Variable input length for LSTM/GRU?

Hi, I’m curious if there is a way to have a variable length feature vector as input into a GRU/LSTM module? I did some light searching and can’t find much discussing input vectors, rather just variable length sequences. Currently, I’m working on a time-series problem where I have data chunked into weeks. So for my case, if I have a sequence length of n, I want sequence elements 1…n-1 to have X features and sequence element n to have X+1 features. Judging by how the GRU model takes in a specific feature input size on creation I am doubtful there is a way to do this directly, however I figured I would confirm my doubts.