class Graph():  #non-directed
    def __init__(self, n, edge, indexed=1):
        self.n = n
        self.edge = edge
        self.indexed = indexed
        self.graph = [[] for _ in range(n)]
        for e in edge:
            self.graph[e[0] - indexed].append((e[1] - indexed, e[2]))
            self.graph[e[1] - indexed].append((e[0] - indexed, e[2]))

    def kruskal(self):  #UnionFind
        edgemap = {self.edge[i]: i for i in range(len(self.edge))}
        sortedge = sorted(self.edge, key=lambda x: x[2])
        uf = UnionFind(self.n)
        res = 0
        restored = []
        for e in sortedge:
            if uf.find(e[0] - self.indexed) != uf.find(e[1] - self.indexed):
                res += e[2]
                restored.append(edgemap[e])
                uf.unite(e[0] - self.indexed, e[1] - self.indexed)
        return res, restored

class UnionFind():
    def __init__(self, n):
        self.n = n
        self.parents = [i for i in range(n)]
        self.rank = [0 for _ in range(n)]
        self.size = [1 for _ in range(n)]

    def find(self, x):
        root = x
        while self.parents[root] != root:
            root = self.parents[root]
        while self.parents[x] != root:
            parent = self.parents[x]
            self.parents[x] = root
            x = parent
        return root

    def unite(self, x, y):
        xroot = self.find(x)
        yroot = self.find(y)
        if xroot == yroot: return
        xrank = self.rank[xroot]
        yrank = self.rank[yroot]
        if xrank < yrank:
            self.parents[xroot] = yroot
            self.size[yroot] += self.size[xroot]
        elif xrank == yrank:
            self.parents[yroot] = xroot
            self.rank[yroot] += 1
            self.size[xroot] += self.size[yroot]
        else:
            self.parents[yroot] = xroot
            self.size[xroot] += self.size[yroot]

L, R = map(int, input().split())

N = R - L + 1
E = []

for i in range(N - 1):
    E.append((i, i + 1, 1))

for i in range(L, R + 1):
    for j in range(2 * i, R + 1, i):
        E.append((i - L, j - L, 0))

g = Graph(N, E, 0)
print(g.kruskal()[0])