deeplizard

deeplizard

Code for PyTorch Deep Learning course

Added 2018-09-24 10:18:33 +0000 UTC

Full access to code files and resources from the PyTorch Deep Learning course
Find out more about this code at https://deeplizard.com/resources

Code for PyTorch Deep Learning course

Comments

Hey Wesley. You are welcome for the course! It's okay for you to use parts of the code for your paper. Please site the website, deeplizard.com in your work. Here is the citation for the PyTorch course: deeplizard [Chris and Mandy]. “Neural Network Programming - Deep Learning with PyTorch” deeplizard, September 3, 2018, deeplizard.com/learn/video/v5cngxo4mIg May 8, 2021. This follows the below MLA format: Author or compiler name (if available). “Posting Title.” Name of Site, Version number (if available), Name of institution/organization affiliated with the site (sponsor or publisher), URL. Date of access. Good luck with your paper! Chris

2021-05-08 07:53:56 +0000 UTC

Hi guys. Thanks so much for the course. I'm doing an academic paper for my university in South Africa and I'll be using some of your code. My supervisor has told me that we can use other peoples code as long as we reference it correctly, the example he used though was from Github. Firstly, are you guys okay with me using parts of the code and if so how would you like me to cite your work? Thanks, Wesley.

2021-05-07 15:41:37 +0000 UTC

Hey Ling - Try artificially balancing your dataset using the data loader. See here: https://discuss.pytorch.org/t/how-to-prevent-overfitting/1902 Also, check the overfitting lesson in the DL Fundamentals course: https://deeplizard.com/learn/video/DEMmkFC6IGM

2020-06-04 05:53:29 +0000 UTC

Hi team, I enjoyed your course. I just tried to build my CNN to classify financial charts, but how comes my result all go to one class (I have three classes)? my data is imbalanced, classes ratio is 1:10:1, I have tried to use weighted cross entropy loss, it is not working. And I picked 5 samples from each class to train model, still all go to one class. I output data in the hidden fully connected layer, data is different for each image. but when it goes to the output layer to output 3 dim data, does not matter how different is the image, they all got the same output nearly. I cannot figure out why? Appreciate your help. Ling

2020-06-04 04:51:34 +0000 UTC

I believe you're talking about this video? https://youtu.be/LhEMXbjGV_4

2019-11-01 06:25:33 +0000 UTC

Hi Deeplizard. I remember that, in one of the sessions, you list the dataset folder as a tree structure. How to do that and which session is that? Thanks

2019-10-31 12:52:36 +0000 UTC

Thank you! I haven't seen that issue before. I see that you got a solution for this from someone on stackoverflow.

2019-10-28 15:03:40 +0000 UTC

Since I do not know how to post an image here. I did it on https://stackoverflow.com/questions/58589349/pytorch-confusion-matrix-plot. If you can anser this, it would be great thanks. Also, if you need me to delete the function, please let me know.

2019-10-28 10:35:21 +0000 UTC

Hi. It is a good series. I have run the confusion code, but there is something wrong with the code, in which the upper and lower line of the plot is not correctly placed

2019-10-28 10:11:47 +0000 UTC

Thank you Aayam! I fixed the typo you found and found an additional one. The notebook runs all cells now. Thanks for pointing that out. 🙏

2019-08-31 11:21:25 +0000 UTC

I bumped to the channel a few months back, (when it was relatively small). Really happy to see it grow while maintaining the quality of the videos. (if not doing things better.) FYI : deeplizard-full-code-fashion-mnist-project-v2.4.zip has a typo in one of the torch.cat examples. trivial but still . . . .

2019-08-25 22:45:21 +0000 UTC

Oh I see. That sounds cool. Send it, and I'll have a look.

2019-07-14 16:38:55 +0000 UTC

I love this example program and its a great springboard for more learning. Specifically I wanted to learn how to log and display information using Tensorboard. I have a jupyter notebook that takes the sample code you provided and adds the code to do this. It shows graphs of the scalars such and # correct, loss, and training time. Also is shows a collection of images for each epoch that show the input to and output of the two convolution layers. If this is something your can use let me know how I can send it to you.

2019-07-14 14:38:23 +0000 UTC

Hey jdgamet - The packages you installed are not required. I'd uninstall those. The resources.plotcm reference is to a local file. A folder called "resources" is contained inside the zip. Inside the "resources" folder there is a Python file called plotcm.py. The import should work as long as the folder is in the same directory as the running program. In Jupyter notebook's case, the "resources" folder should be in the same directory as the notebook file.

2019-06-13 09:40:44 +0000 UTC

I was getting complaints when running the code about missing the module "resources" and when I run $pip install resources It seems to go well. Then it complains that I don't have module "staty" and I install that as well. Now it's complaining that I don't have a module called resources.plotcm

2019-06-12 20:33:30 +0000 UTC

Hey Joaquim, thank you! We are hoping to get back working on the PyTorch series within the next week or two.

2019-03-13 15:07:48 +0000 UTC

Hi, Thanks for the tutorials, they are incredible! Do you have any clue about when the rest of these PyTorch series will be available?

2019-03-13 13:59:28 +0000 UTC

Hey Scott. You are welcome! We haven't covered any custom data loading yet in the PyTorch series, but it's pretty straight forward for image data. As long as the directory structure is correct, torchvision.datasets.ImageFolder just needs the path to the root directory. The structure follows this scheme: root > classes > data Here are some examples: ./root/dog/d1.png ./root/dog/d2.png ./root/cat/c1.png ./root/cat/c2.png Given this structure, we just pass the path for the root directory: dataset = torchvision.datasets.ImageFolder('./root') Then, a PyTorch dataset is returned: sample, label = dataset[0] Let us know if you get it to work.

2019-02-22 05:24:43 +0000 UTC

Thanks for the series. Is there an example of loading custom data with a command like torchvision.datasets.ImageFolder?

2019-02-21 19:29:35 +0000 UTC

More Creators

ploopyloopy

ploopyloopy

gumroad

tenpbl

tenpbl

fanbox

🔞 𝕽𝖆𝖊𝖑 🔞

🔞 𝕽𝖆𝖊𝖑 🔞

gumroad

smf_frieza

smf_frieza

patreon

Ulichan

Ulichan

patreon

tamayura_yoru

tamayura_yoru

patreon

taraadevlin

taraadevlin

patreon

yackimas

yackimas

fanbox

featherfeettickling

featherfeettickling

patreon

s_zergling

s_zergling

patreon

なまもななせ

なまもななせ

fantia

hoshime

hoshime

fanbox

Randy Bishop

Randy Bishop

gumroad

cafekko57

cafekko57

patreon

Aちゃん

Aちゃん

fanbox

circasurvive

circasurvive

patreon

OutForSmokes

OutForSmokes

patreon

lurrak

lurrak

patreon

polygonheart

polygonheart

patreon

raz3r

raz3r

patreon

thatssojordy

thatssojordy

patreon

Trash

Trash

gumroad

Beforu

Beforu

patreon

Tuesday Night Thunder

Tuesday Night Thunder

patreon

FrostyCherri

FrostyCherri

patreon

IosonoOtakuman

IosonoOtakuman

patreon

psicell

psicell

patreon

Dosha pmv

Dosha pmv

patreon

Nicole Ant

Nicole Ant

patreon

Kamiyamaneki

Kamiyamaneki

patreon

FLChrome

FLChrome

patreon

DemoraAvarice

DemoraAvarice

patreon

どくろパイ

どくろパイ

fanbox

eatingtill350

eatingtill350

patreon

みらつ

みらつ

fanbox

mozumichael

mozumichael

fanbox

ぷく

fantia

E.L.V.N

E.L.V.N

gumroad

Michael Guimont

Michael Guimont

gumroad

The Egg

The Egg

patreon