public dataset split

[Ginven-W]

Hi authors,

Thanks for your wonderful work on PathDiff! I am trying to reproduce the experimental results of the PathDiff model using your released pretrained weights on three public datasets: PanNuke, MoNuSAC, and CoNIC.

However, there is a considerable performance gap between my reproduced results and the numbers reported in your paper. I suspect this discrepancy originates from differences in dataset train‑test splitting strategies.

Could you please share the official split information for PanNuke, MoNuSAC, and CoNIC? Thank you very much for your help.

[bhosalems]

Pretrained weights model is for CONIC dataset only. I will share the train/validation/test splits in some time.