Tensorflow 2.14 Keras 2¶
Download drutai 12 predictive models for tensorflow 2.14x Keras 2x.
import drutai
drutai.predict.download_data(
url='https://github.com/2003100127/drutai/releases/download/model/model.zip',
sv_fpn='../../data/drutai/model.zip',
) ____ _ _
| _ \ _ __ _ _| |_ __ _(_)
| | | | '__| | | | __/ _` | |
| |_| | | | |_| | || (_| | |
|____/|_| \__,_|\__\__,_|_|
02/04/2025 01:56:13 logger: downloading data...
02/04/2025 01:56:37 logger: downloaded!Tensorflow 2.19 Keras 3¶
In Keras 3, the approach to loading and calling models has changed significantly compared to Keras 2, especially when used within the TensorFlow ecosystem. Many legacy TensorFlow models that relied on tf.keras.Model subclassing or SavedModel formats may now require structural modifications, explicit serialization logic, or migration to the new .keras format to ensure compatibility. However, when we attempted to make this transition, the process did not go as smoothly as expected. Please check the installation guide.
Error reports
ValueError: File format not supported: filepath=self.model_fp. Keras 3 only supports V3 .keras files and legacy H5 format files (.h5 extension). Note that the legacy SavedModel format is not supported by load_model() in Keras 3. In order to reload a TensorFlow SavedModel as an inference-only layer in Keras 3, use keras.layers.TFSMLayer(self.model_fp, call_endpoint='serving_default') (note that your call_endpoint might have a different name).
Errors continue to reared its head or different ones popped out even though we corrected errors.
Error reports
ValueError: Layer count mismatch when loading weights from file. Model expected 0 layers, found 5 saved layers.
We generated a LSTMCNN model for use in Keras 3. You can access it as shown below. Due to a complicated process of conversion, we did not apply it to all models.
import drutai
drutai.predict.download_data(
url='https://github.com/2003100127/drutai/releases/download/model-k3/lstmcnn_k3x.h5',
sv_fpn='../../data/lstmcnn_k3x.h5',
) ____ _ _
| _ \ _ __ _ _| |_ __ _(_)
| | | | '__| | | | __/ _` | |
| |_| | | | |_| | || (_| | |
|____/|_| \__,_|\__\__,_|_|
02/04/2025 02:00:49 logger: downloading data...
02/04/2025 02:00:50 logger: downloaded!