#include <bits/stdc++.h>
#pragma GCC optimize ("O3")
using namespace std;
clock_t tbegin;
constexpr float TL = 2.9;
int n, k, best = 0;
vector<int> o, ro;
vector<vector<int>> adj;
inline vector<int> intersect(const vector<int>& a, const vector<int>& b) {
vector<int> ans;
for (int i = 0, j = 0; i < int(a.size()) && j < int(b.size());) {
if (a[i] == b[j]) {
ans.push_back(a[i]);
++i, ++j;
} else if (ro[a[i]] < ro[b[j]]) {
++i;
} else {
++j;
}
}
return ans;
}
vector<vector<int>> cur;
void solve(int step, int i = 0) {
best = max(best, step);
if (int(cur.back().size()) <= i ||
step + int(cur.back().size()) <= best ||
best == k ||
clock() - tbegin > TL*CLOCKS_PER_SEC) return;
cur.push_back(intersect(cur.back(), adj[cur.back()[i]]));
solve(step+1);
cur.pop_back();
solve(step, i+1);
}
int main() {
ios_base::sync_with_stdio(false);
cin.tie(NULL);
tbegin = clock();
cin >> n >> k;
o.resize(n);
ro.resize(n);
adj.resize(n);
for (int i = 0; i < n; ++i) {
int d; cin >> d;
for (int a; d--;) {
cin >> a;
adj[i].push_back(a);
}
}
iota(o.begin(), o.end(), 0);
sort(o.begin(), o.end(), [&](int i, int j){
return adj[i].size() > adj[j].size();
});
cur.push_back(o);
for (int i = 0; i < n; ++i) ro[o[i]] = i;
vector<vector<int>> aa(n);
for (int i = 0; i < n; ++i) {
for (int j : adj[i]) {
if (ro[i] < ro[j]) {
aa[i].push_back(j);
}
}
}
adj = aa;
for (auto& x : adj) {
sort(x.begin(), x.end(), [&](int i, int j){return ro[i]<ro[j];});
}
solve(0);
cout << best << "\n";
}
