#pragma region Macros
#pragma GCC optimize("O3")
#include <bits/stdc++.h>
#define ll long long
#define ld long double
#define rep2(i, a, b) for(ll i = a; i <= b; ++i)
#define rep(i, n) for(ll i = 0; i < n; ++i)
#define rep3(i, a, b) for(ll i = a; i >= b; --i)
#define pii pair<int, int>
#define pll pair<ll, ll>
#define pb push_back
#define eb emplace_back
#define vi vector<int>
#define vll vector<ll>
#define vpi vector<pii>
#define vpll vector<pll>
#define overload2(_1, _2, name, ...) name
#define vec(type, name, ...) vector<type> name(__VA_ARGS__)
#define VEC(type, name, size)                                                                                                                                  \
    vector<type> name(size);                                                                                                                                   \
    IN(name)
#define vv(type, name, h, ...) vector<vector<type>> name(h, vector<type>(__VA_ARGS__))
#define VV(type, name, h, w)                                                                                                                                   \
    vector<vector<type>> name(h, vector<type>(w));                                                                                                             \
    IN(name)
#define vvv(type, name, h, w, ...) vector<vector<vector<type>>> name(h, vector<vector<type>>(w, vector<type>(__VA_ARGS__)))
#define fi first
#define se second
#define all(c) begin(c), end(c)
#define ios ios_base::sync_with_stdio(0), cin.tie(0), cout.tie(0);
#define lb(c, x) distance((c).begin(), lower_bound(all(c), (x)))
#define ub(c, x) distance((c).begin(), upper_bound(all(c), (x)))
using namespace std;
template <class T> using pq = priority_queue<T>;
template <class T> using pqg = priority_queue<T, vector<T>, greater<T>>;
#define Size(c) (int)(c).size()
#define INT(...)                                                                                                                                               \
    int __VA_ARGS__;                                                                                                                                           \
    IN(__VA_ARGS__)
#define LL(...)                                                                                                                                                \
    ll __VA_ARGS__;                                                                                                                                            \
    IN(__VA_ARGS__)
#define ULL(...)                                                                                                                                               \
    ull __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 LD(...)                                                                                                                                                \
    ld __VA_ARGS__;                                                                                                                                            \
    IN(__VA_ARGS__)
int scan() { return getchar(); }
template <class T> void scan(T a) { cin >> a; }
void scan(int &a) { cin >> a; }
void scan(long long &a) { cin >> a; }
void scan(char &a) { cin >> a; }
void scan(double &a) { cin >> a; }
void scan(long double &a) { cin >> a; }
void scan(char a[]) { scanf("%s", a); }
void scan(string &a) { cin >> a; }
template <class T> void scan(vector<T> &);
template <class T, size_t size> void scan(array<T, size> &);
template <class T, class L> void scan(pair<T, L> &);
template <class T, size_t size> void scan(T (&)[size]);
template <class T> void scan(vector<T> &a) {
    for(auto &i : a) scan(i);
}
template <class T> void scan(deque<T> &a) {
    for(auto &i : a) scan(i);
}
template <class T, size_t size> void scan(array<T, size> &a) {
    for(auto &i : a) scan(i);
}
template <class T, class L> void scan(pair<T, L> &p) {
    scan(p.first);
    scan(p.second);
}
template <class T, size_t size> void scan(T (&a)[size]) {
    for(auto &i : a) scan(i);
}
template <class T> void scan(T &a) { cin >> a; }
void IN() {}
template <class Head, class... Tail> void IN(Head &head, Tail &... tail) {
    scan(head);
    IN(tail...);
}
string stin() {
    string s;
    cin >> s;
    return s;
}
template <class T, class S> inline bool chmax(T &a, S b) {
    if(a < b) {
        a = b;
        return 1;
    }
    return 0;
}
template <class T, class S> inline bool chmin(T &a, S b) {
    if(a > b) {
        a = b;
        return 1;
    }
    return 0;
}
vi iota(int n) {
    vi a(n);
    iota(all(a), 0);
    return a;
}
template <class T> void UNIQUE(vector<T> &x) {
    sort(all(x));
    x.erase(unique(all(x)), x.end());
}
int in() {
    int x;
    cin >> x;
    return x;
}
ll lin() {
    unsigned long long x;
    cin >> x;
    return x;
}
void print() { putchar(' '); }
void print(bool a) { cout << a; }
void print(int a) { cout << a; }
void print(long long a) { cout << a; }
void print(char a) { cout << a; }
void print(string &a) { cout << a; }
void print(double a) { cout << a; }
template <class T> void print(const vector<T> &);
template <class T, size_t size> void print(const array<T, size> &);
template <class T, class L> void print(const pair<T, L> &p);
template <class T, size_t size> void print(const T (&)[size]);
template <class T> void print(const vector<T> &a) {
    if(a.empty()) return;
    print(a[0]);
    for(auto i = a.begin(); ++i != a.end();) {
        cout << " ";
        print(*i);
    }
    cout << endl;
}
template <class T> void print(const deque<T> &a) {
    if(a.empty()) return;
    print(a[0]);
    for(auto i = a.begin(); ++i != a.end();) {
        cout << " ";
        print(*i);
    }
}
template <class T, size_t size> void print(const array<T, size> &a) {
    print(a[0]);
    for(auto i = a.begin(); ++i != a.end();) {
        cout << " ";
        print(*i);
    }
}
template <class T, class L> void print(const pair<T, L> &p) {
    cout << '(';
    print(p.first);
    cout << ",";
    print(p.second);
    cout << ')';
}
template <class T> void print(set<T> &x) {
    for(auto e : x) print(e), cout << " ";
    cout << endl;
}
template <class T> void print(multiset<T> &x) {
    for(auto e : x) print(e), cout << " ";
    cout << endl;
}
template <class T, size_t size> void print(const T (&a)[size]) {
    print(a[0]);
    for(auto i = a; ++i != end(a);) {
        cout << " ";
        print(*i);
    }
}
template <class T> void print(const T &a) { cout << a; }
int out() {
    putchar('\n');
    return 0;
}
template <class T> int out(const T &t) {
    print(t);
    putchar('\n');
    return 0;
}
template <class Head, class... Tail> int out(const Head &head, const Tail &... tail) {
    print(head);
    putchar(' ');
    out(tail...);
    return 0;
}
ll gcd(ll a, ll b) {
    while(b) {
        ll c = b;
        b = a % b;
        a = c;
    }
    return a;
}
ll lcm(ll a, ll b) {
    if(!a || !b) return 0;
    return a * b / gcd(a, b);
}
vector<pll> factor(ll x) {
    vector<pll> ans;
    for(ll i = 2; i * i <= x; i++)
        if(x % i == 0) {
            ans.push_back({i, 1});
            while((x /= i) % i == 0) ans.back().second++;
        }
    if(x != 1) ans.push_back({x, 1});
    return ans;
}
vector<int> divisor(int x) {
    vector<int> ans;
    for(int i = 1; i * i <= x; i++)
        if(x % i == 0) {
            ans.pb(i);
            if(i * i != x) ans.pb(x / i);
        }
    return ans;
}
template <typename T> void zip(vector<T> &x) {
    vector<T> y = x;
    sort(all(y));
    for(int i = 0; i < x.size(); ++i) { x[i] = lb(y, x[i]); }
}
int popcount(ll x) { return __builtin_popcountll(x); }
mt19937 rng(chrono::steady_clock::now().time_since_epoch().count());
int rnd(int n) { return uniform_int_distribution<int>(0, n - 1)(rng); }
template <typename T> void shuffle(vector<T> &v) {
    rep3(i, v.size() - 1, 1) { swap(v[i], v[rnd(i)]); }
}
#define endl '\n'
vector<string> YES{"NO", "YES"};
vector<string> Yes{"No", "Yes"};
vector<string> yes{"no", "yes"};

#ifdef _LOCAL
#undef endl
#define debug(x)                                                                                                                                               \
    cout << #x << ": ";                                                                                                                                        \
    print(x);                                                                                                                                                  \
    cout << endl;
void err() {}
template <class T> void err(const T &t) {
    print(t);
    cout << " ";
}
template <class Head, class... Tail> void err(const Head &head, const Tail &... tail) {
    print(head);
    putchar(' ');
    out(tail...);
}
#else
#define debug(x)
template <class... T> void err(const T &...) {}
#endif

template <typename T> struct edge {
    int from, to;
    T cost;
    int id;
    edge(int to, T cost) : from(-1), to(to), cost(cost) {}
    edge(int from, int to, T cost) : from(from), to(to), cost(cost) {}
    // edge(int from,int to,T cost,int id):from(from),to(to),cost(cost),id(id){}
    edge &operator=(const int &x) {
        to = x;
        return *this;
    }
    operator int() const { return to; }
};

template <typename T> using Edges = vector<edge<T>>;
template <typename T> using WeightedTree = vector<Edges<T>>;
template <typename T> using Edges = vector<edge<T>>;
template <typename T> using WeightedGraph = vector<Edges<T>>;
template <typename T> using Matrix = vector<vector<T>>;
using Graph = vector<vector<int>>;
using tree = vector<vector<int>>;

tree getTree(int n, int offset = 1) {
    tree res(n);
    for(int i = 0; i < n - 1; i++) {
        int a, b;
        cin >> a >> b;
        a -= offset, b -= offset;
        res[a].emplace_back(b);
        res[b].emplace_back(a);
    }
    return res;
}
template <typename T> WeightedTree<T> getWeightedTree(int n, int offset = 1) {
    WeightedTree<T> res(n);
    for(int i = 0; i < n - 1; i++) {
        int a, b;
        cin >> a >> b;
        a -= offset, b -= offset;
        T c;
        cin >> c;
        res[a].emplace_back(b, c);
        res[b].emplace_back(a, c);
    }
    return res;
}

Graph getGraph(int n, int m, int offset = 1) {
    Graph res(n);
    for(int i = 0; i < m; ++i) {
        int a, b;
        cin >> a >> b;
        a -= offset, b -= offset;
        res[a].emplace_back(b);
        res[b].emplace_back(a);
    }
    return res;
}

template <typename T> WeightedGraph<T> getWeightedGraph(int n, int m, int offset = 1) {
    WeightedGraph<T> res(n);
    for(int i = 0; i < m; ++i) {
        int a, b;
        cin >> a >> b;
        T c;
        cin >> c;
        res[a].emplace_back(b, c);
        res[b].emplace_back(a, c);
    }
}
struct Setup_io {
    Setup_io() {
        ios_base::sync_with_stdio(0), cin.tie(0), cout.tie(0);
        cout << fixed << setprecision(15);
    }
} setup_io;
#pragma endregion

// _________コーディングはここから！！___________

// namespace modular {
// constexpr ll MOD = 998244353;
// const int MAXN = 1100000;
// template <ll Modulus> class modint {
//     using u64 = ll;

//   public:
//     u64 a;

//     constexpr modint(const u64 x = 0) noexcept : a(((x % Modulus) + Modulus) % Modulus) {}
//     constexpr u64 &value() noexcept { return a; }
//     constexpr const u64 &value() const noexcept { return a; }
//     constexpr modint operator+(const modint rhs) const noexcept { return modint(*this) += rhs; }
//     constexpr modint operator-(const modint rhs) const noexcept { return modint(*this) -= rhs; }
//     constexpr modint operator*(const modint rhs) const noexcept { return modint(*this) *= rhs; }
//     template <typename T> constexpr modint operator^(T rhs) const noexcept { return modint(*this) ^= rhs; }
//     constexpr modint &operator+=(const modint rhs) noexcept {
//         a += rhs.a;
//         if(a >= Modulus) { a -= Modulus; }
//         return *this;
//     }
//     constexpr modint &operator-=(const modint rhs) noexcept {
//         if(a < rhs.a) { a += Modulus; }
//         a -= rhs.a;
//         return *this;
//     }
//     constexpr modint &operator*=(const modint rhs) noexcept {
//         a = a * rhs.a % Modulus;
//         return *this;
//     }
//     constexpr bool operator==(const modint rhs) const noexcept { return a == rhs.a; }
//     template <typename T> constexpr modint &operator^=(T n) noexcept {
//         modint<Modulus> res = 1;
//         modint<Modulus> x = a;
//         while(n) {
//             if(n & 1) res *= x;
//             x *= x;
//             n >>= 1;
//         }
//         a = res.a;
//         return *this;
//     }
// };
// #define mint modint<MOD>
// #define vmint vector<mint>
// vmint Inv{0, 1}, Prd{1, 1}, Invprd{1, 1};
// mint inv(int n) {
//     if(Inv.size() > n)
//         return Inv[n];
//     else {
//         for(int i = Inv.size(); i <= n; ++i) Inv.emplace_back(Inv[MOD % i] * (-MOD / i));
//         return Inv[n];
//     }
// }
// mint prd(int n) {
//     if(Prd.size() > n)
//         return Prd[n];
//     else
//         for(int i = Prd.size(); i <= n; ++i) Prd.emplace_back(Prd[i - 1] * i);
//     return Prd[n];
// }
// mint invprd(int n) {
//     if(Invprd.size() > n)
//         return Invprd[n];
//     else
//         for(int i = Invprd.size(); i <= n; ++i) Invprd.emplace_back(Invprd[i - 1] * inv(i));
//     return Invprd[n];
// }
// mint modpow(ll a, ll n) {
//     mint x = a;
//     return x ^= n;
// }
// template <ll T> modint<T> operator/(modint<T> l, modint<T> r) {
//     if(r.a < MAXN) return l * inv(r.a);
//     return l * (r ^ (MOD - 2));
// }
// template <typename T, ll S> modint<S> operator/(T l, modint<S> r) { return modint<S>(l) / r; }
// template <ll T> modint<T> operator/=(modint<T> &l, modint<T> r) { return l = l / r; }
// mint C(int a, int b) {
//     if(a < 0 || b < 0) return 0;
//     if(a < b) return 0;
//     return prd(a) * invprd(b) * invprd(a - b);
// }
// mint P(int a, int b) {
//     if(a < 0 || b < 0) return 0;
//     if(a < b) return 0;
//     return prd(a) * invprd(a - b);
// }
// ostream &operator<<(ostream &os, mint a) {
//     os << a.a;
//     return os;
// }
// template <typename T> ostream &operator<<(ostream &os, vector<T> &a) {
//     for(auto &e : a) os << e << " ";
//     os << endl;
//     return os;
// }
// istream &operator>>(istream &is, mint &a) {
//     ll x;
//     is >> x;
//     a = x;
//     return is;
// }
// } // namespace modular
// Stirling number from https://judge.yosupo.jp/submission/4988

char readchar() {
    static char buf[1 << 15], *ptr = buf, *end = buf;
    if(ptr == end) ptr = buf, end = buf + fread(buf, 1, 1 << 15, stdin);
    return *ptr++;
}
void writechar(char c = 0) {
    static char buf[1 << 15], *ptr = buf;
    if(ptr == end(buf) or c == 0) fwrite(buf, 1, ptr - buf, stdout), ptr = buf;
    *ptr++ = c;
}
template <class Z> Z getint() {
    char c = readchar();
    bool neg = c == '-';
    Z res = neg ? 0 : c - '0';
    while(isdigit(c = readchar())) res = res * 10 + (c - '0');
    return neg ? -res : res;
}
template <class Z> void putint(Z a, char c = '\n') {
    if(a < 0) writechar('-'), a = -a;
    int d[40], i = 0;
    do
        d[i++] = a % 10;
    while(a /= 10);
    while(i--) writechar('0' + d[i]);
    writechar(c);
}
string getstring() {
    string res;
    for(char c; (c = readchar()) != '\n';) res += c;
    return res;
}
void putstring(string s) {
    for(char c : s) writechar(c);
    writechar('\n');
}

template <class T, class Op = multiplies<T>> T power(T a, long long n, Op op = multiplies<T>(), T e = {1}) {
    assert(n >= 0);
    while(n) {
        if(n & 1) e = op(e, a);
        if(n >>= 1) a = op(a, a);
    }
    return e;
}

template <unsigned M> struct modular {
    using m = modular;
    unsigned v;
    modular(long long x = 0) : v((x %= M) < 0 ? x + M : x) {}
    m operator-() const { return m() -= *this; }
    m &operator+=(m b) {
        if((v += b.v) >= M) v -= M;
        return *this;
    }
    m &operator-=(m b) {
        if(v < b.v) v += M;
        v -= b.v;
        return *this;
    }
    m &operator*=(m b) {
        v = (uint64_t)v * b.v % M;
        return *this;
    }
    m &operator/=(m b) { return *this *= power(b, M - 2); }
    friend m operator+(m a, m b) { return a += b; }
    friend m operator-(m a, m b) { return a -= b; }
    friend m operator*(m a, m b) { return a *= b; }
    friend m operator/(m a, m b) { return a /= b; }
    friend bool operator==(m a, m b) { return a.v == b.v; }
    friend string to_string(m a) { return to_string(a.v); }
};

template <unsigned M> void ntt(vector<modular<M>> &a, bool inverse) {
    static vector<modular<M>> dw(30), idw(30);
    if(dw[0] == 0) {
        modular<M> root = 2;
        while(power(root, (M - 1) / 2) == 1) root += 1;
        for(int i = 0; i < 30; ++i) dw[i] = -power(root, (M - 1) >> (i + 2)), idw[i] = 1 / dw[i];
    }
    int n = a.size();
    assert((n & (n - 1)) == 0);
    if(not inverse) {
        for(int m = n; m >>= 1;) {
            modular<M> w = 1;
            for(int s = 0, k = 0; s < n; s += 2 * m) {
                for(int i = s, j = s + m; i < s + m; ++i, ++j) {
                    auto x = a[i], y = a[j] * w;
                    if(x.v >= M) x.v -= M;
                    a[i].v = x.v + y.v, a[j].v = x.v + (M - y.v);
                }
                w *= dw[__builtin_ctz(++k)];
            }
        }
    } else {
        for(int m = 1; m < n; m *= 2) {
            modular<M> w = 1;
            for(int s = 0, k = 0; s < n; s += 2 * m) {
                for(int i = s, j = s + m; i < s + m; ++i, ++j) {
                    auto x = a[i], y = a[j];
                    a[i] = x + y, a[j].v = x.v + (M - y.v), a[j] *= w;
                }
                w *= idw[__builtin_ctz(++k)];
            }
        }
    }
    auto c = 1 / modular<M>(inverse ? n : 1);
    for(auto &&e : a) e *= c;
}
template <unsigned M> vector<modular<M>> operator*(vector<modular<M>> a, vector<modular<M>> b) {
    if(a.empty() or b.empty()) return {};
    int n = a.size(), m = b.size(), sz = 1 << __lg(2 * (n + m - 1) - 1);
    if(min(n, m) < 30) {
        vector<long long> res(n + m - 1);
        for(int i = 0; i < n; ++i)
            for(int j = 0; j < m; ++j) res[i + j] += (a[i] * b[j]).v;
        return {begin(res), end(res)};
    }
    bool eq = a == b;
    a.resize(sz), ntt(a, false);
    if(eq)
        b = a;
    else
        b.resize(sz), ntt(b, false);
    for(int i = 0; i < sz; ++i) a[i] *= b[i];
    ntt(a, true), a.resize(n + m - 1);
    return a;
}

constexpr long long mod = 998244353;
using mint = modular<mod>;

vector<mint> fact, inv_fact, minv;
void prepare(int n) {
    fact.resize(n + 1), inv_fact.resize(n + 1), minv.resize(n + 1);
    for(int i = 0; i <= n; ++i) fact[i] = i ? i * fact[i - 1] : 1;
    inv_fact[n] = power(fact[n], mod - 2);
    for(int i = n; i; --i) inv_fact[i - 1] = i * inv_fact[i];
    for(int i = 1; i <= n; ++i) minv[i] = inv_fact[i] * fact[i - 1];
}
mint binom(int n, int k) {
    if(k < 0 or k > n) return 0;
    return fact[n] * inv_fact[k] * inv_fact[n - k];
}
template <> mint &mint::operator/=(mint r) { return *this *= r.v < minv.size() ? minv[r.v] : power(r, mod - 2); }

int main() {
    int n = getint<int>();
    int k = getint<int>();
    prepare(k + 1);
    vector<mint> a(k + 1), b(k + 1);
    for(int i = 0; i <= k; ++i) {
        a[i] = power(-1, i) * power(mint(i), n) * inv_fact[i];
        b[i] = inv_fact[i];
    }
    auto c = a * b;
    mint ans;
    for(int i = 1; i <= k; ++i) {
        c[i] *= power(-1, i);
        if((n - i) & 1) ans += binom(k, i) * c[i].v * fact[i];
        // putint(c[i].v, " \n"[i == n]);
    }
    putint(ans.v, '\n');
    writechar();
}