Skip to content

error when the model has tf.cond #6

@flyree

Description

@flyree

tf.cond is constructed with switch and merge operations.

There are two issues:

  1. The current code do not support those two operations (I implement those operations on my own).
  2. The fault injected graph seems not correct when there is tf.cond.

This link explains how tf.cond works.

https://towardsdatascience.com/tensorflow-control-flow-tf-cond-903e020e722a

graph_fi
graph_no_fi

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