TF-Lite converter crashes when quantizing GRU layer

[pkgoogle]

Original Issue: tensorflow/tensorflow#76672
Original Author: @bas-aarts

On Apple M1 Pro with MacOs Sonoma 14.1.2 (23B92)
TensorFlow version 2.16.1 (still an issue with TF 2.17.0)

sources + assets:
Gru_quantization_bug.zip

The zip contain a TF saved model and a convert.py script that converts it to a TF-Lite mode, but it crashes

removing either line 11 or 12 from the script makes the compilation pass

[pkgoogle]

Hi @bas-aarts

You can convert a GRU model with https://github.com/google-ai-edge/ai-edge-torch. Does that work for you?

[NidaaElsayed]

Hello, is there a way to convert a Keras LSTM model to a quantized Unidirectional Sequence LSTM operator without crashing ?