Deploy AI in google cloud

[sani20]

I want to deploy an AI model in google cloud but I am getting the following error:

Error loading Day 3 model: File format not supported: filepath=/tmp/tmpmt2bwnrg. 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(/tmp/tmpmt2bwnrg, call_endpoint='serving_default')` (note that your `call_endpoint` might have a different name).

[mokit]

Hi, @sani20.

Could you please let me know which file format you're using to save the model? Based on the logs, it appears you're using Keras 3 and the model format is not compatible with this version. Specifically, Keras 3 only supports the following formats:

• .keras files (the newer Keras format)
• Legacy .h5 format (from older versions of Keras)

Regards,
Mokit

[sani20]

Hi Mokit,
Thank you for your response. I saved the model with the following format and uploaded it in bucke:

model_day2.save(model_day2_path, save_format="keras")

Regards,

Sanaz

[mokit]

Instead of this:

model_day2.save(model_day2_path, save_format="keras")

Can you're saving the model as a .keras file:

model_day2.save(model_day2_path + '.keras')

When loading the model, you can follow the format below:

from tensorflow.keras.models import load_model

model = load_model(model_day2_path + '.keras')

Regards,
Mokit