Skip to content

TensorFlow savedmodel input dimension does not match #9

Open
@ZhanqiuHu

Description

@ZhanqiuHu

When I used the saved TensorFlow query tower model created using offline-batch-recsys/Offline-Batch-Recommender-System.ipynb for Triton inference server, I got an error saying

1-user-embeddings      UNAVAILABLE: Invalid argument: model '1-user-embeddings', tensor 'user_age': the model expects 1 dimensions ( shape [-1]) but the model configuration specifies 2 dimensions (shape [-1,1])

I'm wondering if there is a way to create the query tower model so that the input shape becomes (shape [-1,1]) instead of shape [-1].

Thanks!

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions