Conversation
Added links to TFT samples.
| incorporate the exported TensorFlow graph into their serving model, thus | ||
| avoiding skew between the served model and the training data. | ||
|
|
||
| # tf.Transform Samples |
There was a problem hiding this comment.
Perhaps discuss with @KesterTong regarding what the best place for this is?
Should it be here or in the Getting Started Section or the more examples section?
There was a problem hiding this comment.
I am trying to optimize for quick and easy discovery for samples @katsiapis @KesterTong Thoughts?
There was a problem hiding this comment.
Those samples are in the context of using Google Cloud ML Engine and also demonstrate how to use tf.Transform with Google Cloud Dataflow, while the samples in this repo give more detail on how to use tf.Transform in general.
Therefore I think that in the intro we should add a new paragraph (not a new section) here saying "tf.Transform works seamlessly with Google CloudML Engine. See *examples* for examples of using tf.Transform with Google CloudML Engine"
And in the "Getting Started" section I would add sentence along the lines of "For examples of how to run tf.Transform on Google Cloud Dataflow, see *examples*. These examples also demonstrate training with Google CloudML Engine.
| @@ -9,6 +9,10 @@ to their data as a TensorFlow graph. This is important as the user can then | |||
| incorporate the exported TensorFlow graph into their serving model, thus | |||
| avoiding skew between the served model and the training data. | |||
|
|
|||
There was a problem hiding this comment.
Perhaps we should also link to the Research Blog Post for TFT? @KesterTong what do you think? Any suggestions for location?
There was a problem hiding this comment.
In general I feel more cross-linking the better :) But here I will leave it to the esteemed reviewers.
There was a problem hiding this comment.
This README supercedes as a description of tf.Transform, so I think a link is not useful.
There was a problem hiding this comment.
I mean that this README supersedes the blog post.
There was a problem hiding this comment.
Also along the lines of more cross-linking, as far as I am aware Google Cloud ML Engine only mentions tf.Transform in the migration guide. Is this something that is being updated?
3 participants
Added links to TFT samples. @elmer-garduno @zoyahav