kahmed10
kahmed10
Thanks for the update. Please close this issue when the documentation gets updated.
@rnaidu02 while this PR still hasn't been merged, can you also fix a typo in https://github.com/mlcommons/inference/blob/master/language/bert/README.md? `The inputs to the SUT are input_ids, input_make, and segment_ids. The output from SUT...
> You can try to set LD_LIBRARY_PATH to /opt/rocm/lib explicitly. Which version of rocm are you using? I found libhsa-runtime64.so.1 under /opt/rocm/hsa/lib, and setting LD_LIBRARY_PATH to that location fixed the...
I would like to see additional documentation and possible refactoring since the parse_gridsample.cpp is quite large. Not blocking for merge.
Looks to be fixed from develop already. This PR adds more tests that are passing.
This gives perf boosts on VAE. I'll see if I can add unit tests to this PR so that it can get merged in.
We may also need to update the gradio app to support more images in the UI. That can be a separate PR if it's more involved.
@richagadgil can you please open a new PR to address the unresolved comments here?
Example test output: ``` MIGRAPHX_TRACE_EVAL=2 ./bin/test_verify_onnx resize_downsample_linear_dyn_test [ RUN ] resize_downsample_linear_dyn_test Run instruction: X = @param:X -> dynamic, half_type, {[ 1, 1, {} ], [ 1, 1, {} ], [...
All that's left is adding the full model in DLM.