search

tenstorrent / tt-metal

:metal: TT-NN operator library, and TT-Metalium low level kernel programming model.
Apache License 2.0
421 stars 54 forks source link

'ModuleNotFoundError' #9882

Closed colchocor closed 2 months ago

colchocor commented 3 months ago

I'm working through trying to test out the Grayskull e150 card and I'm running into a ModuleNotFoundError. I'm trying to run the Resnet Demo in tt-metal. Let me know what I can do. Thanks.

Image

https://github.com/tenstorrent/tt-metal/tree/main/models/demos/resnet

cfjchu commented 3 months ago

@tt-aho can you take a look?

TT-BrianLiu commented 3 months ago

Your missing the python path. Here are some common exports:

export TT_METAL_ENV=dev && export TT_METAL_HOME=$PWD && export ARCH_NAME=grayskull && export PYTHONPATH=$TT_METAL_HOME && source build/python_env/bin/activate
tt-aho commented 3 months ago

Hey @colchocor, have you set up the environment and env vars as described in https://github.com/tenstorrent/tt-metal/blob/main/INSTALLING.md, specifically TT_METAL_HOME and PYTHONPATH should be the tt-metal root directory.

I don't see the module not found error but we've deprecated support for batch 8 with this implementation of Resnet, so the demo will fail for a different issue. I am working on uplifting this to work with our latest batch 20 implementation

tt-aho commented 3 months ago

I've updated the demo and associated readme with new instructions/commands on latest main commit 3f6dab2dc4c3eba3380533f68fd209768807691f. Let me know if you still see any issues running the demo.

ayerofieiev-tt commented 2 months ago

Closing this ticket for inactivity. Please reopen if this issue is still not resolved. Thank you!