NVIDIA-Merlin / systems

Merlin Systems provides tools for combining recommendation models with other elements of production recommender systems (like feature stores, nearest neighbor search, and exploration strategies) into end-to-end recommendation pipelines that can be served with Triton Inference Server.
Apache License 2.0
90 stars 30 forks source link

[TMP] - Debug Implicit Operator #357

Closed oliverholworthy closed 1 year ago

oliverholworthy commented 1 year ago

Placeholder PR for debugging the implicit test that is failing with out-of-memory on GPU

github-actions[bot] commented 1 year ago

Documentation preview

https://nvidia-merlin.github.io/systems/review/pr-357

oliverholworthy commented 1 year ago

Tried splitting the parametrizeed test out into separate tests for each combination to check if it was something to do with pytest parametrize not cleaning up after each test run. It turns out to still fail like this, so it seems there is some memory being managed in implicit possibly that is not being cleaned up automatically.