Answer from cs61c-el (David Tung 16479725) for Question 1
TLB miss rates are much smaller because, first of all, there are just much fewer mappings from a virtual address to a physical address than everything that could be in memory.  Since there are fewer things that we can be looking for, the miss rate decreases.  Second of all, TLB caches are fully associative, decreasing the miss rate.