Course 5 (Sequence Models), Week 1 Assignment 3

Hello all, I have a question about the model overview: what is the one-hot vectors X representing?

The description says that X has shape (m, Tx, 90), where:

  • m is number of training examples
  • Tx is a snippet of 30 musical values
  • 90 dim is a one-hot vector

It’s not clear to me what this one-hot vector is representing. Before reading this description, I thought in this assignment each X input would be an audio snippet, so the system would learn how to generate the following snippet Y=X<t+1>.


1 Like

Which course are you asking about?

You have posted in the “AI Questions” area, which is a general topic. If you have a question on a specific course, it is better if you post in the forum for that course.

In the “Course Q&A” topic, each course has a dedicated forum. You pick your course, and add a tag for the week number.

You can move your thread there by using the “pencil” icon in the thread title.

1 Like

Thanks for the instructions, I just moved it to the course area.

1 Like

Hello @Frederico_Severgnini,

It was a representation of the tone at each time step. Section 1.1 is explaining that idea, and under which you can also find a variable called indices_values that defines what each one of those 90 musical values are.

We might expect an audio is just a 1D time series, but we could consider the assignment is using a different representation that details out exactly what muscial value is used at each time step.

At the end of the notebook there is a generate_music function. If you check out its code in, it should somehow combine those 90 values to make an audio clip.



Hello @rmwkwok,
Thank you for your reply. Does this means that there are 90 values in total, but at each training example we only work with a sequence of 30 of them (indices X<1> to X ^Tx )?

1 Like

There are 30 time steps, and each time step has 90 one-hot components, so for each sample, there are 30×90 values.





No problem!

1 Like