Problem with cloud(cirrus) detection

Hello, I tried to use MAJA4.7.1 to get S2L2A. At the beginning, I tried 50TMP and 47SLD from December 2022 to January 2023, but the clear sky and snow were both identified as clouds and most of them were identified as cirrus clouds. I’ve also started an issue on Github, the feedback they gave me was that the time range I was using had higher reflectance intensity in the 10 band of S2, which triggers the cirrus and cloud detection thresholds over a large area, let me change it from November 2021 or February 2023, but I still have the same problem after changing the time as follows.

The red mask represent cloud (I didn’t change the format of L2A, I just open it in SNAP). What should I do in this case?