You can also generate a lower quality image and use an AI upscaler to increase the size of your artistic QR code.

You’re right. I read past the “I want to learn ML” and went straight to “do something useful with the data”.

If the goal is to understand how modern LLMs work, it’s also good to read up on RNNs and LSTMs. For this, 3Blue1Brown does an amazing job, and even posted an in-depth video about transformers. I’d watch that next, followed by implementing a simple transformer in PyTorch (perhaps using the existing blocks).

You could argue that it’s important to design everything from scratch first, but it’s easier to first go high level, see how the network behaves, and then attempt to implement it yourself based on the paper. It is up to OP how comfortable he is with the topic though 😁

Depending on how much compute you have available, you can look into finetuning models from HuggingFace (e.g. Llama 3, or a smaller Phi model). Look into LoRA, and try to learn how the model you choose calculates the loss.

There are various ways to train, and usually involves masking the input by replacing random input tokens with the mask token. I won’t go into too much detail with this, because it’s a lot to explain, and I suggest you read an article on this (link1 or link2)

That is good to know. Tried the free version of Roll20 before, and it definitely felt lacking in certain areas. Oh, and thanks for letting me know about the sale! I’ll definitely keep an eye out for that one :)

With the way current LLMs operate? The short answer is no. Most machine learning models can learn the probability distribution by performing backward propagation, which involves “trickling down” errors from the output node all the way back to the input. More specifically, the computer calculates the derivatives of each layer and uses that to slowly nudge the model towards the correct answer by updating the values in each neural layer. Of course, things like the attention mechanism resemble the way humans pay attention, but the underlying processes are vastly different.

In the brain, things don’t really work like that. Neurons don’t perform backpropagation, and, if I remember correctly, instead build proteins to improve the conductivity along the axons. This allows us to improve connectivity in a neuron the more current passes through it. Similarly, when multiple neurons in a close region fire together, they sort of wire together. New connections between neurons can appear from this process, which neuroscientists refer to as neuroplasticity.

When it comes to the Doom example you’ve given, that approach relies on the fact that you can encode the visual information to signals. It is a reinforcement learning problem where the action space is small, and the reward function is pretty straight forward. When it comes to LLMs, the usual vocabulary size of the more popular models is between 30-60k tokens (these are small parts of a word, for example “#ing” in “writing”). That means, you would need a way to encode the input of each to feed to the biological neural net, and unless you encode it as a phonetic representation of the word, you’re going to need a lot of neurons to mimic the behaviour of the computer-version of LLMs, which is not really feasible. Oh, and let’s not forget that you would need to formalize the output of the network and find a way to measure that! How would we know which neuron produces the output for a specific part of a sentence?

We humans are capable of learning language, mainly due to this skill being encoded in our DNA. It is a very complex problem that requires the interaction between multiple specialized areas: e.g. Broca’s (for speech), Wernicke’s (understanding and producing language), certain bits in the lower temporal cortex that handle categorization of words and other tasks, plus a way to encode memories using the hippocampus. The body generates these areas using the genetic code, which has been iteratively improved over many millennia. If you dive really deep into this subject, you’ll start seeing some scientists that argue that consciousness is not really a thing and that we are a product of our genes and the surrounding environment, that we act in predefined ways.

Therefore, you wouldn’t be able to call a small neuron array conscious. It only elicits a simple chemical process, which appears when you supply enough current for a few neurons to reach the threshold potential of -55 mV. To have things like emotion, body autonomy and many other things that one would think of when talking about consciousness, you would need a lot more components.

I got NFS Most Wanted (2005) working in Wine, and was somewhat impressed how easy it was at the time. Game worked quite well, and would only crash once in a while with some cryptic errors that I don’t remember. Made me hopeful for the future of linux gaming :)

Yeah, it’s the Osaifu-Keitai. Apple has it enabled for all phones on the market, while Android phone manufacturers avoid adding it to theirs outside Japan because they would have to pay fees to Sony for it. The funny part is that Sony itself doesn’t enable it for phones outside Japan, even though FeliCa is a subsidiary of Sony :D Another funny bit is that some phones, like the Pixel, are capable of running it on phones made for other markets. Some users were able to force the Osaifu-Keitai app to think the phone was made in Japan, and that was all it took to enable it (although you’d have to root your phone + the manufacturer should have released their phones in Japan, to ensure the chip is capable). So, yeah, although a few years ago it might have been a specific chip being needed in the phone, nowadays it’s mostly software that doesn’t allow you to use the one you have while in Japan.

All in all, PASMO/Suica/etc is basically a very limited debit card company haha. I guess Japanese people enjoy using it mainly because it puts a cap on how much they can spend (iirc, about 100 euros allowed at once on the card). Japan is a highly consumerist society, so this format was probably adopted (instead of credit/debit cards) mainly to combat it somewhat :D

And for some reason you still can’t charge transport cards online or with a credit/debit card if you don’t have a japanese phone. Think that’s coming in 2035 at this rate? 🤣

Wow, some of the comments on that article saying Google should have made Android closed source are mindboggling. They realize they never would have had their current worldwide marketshare if they did that, no?

But maybe if they did, we would have had more people working on true linux phones 🤔 I’m a bit torn on this one haha.

The Framework 13 inch model should be plenty, especially if you want to dev on the go. Much more lightweight and smaller, and you can connect it to external monitors if the screen size is not big enough. Also, you shouldn’t have issues running Linux on either laptops.

Instead of going for the 16 version, I would use the extra 900-1000 euros (that’s the amount I saw I could save between the two almost maxed-out models) to make a dedicated server or mini-cluster to run your workloads. Deploy Kubernetes or Proxmox on it, and you’ll also get some more practice on it outside work if you want to run stuff for your home lab. That is only if you don’t want to game on your laptop, but I’d still put that money aside to make a desktop.

He’s the chosen one, of course it’s supposed to look like that!

It’s amazing that Linux gaming is becoming a thing that’s better sometimes than Windows gaming (minus the getting banned part in some games). I also like that AMD is making some big pushes on open source drivers, plus their ROCm open-source alternative to CUDA.

This is a great time for Linux users! :)

Same. It sucks that most banks wouldn’t jump on this train :(

What a stupid article. It’s like saying “stop using electric vehicles because you can’t use gas stations”. I don’t understand why he’s so adamant about this? It’s not like Wayland had about 20 years of extra time to develop like X11. People keep working on it, and it takes time to polish things.

If I understand correctly, this would just provide the binary for paru without really making it accessible to the user, no?

In the end, I just decided to copy all my dotfiles to /home/user and running an additional script after reboot. For some reason, if you try to arch-chroot and install it tries to build everything as root (even after providing “-u user” to arch-chroot).

It’s a cooling block for the GPU, and apparently it can knock off up to 20 degrees. However, it is crazy expensive, like 900$.

I’ve looked into this before, and it really depends on the type of RFID they use. Older versions have been cracked, but newer ones can’t be copied over (easily or at all).

If your company is serious about security, you will not be able to put the content of the card on your phone. What newer, more secure versions of RFID do is receive a code from the reader system, replies to it internally, and then sends back the answer. Even if you try to copy this over, you will not be able to open the doors of your facility.

I think the first step should be to use one of these apps that can read RFID and see what protocol your card uses. If it’s an unsecure one (i.e., only pushes out a code and checks it in their database that it’s yours), you could probably try to copy it over. However, if it’s not, you could also just dissolve the card with some acetone and place the resulting wires in your phone’s case, near the bottom. Like that, it shouldn’t interfere with your phone’s NFC, as that one is usually next to the top area of your phone.

Framework laptops are getting better. Not Apple levels good, but it certainly beats them in average longevity.

The only hope with Apple is having the EU step in again to stop this kind of bullcrap.

You can do it here too! Just tag @remindme@mstdn.social :)

I also recommend Librewolf. It’s very good at what it does. They also have a list of recommended addons to enhance your privacy.