Hello CRMArena Authors,
Thank you for releasing this benchmark—its high fidelity and realistic object relationships are genuinely impressive.
I am currently creating a benchmark to evaluate privacy-preserving capabilities of LLMs in collaborative scenarios. I noticed in Figure 8 of your paper that your sandbox environment uses an elaborately designed hierarchical data generation pipeline. However, I wasn't able to find reproducibility instructions or source code related to this pipeline.
Would it be possible to share technical details (or source code) regarding the sandbox environment and data generation process? Access to these resources would greatly expedite our research
Best regards,
Wenjie