Open levsnv opened 3 years ago
@levsnv I now see that the bug remains in the model builder API.
If, in FIL, we decide to convert empty categorical nodes to dummy numerical nodes (#4314 #4315), this bug will not longer be exercised by FIL and will not require the workaround. Whether that's still a model builder API issue or not is up to you.
@levsnv I'll still try to fix this bug.
This issue has been labeled inactive-30d
due to no recent activity in the past 30 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed. This issue will be labeled inactive-90d
if there is no activity in the next 60 days.
This issue has been labeled inactive-90d
due to no recent activity in the past 90 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed.
Describe the bug Although fixes have been merged into cuml for Treelite to handle empty categorical nodes, the fix doesn't seem to work in cuml. After committing model using model builder C++ API, it crashes with "nid out of range".
Steps/Code to reproduce bug Once https://github.com/rapidsai/cuml/pull/4173 is merged, one will be able to reproduce the issue on branch-21.10.
A good CI run: https://gpuci.gpuopenanalytics.com/job/rapidsai/job/gpuci/job/cuml/job/prb/job/cuml-gpu-test/CUDA=11.2,GPU_LABEL=gpu,LINUX_VER=ubuntu18.04,PYTHON=3.8/1402/ A bad CI run: https://gpuci.gpuopenanalytics.com/job/rapidsai/job/gpuci/job/cuml/job/prb/job/cuml-gpu-test/CUDA=11.2,GPU_LABEL=gpu,LINUX_VER=ubuntu18.04,PYTHON=3.8/1403/ Difference:
Expected behavior A clear and concise description of what you expected to happen.
Environment details (please complete the following information):
long version: