style(nyz): polish flake8 style

Author: PaParaZz1

ding/reward_model/__init__.py

@@ -15,4 +15,4 @@
 from .icm_reward_model import ICMRewardModel
 # LLM/VLM reward model and verifier
 from .math_reward_model import MathRewardModel
-from .math_rule_reward_model import MathRuleRewardModel
+from .math_rule_reward_model import MathRuleRewardModel

ding/reward_model/math_reward_model.py

@@ -42,4 +42,4 @@ def collect_data(self, data: list) -> None:
         pass
 
     def clear_data(self) -> None:
-        pass
+        pass

ding/reward_model/math_rule_reward_model.py

@@ -41,7 +41,8 @@ def estimate(self, data: List[str]) -> List[Dict]:
         """
         # 1. parse the query to get question and predicted answer
         # 2. get the ground truth answer according to the question
-        # 3. calculate the reward based on the predicted answer and the ground truth answer (format error -2, answer error -1, correct 1)
+        # 3. calculate the reward based on the predicted answer and the ground truth answer
+        # (format error -2, answer error -1, correct 1)
         pass
 
     # rule-based reward model does not need training, thus the following methods are empty