Month: January 2024

January 25 2024, 22:27

Check out my workstation. Two 17” pro M3 max and M1, Apple Cinema HD 30”, LG HDR WQHD 34.5” screens, and two nearly identical das keyboard ultimates, and two identical Corsair M65 mice. It’s worth mentioning that Apple Silicon effortlessly handles three displays with a total resolution of 9456×2234. Trying to keep up with everything!

January 25 2024, 16:27

Interesting from Anton Repushko:

“A lengthy text about how I’ve been going insane from despair over the last 8 months and started learning Ancient Greek.

Posts about incomprehensible white letters and jokes about lazy ancient Greek scribes have already flickered above. It’s time to put everything together into a coherent picture and explain myself. I can’t publish and explain the technical solution in detail due to the rules (it’s under review), but I’ll share various other things.

The competition is called the Vesuvius Challenge. When Vesuvius erupted and destroyed Pompeii, it also destroyed the city of Herculaneum at its base. We are interested in Herculaneum because there they excavated a villa that had a very rich library. And in this library, they have already found/expect to find about 4-5 thousand scrolls. Important clarification: from Antiquity, very few texts have survived to us: the Odyssey, the Iliad, all philosophy – just a small percentage of all texts that existed back then. Imagine how much knowledge and history are stored in these thousands of scrolls. But of course, there is a problem: they look like potatoes charred on coals due to conservation in lava and ash (but thanks to this they lay in the ground for 2000 years).

Enter Professor from Kentucky, Dr. Brent Seales, who spent his entire career pushing the following method: let’s make 3D CT scans of these scrolls and then try to cleverly restore and read them. The plan is reliable, like a Swiss watch, if it weren’t for one thing: the inks there are carbon-based and are invisible on MRI results. There were also attempts to scan the same scroll at different beam phases and by subtracting the results of this get some letters. The letters were enough for an article in Nature, but it looks pretty awful. What to do with this next?

Enter ex-CEO of Github, Nat Friedman: a philanthropist, organizes various AI grants, and invests in AI startups. A very nice fellow who helped organize this competition and stuffed the prize fund with money. Here’s his private fund and so far, to me, he seems like a healthy-person’s Elon Musk.

The competition itself started last February with a prize fund of 1M USD. It began on Kaggle, and then it all continued on Discord. There are 3D CT scans of the scrolls and the text needs to be read from there. The process of virtual unfolding of text proposed by the organizers is quite complicated and tricky, and now there’s about 7-8Tb of data, which sets a pretty high technical entry barrier.

After the Kaggle competition, things quieted down until in August, two students (one from the states, the other from Berlin) independently found some unclear reliefs with their eyes (!) that confirmed the initial idea: despite the ink being invisible, the paper remained deformed at the place of the written letters. Thus, the task became trying to programmatically learn to identify these deformed places (sometimes invisible to the eye) and read the letters. Around early August, that’s when I dived head-first into the competition.

So many sleepless nights, worries, and everything else throughout these months. I came and pestered anyone who could help me with advice/ideas (thank you all, friends. I couldn’t have done it without you). When discussing news and what happened during the day with my girlfriend, I thought dozens of times that nothing at all happened to me: I was solving the scrolls, just like yesterday, and will continue tomorrow.

And despite not meeting the official criteria for the main prize (four texts with 140 readable characters each), I think no one did. According to the organizers, they received about a dozen submissions. And for about a month now, we and the other participants have been waiting for the results of the papyrologists’ analysis. During these six months, I’ve met a lot of interesting people, won an intermediate prize of 5200 USD, started learning ancient Greek and greatly improved in solving such unclear problems.

Interestingly, during the competition, I felt incredibly fulfilled and it’s a very good alternative to all the races with LLM-research: you have a complex unclear problem, which intuitively seems solvable with current means. And the whole history of these scrolls (which were passed from hand to hand for decades until they could be read with new technical means) is a great example of human cooperation for something that can’t be measured in money. I really like it.

The organizers have already announced a new stage of the competition sometime in February, so there will be even more posts about it. Below is a picture with roughly the final results achieved: individual words and prepositions are readable, but without knowing Ancient Greek, it’s hard to read more. Let’s see what happens next.”

January 22 2024, 21:42

I continue to read Ed Yong’s “An Immerse World” about how animals perceive the world.

The chapter on hearing has many interesting insights. For instance, owls are known to have large eyes that unusually for birds, face forward. It was previously believed that it was their sharp, night-adapted vision that allowed owls to hunt in the dark of night. By the mid-20th century, it was understood that ears aid owls to a much greater extent. Have you ever seen owl ears? The so-called ears, often visible at the top of various owl’s heads, though located in the right place, as in many mammals, are actually not related to hearing. The real ears are situated on the edge of the facial disc, as shown in the first picture. They are simply huge! The narrow and dense feathers of the facial disc, tightly fitted together, form a parabolic antenna designed to capture the faintest sounds and rustles, directing them to the ear and further into the corresponding brain areas.

Moreover, you can see the back part of the eyeball through the ear hole in an owl. But there’s another trick, used by only a few animals in the world apart from owls – the asymmetry of the auditory openings — the left ear is noticeably higher than the right. For some owls, for whom hearing is especially important, this asymmetry is even reflected in the skull. By the time difference in signals arriving at the right and left ear, owls can determine the position of a sought-after mouse (even under snow) with accuracy up to 1°. Experiments have established that the owl’s brain can detect a signal arrival difference of 0.00003 seconds. This precision so amazed the scientists and, as usual, the military joined them, that entire bionics laboratories were deployed in the last century to unravel the mechanism of owls’ super-precision location.

To not interfere with their super-sensitive ears, the owl’s feathers and its flight are absolutely silent. And most rodents do not hear an owl approaching from behind. But there’s one rodent – the kangaroo rat, found in North America, which has an enormous middle ear. There, frequencies produced by a flying owl are amplified, and the rodent manages to escape. Overall, this is fairly simple, because owls are not adept at fast maneuvering, while rodents are.

Another interesting story is about the Ormia ochracea fly, also found in the USA, also a nocturnal hunter. It’s notable for its parasitism on crickets and exceptionally precise directional hearing. The female is attracted by the male cricket’s song (only the male’s) and lays larvae on or nearby him. Then, the larvae start to parasitize the cricket. They penetrate inside his body and feed on his tissues, which ultimately leads, of course, to the cricket’s death, much to the joy of the larvae.

So, these flies are known for their unique “ears,” which are complex structures inside the front segment of the fly’s body, near the base of its front legs. The fly is too small for the time difference in receiving sound waves by the left and right “ear” to be calculated in the usual way, as in other animals. Just to give an idea, the distance between its “ears” – is as tiny as the dot at the end of this sentence, which is very small. However, it manages to determine the direction of sound sources with astonishing accuracy — up to 1 degree. That is, it turns towards the cricket’s sound with that precision, and then flies straight towards him. How does it do this? It turns out that the eardrums of the opposite ears are directly mechanically connected by a lever, which creates about a 50ms delay on the incoming signal, and this is processed by its brain sufficiently for a turn. Incidentally, the turn happens unconsciously.

When these flies ended up in Hawaii, 30% of the male cricket population was infested with their larvae and were doomed. What happened next was this: within just 20 generations, a mutation that physically changed the organs with which crickets chirp spread, and these updated crickets became undetectable to the fly. But truth be told, they also had difficulties with female individuals, who stopped noticing them. What happened next was this – these cricket-invalids from birth (but protected from the flies) started hanging around those that could still sing. Because their wing movements produce a female (who of course cannot hear them, since they’re invalids, but hears the healthy ones nearby), evolution does not “turn off” the movements that should produce sound, but physiologically do not.

Another interesting story about Zebra Finch birds. Well, let’s start with the fact that they (and not just them, but birds in general, just they do it very cool) turned out to be able to hear components of a “song” lasting 1 millisecond. But here’s something interesting. For example, this bird’s song could be recorded as A-B-C-D-E. When scientists Beth Vernaleo and Robert Dudling inverted the middle part, so it became A-B-Ɔ-D-E, the zebra finch always heard the difference, while to our ear there’s no difference at all, even after extensive training. But when the team slightly increased the gap between these “syllables” of the song, it sounded 100% different to a human, but the bird recognized it as its own.

What they did next is interesting. Shelby Lawson and Adam Fishbein decided to randomly swap the “syllable-notes”. This changes the entire song to something unrecognizable! but it turned out that for the bird it was THE SAME SONG. That is, the order does not matter to them. The components are important. It seems that they perceive the song quite differently than we do. For us, it is something prolonged over time. For them, they are “checkboxes”.

It also turned out that if you take two seemingly identical songs from different birds – all the notes the same, they sing very very similarly, and replace say B with B from another bird, leaving A,C,D,E original. This replacement is detected by the finches. It seems that for them, each “note” has its own hues, which mean something. For us, it’s just a note.

Zebra Finches are monogamous, and live with a partner all their life. With the help of this singing, they hear each other across half a forest and understand where to fly. To us, they all sing the same song. If you think about it, the world is perceived by different animals so differently, that we simply cannot imagine this sensation.