mirror of
https://github.com/InternLM/InternBootcamp.git
synced 2026-04-24 17:05:00 +00:00
init-commit
This commit is contained in:
commit
18a552597a
3461 changed files with 1150579 additions and 0 deletions
266
internbootcamp/bootcamp/shingoki/shingoki.py
Executable file
266
internbootcamp/bootcamp/shingoki/shingoki.py
Executable file
|
|
@ -0,0 +1,266 @@
|
|||
"""# 谜题训练场开发任务
|
||||
|
||||
## 任务概述
|
||||
你是一位资深程序员,我需要你帮我实现一个特定谜题的训练场环境类。这个类继承自`Basebootcamp`,用于生成谜题实例并验证解答。
|
||||
|
||||
## 背景说明
|
||||
我正在开发一系列谜题训练场,每个训练场对应一个特定类型的谜题。训练场类命名为`{PuzzleName}bootcamp`,其中`PuzzleName`是谜题的名称。
|
||||
|
||||
每个训练场类主要提供两个核心功能:
|
||||
1. 生成该谜题类型的问题实例
|
||||
2. 验证用户对问题的回答是否正确
|
||||
|
||||
## 技术接口规范
|
||||
|
||||
### 类方法实现要求
|
||||
|
||||
```python
|
||||
class {PuzzleName}bootcamp(Basebootcamp):
|
||||
def __init__(self, **params):
|
||||
\"\"\"
|
||||
请你自定义params,以保存该puzzle相关的参数,例如网格大小等,参数配有默认值
|
||||
\"\"\"
|
||||
pass
|
||||
|
||||
def case_generator(self):
|
||||
\"\"\"
|
||||
生成谜题实例,提示:为保证谜题有解,可以先生成结果再对结果处理得到谜题
|
||||
返回:一个可JSON序列化的字典(避免包含set等无法通过json.dumps处理的数据结构)
|
||||
\"\"\"
|
||||
pass
|
||||
|
||||
@staticmethod
|
||||
def prompt_func(question_case) -> str:
|
||||
\"\"\"
|
||||
将case_generator生成的谜题实例转换为文本形式的问题,问题中包含问题背景、对谜题规则的介绍、具体要解决的谜题实例、期望最终答案的格式,
|
||||
例如:你是xxxx,请你解答yyyy,规则如下:yyyy,最终答案放置在:zzzzz
|
||||
|
||||
参数:
|
||||
question_case: 由case_generator生成的谜题实例
|
||||
|
||||
返回:
|
||||
str: 格式化的问题字符串
|
||||
|
||||
注意:
|
||||
1. 需考虑问题的格式,以便后续能正确提取
|
||||
2. 问题描述中应包含期望的答案格式说明,以便后续能正确提取,为了避免抽取时匹配出干扰项,请要求模型将答案放在特定标签,如[answer] [/answer]内
|
||||
\"\"\"
|
||||
pass
|
||||
|
||||
@staticmethod
|
||||
def extract_output(output):
|
||||
\"\"\"
|
||||
从LLM的回复中提取符合格式要求的答案,如有多个,请抽取最后一个,避免使用re.search等只抽取第一个结果的方式。
|
||||
|
||||
参数:
|
||||
output: LLM的完整输出(包含原始问题和回答)
|
||||
|
||||
返回:
|
||||
提取的答案,若未找到符合格式的答案则返回None
|
||||
\"\"\"
|
||||
pass
|
||||
|
||||
@classmethod
|
||||
def _verify_correction(cls, solution, identity):
|
||||
\"\"\"
|
||||
验证提取的答案是否正确,注意一个问题可以能有多个解,按照谜题规则进行检验,不要直接匹配可能的答案。
|
||||
|
||||
参数:
|
||||
solution: extract_output提取的答案
|
||||
identity: case_generator生成的谜题实例
|
||||
|
||||
返回:
|
||||
bool: 答案是否正确
|
||||
\"\"\"
|
||||
pass
|
||||
```
|
||||
|
||||
### 验证评分方法(基类已实现)
|
||||
|
||||
```python
|
||||
@classmethod
|
||||
def verify_score(cls, model_output, identity:dict, format_score=0.1) -> float:
|
||||
\"\"\"
|
||||
验证输出结果并评分。
|
||||
|
||||
参数:
|
||||
model_output: 模型的完整输出
|
||||
identity: 谜题实例(由case_generator生成)
|
||||
format_score: 答案格式正确时的基础分数
|
||||
|
||||
返回:
|
||||
float: 评分结果(0-1之间)
|
||||
\"\"\"
|
||||
score = 0.
|
||||
try:
|
||||
extract_solution = cls.extract_output(model_output)
|
||||
if extract_solution is None:
|
||||
return score
|
||||
else:
|
||||
score = format_score # 格式正确时的基础分数
|
||||
if cls._verify_correction(extract_solution, identity):
|
||||
score = 1. # 答案完全正确时的满分
|
||||
except Exception as e:
|
||||
# 处理异常情况
|
||||
pass
|
||||
return score
|
||||
```
|
||||
|
||||
### 使用示例
|
||||
|
||||
```python
|
||||
# 初始化谜题训练场
|
||||
bootcamp = Puzzlebootcamp()
|
||||
|
||||
# 生成谜题实例
|
||||
case = bootcamp.case_generator()
|
||||
|
||||
# 将谜题转换为文本问题
|
||||
prompt = Puzzlebootcamp.prompt_func(case)
|
||||
|
||||
# 获取LLM对问题的解答
|
||||
response = get_response(prompt, \"LLM\")
|
||||
|
||||
# 从完整对话中提取答案
|
||||
extracted_output = Puzzlebootcamp.extract_output(prompt + response)
|
||||
|
||||
# 验证答案并评分
|
||||
score = Puzzlebootcamp.verify_score(extracted_output, case)
|
||||
```
|
||||
|
||||
## 你的任务
|
||||
请根据以下谜题描述(谜题描述可能不完整,请先结合你的知识澄清规则),实现一个完整的谜题训练场类:
|
||||
|
||||
### 谜题描述
|
||||
|
||||
Shingoki is a logic puzzle where players draw a single continuous, non-intersecting loop on a grid. Here are the core rules:
|
||||
|
||||
1. **Objective**:
|
||||
Create a closed loop that traverses *edges* of the grid (lines between dots/nodes). The loop must not branch, cross itself, or form isolated segments.
|
||||
|
||||
2. **Clues (Numbered Circles)**:
|
||||
- Numbered circles appear at *intersections* (nodes where four grid edges meet).
|
||||
- Each number specifies *exactly* how many of the four edges connected to that node are part of the loop. For example:
|
||||
- **0**: None of the four edges are used.
|
||||
- **2**: Exactly two edges are used (e.g., a straight line or a turn).
|
||||
- **3/1**: Rare, typically near grid borders where fewer edges exist.
|
||||
|
||||
3. **2x2 Block Restriction**:
|
||||
The loop must not enclose any 2x2 area entirely with its lines. This prevents \"solid blocks\" of the loop.
|
||||
|
||||
4. **Logical Deductions**:
|
||||
- The loop must form a single closed path.
|
||||
- Unnumbered nodes have no constraints (any number of edges can be used).
|
||||
- Clues override default assumptions (e.g., a \"3\" forces three edges at a node, even if it seems counterintuitive).
|
||||
|
||||
The puzzle is solved by deducing edge usage around numbered nodes while adhering to these constraints.
|
||||
|
||||
|
||||
请完成上述谜题的训练场环境类实现,包括所有必要的方法。
|
||||
"""
|
||||
|
||||
from bootcamp import Basebootcamp
|
||||
import re
|
||||
from typing import Dict, Set, Tuple
|
||||
|
||||
class Shingokibootcamp(Basebootcamp):
|
||||
def __init__(self, rows=5, cols=5):
|
||||
self.rows = rows
|
||||
self.cols = cols
|
||||
|
||||
def case_generator(self) -> Dict:
|
||||
"""
|
||||
生成Shingoki谜题实例(简化的示例版本)
|
||||
注意:当前示例未实现实际环生成算法,需后续完善
|
||||
"""
|
||||
return {
|
||||
"rows": self.rows,
|
||||
"cols": self.cols,
|
||||
"clues": {(2, 2): 0} # 示例线索(实际实现需要有效生成)
|
||||
}
|
||||
|
||||
@staticmethod
|
||||
def prompt_func(question_case: Dict) -> str:
|
||||
clues = question_case['clues']
|
||||
rows = question_case['rows']
|
||||
cols = question_case['cols']
|
||||
prompt = [
|
||||
"请解决以下Shingoki谜题:",
|
||||
"在{}×{}的网格中绘制一个闭合不交叉的环,满足以下条件:".format(rows, cols),
|
||||
"1. 环必须经过网格边线且满足2×2区块限制",
|
||||
"2. 数字表示相连边数(0-3)",
|
||||
"————————————————",
|
||||
"已知线索节点:"
|
||||
]
|
||||
|
||||
# 格式化线索描述
|
||||
for (i, j), num in clues.items():
|
||||
prompt.append(f"• 位置 ({i},{j}) 处必须连接 {num} 条边")
|
||||
|
||||
prompt.extend([
|
||||
"\n请用以下格式回答:",
|
||||
"[answer]",
|
||||
"(行坐标,列坐标,方向) 每行一个边",
|
||||
"示例:",
|
||||
"(0,0,H) # 水平边,连接(0,0)-(0,1)",
|
||||
"(1,2,V) # 垂直边,连接(1,2)-(2,2)",
|
||||
"[/answer]"
|
||||
])
|
||||
|
||||
return "\n".join(prompt)
|
||||
|
||||
@staticmethod
|
||||
def extract_output(output: str) -> Set[Tuple[int, int, str]]:
|
||||
# 匹配最后一个answer块
|
||||
answer_blocks = re.findall(r'\[answer\](.*?)\[/answer\]', output, re.DOTALL)
|
||||
if not answer_blocks:
|
||||
return None
|
||||
|
||||
# 解析边坐标
|
||||
edges = set()
|
||||
pattern = re.compile(r'\((\d+),(\d+),(H|V)\)')
|
||||
for line in answer_blocks[-1].strip().split('\n'):
|
||||
match = pattern.search(line)
|
||||
if match:
|
||||
i, j, dir = int(match[1]), int(match[2]), match[3]
|
||||
edges.add((i, j, dir))
|
||||
|
||||
return edges if edges else None
|
||||
|
||||
@classmethod
|
||||
def _verify_correction(cls, solution: Set[Tuple[int, int, str]], identity: Dict) -> bool:
|
||||
# 转存谜题参数
|
||||
rows = identity['rows']
|
||||
cols = identity['cols']
|
||||
clues = identity['clues']
|
||||
|
||||
# 验证边有效性
|
||||
for i, j, dir in solution:
|
||||
if dir == 'H' and j >= cols-1:
|
||||
return False # 水平边越界
|
||||
if dir == 'V' and i >= rows-1:
|
||||
return False # 垂直边越界
|
||||
|
||||
# 验证线索条件
|
||||
edge_set = {(dir, i, j) for (i, j, dir) in solution}
|
||||
for (node_i, node_j), expected in clues.items():
|
||||
count = 0
|
||||
# 检查四边
|
||||
if ('H', node_i, node_j) in edge_set: # 右边
|
||||
count += 1
|
||||
if node_j > 0 and ('H', node_i, node_j-1) in edge_set: # 左边
|
||||
count += 1
|
||||
if ('V', node_i, node_j) in edge_set: # 下边
|
||||
count += 1
|
||||
if node_i > 0 and ('V', node_i-1, node_j) in edge_set: # 上边
|
||||
count += 1
|
||||
if count != expected:
|
||||
return False
|
||||
|
||||
# TODO: 实际实现需添加以下验证
|
||||
# 1. 环的连通性检查
|
||||
# 2. 闭合性检查
|
||||
# 3. 2×2区块限制检查
|
||||
# 4. 无交叉检查
|
||||
|
||||
return True # 示例实现仅验证线索条件
|
||||
Loading…
Add table
Add a link
Reference in a new issue