Question about 'sequence-to-one' regression using LSTM layers

I have constructed a 'sequence-to-one' LSTM network, the structure is shown in the first figure .

I want to add the last time step of the input sequence to the output of the LSTM layer('outputMode','Last'), but the data format of them are different. LSTM has feature format 'CB', while the input_tailor is sequence 'CBT', shown in the second figure:

How to transfer the sequence data format to feature or feature to sequence?

Or: Have other solutions?