I am trying to train Attention based RNN model for video summarization. When I run the code on CPU the training loop completes. But if I run the same code on GPU it is giving the following error.
InvalidArgumentError: Unable to broadcast tensor of shape [512] to tensor of shape [32,512,1,1] [[node decoder_9/attention_model_images_9/dense_77/BiasAdd (defined at tmp/ipykernel_34/4173017652.py:13) ]] [Op:__inference_train_step_360921]
Function call stack: train_step
The tensorflow version is same in both cases(2.6.2).
I can share the specific part of the code if required.
Thanks in Advance.