Data Discrepancies: Inconsistent and Unreasonable Evaluation Results
First and foremost, I would like to express my sincere gratitude for your open-source work. Your contributions to the community are truly appreciated.
I've noticed some inconsistencies in the evaluation data, as shown in the attached image. The results appear to be unreasonable, which leads me to believe there might be a spelling error or perhaps a data pasting mistake in the evaluation process.
I want to emphasize that this post is not meant to criticize or cause any harm. My intention is solely to bring this to your attention for the benefit of the project. If you feel this post could negatively impact your team's work in any way, please feel free to delete it.
Thank you for your time and consideration.
Hello, thank you so much for your interest in our work.
The first image represents the re-ranking of the top-100 search results for bge-en-large, whereas the second image is the re-ranking of the top-100 search results for e5-mistral-7b; thus, their values differ. Because the initial search results from e5-mistral-7b are better, the re-ranking of these results is also superior.