Phil

Results 7 comments of Phil

TabPFNClassifier causes segmentation faults

Thanks for the detailed investigation and for finding a workaround, really appreciate it! We realize that at the moment the model is quite memory hungry and activation tensor allocations depend...

TabPFNClassifier causes segmentation faults

I've seen segfaults before as well, so while not common it's also not uncommon :) Back then all I could find was "that can have many reasons", like the ones...

added an adaptive batching mechanism to handle large test sets based on estimated memory usage

Closing due to inactivity.

Parse model version from model_path

Note: https://github.com/PriorLabs/TabPFN-private/pull/227 Also needs to be on the server before we can switch to the gated version (otherwise the checkpoint cleaned for deployment fails on load)

Parse model version from model_path

Ah sorry, the linked PR is already the copy over from the public repo, so cutting over should be good!

segfault - memory violation

Thanks for reporting! Would you mind trying the workaround suggested in https://github.com/PriorLabs/TabPFN/issues/404 of calling `torch.cuda.empty_cache()` before each predict call and reporting back if this fixes things?

segfault - memory violation

Ah, sorry, I missed that :) Could you share which dtype you were using?