Image Generation — across courses
cmu-10799-diffusion-flow
- L10 @ 3514s
>> and yeah um experimentally we achieve still there one step result on imaget still there one step result on imaget still there one step result on imaget and it can also surpass uh dn and xit and it can also surpass uh dn and xit and it can also surpass uh dn and xit with four step um and we can get similar with four step um and we can get similar with four step um and we can get similar similar results uh compared to baseline similar results uh compared to baseline similar results uh compared to baseline for imaging at 512 and 512. Uh here are for imaging at 512 and 512. Uh here are for imaging at 512 and 512. Uh here are some samples from both the 256 some samples from both the 256 some samples from both the 256 resolution and 512 resolution and these resolution and 512 resolution and these resolution and 512 resolution and these are produced with just uh one step are produced with just uh one step are produced with just uh one step and and and but you know since since nowadays every but you know since since nowadays every but you know since since nowadays every every method you know can perform well every method you know can perform well every method you know can perform well on imageet so so we had this follow-up on imageet so so we had this follow-up on imageet so so we had this follow-up effort to try to scale it to 10 10 effort to try to scale it to 10 10 effort to try to scale it to 10 10 billion plus parameter scale uh for text billion plus parameter scale uh for text billion plus parameter scale uh for text to image and this is what I and this is to image and this is what I and this is to image and this is what I and this is kind of what I wanted to show for for kind of what I wanted to show for for kind of what I wanted to show for for this work is that TVM works well at this work is that TVM works well at this work is that TVM works well at scale and this model is is trained with
- L6 @ 4059s
Max. Please come to come in person. If Max. Please come to come in person. If you want to ask questions, it's going to you want to ask questions, it's going to you want to ask questions, it's going to be really fun. I promise. Uh, and then be really fun. I promise. Uh, and then be really fun. I promise. Uh, and then the Thursday we're going to be talking the Thursday we're going to be talking the Thursday we're going to be talking about text to image models and uh, how about text to image models and uh, how about text to image models and uh, how to make your model quality sota or at to make your model quality sota or at to make your model quality sota or at least this is what people think that least this is what people think that least this is what people think that that that they do that they do. And uh that that they do that they do. And uh that that they do that they do. And uh uh in the following week we're going to uh in the following week we're going to uh in the following week we're going to be talking about essentially how to make be talking about essentially how to make be talking about essentially how to make your model like just like like lightning your model like just like like lightning your model like just like like lightning fast like just like how to sample from fast like just like how to sample from fast like just like how to sample from these like iterative uh uh sampling these like iterative uh uh sampling these like iterative uh uh sampling procedures with only one step. Is it procedures with only one step. Is it procedures with only one step. Is it even possible? Then we're going to do it even possible? Then we're going to do it even possible? Then we're going to do it by using something like distillation. by using something like distillation. by using something like distillation. And then the the final class in the next And then the the final class in the next And then the the final class in the next is not final class but like the the is not final class but like the the
diffusion-principles-vizuara
- L11 @ 276s
I have structured this course and how I plan to teach it. It'll be very similar plan to teach it. It'll be very similar plan to teach it. It'll be very similar to Vijuara's other teaching material and to Vijuara's other teaching material and to Vijuara's other teaching material and I'll be teaching these series of I'll be teaching these series of I'll be teaching these series of lectures and it'll be uploaded on lectures and it'll be uploaded on lectures and it'll be uploaded on YouTube completely for free. Now let's YouTube completely for free. Now let's YouTube completely for free. Now let's dive in and understand the curriculum in dive in and understand the curriculum in dive in and understand the curriculum in detail. Thank you. Let us try to understand how this course Let us try to understand how this course is going to be structured. The title of is going to be structured. The title of is going to be structured. The title of this course is the principles of this course is the principles of this course is the principles of diffusion models from origins to advances. As I said before in the introduction, As I said before in the introduction, As I said before in the introduction, diffusion models has really diffusion models has really diffusion models has really revolutionized the text to image revolutionized the text to image revolutionized the text to image synthesis as they produce images which synthesis as they produce images which synthesis as they produce images which are very realistic and some examples we are very realistic and some examples we are very realistic and some examples we see are Delhi 3 and stable diffusion. see are Delhi 3 and stable diffusion. see are Delhi 3 and stable diffusion. And apart from these applications uh And apart from these applications uh And apart from these applications uh which also include midjourney, there are which also include midjourney, there are which also include midjourney, there are also other domains where diffusion also other domains where diffusion also other domains where diffusion models find models find models find a lot of applications. a lot of applications. a lot of applications. The next domain that we are going to The next domain that we are going to The next domain that we are going to discuss is that of protein