Sequence Models (C5) W4 A(1) Ex4 - dimension of x in EncoderLayer.call

The comment in the “call” function of class EncoderLayer says:
x – Tensor of shape (batch_size, input_seq_len, fully_connected_dim)

Shouldn’t the last dimension be “embedding_dim” considering that the input to the encorder layer are word embedding vectors?

You are right.
Thanks for bringing this up. I’ve reported this for the staff to fix.
x is of shape (batch_size, sequence_length, embedding_dim)

1 Like