ValueError when evaluating tuning model using Mtf library

[akku779]

I'm trying to evaluate my model using the Mtf model library. When I run the evaluation function, my model outputs this error:

ValueError: Found input variables with inconsistent numbers of samples: [0, 104]

This is the function I am running:
model.batch_size = train_batch_size * 4 model.eval( mixture_or_task_name="super_glue_wsc_v102_simple_eval", #checkpoint_steps="all" )

My model has already been fine-tuned and the evaluation function does write a file which contains the output predictions in my GCS bucket. I need the model to output the accuracy score but it keeps throwing this error. Any ideas of what is going on?