Ean Garvey
Ean Garvey
This is resolved by https://github.com/nod-ai/SHARK/pull/199.
Yes, we have found that the problem was not with IREE.
#169 Just need to add to pytorch tests now.
Is this still necessary now that we have generate_sharktank? Perhaps we can add options to generate_sharktank to save the command line string for iree-compile if we want equivalent functionalities between...
until patch is merged checkout branch ean-bench to reproduce
perhaps the solution to this will be keeping default shark_args.num_iterations = 1 and increasing only for benchmarks.
``` Traceback (most recent call last): File "/home/ean/SHARK/generate_sharktank.py", line 222, in save_torch_model(args.torch_model_csv) File "/home/ean/SHARK/generate_sharktank.py", line 75, in save_torch_model mlir_importer.import_debug( File "/home/ean/SHARK/shark/shark_importer.py", line 163, in import_debug imported_mlir = self.import_mlir( File "/home/ean/SHARK/shark/shark_importer.py",...
We are experiencing the same error with mpnet on Vulkan: ``` E AssertionError: E Not equal to tolerance rtol=0.01, atol=0.001 E E Mismatched elements: 237596 / 3907456 (6.08%) E Max...
This looks great. Chef's kiss 👨🍳🤌
Sounds good. I can look into this.