Applied Deep Learning with PyTorch - Full Course
video description
Date: 2022-03-14
Related videos
Comments and reviews: 10
Barry
Sorry to point these out. They seemed to jump out at me. The maximum of a sigmoid is exactly 0.25. f(x) = sigmoid(x). f''(x) = sigmoid(x) ( 1 - 2 sigmoid(x) ) ( 1 - sigmoid(x) ) = 0 when sigmoid derivative is maximum. This can only happen when sigmoid(x) = 0.5. f'(x) = sigmoid(x) ( 1 - sigmoid(x)) implies the maximum derivative is 1 quarter. The worst case is not when the derivative is maximum 4 times in a row. The worst case is when the gradient is so low it is treated as zero within machine precision. The graph of the derivative of relu should not be continuous.
reply
Sorry to point these out. They seemed to jump out at me. The maximum of a sigmoid is exactly 0.25. f(x) = sigmoid(x). f''(x) = sigmoid(x) ( 1 - 2 sigmoid(x) ) ( 1 - sigmoid(x) ) = 0 when sigmoid derivative is maximum. This can only happen when sigmoid(x) = 0.5. f'(x) = sigmoid(x) ( 1 - sigmoid(x)) implies the maximum derivative is 1 quarter. The worst case is not when the derivative is maximum 4 times in a row. The worst case is when the gradient is so low it is treated as zero within machine precision. The graph of the derivative of relu should not be continuous.
reply
BuGr33d0_Hunter
This was amazingly done. So much information packed into an amazing lecture. Executed very passionately outstanding. I applaud all the work and effort you put into this. This knowledge will definitely help further down my programming in python for security analyst career. Much appreciated. If I could please have your contact info, to maybe help on some projects, please let me know. Have a wonderful month. -
reply
This was amazingly done. So much information packed into an amazing lecture. Executed very passionately outstanding. I applaud all the work and effort you put into this. This knowledge will definitely help further down my programming in python for security analyst career. Much appreciated. If I could please have your contact info, to maybe help on some projects, please let me know. Have a wonderful month. -
reply
Mayank
Dude! you didn't finish the video. There are 2 big training functions that should have run after your explanation. Also, you explained the problem using LSTM but you implemented GRU with a dense and confusing diagram. It makes me wonder if people who are commenting good here have even watched the video passed the introduction.
reply
Dude! you didn't finish the video. There are 2 big training functions that should have run after your explanation. Also, you explained the problem using LSTM but you implemented GRU with a dense and confusing diagram. It makes me wonder if people who are commenting good here have even watched the video passed the introduction.
reply
Homelife
Brilliant Course! The instructor has a wonderful English accent, but he speaks so fast. I would recommend watching the video at half of the speed. But still OK! Thank you FreeCodeCamp for providing such a wonderful course, and thanks for the instructor for sharing his beautiful explanations for free!
reply
Brilliant Course! The instructor has a wonderful English accent, but he speaks so fast. I would recommend watching the video at half of the speed. But still OK! Thank you FreeCodeCamp for providing such a wonderful course, and thanks for the instructor for sharing his beautiful explanations for free!
reply
rajat
Hi...video is great...however it seems as if you were being superficial during preprocessing part...i think that is the mot important aspect of process.....please be more rigorous in explaining the steps...why r u creating dictionaries in a certain format...what kind of input u r expecting....etc
reply
Hi...video is great...however it seems as if you were being superficial during preprocessing part...i think that is the mot important aspect of process.....please be more rigorous in explaining the steps...why r u creating dictionaries in a certain format...what kind of input u r expecting....etc
reply
Kareem
In 2:45:10, this means that you throw away the entire pair of sentences only because they have a word that's not frequent. Isn' this a big loss in the data. Can't we weight the the words instead?
reply
In 2:45:10, this means that you throw away the entire pair of sentences only because they have a word that's not frequent. Isn' this a big loss in the data. Can't we weight the the words instead?
reply
Venkat
Guys can anyone help me , I finished training my model then i closed my jupyter notebook , then after restarting i lost all trained data , so can you tell me how to save the model??
reply
Guys can anyone help me , I finished training my model then i closed my jupyter notebook , then after restarting i lost all trained data , so can you tell me how to save the model??
reply
Carlos
Learning about Graph Theory (Directed Acyclic Graphs) before learning deep learning frameworks can help you understand the -neural network- jargon (vocabulary) better.
reply
Learning about Graph Theory (Directed Acyclic Graphs) before learning deep learning frameworks can help you understand the -neural network- jargon (vocabulary) better.
reply
Amardeep
Thanks for this amazing video you ccovered most of the crirical topics thanks a ton,there arent many resources for encoder and decoder,again thanks a lot!!!
reply
Thanks for this amazing video you ccovered most of the crirical topics thanks a ton,there arent many resources for encoder and decoder,again thanks a lot!!!
reply
Saswata
The way you explained everything in great detail is just mesmerizing........No words for this speechless awesome video......Thanks a ton
reply
The way you explained everything in great detail is just mesmerizing........No words for this speechless awesome video......Thanks a ton
reply
Add a review, comment
Other channel videos
