mirror of
https://github.com/InternLM/InternBootcamp.git
synced 2026-04-23 16:55:02 +00:00
265 lines
9.2 KiB
Python
Executable file
265 lines
9.2 KiB
Python
Executable file
"""# 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
|
||
|
||
Nonograms, also called \"Paint by Numbers,\" are logic puzzles where you reveal a hidden image by filling cells in a grid according to numerical clues. Here are the general rules:
|
||
|
||
1. **Grid Structure**:
|
||
- The puzzle consists of a rectangular grid (e.g., 10×10, 15×15, etc.).
|
||
- Each **row** and **column** has a sequence of numbers (clues) at its edge.
|
||
|
||
2. **Clue Interpretation**:
|
||
- Clues indicate groups of **consecutively filled cells** in that row/column.
|
||
Example: A clue of `3 2` means the row/column contains a block of 3 filled cells, followed by **at least one empty cell**, then a block of 2 filled cells.
|
||
- The **order of clues** matches the order of blocks (left-to-right for rows, top-to-bottom for columns).
|
||
- Empty cells can be marked with an \"X\" or left blank, depending on the puzzle variant.
|
||
|
||
3. **Rules for Filling**:
|
||
- **Exact blocks**: The numbers must correspond **exactly** to the filled cells.
|
||
Example: If a row has a clue `5`, the entire row must be filled with 5 contiguous cells.
|
||
- **No overlaps**: Blocks of filled cells cannot overlap unless the clues explicitly allow it (rare).
|
||
- **Separation**: Blocks in the same row/column must be separated by **at least one empty cell**.
|
||
|
||
4. **Solving Logic**:
|
||
- Use **cross-referencing** between row and column clues to deduce filled cells.
|
||
- Eliminate impossible configurations using overlaps or forced gaps.
|
||
|
||
5. **Victory Condition**:
|
||
- The puzzle is solved when all filled cells match the clues for every row and column, revealing the hidden image.
|
||
|
||
Nonograms require no guessing—only logical deduction based on the clues and grid 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 random
|
||
import re
|
||
|
||
class Nonogramsbootcamp(Basebootcamp):
|
||
def __init__(self, **params):
|
||
self.rows = params.get('rows', 5)
|
||
self.cols = params.get('cols', 5)
|
||
self.fill_prob = params.get('fill_prob', 0.3)
|
||
|
||
def case_generator(self):
|
||
# Generate solution grid
|
||
solution = [
|
||
[random.random() < self.fill_prob for _ in range(self.cols)]
|
||
for _ in range(self.rows)
|
||
]
|
||
|
||
# Calculate clues
|
||
rows_clues = [self._get_clues(row) for row in solution]
|
||
cols_clues = [self._get_clues([solution[r][c] for r in range(self.rows)])
|
||
for c in range(self.cols)]
|
||
|
||
return {'rows': rows_clues, 'columns': cols_clues}
|
||
|
||
@staticmethod
|
||
def prompt_func(question_case) -> str:
|
||
prompt = """你正在解决一个Nonogram谜题。根据行和列的数字线索填充网格：
|
||
|
||
规则说明：
|
||
1. 数字表示连续填充的单元格块，块间至少间隔一个空格
|
||
2. 行线索从左到右排列，列线索从上到下排列
|
||
3. 用'X'表示填充，用空格或'.'表示空白
|
||
|
||
行线索：
|
||
""" + "\n".join(
|
||
f"第{i+1}行: {clues if clues else '无'}"
|
||
for i, clues in enumerate(question_case['rows'])
|
||
) + "\n\n列线索：\n" + "\n".join(
|
||
f"第{i+1}列: {clues if clues else '无'}"
|
||
for i, clues in enumerate(question_case['columns'])
|
||
) + """
|
||
|
||
请将最终答案放在[answer]标签内，每行用'X'和空格表示填充状态：
|
||
示例：
|
||
[answer]
|
||
XX X
|
||
XXX
|
||
X X
|
||
[/answer]"""
|
||
return prompt
|
||
|
||
@staticmethod
|
||
def extract_output(output):
|
||
matches = re.findall(r'\[answer\](.*?)\[\/answer\]', output, re.DOTALL)
|
||
if not matches:
|
||
return None
|
||
grid_str = matches[-1].strip()
|
||
solution = []
|
||
for line in grid_str.split('\n'):
|
||
line = line.strip()
|
||
if not line:
|
||
continue
|
||
solution.append([c.upper() == 'X' for c in line if not c.isspace() or c == '.'])
|
||
return solution
|
||
|
||
@classmethod
|
||
def _verify_correction(cls, solution, identity):
|
||
# Validate grid dimensions
|
||
if len(solution) != len(identity['rows']):
|
||
return False
|
||
if any(len(row) != len(identity['columns']) for row in solution):
|
||
return False
|
||
|
||
# Check row clues
|
||
for i, row in enumerate(solution):
|
||
if cls._get_clues(row) != identity['rows'][i]:
|
||
return False
|
||
|
||
# Check column clues
|
||
for j in range(len(identity['columns'])):
|
||
col = [solution[i][j] for i in range(len(solution))]
|
||
if cls._get_clues(col) != identity['columns'][j]:
|
||
return False
|
||
|
||
return True
|
||
|
||
@staticmethod
|
||
def _get_clues(line):
|
||
clues = []
|
||
current = 0
|
||
for cell in line:
|
||
if cell:
|
||
current += 1
|
||
elif current > 0:
|
||
clues.append(current)
|
||
current = 0
|
||
if current > 0:
|
||
clues.append(current)
|
||
return clues
|