poj_2503 Babelfish（ELF哈希）

关注

原创最新推荐文章于 2026-06-01 08:15:33 发布 356 阅读 AI 写同款· GEO 优化 ›

YOLO26算法车道线检测+训练好的模型+360张数据集+pyqt可视化界面.zip 附训练好的 YOLO 权重与 PyQt 可视化检测界面，页面底部含可视化效果预览，毕设/课设开箱即用。【数据集概况】 · 检测类别（中文）：[未获取()] 【训练曲线与评估图】【模型训练配置】参数 | 值模型 | yolo26n 训练轮数 | 100 epochs 输入尺寸 | 640x640 批次大小 | 24 优化器 | auto 初始学习率 | 0.01 训练设备【关键指标汇总】训练了 55 个 epoch，最终轮指标：指标 | 数值 mAP50 | **0.9512** mAP50-95 | 0.6746 Precision | 0.8796 Recall | 0.8825 train/box_loss | 0.8608 train/cls_loss | 0.7232 val/box_loss | 0.9955 val/cls_loss | 0.5511 【训练过程分析】这轮训练我经历了从初始到稳定的曲折过程。前 20 个epoch，loss 曲线呈现了陡峭的下降趋势，mAP50 指标从 0.8 开始逐步攀升，仿佛在寻找一个合适的平衡点。然而，从第 21 个epoch 开始，mAP50 的提升速率显著放缓，整体曲线显得有些曲折，有时甚至停滞不前，让我一度以为训练出现了问题。直到第 35 个epoch，loss 开始逐渐趋于平稳，这标志着模型已经找到了一个较好的收敛区间。到了最后的 20 个 epoch，尽管 mAP50 依旧在稳步上升，但提升幅度已经不大，这意味着模型已经接近其性能天花板。在这个过程中，我重点关注了 box_loss 和 cls_loss 的变化。初期阶段，box_loss 快速下降，从 0.9 逐渐降至 0.86，而 cls_loss 则从 0.75 逐步下降至... 使用场景：毕业设计、课程设计、深度学习入门、工业场景落地验证。立即下载

Babelfish

Time Limit: 3000MS		Memory Limit: 65536K
Total Submissions: 41750		Accepted: 17700

Description

You have just moved from Waterloo to a big city. The people here speak an incomprehensible dialect of a foreign language. Fortunately, you have a dictionary to help you understand them.

Input

Input consists of up to 100,000 dictionary entries, followed by a blank line, followed by a message of up to 100,000 words. Each dictionary entry is a line containing an English word, followed by a space and a foreign language word. No foreign word appears more than once in the dictionary. The message is a sequence of words in the foreign language, one word on each line. Each word in the input is a sequence of at most 10 lowercase letters.

Output

Output is the message translated to English, one word per line. Foreign words not in the dictionary should be translated as "eh".

Sample Input

dog ogday
cat atcay
pig igpay
froot ootfray
loops oopslay

atcay
ittenkay
oopslay

Sample Output

cat
eh
loops

Hint

Huge input and output,scanf and printf are recommended.

查词典题，不过我因为把手误把 int 打成 char WA了几次。。

字符串哈希用ELFHash会快上很多，至少是我自己写的哈希的两倍快。

#include <iostream>
#include <cstdio>
#include <cstdlib>
#include <cstring>
#include <cmath>
#include <stack>
#include <bitset>
#include <queue>
#include <set>
#include <map>
#include <string>
#include <algorithm>
#define FOP freopen("data.txt","r",stdin)
#define inf 0x3f3f3f3f
#define maxn 1000010
#define mod 1000007
#define PI acos(-1.0)
#define LL long long
using namespace std;

struct Node
{
    int pos;
    int next;
}node[maxn];

int cot;
int hashTable[maxn];

int n = 0;
char word[100010][12];
char vals[100010][12];

void initHash()
{
    cot = 0;
    memset(hashTable, -1, sizeof(hashTable));
}

bool cmp(char *a, char *b)
{
    return strcmp(a, b) == 0;
}

int getHash(char *a)
{
    int res = 0;
    for(int i = 0; a[i] != '\0'; i++) res += a[i] * (i+1);
    return res % mod;
}

int ELFHash(char *key)
{
    unsigned long h = 0;
    while(*key)
    {
        h = (h<<4) + (*key++);
        unsigned long g = h&0Xf0000000L;
        if(g) h ^= g>>24;
        h &= ~g;
    }
    return h%mod;
}

void insertHash(char *val, int pos)
{
    node[cot].pos = pos;
    //int key = getHash(val);
    int key = ELFHash(val);
    node[cot].next = hashTable[key];
    hashTable[key] = cot++;
}

int searchHash(char *val)
{
    //int key = getHash(val);
    int key = ELFHash(val);
    int next = hashTable[key];
    while(next != -1)
    {
        int t =  node[next].pos;
        if(cmp(val, vals[t])) return t;
        next = node[next].next;
    }
    return -1;
}


int main()
{
    //FOP;
    initHash();
    char c[100];
    char val[12];

    while(gets(c) && c[0] != '\0')
    {
        sscanf(c, "%s %s", word[n], vals[n]);
        insertHash(vals[n], n);
        n++;
    }

    while(~scanf("%s", val))
    {
        int t = searchHash(val);
        if(t != -1) printf("%s\n", word[t]);
        else printf("eh\n");
    }
    return 0;
}

收录于