recommenders
recommenders copied to clipboard
Question: examples/best practices for integrating tfrs with tfx
I am currently working on integrating a tfrs model into an existing TFX pipeline to provide on-service recommendations. I am quite new to both TFX and tfrs and am not seeing any resources on integrating the two. I want to be sure I am implementing best practices–in particular with TFT and TFMA. Does anyone know of existing docs that may help me with this, or better yet, an existing example pipelines? I posted a similar question on the forum but have not received any feedback. Maybe @maciejkula, you would have some insight?
Any help would be much appreciated!
any updates on this?
Hello, can you send the detailed code for saving the model? I want to save the model now, but I don't know how to implement it
TFX ExampleGen does not currently support TFDS (https://github.com/tensorflow/tfx/issues/2027). This is a limitation on TFX.
One hacky way to do it is to just feed an empty csv to CsvExampleGen but do all the data ingestion in the trainer. Or you can create a custom ExampleGen (I think).
Once you figure out the ExampleGen part, the rest of the pipeline should just work.
We have released a ranking tutorial with TFX.
@windmaple will there be a TFRS-TFX tutorial using all components?
Our TFX colleague @rcrowe-google is working on one
This new TFX+TFRS retrieval tutorial (https://www.tensorflow.org/tfx/tutorials/tfx/recommenders) has finally landed. It uses all major TFX components.
Thanks to the hard work from @rcrowe-google.