Update data_eval.py

This commit is contained in:
chywang
2025-07-18 16:23:52 +08:00
committed by GitHub
parent cddeb27960
commit cd0a9c656f

View File

@@ -40,7 +40,7 @@ def build_cot_prompts(instruction, output):
"6-7 Extensive verbosity with comprehensive justification and exploration of complex connections.\n"
"8-9 High verbosity with deep, exhaustive exploration of reasoning; involves extensive elaboration, nested justifications, "
"and consideration of counterarguments or alternative perspectives.\n"
"Given Problem, Answer with hain-of-Thought, you will:\n"
"Given Problem, Answer with Chain-of-Thought, you will:\n"
"1. Analyze the Reasoning Verbosity\n"
"2. Determine score using the above criteria\n"
"3. Output ONLY the integer score (0-9), place your score in <score></score>\n"
@@ -59,7 +59,7 @@ def build_cot_prompts(instruction, output):
"4-5 Early-undergraduate logic/algebra; one non-obvious insight.\n"
"6-7 Advanced undergraduate techniques (determinants, dynamic programming, layered code reasoning, etc).\n"
"8-9 Graduate-level abstraction, nested proofs, intricate algorithmic analysis.\n"
"Given Problem, Answer with hain-of-Thought, you will:\n"
"Given Problem, Answer with Chain-of-Thought, you will:\n"
"1. Analyze the Cognitive Difficulty\n"
"2. Determine score using the above criteria\n"
"3. Output ONLY the integer score (0-9), place your score in <score></score>\n"
@@ -69,7 +69,7 @@ def build_cot_prompts(instruction, output):
lc_prompt_template = (
"You are a rigorous logical validator analyzing problem-solving components. "
"Your task is to separately assess the validity of the reasoning process and final solution. "
"Given Problem, Answer with hain-of-Thought, you will:\n"
"Given Problem, Answer with Chain-of-Thought, you will:\n"
"1. Verify stepwise logical coherence and soundness\n"
"2. Confirm all critical problem constraints are properly addressed\n"
"3. Check for self-contradictions or unsupported leaps in logic\n"
@@ -265,4 +265,4 @@ def main():
if __name__ == "__main__":
main()
main()