CandleStack
← All problem types

Model check failed

The uploaded file did not pass a check of the model, so it was not kept.

Details

  • Status422 Unprocessable Content
  • Typehttps://candlestack.tech/problems/model-check-failed

When it happens

  • The opens check failed. The file is not a real Keras model, it is damaged, it is empty, or it holds custom code such as a Lambda or a custom layer, which CandleStack never runs.
  • The weights check failed. The weights of the model cannot be read, or the model holds no trained weights.
  • The check took too long, stopped before it finished, or too many models were being checked at once.

What to do

  1. 1Read detail, which says what is wrong with the file.
  2. 2Save the model again with model.save("model.keras") and upload that file.
  3. 3Replace custom layers and Lambda layers with built-in Keras layers.
  4. 4If detail says many models are being checked, upload the file again in a few minutes.

In the response

  • checkThe check that failed, opens or weights.