training data, 10k are considered for test data.
This is THE "OG" computer vision dataset.
Playing with it is the de-facto computer vision hello world.
It was on this dataset that Yann LeCun made great progress with the LeNet model. Running LeNet on MNIST has to be the most classic computer vision thing ever. See e.g. activatedgeek/LeNet-5 for a minimal and modern PyTorch educational implementation.
But it is important to note that as of the 2010's, the benchmark had become too easy for many applications. It is perhaps fair to say that the next big dataset revolution of the same importance was with ImageNet.
The dataset could be downloaded from yann.lecun.com/exdb/mnist/ but as of March 2025 it was down and seems to have broken from time to time randomly, so Wayback Machine to the rescue:but doing so is kind of pointless as both files use some crazy single-file custom binary format to store all images and labels. OMG!
wget \
https://web.archive.org/web/20120828222752/http://yann.lecun.com/exdb/mnist/train-images-idx3-ubyte.gz \
https://web.archive.org/web/20120828182504/http://yann.lecun.com/exdb/mnist/train-labels-idx1-ubyte.gz \
https://web.archive.org/web/20240323235739/http://yann.lecun.com/exdb/mnist/t10k-images-idx3-ubyte.gz \
https://web.archive.org/web/20240328174015/http://yann.lecun.com/exdb/mnist/t10k-labels-idx1-ubyte.gz
OK-ish data explorer: knowyourdata-tfds.withgoogle.com/#tab=STATS&dataset=mnist
gothinkster/realworld implementations based on Express.js.
If you are a pussy and work a soul crushing job, this is one way to lie to yourself that your life is still worth living: do one cool thing every day.
Find a time in which your mind hasn't yet been destroyed by useless work, usually in the morning before work, and do one thing you actually like in life.
Work a little less well for you boss, and a little better for yourself. Ross Ulbricht:Selling drugs online is not advisable however.
Even better, try to reach an official agreement with your employer to work 20% less than the standard work week. For example, you could work one day less every week, and do whatever you want on that day. It is not possible to push your passion to weekends, because your brain is too tired. "You keep all non-company-related IP you develop on that time" is a key clause obviously.
On a related note, good employers must allow employees to do whichever the fuck "crazy projects", "needed refactorings or other efficiency gains" and "learn things deeply" at least 20% of their time if employees want that: en.wikipedia.org/wiki/20%25_Project. Employees must choose if they want to do it one day a week or two hours per day. One day per month initiatives are bullshit. Another related name: genius hour.
Highly relevant on this topic: Video "What Predicts Academic Ability? by Jordan B Peterson (2017)".
I did it for me, Skyler
. Source. Pursuing a dream part time can make you feel afraid and tired. But at least, you will feel alive.Maybe you will be fired, but long term, having tried, or even succeeded your dream, or a one of its side effects, will be infinitely more satisfying.
The same goes for school, and maybe even more so because your parents can still support you there. Some Gods who actually followed this advice and didn't end up living under a bridge:
- George M. Church "[We] hope that whatever problems... contributed to your lack of success... at Duke will not keep you from a successful pursuit of a productive career." Lol, as of 2019 the dude is the most famous biotechnologist in the world, those "problems" certainly didn't keep him back.
- Freeman Dyson proved the equivalence of the three existing versions of quantum electrodynamics theories that were around at his time, and he has always been proud of not having a PhD!
- Ramanujan, from Wikipedia:
He received a scholarship to study at Government Arts College, Kumbakonam, but was so intent on mathematics that he could not focus on any other subjects and failed most of them, losing his scholarship in the process.
- Person that Ciro met personally and shall remain anonymous for now for his privacy: once Ciro was at a bar with work colleagues casually, it was cramped, and an older dude sat next to his group.The dude then started a conversation with Ciro, and soon he explained that he was a mathematician and software engineer.As a Mathematician, he had contributed to the classification of finite simple groups, and had a short Wiki page because of that.He never did a PhD, and said that academia was a waste of time, and that you can get as much done by working part time a decent job and doing your research part time, since you skip all the bullshit of academia like this.Yet, he was still invited by collaborating professors to give classes on his research subject in one of the most prestigious universities in the world. Students would call him Doctor X., and he would correct them: Mister X.As a software engineer, he had done a lot of hardcore assembly level optimizations for x86 for some mathematical libraries related to his mathematics interests. He started talking microarchitecture with Ciro's colleagues.
Pinned article: Introduction to the OurBigBook Project
Welcome to the OurBigBook Project! Our goal is to create the perfect publishing platform for STEM subjects, and get university-level students to write the best free STEM tutorials ever.
Everyone is welcome to create an account and play with the site: ourbigbook.com/go/register. We belive that students themselves can write amazing tutorials, but teachers are welcome too. You can write about anything you want, it doesn't have to be STEM or even educational. Silly test content is very welcome and you won't be penalized in any way. Just keep it legal!
Intro to OurBigBook
. Source. We have two killer features:
- topics: topics group articles by different users with the same title, e.g. here is the topic for the "Fundamental Theorem of Calculus" ourbigbook.com/go/topic/fundamental-theorem-of-calculusArticles of different users are sorted by upvote within each article page. This feature is a bit like:
- a Wikipedia where each user can have their own version of each article
- a Q&A website like Stack Overflow, where multiple people can give their views on a given topic, and the best ones are sorted by upvote. Except you don't need to wait for someone to ask first, and any topic goes, no matter how narrow or broad
This feature makes it possible for readers to find better explanations of any topic created by other writers. And it allows writers to create an explanation in a place that readers might actually find it.
Figure 1. Screenshot of the "Derivative" topic page. View it live at: ourbigbook.com/go/topic/derivativeVideo 2. OurBigBook Web topics demo. Source. - local editing: you can store all your personal knowledge base content locally in a plaintext markup format that can be edited locally and published either:This way you can be sure that even if OurBigBook.com were to go down one day (which we have no plans to do as it is quite cheap to host!), your content will still be perfectly readable as a static site.
- to OurBigBook.com to get awesome multi-user features like topics and likes
- as HTML files to a static website, which you can host yourself for free on many external providers like GitHub Pages, and remain in full control
Figure 3. Visual Studio Code extension installation.
Figure 4. Visual Studio Code extension tree navigation.
Figure 5. Web editor. You can also edit articles on the Web editor without installing anything locally.Video 3. Edit locally and publish demo. Source. This shows editing OurBigBook Markup and publishing it using the Visual Studio Code extension.Video 4. OurBigBook Visual Studio Code extension editing and navigation demo. Source. 
- Infinitely deep tables of contents:
All our software is open source and hosted at: github.com/ourbigbook/ourbigbook
Further documentation can be found at: docs.ourbigbook.com
Feel free to reach our to us for any help or suggestions: docs.ourbigbook.com/#contact