max engine leaking?
Apple M2 Max 96G
Was trying to run a text2text onnx model, but ram usage just keeps rising (see video).
I am using https://github.com/bloomberg/memray as a profiler. Seems like it is the engine specifically
GitHub
GitHub - bloomberg/memray: Memray is a memory profiler for Python
Memray is a memory profiler for Python. Contribute to bloomberg/memray development by creating an account on GitHub.
2 Replies
Can anyone please take a look? I am probably doing something wrong or profiling the wrong thing
Hey, could you create an issue and fill in the details https://github.com/modularml/max/issues?
GitHub
Issues · modularml/max
A collection of sample programs, notebooks, and tools which highlight the power of the MAX Platform - Issues · modularml/max