Skip to content

[BUG] Can not apply on conv1D layer #330

Description

@niki-lei

Describe the bug

I try to apply the newest version of innvestigate on 1-dimensional data on tensorflow 2.15. The same model structure with conv2D can work perfectly, while I change the model to 1D and feed in 1D data, it shows "Placeholder error":

'''
InvalidArgumentError: You must feed a value for placeholder tensor 'input_layer' with dtype float and shape [?,28,1] [[{{node input_layer}}]]
'''

Steps to reproduce the bug

This is the jupyter notebook script to replicate the bug.

Platform information

  • OS: [Windows 11]
  • Python version: [3.11]
  • iNNvestigate version: [ 2.0.1]
  • TensorFlow version: [2.15]

Metadata

Metadata

Assignees

No one assigned

    Labels

    triageBug report that needs assessment

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions