Hi authors,
I encountered two critical issues when reproducing MedMemoryBench:
- Git LFS download failed with the error:
This repository exceeded its LFS budget. I can not fetch the real dataset files via git lfs fetch/pull, only LFS pointer files are obtained locally.
- When I run the evaluation pipeline using incomplete pointer JSON files, the program throws an error:
AttributeError: 'str' object has no attribute 'get'
File: benchmarks/medmemorybench/dataset.py
Line: timestamp = event_info.get("date") if event_info else None
It is caused by malformed dataset content. The event_info field is a string instead of a dictionary in incomplete files.
Could you please provide the complete data/MedMemoryBench dataset as a zip archive (uploaded to Releases or cloud drive)?
Without the full dataset, I cannot complete the benchmark evaluation.
Thank you very much for your help!
Hi authors,
I encountered two critical issues when reproducing MedMemoryBench:
This repository exceeded its LFS budget. I can not fetch the real dataset files viagit lfs fetch/pull, only LFS pointer files are obtained locally.It is caused by malformed dataset content. The
event_infofield is a string instead of a dictionary in incomplete files.Could you please provide the complete
data/MedMemoryBenchdataset as a zip archive (uploaded to Releases or cloud drive)?Without the full dataset, I cannot complete the benchmark evaluation.
Thank you very much for your help!