#include <bits/stdc++.h>
using namespace std;

template <class F> class y_combinator {
    F f;

   public:
    y_combinator(F&& f) : f(std::forward<F>(f)) {}
    template <class... Args> auto operator()(Args&&... args) const { return f(*this, std::forward<Args>(args)...); }
};

constexpr int dx[8] = {1, 0, -1, 0, 1, 1, -1, -1};
constexpr int dy[8] = {0, 1, 0, -1, 1, -1, 1, -1};

using ll = long long;
using u32 = unsigned int;
using u64 = unsigned long long;

using vi = vector<int>;
using vl = vector<ll>;
using pii = pair<int, int>;
using pll = pair<ll, ll>;
template <class T> using vc = vector<T>;
template <class T> using vvc = vector<vc<T>>;
template <class T, class U = std::less<T>> using prique = priority_queue<T, vector<T>, U>;

#define overload(a, b, c, d, e, ...) e
#define len(x) (ll)(x.size())
#define all(x) (x).begin(), (x).end()
#define rall(x) (x).rbegin(), (x).rend()
#define rep1(n) for (ll _ = 0; _ < ll(n); _++)
#define rep2(i, n) for (ll i = 0; i < ll(n); i++)
#define rep3(i, a, b) for (ll i = ll(a); i < ll(b); i++)
#define rep4(i, a, b, c) for (ll i = ll(a); i < ll(b); i += ll(c))
#define rep(...) overload(__VA_ARGS__, rep4, rep3, rep2, rep1)(__VA_ARGS__)
#define rrep(i, n) for (ll i = ll(n) - 1; i >= 0; i--)

template <class T = ll, class U> vector<T> cumsum(const vector<U>& a) { vector<T> res(a.size() + 1); rep(i, a.size()) res[i + 1] = res[i] + a[i]; return res; }
template <class T> void dedup(vector<T>& a) { sort(all(a)), a.erase(unique(all(a)), a.end()); }
template <class T, class S> bool chmax(T& a, const S& b) { return a < static_cast<T>(b) ? a = static_cast<T>(b), 1 : 0; }
template <class T, class S> bool chmin(T& a, const S& b) { return a > static_cast<T>(b) ? a = static_cast<T>(b), 1 : 0; }

namespace cppio {

template <typename T> struct is_tuple_like : false_type {};
template <typename... Ts> struct is_tuple_like<tuple<Ts...>> : true_type {};
template <typename T, typename U> struct is_tuple_like<pair<T, U>> : true_type {};
template <typename T, typename = void> struct is_container : false_type {};
template <typename T> struct is_container<T, void_t<decltype(begin(declval<T&>()))>> : bool_constant<!is_same_v<decay_t<T>, string> && !is_same_v<decay_t<T>, char*> && !is_same_v<decay_t<T>, const char*> && !is_array_v<remove_reference_t<T>>> {};
template <typename T> void _in(T& x) { if constexpr (is_tuple_like<T>::value) apply([](auto&... elems) { (_in(elems), ...); }, x); else if constexpr (is_container<T>::value) { for (auto& e : x) _in(e); } else cin >> x; }
template <typename... Ts> void _in_all(Ts&... args) { (_in(args), ...); }
template <typename T> void _out(const T& x) { bool first = true; if constexpr (is_tuple_like<T>::value) { apply([&](const auto&... elems) { ((cout << (first ? "" : " "), _out(elems), first = false), ...); }, x); } else if constexpr (is_container<T>::value) { for (const auto& e : x) { if (!first) cout << ' '; _out(e), first = false; } } else cout << x; }
template <typename... Ts> void _print(const Ts&... args) { bool first = true; ((cout << (first ? "" : " "), _out(args), first = false), ...); }
template <typename... Ts> void _out_no_el(const Ts&... args) { _print(args...); }
template <typename... Ts> void _out_all(const Ts&... args) { _out_no_el(args...), cout << '\n'; }

}  // namespace cppio

#define IN(...) cppio::_in_all(__VA_ARGS__)
#define OUT(...) cppio::_out_all(__VA_ARGS__)

#define INT(...) int __VA_ARGS__; IN(__VA_ARGS__)
#define LL(...)  ll __VA_ARGS__; IN(__VA_ARGS__)
#define STR(...) string __VA_ARGS__; IN(__VA_ARGS__)
#define CHR(...) char __VA_ARGS__; IN(__VA_ARGS__)
#define DBL(...) double __VA_ARGS__; IN(__VA_ARGS__)
#define VEC(type, name, size) vector<type> name(size); IN(name)
#define VV(type, name, h, w)  vector<vector<type>> name(h, vector<type>(w)); IN(name)
#define RETURN(...) do { __VA_ARGS__; return; } while (0)

bool Yes(bool b = true) { return OUT(b ? "Yes" : "No"), b; }
bool No(bool b = true) { return Yes(!b), b; }
bool YES(bool b = true) { return OUT(b ? "YES" : "NO"), b; }
bool NO(bool b = true) { return YES(!b), b; }

#ifdef ONLINE_JUDGE
#define debug(...) (void(0))
#endif

template <class T> T lagrange_eval(const vector<T>& xs, const vector<T>& ys, const T& X) {
    assert(xs.size() == ys.size());
    int n = xs.size();
    for (int i = 0; i < n; i++) {
        if (xs[i] == X) return ys[i];
    }
    T ans = 0, coef = 1;
    for (int i = 0; i < n; i++) coef *= X - xs[i];
    for (int i = 0; i < n; i++) {
        T denom = X - xs[i];
        for (int j = 0; j < n; j++) {
            if (i != j) denom *= xs[i] - xs[j];
        }
        ans += ys[i] / denom;
    }
    return ans * coef;
}
template <class Mint> Mint lagrange_eval_ap(const Mint& a, const Mint& d, const vector<Mint>& ys, const Mint& X) {
    int n = ys.size();
    for (int i = 0; i < n; i++) {
        if (X == a + d * i) return ys[i];
    }
    Mint ans = 0, pref = 1;
    vector<Mint> ifac(n, 1), suff(n + 1, 1);
    for (int i = 1; i < n; i++) ifac[n - 1] *= i;
    ifac[n - 1] = 1 / ifac[n - 1];
    for (int i = n - 1; i >= 0; i--) {
        if (i) ifac[i - 1] = ifac[i] * i;
        suff[i] = suff[i + 1] * (X - a - d * i);
    }
    for (int i = 0; i < n; i++) {
        Mint curr = pref * suff[i + 1] * ifac[i] * ifac[n - 1 - i] * ys[i];
        if ((n - 1 - i) & 1) curr = -curr;
        ans += curr;
        pref *= X - a - d * i;
    }
    ans *= d.pow(n - 1).inv();
    return ans;
}

template <class Mint> Mint lagrange_eval_ap(const vector<Mint>& ys, const Mint& X) {
    return lagrange_eval_ap<Mint>(0, 1, ys, X);
}

#include <atcoder/modint>
using mint = atcoder::modint1000000007;
void run_case() {
    LL(N, K);
    vector<mint> ys(K + 2);
    mint ans = 0;
    rep(i, K + 2) {
        ans += mint::raw(i).pow(K);
        ys[i] = ans;
    }
    OUT(lagrange_eval_ap<mint>(ys, N).val());
}
int main() {
    std::ios_base::sync_with_stdio(false);
    std::cin.tie(nullptr);
    std::fixed(std::cout).precision(16);
    int t = 1;
    while (t--) run_case();
}