MSF tensorflow config_error ai_generated true

ValueError: Unknown model format: 'h5'. Supported formats: 'tf', 'keras'

ID: tensorflow/unknown-model-save-format

Also available as: JSON · Markdown · 中文
90%Fix Rate
88%Confidence
1Evidence
2023-11-05First Seen

Version Compatibility

VersionStatusIntroducedDeprecatedNotes
tensorflow 2.13.0 active — — —
tensorflow 2.14.0 active — — —

Root Cause

Saving a model with .h5 extension is not supported in TF2's default save format; only SavedModel or Keras native format is allowed.

generic

中文

在TF2默认保存格式中不支持使用.h5扩展名保存模型；只允许SavedModel或Keras原生格式。

Official Documentation

https://www.tensorflow.org/guide/keras/save_and_serialize

Workarounds

  1. 95% success Use model.save('my_model.keras') with the native Keras format. 
    Use model.save('my_model.keras') with the native Keras format.
  2. 90% success Use model.save('my_model', save_format='tf') for SavedModel format. 
    Use model.save('my_model', save_format='tf') for SavedModel format.
  3. 85% success Explicitly set save_format='h5' in model.save() to force HDF5: model.save('my_model.h5', save_format='h5') 
    Explicitly set save_format='h5' in model.save() to force HDF5: model.save('my_model.h5', save_format='h5')

中文步骤

  1. Use model.save('my_model.keras') with the native Keras format.
  2. Use model.save('my_model', save_format='tf') for SavedModel format.
  3. Explicitly set save_format='h5' in model.save() to force HDF5: model.save('my_model.h5', save_format='h5')

Dead Ends

Common approaches that don't work:

  1. 60% fail

    Renaming file extension to .tf doesn't change underlying format.

  2. 90% fail

    Installing h5py doesn't add .h5 support to TF2 save API.