NVTabular
NVTabular copied to clipboard
NVIDIA-Merlin
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.
**What needs doing** Current existing examples are covering multiple concepts. They should be moved and/or removed. Based on: https://docs.google.com/spreadsheets/d/1ncCGvDZtpUTbtI7VGZpJTk3RjcT5Ww9S_fr7YTF_yFw/edit#gid=0 - [ ] MovieLens - [ ] Scaling Criteo - [...
**What is your question?** I have a dataset that includes a column feature of pre-trained embeddings. I couldn't find any documentations or examples on how this column should be passed...
For the example notebooks in NVTabular that are run as integration tests: - [ ] The metrics for asvdb that should be captured: - [x] Training Time - [x] Max...
Example code: ``` import json import os import pathlib from google.protobuf import text_format from numpy import true_divide import nvtabular as nvt from nvtabular.inference.triton import export_hugectr_ensemble # Temporary fix to https://github.com/NVIDIA-Merlin/NVTabular/issues/1221...
**Is your feature request related to a problem? Please describe.** I wish NVTabular could help me roll up the last ten or so item ids that a user interacted with....
- [ ] gpu - [ ] cpu - [ ] multi-gpu: - 2x - 3x
NVTabular should create ops it easier the process of aggregating some past user interactions to represent users preferences. Some possible approaches would be: - [ ] Including features of the...
Coverage improved: * From 36.6% to 38.4% by https://github.com/NVIDIA-Merlin/NVTabular/pull/1475 Modules with missing docstrings: - [ ] ops - [ ] tools - [x] workflow Modules to be deprecated: - [...
Create a function that encapsulates remove targets from a workflow because this is something that will need to be done when migrating your Feature engineering to serve inference.
Metadata
Owner
Metadata
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.