init-commit

internbootcamp/bootcamp/kor_operation_unicodeffe0/kor_operation_unicodeffe0.py

@@ -0,0 +1,342 @@
+"""# 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
+from bootcamp import Basebootcamp
+
+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
+        注意：请参照提供的谜题描述进行复述，规则应当描述详细，包括任务背景、具体任务操作规则、对题目格式和答案格式的含义介绍等，
+
+        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（如双括号）内，例如[[your answer here]]
+        \"\"\"
+        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
+a￠b=\log_{b}{a}+\log_{a}{b}.
+a and b are positive integers.Example questions are as follows:
+
+<example 0>
+Compute 2￠8.
+If the answer is a fraction, write it in 'a/b' text format.Decimals are not allowed.
+Please wrap the answer in double square brackets, like this: [[your answer]].
+</example 0>
+
+<example 1>
+Compute 3￠9.
+If the answer is a fraction, write it in 'a/b' text format.Decimals are not allowed.
+Please wrap the answer in double square brackets, like this: [[your answer]].
+</example 1>
+
+<example 2>
+Compute 2￠2￠4.
+If the answer is a fraction, write it in 'a/b' text format.Decimals are not allowed.
+Please wrap the answer in double square brackets, like this: [[your answer]].
+</example 2>
+
+<example 3>
+Compute 5￠8.
+If the answer cannot be reduced to an integer or fraction then retain the form \log_{b}{a}+\log_{a}{b}.
+Please provide your answer in LaTeX format. Please wrap the answer in double square brackets, like this: [[your answer]].
+</example 3>
+
+<example 4>
+Compute 16￠256.
+If the answer is a fraction, write it in 'a/b' text format.Decimals are not allowed.
+Please wrap the answer in double square brackets, like this: [[your answer]].
+</example 4>
+
+<example 5>
+Compute 9￠243.
+If the answer is a fraction, write it in 'a/b' text format.Decimals are not allowed.
+Please wrap the answer in double square brackets, like this: [[your answer]].
+</example 5>
+
+<example 6>
+If 512￠X=82/9, find X.
+The answer should only be given as a number.
+Please wrap the answer in double square brackets, like this: [[your answer]].
+</example 6>
+
+<example 7>
+If 3￠X=17/4, find X.
+The answer should only be given as a number.
+Please wrap the answer in double square brackets, like this: [[your answer]].
+</example 7>
+
+<example 8>
+If X￠256=73/24, find X.
+The answer should only be given as a number.
+Please wrap the answer in double square brackets, like this: [[your answer]].
+</example 8>
+
+<example 9>
+If 3￠3￠X=5/2, find X.
+The answer should only be given as a number.
+Please wrap the answer in double square brackets, like this: [[your answer]].
+</example 9>
+
+
+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 math
+from fractions import Fraction
+import re
+from bootcamp import Basebootcamp
+
+class KorOperationUnicodeffe0bootcamp(Basebootcamp):
+    def __init__(self, min_value=2, max_value=256, problem_types=None, allow_multilevel=True):
+        self.min = min_value
+        self.max = max_value
+        self.problem_types = problem_types or ['compute', 'solve']
+        self.allow_multilevel = allow_multilevel
+
+    @staticmethod
+    def is_power(a, b):
+        """判断两个数是否互为整数幂"""
+        if a == 1 or b == 1:
+            return a == b
+        try:
+            exponent = math.log(a, b)
+            return abs(exponent - round(exponent)) < 1e-10 and b**round(exponent) == a
+        except ValueError:
+            pass
+        try:
+            exponent = math.log(b, a)
+            return abs(exponent - round(exponent)) < 1e-10 and a**round(exponent) == b
+        except ValueError:
+            return False
+
+    def case_generator(self):
+        problem_type = random.choice(self.problem_types)
+        
+        def generate_expressible():
+            k = random.randint(1,4)
+            b = random.randint(self.min, self.max)
+            a = b ** k
+            while a > self.max:
+                k = random.randint(1,4)
+                b = random.randint(self.min, self.max)
+                a = b ** k
+            return a, b, Fraction(k**2 +1, k)
+
+        if problem_type == 'compute':
+            if random.random() < 0.3:  
+                a = random.randint(self.min, self.max)
+                b = random.randint(self.min, self.max)
+                while self.is_power(a, b) or a == b:
+                    a = random.randint(self.min, self.max)
+                    b = random.randint(self.min, self.max)
+                return {
+                    "type": "compute",
+                    "expression": [a, b],
+                    "log_expr": f"\\log_{{{b}}}{{{a}}} + \\log_{{{a}}}{{{b}}}",
+                    "is_expressible": False
+                }
+            
+            if self.allow_multilevel and random.random() < 0.5:
+                a, base, frac = generate_expressible()
+                m = random.randint(1,4)
+                c = base ** m
+                return {
+                    "type": "compute",
+                    "expression": [a, base, c],
+                    "target": str(Fraction(m**2 +1, m)),
+                    "is_expressible": True
+                }
+            else:
+                a, b, frac = generate_expressible()
+                return {
+                    "type": "compute",
+                    "expression": [a, b],
+                    "target": f"{frac.numerator}/{frac.denominator}",
+                    "is_expressible": True
+                }
+                
+        else:  # solve类型完整实现
+            # 生成单层求解案例
+            position = random.choice([0, 1])
+            X = random.randint(self.min, self.max)
+            k = random.randint(1,4)
+            if position == 0:
+                b = X ** k
+                equation = ['X', b]
+            else:
+                a = X ** k
+                equation = [a, 'X']
+            target = Fraction(k**2 +1, k)
+            return {
+                "type": "solve",
+                "equation": equation,
+                "target": f"{target.numerator}/{target.denominator}",
+                "solution": X
+            }
+
+    @staticmethod
+    def prompt_func(question_case):
+        definition = """a￠b=\log_{b}{a}+\log_{a}{b}.
+a and b are positive integers.
+"""
+        if question_case['type'] == 'compute':
+            expr = '￠'.join(map(str, question_case['expression']))
+            prompt = f"Compute {expr}.\n"
+            if not question_case.get('is_expressible', True):
+                prompt += "If the answer cannot be reduced to an integer or fraction then retain the form \\log_{b}{a}+\\log_{a}{b}.\n"
+                prompt += "Please provide your answer in LaTeX format. "
+            else:
+                prompt += "If the answer is a fraction, write it in 'a/b' text format. Decimals are not allowed.\n"
+            prompt += "Please wrap the answer in double square brackets, like this: [[your answer]]."
+            return definition + prompt
+        else:
+            equation_str = '￠'.join(
+                [str(x) if x != 'X' else 'X' for x in question_case['equation']]
+            )
+            prompt = f"If {equation_str} = {question_case['target']}, find X.\n"
+            prompt += "The answer should only be given as a number.\n"
+            prompt += "Please wrap the answer in double square brackets, like this: [[your answer]]."
+            return definition + prompt
+
+    @staticmethod
+    def extract_output(output):
+        matches = re.findall(r'\[\[(.*?)\]\]', output)
+        return matches[-1] if matches else None
+
+    @classmethod
+    def _verify_correction(cls, solution, identity):
+        try:
+            if identity['type'] == 'compute':
+                if identity.get('is_expressible', True):
+                    return Fraction(solution) == Fraction(identity['target'])
+                else:
+                    given = re.sub(r'\s+', '', solution)
+                    expected = re.sub(r'\s+', '', identity['log_expr'])
+                    return given == expected
+            else:
+                return int(solution) == identity['solution']
+        except:
+            return False