Shape of minibatch - Week 1 DLS5

Course Q&A Deep Learning Specialization Sequence Models
1

If there are 20 mini batches, And, there are 10 time steps with each time step consisting of 5000 (one-hot encoded language) units.

I was expecting the dimension to be (20, 10, 5000) i.e, 20 batches of 10 time step (row) with 5000 columns. So, I was expecting the shape of mini batch to be (m, Tx, nx)

But exercise states it to be (nx, m, Tx) – (5000, 20, 10)

I am not able to understand – Please help

2

Your understanding of how tensorflow wants the data to be shaped is correct.

The assignment does things manually. Since the markdown clearly states the data shape, please follow the assignment convention.