sdlion

I guess that's why is an active research area, and an interdisciplinary one, what is intelligence? Is what have you practiced a ton of times? Is what you have purposely and efficiently practice? What is the implication of "purposely" doing something and how much memory is involved into it? If memory is involved how much of it is relevant? What is problem solving or wisdom? creativity? How much diversity in your memory do you need for creativity?

So what should you cull and how much? There's already techniques in CNNs to trim unused or less active neuron paths to reduce a model's size, but how do you (and how much) do it in a general LLM? A product that will be used without supervision from a child to an 80yo elder?

If I recall correctly, electrolytic capacitors have to get "burned in" during manufacturing to make them work.

A current is passed through the capacitor and a thin film of oxide is built up in one of the terminals, according to the polarity. This is why electrolytic caps have polarity, if you use them with their polarity inverted, you flake off that oxide layer and thus short them out.

A free running current in that electrolyte boils it off, and you get an exploding cap

As a note, for Japanese text deepl is widely used even by Japanese people. From eng to jpn it may not choose properly nuanced words though, but it largely produces acceptable translations.

In my experience, being a workplace or even academia, there's as many types of work environments as types of teachers.

In the end, management (or founders if it's a startup) set the tone and who they end up hiring (thus, who will manage your work). And management is just normal people in the end.

I guess because at server farm level, performance/efficiency translate to real million USD savings. In general, at scale ends (the cloud and the embedded) this matters a lot. In resource limited environments like raspberry pi, this design philosophy wins over many users between DIY and the private industry.

For training, more GPU RAM will allow you train with greater resolutions in less time and better performance. Before feeding data to a model, it needs to be resized to a "network dimension" (YOLOv4 default is 416x416 px if I recall correctly). For training, it will group several samples and train with them at the same time, in "batches". For better generalization you want bigger batches (so more different images are feed at the same time). With a 3060 (non-Ti) you'll have 12GB of GPU RAM, with that I think you can run the default settings (network size, batch size and subdivision of batches) for the YOLOv4 model. If you want to go to 512px, you might have to increase the subdivision (create more subbatches) or reduce the batch size. If I recall correctly, you could find 3070 with less than 12GB of RAM, so in trying to purse faster training times (I'm not talking about inference, using the model to actually recognize something) you might not be able to train with a broader range of options that can improve your accuracy.

There's a reason to not compare it with the YOLOv4 family? If I recall correctly, the advantage of YOLOv5 over YOLOv4 is still disputed and YOLOv4-tiny seems widely used.

I went to a technical high school in Mexico. Here's divided as 6 years of grade school, 3 of Middle school and 3 of High school.

After finishing this "bachelor" high school you get a technician degree. Mine was an electronics technician. And well, I met some of my best friends there and equals even when most of them went to different majors in college.

I think is great when you can get specialized subjects in such environment since people with similar interests can interact between them more easily.

Granted, there were bullied people, others that just wouldn't fit, but for the most of us (on classrooms over 40 students) we went along fine.

And certainly I felt it a better environment than middle school (12-15yo). Friends that really didn't like it, weren't that interested in electronics or science, but still we got along.

But well, in the end, Mexico is a place where we care about our relationships. Maybe too much.

I don't think the concept of high school is wrong. On any country contacts are important and usually change your life, and any social hub in life is a chance. It's just that, environment and motivation matters.

Graphics also matter. My 6yo Inspiron let me develop with 8GB of RAM, last year I bought a new one with Ryzen5 set the very same link setup and even used the same SSD but it would began to freeze or crash...

The APU was reserving 3GB of RAM for the GPU. The next day I went for 8GB more.

I think I did that, then have problems with the windows installer. A driver prompt. I regularly have to give services to laptops, so I prefer options that "just works". This is also the reason I don't usually use the Windows installation media creator, because is slow and won't burn my several ISOs. There was a time the new Win10 iso installation process would fail on any configuration I would try... I lost somewhat around THREE DAYS trying to solve each f*"** cryptic error message and an old iso did the trick. Just a few hours updating and it was done.

Sadly most of my clients won't just accept the option of installing Linux ;)

There are available eInk with a DSI interface and Linux drivers? It shouldn't be that hard to build a DIY eInk slim tablet with a PiZero or for some, with a CM4.

Same problem with latin american keyboards found just a few kilometers south the US. If I remember correctly, the default modifier key in VIM is next to imposible to press on a l-a keyboard. It was ` ?