sushreebarsa

@selvakarna `AttributeError: type object 'TFLiteConverterV2' has no attribute` '`from_frozen_graph`' error because the from_frozen_graph method is not available in TensorFlow 2.x's TFLiteConverterV2 class. If your model is a frozen graph saved...

@selvakarna Could you use tf.saved_model.load to load the frozen graph as a SavedModel and then convert it to TFLite? Please let us know the outcome? Thank you!

@NOORLEICESTER `numpy.core._exceptions._ArrayMemoryError` on the HPC system but not on your laptop indicates a limitation in the available memory on the HPC cluster compared to your laptop. HPC environments often prioritize...

@NOORLEICESTER In order to expedite the trouble-shooting process, please provide a code snippet to reproduce the issue reported here. Thank you!

@pzy2000 If the lambda function or dropout layer is causing the NaNs, can we consider removing them and potentially adding alternative functionalities elsewhere in the network ? The pre-processing data...

@pzy2000 Could you please let us know if you have tried it in the latest TF? I tried it but faced a different issue, so could you please share all...

@pzy2000 The error occurs when the keras.models.load_model function attempts to deserialize a custom Lambda layer configuration saved in the model. It might be occurring for keras model loading. Please refer...

@waarmond Kindly double-check the target name you're using. The error message suggests "build_pip_package.py" might exist. Could you ensure you're using the correct target name with a colon (":") separating the...

@waarmond Thank you for your response! Could you please confirm if we can close this issue as the build from the source guide looks good to you. So that we...

@rasitsimsek One workaround could be to wrap the existing EditDistance class with a custom constructor that takes the desired parameters and sets the options internally. Please upgrade to the latest...