mse, mae = compute_metrics(series_valid, rnn_forecast)
is throwing off an error, specifically:
InvalidArgumentError: Incompatible shapes: [1150,64] vs. [1150] [Op:SquaredDifference]
Please print the model summary and notice the shape of the final layer. Shapes of forecast and true_series must match.
2 Likes
still having problems, can i send you my notebook so that you can identify where i am going wrong?
Please click my name and message your notebook as an attachment.
1 Like
Here’s a hint:
The output shape of the last layer should be (None, 1).
Yours is (None, None, 1)
Please think about where the correction(s) should be made.
2 Likes
return_sequences, got it
I’m also using return_sequences = True for the first LSTM layer but for whatever reason I still have an extra dimension. What else could be causing this incorrect shape?
Nevermind, I got it to work. I was loading an earlier version of my DNN
mine is (none,none,1) but it is still showing same error please reply
Please click my name and message your notebook as an attachment.
i’ve did the same thing but got the same error:
There was a problem grading your submission. Details:
Incompatible shapes: [1150,64] vs. [1150] [Op:SquaredDifference]
Please pay attention to model output shape and fix the NN.
1 Like