From 6ea100d658052602bef791db2b14aba7f99fc6fa Mon Sep 17 00:00:00 2001
From: YeonwooSung <neos960518@gmail.com>
Date: Sun, 16 Feb 2025 18:07:20 +0900
Subject: [PATCH] fix: Fix up regex str for better filtering

---
 LLMs/training/train_grpo.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/LLMs/training/train_grpo.py b/LLMs/training/train_grpo.py
index 0d844e0..819d796 100644
--- a/LLMs/training/train_grpo.py
+++ b/LLMs/training/train_grpo.py
@@ -77,7 +77,8 @@ def strict_format_reward_func(completions, **kwargs) -> list[float]:
 
 def soft_format_reward_func(completions, **kwargs) -> list[float]:
     """Reward function that checks if the completion has a specific format."""
-    pattern = r"<reasoning>.*?</reasoning>\s*<answer>.*?</answer>"
+    #pattern = r"<reasoning>.*?</reasoning>\s*<answer>.*?</answer>"
+    pattern = r"<reasoning>[\s\S]*</reasoning>\s*<answer>.*?</answer>"
     responses = [completion[0]["content"] for completion in completions]
     matches = [re.match(pattern, r, flags=re.DOTALL) for r in responses] 
     return [0.5 if match else 0.0 for match in matches]