Hi authors, thanks for the great work on PersonaMem — this is a very insightful benchmark for studying dynamic user profiling and personalization in long-context settings.
I have a question regarding the baseline experiments reported in the paper. In addition to evaluating various LLMs, the paper also compares against several memory / personalization baselines (e.g., retrieval-based or external memory methods).
I was wondering:
Do you plan to release the evaluation code or scripts used to test these baseline systems on PersonaMem?
If full code release is not possible, would it be feasible to share implementation details, pseudo-code, or configuration settings that would help reproduce the baseline results?
Thanks again for the excellent work, and looking forward to your response!
Best regards
Hi authors, thanks for the great work on PersonaMem — this is a very insightful benchmark for studying dynamic user profiling and personalization in long-context settings.
I have a question regarding the baseline experiments reported in the paper. In addition to evaluating various LLMs, the paper also compares against several memory / personalization baselines (e.g., retrieval-based or external memory methods).
I was wondering:
Do you plan to release the evaluation code or scripts used to test these baseline systems on PersonaMem?
If full code release is not possible, would it be feasible to share implementation details, pseudo-code, or configuration settings that would help reproduce the baseline results?
Thanks again for the excellent work, and looking forward to your response!
Best regards