what is the difference between DDPM vs cDDPM?

Well. Your mic is definitely mediocre.

00:00 Sorry, I made a mistake in the title (𝗖𝗮𝘁𝗕𝗼𝗼𝘀𝘁 𝗖𝗹𝗮𝘀𝘀𝗶𝗳𝗶𝗲𝗿 using Scikit-Learn) of this video in the Jupyter Notebook. Scikit-Learn doesn't have this algorithm. Sorry for the mistake.

Work on your audio

Hi there! Great job, man. Your tutorials are amazing. Please keep going and upload more tutorials. If you don't mind, I have a suggestion for you. I believe it would be more beneficial if you explained each section or line while coding. For example, clarify the purpose of each section or each line. One more thing, your voice is not clear. The typing noise is louder than your voice. Thank you so much.

What if we have Tabular 1D data. Can you please guide how can we implement conditional DDPM on 1D data. Thanks

yo bro that video is awesome. can you make more videos like this!

Do you know of any way to do this for coco annotated data?

挺好，如果讲解的声音大一点就更好了。

good job

How to save the images in another folder after getting the patches

Thank you for sharing i really aprecciate it, i would try to train the model using a 2D latent space, do you think this architecture will also work for CelebA dataset?

36:13 In the scaling term, I accidentally wrote "beta_t" instead of "beta_t_square". I corrected it in the slide. Check out the GitHub address.

Can a PyTorch identify handwritten numbers from 0-99, the dataset is spliced into 0-99 using mnist

Can you help me design a CNN model? I already have a data set

Do you know how to concatenate datasets

You mentioned around 16:50 that you weren't sure why train loss was much higher than test loss. The reason is because of the L1 term. The returned loss value for train includes the L1 term. The returned loss value test does not. If you want comparable values between train and test, you need to either include the L1 term in the test batch function, or you need to only return the classification loss from the train batch function. Otherwise, a good explanation!

Thank you very much for all your time into these lessons. I have found it more helpful than lectures by MIT professors.

This help me, and your git hub code, Thanks

thank you! this was very helpful!

HI again, I want to kindly ask if you could consider doing a video about 1)the selection of clusters by computing the eigengap scores and plotting them as an eigengap plot.2) the use of normalized mutual information (NMI) score and Rand index to quantify the overlap between discovered and ground truth clusters. Thanks in advance.

Thank you! This is the best patchify example I've found.

I'm grateful for your video, and I'm presently exploring spectral clustering for data analysis for my Ph.D. dissertation in agriculture. Given my limited experience in this area, I'm curious if you could kindly consider sharing the scripts employed in your video and to share more videos about how to identify cluster sizes and how to validate them and how to do Character analysis of identified clusters too . thanks in advance .

This is a good channel and clearly explined. Can I get the code in the Github Repository?

At 06:19, I performed preprocessing but forgot to use the scaled features later. I have corrected this mistake in the code that I shared on GitHub. Check that out.

Within 10:11 - 10:43, I scaled the features but eventually forgot to use them later. It's better not to scale the features for this example. It seems unscaled features work better in calculating AIC.

Thank you

Thanks a lot. Looking forward to more videos.

BTW good content related to Data Science

how can I get this jupyter file.

26:05 ---> I made a mistake here; Train loss: {train_per_epoch_loss} should be the correct line but I wrongly wrote Train loss: {test_per_epoch_loss}. Remember to correct this portion.