NVIDIA-Merlin / NVTabular

NVTabular is a feature engineering and preprocessing library for tabular data designed to quickly and easily manipulate terabyte scale datasets used to train deep learning based recommender systems.
Apache License 2.0
1.04k stars 143 forks source link

nvtabular import error :Kaggle notebook #1731

Closed DilipKumar3 closed 1 year ago

DilipKumar3 commented 1 year ago

Error on importing nvtabular in kaggle notebook imageKindly let us know on how to resolve this

bschifferer commented 1 year ago

@DilipKumar3 can you provide the notebook you tried on the Kaggle environment? How did you install nvtabular?

karlhigley commented 1 year ago

From the error message, it sounds like there's a newer version of NVTabular running against an older version of merlin-core. What versions of those libraries are installed?

oliverholworthy commented 1 year ago

This is likely to be because the Kaggle environment is running Python 3.7 and we support a min version of Python 3.8. The bug on our side is that we didn't update the python_requires attribute in our published package for NVTabular, to reflect the python version requirement.

Installing an older version of NVTabular may work (pip install nvtabular==1.4.0)

DilipKumar3 commented 1 year ago

pip install nvtabular==1.4.0

@bschifferer im using _pip install nvtabula_r for installing nvtabular.

bschifferer commented 1 year ago

pip install nvtabular==1.4.0

@bschifferer im using _pip install nvtabula_r for installing nvtabular.

Hello @DilipKumar3 , can you provide more information what you want to tryout with NVTabular?

The main issue is that Kaggle environment uses Python 3.7 as pointed out by Oliver and Karl. Some Kagglers provided getting started notebook for Merlin on Kaggle in the Otto Competition: https://www.kaggle.com/code/snnclsr/transformers4rec-synthetic-data-example https://www.kaggle.com/code/radek1/training-an-xgboost-ranker-on-the-gpu https://www.kaggle.com/code/theoviel/pretraining-with-merlin-s-transformers4rec

I am not sure, if you want to use the same competition. They installed it with

!pip install -q transformers4rec[pytorch,nvtabular]
!pip install -q -U nvtabular==1.3.3

does that work for you?

DilipKumar3 commented 1 year ago

like mentioned i tried by installed specific version and it worked . thanks for the help.