#include <bits/stdc++.h>
// clang-format off
using namespace std;
void print0(){}; template<typename H,typename... T> void print0(H h,T... t){cout<<h;print0(t...);}
void print(){print0("\n");}; template<typename H,typename... T>void print(H h,T... t){print0(h);if(sizeof...(T)>0)print0(" ");print(t...);}
#define debug1(a) { cerr<<#a<<":"<<a<<endl; }
#define debug2(a,b) { cerr<<#a<<":"<<a<<" "<<#b<<":"<<b<<endl; }
#define debug3(a,b,c) { cerr<<#a<<":"<<a<<" "<<#b<<":"<<b<<" "<<#c<<":"<<c<<endl; }
#define debug4(a,b,c,d) { cerr<<#a<<":"<<a<<" "<<#b<<":"<<b<<" "<<#c<<":"<<c<<" "<<#d<<":"<<d<<endl; }
// clang-format on

namespace marathon {
mt19937 engine(0);
clock_t start_time;
double now() {
    return 1000.0 * (clock() - start_time) / CLOCKS_PER_SEC;
}
void marathon_init() {
    start_time = clock();
    random_device seed_gen;
    engine.seed(seed_gen());
}
int randint(int mn, int mx) {
    int rng = mx - mn + 1;
    return mn + (engine() % rng);
}
bool anneal_accept(double new_score, double old_score, double cur_time, double begin_time, double end_time, double begin_temp, double end_temp) {
    const int ANNEAL_RND = 1e8;
    const double ANNEAL_EPS = 1e-6;
    double temp = (begin_temp * (end_time - cur_time) + end_temp * (cur_time - begin_time)) / (end_time - begin_time);
    return (exp((new_score - old_score) / temp) > double(engine() % ANNEAL_RND) / ANNEAL_RND + ANNEAL_EPS);
}
}  // namespace marathon

const int N2 = 36;
const int N = 6;
const int SCORE_BASE = 3;
const double END_TIME = 1900;

struct six {
    int val[6];
};

int calc_score(vector<vector<int>> &perm, vector<six> &rolls) {
    int score = 0;
    for (int i = 0; i < N; i++) {
        for (int d = 0; d < N; d++) {
            int cnt = 0;
            for (int j = 0; j < N; j++) {
                int rollid = perm[i][j];
                int roll_cnt = rolls[rollid].val[d];
                if (!roll_cnt) {
                    cnt = 0;
                    break;
                }
                cnt += roll_cnt;
            }
            if (!cnt) continue;
            score += cnt - N + SCORE_BASE;
        }
    }
    for (int j = 0; j < N; j++) {
        for (int d = 0; d < N; d++) {
            int cnt = 0;
            for (int i = 0; i < N; i++) {
                int rollid = perm[i][j];
                int roll_cnt = rolls[rollid].val[d];
                if (!roll_cnt) {
                    cnt = 0;
                    break;
                }
                cnt += roll_cnt;
            }
            if (!cnt) continue;
            score += cnt - N + SCORE_BASE;
        }
    }
    return score;
}

void anneal(vector<six> rolls) {
    vector<vector<int>> perm(N, vector<int>(N));
    for (int i = 0; i < N; i++) {
        for (int j = 0; j < N; j++) {
            perm[i][j] = i * N + j;
        }
    }
    int old_score = calc_score(perm, rolls);

    double begin_time = marathon::now();
    double end_time = END_TIME;
    double begin_temp = 2.0;
    double end_temp = begin_temp * 0.1;

    int anneal_iter = 0;
    int anneal_accept = 0;
    while (marathon::now() < END_TIME) {
        anneal_iter++;
        auto nperm = perm;  // TODO 差分計算
        int x = 0;
        int y = 0;
        while (x == y) {
            x = marathon::engine() % N2;
            y = marathon::engine() % N2;
        }
        int xi = x / N;
        int xj = x % N;
        int yi = y / N;
        int yj = y % N;
        swap(nperm[xi][xj], nperm[yi][yj]);
        int new_score = calc_score(nperm, rolls);
        if (marathon::anneal_accept(new_score, old_score, marathon::now(), begin_time, end_time, begin_temp, end_temp)) {
            anneal_accept++;
            perm = nperm;
            old_score = new_score;
        }
    }
    debug3(anneal_iter, anneal_accept, old_score);

    vector<pair<int, int>> revperm(N2);
    for (int i = 0; i < N; i++) {
        for (int j = 0; j < N; j++) {
            revperm[perm[i][j]] = {i, j};
        }
    }
    for (int i = 0; i < N2; i++) {
        print(revperm[i].first + 1, revperm[i].second + 1);
    }
}

int main() {
    marathon::marathon_init();
    vector<six> rolls(N2);
    for (int i = 0; i < N2; i++) {
        six s;
        for (int k = 0; k < N; k++) {
            s.val[k] = 0;
        }
        for (int j = 0; j < N; j++) {
            int d;
            cin >> d;
            d--;
            s.val[d]++;
        }
        rolls[i] = s;
    }
    anneal(rolls);
    return 0;
}