[FEA] Incremental Pre-processing

Open
#798 4 comments 1 reaction 0 assignees View on GitHub

Nobody has claimed this yet.

Assessment

Difficulty
5/5
Estimated time
Over a week
Newbie friendliness
25/100
Issue type
Feature
Clarity
Mostly clear
Activity status
Stale
Tech stack
python

Research direction

No files or tests are mentioned. Start by tracing NVTabular's categorical preprocessing and mapping entry points, then review how existing encoded values are persisted and extended. Done means previous mappings remain stable and new categorical values receive contiguous ids suitable for appended embedding tables.

Written by the indexing model from the issue text.

Description

Iterative Deployment session-based

Is your feature request related to a problem? Please describe.
It is common in industry to train recommender systems models incrementally, i.e., taking a model trained with past data and fine-tuning with new data. In such cases, new values on existing categorical features need to be encoded as contiguous ids on top of the existing ones, whose embeddings will be appended to the pre-trained embedding tables.

Describe the solution you'd like
NVTabular should support incremental pre-processing, by keeping the previous mapping between raw values and encoded values for categorical features (so that they match the position of the pre-trained embeddings), and assigning new values as contiguous item ids.

Dominant language
Python
Stars
1.2k
Forks
149
PR merge metrics
No merged PRs in 30d

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from NVIDIA-Merlin/NVTabular

All issues in NVIDIA-Merlin/NVTabular

Similar issues

More Python issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.