and what successes kandinsky is recognized for, this is what he does better than his analogues, let’s say, analogues are wrong according to his word, well, it looks like neural networkse the same problem, generating images from text, for example, what we’re cooler in, although kandinsky’s videos are generated, now uh-huh, now, well, that is , already, as it were, we’ve just recently introduced. and you can imagine how many terabytes of data storage is needed there, well , there is a supercomputer, it’s like it’s all learned on a supercomputer and stored nearby in special storage facilities, they are called c3, but these are these big, big data storages that are really for, say, in order for the model to be able to see the dataset, they need to be put there, and not just put, they need to be removed. all low-quality pairs, that is, filter out all low-quality pairs, well, if you go to the internet, then in general half of the data there will probably not be of very good quality, there is a lot of rubbish there, but that’s it, the picture does not correspond to the description, which it’s written down there, or maybe the picture contains a watermark, yeah, it’s probably

the creative industry podcast, our guest today is denis dimitrov, the man who created the kandinsky neural networkanov are still with you. and what successes are recognized for kandinsky?

you can draw a sketch and ask to make it realistic, this is also a task for a neural network as a whole, such as kandinskyral network is named after the great russian artist, who came up with the idea that it would be named after? and you know, we’ve actually been working on this problem for a long time, about 3 years already, and our first model was called this, it’s rudali, she came up with it herself, in short, yes, and we, we sort of thought how to animate it, that let's say, to make her more recognizable, and we decided, at that moment, 3 years ago, she created abstract objects, and we kind of thought, what kind of russian artists are there who painted in the style of abstractionism? art means that you need to do and come up with some new details, new features, make slight changes to existing architectures, thereby making... the algorithm the best in terms of the quality of understanding the text of this industrial product, the visual quality of the image, in fact , when we did kandinsky and are doing it, we set the first scientific and technical task, firstly to create something new, secondly to imp

creative industry podcast, our guest today is denis dimitrov, and the man who created the kandinsky neural networkuccesses are recognized for kandinsky already has this, what he does better than his analogues, let’s say so. well, it looks like neural networks that solve the same problem, generating images from text, for example, where we are cooler, although kandinsky videos are now generated, uh-huh, now, well, that is , it’s already, as it were, we recently presented a model, which generates the video, they called it kandinsky video, although it’s probably not... but these are these big, big data warehouses that really , say, for the dataset model to be able to see, they need to be put there, and not just put it, you need to remove all low-quality pairs, that is, filter, remove everything, well, if you go to the internet, then in general half of the data there will probably not be of very good quality, there is a lot of rubbish there , but that’s exactly the picture does not correspond to the description that is written there below. this also needs to be dealt with, that is, there are a huge num

dimitrov is our guest in the creative industry podcast on channel one, the man who created the kandinsky neural networkful team, and the main conclusion, even if we don’t quite understand everything, in general, in the end, we understand, there’s still nothing to be afraid of , look how nice, smart people are, doing those things that we may be afraid of for some reason, not we must be afraid, because this is the competitiveness of our country, and it is in the reliable hands of people like denis. denis, thank you very much, we are waiting for the next stage. the antennas are on your head, with you there are still analog, analog versions of the presenters, elena kiper, producer and video director, and roman karmanov, general director of the presidential fund for cultural initiatives. it was a creative industry podcast on channel one. see you. this is a podcast of witnesses from einstein and its leading film historians natalya ryabshchikova and stanislav didinsky, we talk about how to watch your favorite, half-forgotten, unknown soviet cinema now, how to find new meanings in it and how to get even more p

the creative industry podcast, our guest today is denis dimitrov, the man who created the kandinsky neural networkbetter, than analogues, let's say, analogues analogues is probably the wrong word, well, it looks like neural networks that solve the same problem. well, that is, it’s as if we recently introduced a model that generates video, we called it kandinsky video, although it probably should have been called something like torkovsky, well, we called it kandinsky video, we will continue this line, but what does it better, the domestic one knows better the cultural code, of course, is what we actually set before the model. what task, and we will certainly develop this further, what is the problem with russian data , knowledge of the russian domain is that data russians, of course there are fewer of them. what does the model learn from, it learns from data from the internet, that is, from open description image pairs, it seems to capture the data at the time of the request, naturally there is a team that deals with the data, and that is, the data is already contained in it, it does not search in

recognized for kandinsky is already this, what he does better than his analogues, let’s say, the analogues are wrong according to his word, well, it looks like it means on neural networks generated now, yeah, now, well, that is, it’s already, as it were, we’ve just recently presented a model that generates video, called kandinsky video. although we probably should have called it something torkovsky, well, they called kandinsky video, we will continue this line, but what does it better, knows the domestic cultural code better, of course, this is what we actually set before the model, which task, and we will certainly develop this further, what is the problem with russian data , knowledge of the russian domain, is that russian data, of course there is less of it, what the model learns from, learns from data from the internet, that is, from open pairs of image descriptions , it seems to capture the data at the time of the request, no, naturally, there is a team that deals with the data, and that is, the data is already contained in it, it is not searching on the internet at this moment, no, no, in no case, it is creates, in this way unique they are stored, this voluminous

dimitrov, our guest in the creative industry podcast on channel one is the man who created the kandinsky neural networkhere’s nothing to be afraid of after all, look, what nice, smart people who do those things. first channel. see you. dear graduates, we congratulate you on your graduation from school. now you can behave badly. but the most important thing is that your dreams become goals. and you, never doubting yourself, rushed to achieve your plans. i will come for you tomorrow after seven, i am the blue sun, i appear less often, less often, it hurts me so much to watch how beautifully you lie, you are on the body of the river, you feed on me, because my kiss, like a knife, i cut you, earth, kings of me . snow, love will save world. sleep, tomorrow i will rise for you with the first ray, i want to be weaker, i want to be with you more and more often and closer. i will burn you, but i will already crush you with my hot shoulder. to the white snowy lips and i will never, never see again, hey, earth, flood me. beyond the snow, love will save the world. hello everyone, this is a podcast 20 years later,

we continue the creative industry podcast, our guests today are the man who created the kandinsky neural networkho are still with you. i’m wondering if i’ve generated a masterpiece, then when i’ve created his masterpiece, it’s just a picture. impossible, yes, but how it was created, this is how kandinsky creates - pictures, i can also tell you on my fingers, imagine how a person creates a picture, any, any image, any painting, he has some kind of white canvas, well not necessarily white, just white paper, some kind of blank, napkin, napkin, yes, there is a tablet screen, he definitely has some kind of idea, he starts sketching out some details, with a pen, uh, stylus, whatever . adds details, maybe removes something, well, a person also has patterns in his head, that is, he has memorized something, he has, now he has patterns, this is after training, so step by step he simply creates a picture how kandinsky creates a model, in principle an analogy can be drawn, he begins to draw a picture from the so-called white noise, what is it, well, this is when there is no signal on the tv, there is such i

generation, we continue the creative industry podcast, our guest today is the man who created the kandinsky neural networky fingers, just imagine like a person creates a picture, any, any image, any painting, he has some kind of white canvas, well, not necessarily white, just white paper. some kind of empty, napkin, napkin, yes, there is a tablet screen, it has some kind of idea, necessarily, he starts sketching out some details, with a pen, a stylus, whatever, he adds details, maybe removes something, well, a person also has patterns in his head, that is, he will remember something, he, he he has a pattern, this is after training, step by step, he simply creates a picture, like creating a kandinsky model, in principle you can draw an analogy, he begins to draw a picture with the so-called white noise, what is it? well, this is when there is no signal on the tv, there is interference, just some kind of matrix naturally, this is your request, this is the same idea. the final picture is lysed, so to speak, appears, so to speak, for example, you want to make a pug in space, yeah, so you always have white noi

we continue the creative industry podcast, our guest today is the man who created the kandinsky neural networket’s say i send it, someone ends up using it, is it possible to open it , see everything the data, how the iteration took place, that is, or it has already... everything is blocked, this is actually, no, how the generation took place, you can see, of course, that is, what kind of specialist, not i, maybe a specialist, from what data it was created, impossible, yes, but how it was created, that’s how kandinsky creates pictures, i can also tell you on my fingers, imagine how a person creates a picture, any, any image, any painting, he has some kind of white canvas, but not necessarily white. just white paper, some kind of empty, napkin, napkin, yes, there is a tablet screen, he has some kind of idea, of course, he starts sketching out some details, with a pen, a stylus, whatever, he adds details, maybe removes something, a person also has patterns in his head, that is, he remembers something, he has it, he has he has patterns, this is after training, step by step, he simply creates a pic

you can draw a sketch and ask to make it realistic, this is also a task for a neural network as a whole, such as kandinskyon this task for a long time, about 3 years already, and our first model it was called that, it was rudali, she came up with it herself, in short, right? there are also neural networks, which also generally, well , work somewhere in this area, why create different models, there are two reasons, the first is that it’s kind of not true, let’s say the truth in the last one, this is generally machine learning - this is engineering, on interfaces with mathematics, and like any other art of engineering, it implies that you need to do and come up with some new details, new features, slightly make changes to existing architectures, thereby... do the algorithm is the best in terms of the quality of understanding the text, this industrial product, the visual quality of the image, in fact, when we did kandinsky and are doing it, we set the first scientific and technical task, firstly to create something new, and secondly, to improve current architecture, even western, there is huge scope for impr

example, you can draw a sketch and ask them to make it realistic, also a task for a neural network as a whole, such as kandinskyafter the great artist? and like any, any engineering art, it implies that you need to do and come up with some new details, new features, make a little changes to existing architectures, thereby making the algorithm the best in terms of the quality of text understanding, here this product, the visual quality of the image, and in fact, when we did kandinsky and are doing it, we set the first such scientific and technical task, firstly, to create something new, secondly, to improve the current architecture, even western, space for...

s voice could have been like. and what would the brutal pushkin post, a question for the neural network kandinsky example with a pack of popcorn. and he would now be sawing shorts and rails, the creator of another non-network, vladimir, is sure. i remember a wonderful moment, you appeared before me like a fleeting vision, like a genius of pure beauty. by the way, according to family legend, he and pushkin have common ancestors. pushkin’s great-grandfather, peter the great’s macaw hanibal, yes, he had many sons, one osip, who was actually pushkin’s grandfather, and he also had a son isaac. we have a belief in our family that this is our ancestor and are transmitted. for all men, some signs, yes, external similarity, curly hair, now with the help of artificial intelligence they help the ancestor illustrate his works, anagen, anagen in the forest, it’s somehow strange that he doesn’t seem to be particularly in the forest, dubrovsky, dubrovsky, everything right, good, and where is anegin, on the far right, no, that’s right, yes, starts, young, certainly the hero of the day would have been celebrated

you can draw a sketch and ask to make it realistic, this is also a task for a neural network in in general, such as kandinskyyes, who came up with what? art means that you need to do and come up with some new details, new features, make a little changes to existing architectures, thereby making the algorithm better, in terms of the quality of understanding the text, this industrial product, the visual quality of the image, in fact, when we did kandinsky and do it, we set the first scientific and technical task, firstly to create something new, secondly improve. rich architecture, even western, there is huge scope for improvement, and kandinsky - there are 2 lines, here is kandinsky 202122, these are three models, they have a unique architecture. their task is simply to generate everything in general, but we can set the task of showing the model a lot of our data of such a domestic cultural code, and this applies not only to the domestic cultural code, this applies to any types of data that we want to show to the model, and for this, of course, you need to make your own developments, we continue the creative