D$)[3$ pts$]$ You are building a BPE vocabulary V$1$ based on some corpus C$.$ Assume that

the learning phase stops only when there are no possible merges left $($not bounded

by the number of merges k $).$

Now, if you did it again using a lowercased C to get new BPE vocabulary V$2,$ how would

V$1$ and V$2$ differ $($if at all$)?$ Explain.