search

tenstorrent / tt-metal

:metal: TT-NN operator library, and TT-Metalium low level kernel programming model.
Apache License 2.0
456 stars 67 forks source link

ttnn conv2d: Fails in Blazepose model #10349

Open Sudharsan-V opened 3 months ago

Sudharsan-V commented 3 months ago

Describe the bug While working on Blazepose model, we observed that ttnn conv2d fails to produce result for the configuration attached below.

To Reproduce Steps to reproduce the behavior:

  1. pick the test from the commit
  2. Run pytest <path to file>

Expected behavior As the above shared configuration works in pytorch, it is expected to run in ttnn aswell.

Error


E           RuntimeError: TT_THROW @ ../ttnn/cpp/ttnn/operations/matmul/matmul.cpp:60: tt::exception
E           info:
E           ttnn.matmul: The width of the first tensor must be equal to the height of the second tensor
E           backtrace:
E            --- ttnn::operations::matmul::matmul(tt::tt_metal::Tensor const&, tt::tt_metal::Tensor const&, std::__1::optional<tt::tt_metal::Tensor const> const&, tt::operations::primary::Matmul const&)
Sudharsan-V commented 3 months ago

cc: @boris-drazic