GoogleCloudPlatform / GoogleCloudPlatform/bigquery-ml-utils
Support non-TRANSFORM model Predictors
Open
- Dominant language
- Jupyter Notebook
- Stars
- 45
- Forks
- 16
- PR merge metrics
- No merged PRs in 30d
Description
Would it be possible to make a tensorflow predictor that did not expect the model to be queried with the TRANSFORM command, but rather the PREDICT command? This library is great for locally using exported models from BQML, but I am manually having to use the Predictor._get_transform_result() function to bypass 2-stage TRANSFORM+PREDICT logic.
Thanks!
Contributor guide
Assessment
This issue has not been assessed yet.