DocumentationAPI Reference
Back to ConsoleLog In

My model fails to benchmark

We encourage you to join our Slack channel, where you can consult with Deci’s support team in real-time, and exchange ideas with industry experts.

Cause 1: You entered the wrong input dimensions when configuring the model in Deci’s Add Model wizard.

  • Solution 1: On the Lab page, hover over the failed model and click the Edit model button on its right side. After the Edit model wizard launches, enter the correct input dimensions. Note that upon failure notification, Deci recommends the correct model input dimensions to be entered.

Cause 2: Your model is bound to a specific batch size (static batch size).

  • Solution 1: Recompile the model with Dynamic Batch Size enabled.

Cause 3: You are using a TensorFlow2 model, and the saved_model.pb file is either missing or not correctly located within the zipped folder.

  • Solution 1: Check whether the tensorFlow2 model files have been zipped correctly, as shown below.
  • Zip file
    • saved_model.pb file
    • Variables folder
      • variables.index file
      • variables.data-00000-of-00001 file

Did this page help you?