TF 2.16 Incorrect model with TextVectorization in SavedModel format contains float input type instead of string type

[rkazants]

Issue type

Bug

Have you reproduced the bug with TensorFlow Nightly?

Yes

Source

source

TensorFlow version

2.16.1

Custom code

Yes

OS platform and distribution

No response

Mobile device

No response

Python version

No response

Bazel version

No response

GCC/compiler version

No response

CUDA/cuDNN version

No response

GPU model and memory

No response

Current behavior?

Model with TextVectorization layer after saving into SavedModel format using tf.saved_model.save contains float32 input instead of string type

Standalone code to reproduce the issue

import numpy as np
import tensorflow as tf
from tensorflow.keras.layers import Input, TextVectorization
from tensorflow.keras.models import Model

vocabulary = ['hello', 'there', 'random', 'vocab', 'check']
length = 64

text_input = Input(dtype=tf.string, shape=(1,), name='text_input')
vectorize_layer = TextVectorization(vocabulary=vocabulary, output_mode='int', name='vectorization_layer',
                                    output_sequence_length=length)(text_input)
model = Model(inputs={'text_input': text_input}, outputs=[vectorize_layer])

input_data = np.array([['hello here and there']], dtype=str)
model({'text_input': input_data})

tf.saved_model.save(model, 'my_model')
loaded_model = tf.keras.layers.TFSMLayer('my_model', call_endpoint="serving_default")
loaded_model(input_data)

Relevant log output

ValueError: could not convert string to float: 'hello here and there'

[tilakrayal]

@rkazants,
Thank you for reporting the issue. Looks like this issue is more related to Keras3.0. Kindly find the gist of it here. Could you please create a new issue on keras-team/keras for the quick resolution. Thank you!

[rkazants]

@rkazants, Thank you for reporting the issue. Looks like this issue is more related to Keras3.0. Kindly find the gist of it here. Could you please create a new issue on keras-team/keras for the quick resolution. Thank you!

Filled here: keras-team/keras#19396

[tilakrayal]

@rkazants,
Could you please feel free to move this issue to closed status, since it is already being tracked there. Thank you!

[google-ml-butler]

Are you satisfied with the resolution of your issue?
Yes
No

[google-ml-butler]

Are you satisfied with the resolution of your issue?
Yes
No