#include <bits/stdc++.h>
using namespace std;
using p = pair<int, int>;
int n, m, te, bits[20] = {0};
p dp[1048576][21];
bool poss[1048576] = {0};
vector<int> council, cnts, dpCouncils;
int main() {
for(auto&i: dp) for(auto&j: i) j = {0, 0};
cin.tie(0)->sync_with_stdio(0);
cin >> n >> m;
council.resize(n);
for(auto&i: council) {
i = 0;
for(int j = 0; j < m; j ++) {
cin >> te;
if(te) {
i += (1 << j);
bits[j] ++;
}
}
poss[((1 << m) - 1) & (~i)] = 1;
}
for(auto&i: council) {
int ni = 0;
int cnt = 0;
for(int j = 0; j < m; j ++) {
int cb = bits[j];
if(i & (1 << j)) cb --;
if(cb - 1 >= (n) / 2) {
cnt ++;
}
else if(cb >= (n) / 2) {
ni += (1 << j);
}
}
dpCouncils.push_back(ni);
cnts.push_back(cnt);
}
for(int i = 0; i < (1 << m); i ++) {
dp[i][0] = {(poss[i] ? __builtin_popcount(i) : 0), 0};
if((poss[i] ? __builtin_popcount(i) : 0) == 10) {
// cout << i << "\n";
}
}
for(int j = 1; j <= m; j ++) {
for(int i = 0; i < (1 << m); i ++) {
int max1 = dp[i][j - 1].first;
int max2 = dp[i][j - 1].second;
auto c = ((i & (1 << (j - 1))) ? dp[i - (1 << (j - 1))][j - 1] : pair<int, int>(dp[i + (1 << (j - 1))][j - 1].first - 1, dp[i + (1 << (j - 1))][j - 1].second - 1));
if(max1 >= c.first) max2 = max(max2, c.first);
else {
max2 = max(max1, c.second);
max1 = c.first;
}
dp[i][j] = {max1, max2};
}
}
for(int i = 0; i < n; i ++) {
int maxV = 0;
for(int j = 0; j < m; j ++) {
int cb = bits[j];
if(council[i] & (1 << j)) cb --;
if(cb - 1 >= (n) / 2) {
}
else if(cb >= (n) / 2) {
if(!(council[i] & (1 << j))) maxV ++;
}
}
// cout << maxV << "\n";
// cout << cnts[i] << "\n";
cout << cnts[i] + ((maxV == dp[dpCouncils[i]][m].first) ? dp[dpCouncils[i]][m].second : dp[dpCouncils[i]][m].first) << "\n";
assert(cnts[i] + (maxV == dp[dpCouncils[i]][m].first ? dp[dpCouncils[i]][m].second : dp[dpCouncils[i]][m].first) <= m);
// cout << cnts[i] << "\n";
}
}
