Problem in mnist.tflite file about tensorflow

[gaikwadrahul8]

Hi, @DanielVacha-dv

Thank you for bringing this issue to our attention, I believe you're following the exact steps mentioned in this codelab tutorial in that it's using the older version of TensorFlow so could you please try with latest version by adding below line, if you want to use GPU then please add this line implementation 'org.tensorflow:tensorflow-lite-gpu:+' or else ignore it

dependencies {
    implementation 'org.tensorflow:tensorflow-lite:+'
    implementation 'org.tensorflow:tensorflow-lite-gpu:+'      // Import the GPU delegate plugin Library for GPU inference
}

We do have TensorFlow Lite Digit Classification Demo Application if you want to give it try

Please let us know after trying with latest version is it resolving your issue or not ? if issue still persists please let us know with error log and steps which you followed before encountering the error message that will help us to investigate your issue further.

Thank you for your cooperation and patience.

from tensorflow.

[DanielVacha-dv]

Hello gaikwadrahul8
Thank you for your reply. You're right, I tried to follow the instructions here https://developer.android.com/codelabs/digit-classifier-tflite?hl=en#0
and also in https://colab.research.google.com/github/tensorflow/examples/blob/master/lite/codelabs/digit_classifier/ml/step2_train_ml_model.ipynb#scrollTo=C4ASalaLIbu2 as accurately as I could.
After cloning the project from the Git repository, I immediately ran the variant in the finish directory and added the mnist.tflite file to the desired location. I had problems during the build, so I changed the parameters in the build.gradle file.
The next day after sending the email, I couldn't be satisfied with the failure and that's why I started, at least superficially, into the issue of creating models and asked Gemini as well.
Gemini advised me to edit the part with the model settings
"converter.target_spec.supported_ops = [
tf.lite.OpsSet.TFLITE_BUILTINS, # enable LiteRT ops.
tf.lite.OpsSet.SELECT_TF_OPS # enable TensorFlow ops.
]" but it still didn't work. Finally, out of desperation, I tried not to save the model with tflite_quantized_model , f.write(tflite_quantized_model) but with this tflite_float_model. Only then I was able to run the application without crashing.
Following your advice, I cloned the https://github.com/tensorflow/examples.git repository again today to a new location and tried the recommended steps, but that also resulted in a translation error. I don't know if it's helpful, but I'm attaching the build.gradle file that allowed me to run the application and also the model (mnist.tflite) that contains tflite_float_model.
Thank you for your cooperation
Regards, Daniel
build.gradle.txt
mnist.tflite.txt