Error from Tuner component: ValueError: Missing data for input “image_xf_input”

[Aly_SEGNANE]

Hi there,

I’m new here and I don’t if it is the right place to post this but I’m going to throw it out in case anyone here can help. It’d be appreciated!!!

Well, Here is what I did:

I was learning about TFX pipeline and wanted to get my hands a little bit dirty with an image classification project using the mnist dataset from tfds.

• Everything worked pretty well until the Transform component. Find the code below:

• Then I wanted to fine-tuned my model which is a modified version of a ResNet. I build a ResidualBlock class stored in residual module file that I call in both Tuner and Trainer components. Here is the most relevant part of the code including:

the _input_fn():

and the model_builder functions:

But I got this error message when trying to run the Tuner component:

ValueError: Missing data for input “image_xf_input”. You passed a data dictionary with keys [‘image_xf’]. Expected the following keys: [‘image_xf_input’]

I’ve tried to change the _transformed_name() function to add _input but I didn’t work.

Looking forward to hearing from you!!!

[jasmine_dhantule]

Hi @Aly_SEGNANE, thanks for posting this issue.

Could you try adjusting the key in your input dictionary to match the expected one like this

data = {
    "image_xf_input": your_image_data_here
}

Let me know if that works, if it doesn’t please provide a reproducible colab to investigate the issue. Thanks!