Girls' research

题目描述:

One day, sailormoon girls are so delighted that they intend to research about palindromic strings. Operation contains two steps:
First step: girls will write a long string (only contains lower case) on the paper. For example, “abcde”, but ‘a’ inside is not the real ‘a’, that means if we define the ‘b’ is the real ‘a’, then we can infer that ‘c’ is the real ‘b’, ‘d’ is the real ‘c’ ……, ‘a’ is the real ‘z’. According to this, string “abcde” changes to “bcdef”.
Second step: girls will find out the longest palindromic string in the given string, the length of palindromic string must be equal or more than 2.

输入:

Input contains multiple cases.
Each case contains two parts, a character and a string, they are separated by one space, the character representing the real ‘a’ is and the length of the string will not exceed 200000.All input must be lowercase.
If the length of string is len, it is marked from 0 to len-1.

输出:

Please execute the operation following the two steps.
If you find one, output the start position and end position of palindromic string in a line, next line output the real palindromic string, or output “No solution!”.
If there are several answers available, please choose the string which first appears.

样例输入:

b babd
a abcd

样例输出:

0 2
aza
No solution!

code:

题目大意:找到最长的那个回文字符串
思路:manacher算法

#include<iostream>
#include<stdio.h>
#include<string.h>
using namespace std;
char feng[420000];
char jiang[460000];
int huang[460000];
void init()
{
	int j=0;
	int len=strlen(feng);
	jiang[j++]='$';
	jiang[j++]='#';
	for(int i=0;i<len;i++)
	{
		jiang[j++]=feng[i];
		jiang[j++]='#';
	}
	jiang[j]='\0';
}
void manacher(int what)
{
	init();
	int len=strlen(jiang);
	int mx=0,maxx=-1,id=0;
	int start,over;
	for(int i=1;i<len;i++)
	{
		if(i<mx){
			huang[i]=min(huang[id*2-i],mx-i);
		}
		else{
			huang[i]=1;
		}
		while(jiang[i-huang[i]]==jiang[i+huang[i]])huang[i]++;
		if(huang[i]+i>mx){
			mx=huang[i]+i;
			id=i;
		}
		if(huang[i]>huang[maxx])
		{
			maxx=i;
		}
	}
	if(huang[maxx]>2)
	{
		start=(maxx-huang[maxx])/2;
		over=(maxx+huang[maxx])/2;
		over-=2; 
		printf("%d %d\n",start,over);
		for(int i=start;i<=over;i++)
		{
			feng[i]+=what;
			if(feng[i]<'a')feng[i]+=26;
			printf("%c",feng[i]);
		}
		printf("\n");
	}
	else printf("No solution!\n");
}
int main()
{
	char a[3];
	while(scanf("%s%s",a,feng)!=EOF)
	{
		int what='a'-a[0];
		manacher(what);
	}
	return 0;
}
idx": 5, "question": "Number of girls is twice of boys in a class of 60 students. Therefore,\nNumber of girls = 40\nNumber of people ahead of Kamal = 16\nNumber of girls ahead of Kamal = 9\nNumber of boys ahead of Kamal = 16 - 9 = 7\nHence, 12 boys are after Kamal in terms of their ranking.", "answer": "Number of girls in the class\n\n1. Total number of students is 60.\n2. Number of girls is twice the number of boys.\n\nLet \\( B \\) be the number of boys. Then the number of girls \\( G = 2B \\). \nTotal students: \\( B + G = 60 \\). \nSubstitute \\( G = 2B \\): \\( B + 2B = 60 \\). \nSolve for \\( B \\): \\( 3B = 60 \\implies B = 20 \\). \nThen \\( G = 2B = 40 \\).\n\nresult: Number of girls = 40\n\n```mathematica\n\nB = 60 / 3;\nG = 2 * B;\n\n```\n\nNumber of boys ahead of Kamal\n\n1. Total number of people ahead of Kamal is 16.\n2. Number of girls ahead of Kamal is 9.\n\nNumber of boys ahead of Kamal = Total people ahead of Kamal - Girls ahead of Kamal \n\\( \\text{Boys ahead} = 16 - 9 = 7 \\).\n\nresult: Number of boys ahead of Kamal = 7\n\n```mathematica\n\nboysAhead = 16 - 9;\n\n```\n\nNumber of boys after Kamal\n\n1. Total number of boys is 20 (from Computation 1).\n2. Number of boys ahead of Kamal is 7.\n\nNumber of boys after Kamal = Total boys - Boys ahead of Kamal - Kamal himself \n\\( \\text{Boys after} = 20 - 7 - 1 = 12 \\).\n\nresult: Number of boys after Kamal = 12\n\n```mathematica\n\ntotalBoys = 20;\nboysAhead = 7;\nboysAfter = totalBoys - boysAhead - 1;\nPrint[\"Number of boys after Kamal = \", boysAfter];\n```\n\n--- \n\nOnly the last Mathematica code snippet includes a `Print` statement to output the final result. The other snippets are executable individually but do not produce output.\n```"}我想用前面这种数据去微调llama,当我输入问题时,能给我答案 #!/usr/bin/env python3 import re import torch import logging logger = logging.getLogger() IGNORE_INDEX = -100 class Processor: def group_texts(self, examples, tokenizer, max_len): input_ids, labels = [], [] final_input_ids, final_labels = [], [] for idx in range(len(examples['input_ids'])): _input_ids = examples['input_ids'][idx] _labels = examples['input_ids'][idx] examples['input_ids'][idx] = None if len(_input_ids) > max_len: # if single sample longer than max_len, break into several devided_input_ids, devided_labels = [], [] for i in range(0, len(_input_ids), max_len): devided_input_ids = _input_ids[i: i + max_len] devided_labels = _labels[i: i + max_len] if len(devided_input_ids) < max_len: devided_pad_num = max_len - len(devided_input_ids) devided_input_ids += [tokenizer.pad_token_id] * devided_pad_num devided_labels += [IGNORE_INDEX] * devided_pad_num final_input_ids.append(devided_input_ids) final_labels.append(devided_labels) continue # if single sample shorter than max_len, combine together if len(input_ids) + len(_input_ids) > max_len: pad_num = max_len - len(input_ids) final_input_ids.append(input_ids + [tokenizer.pad_token_id] * pad_num) final_labels.append(labels + [IGNORE_INDEX] * pad_num) input_ids, labels = [], [] input_ids.extend(_input_ids) labels.extend(_labels) if len(input_ids) > 0: pad_num = max_len - len(input_ids) final_input_ids.append(input_ids + [tokenizer.pad_token_id] * pad_num) final_labels.append(labels + [IGNORE_INDEX] * pad_num) return { "input_ids": torch.tensor(final_input_ids).long(), "labels": torch.tensor(final_labels).long() } def process_tokenize(self, exmaples, tokenizer): """ tokenize samples and add bos and eos tokens """ inputs = tokenizer(exmaples['text'], truncation=False, padding=False) input_ids, labels = [], [] for input_id in inputs['input_ids']: input_ids.append([tokenizer.bos_token_id] + input_id + [tokenizer.eos_token_id]) return { "input_ids": input_ids, }参照这个代码,我要怎样进行分析处理
最新发布
06-20
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值