manujosephv / pytorch_tabular

A standard framework for modelling Deep Learning Models for tabular data
https://pytorch-tabular.readthedocs.io/
MIT License
1.39k stars 140 forks source link

Denoising AutoEncoder link URL is broken #502

Open ramakrse opened 2 weeks ago

ramakrse commented 2 weeks ago

In the readme document Denoising AutoEncoder is broken. Request a fix

dosubot[bot] commented 2 weeks ago

The link to the Denoising AutoEncoder in the README document is not broken. It correctly points to a Kaggle notebook: Denoising AutoEncoder [1].

To continue talking to Dosu, mention @dosu.

ramakrse commented 2 weeks ago

@dosu https://www.kaggle.com/code/faisalalsrheed/denoising-autoencoders-dae-for-tabular-data does not exist. It says that We can't find that page

dosubot[bot] commented 2 weeks ago

The link to the Denoising AutoEncoder in the README document appears to be broken, as the page on Kaggle cannot be found. Unfortunately, I don't have an alternative link or resource for the Denoising AutoEncoder. You might want to check the repository's changelog or update history to see if there have been any recent changes or updates regarding this link [1].

To continue talking to Dosu, mention @dosu.

manujosephv commented 2 weeks ago

@ramakrse For some reason the kaggle notebook that was linked was taken down. I found another one which is also relevant. https://www.kaggle.com/code/springmanndaniel/1st-place-turn-your-data-into-daeta