[GitHub] [tvm] petuca commented on issue #10651: [Bug][ONNX] MatMul in dense_alter_op

[GitBox <gi...@apache.org>]

petuca commented on issue #10651:
URL: https://github.com/apache/tvm/issues/10651#issuecomment-1112248595

   I successfully managed to reproduce the issue just like you.
   I tried to use your ideas for solving this problem. First I tried to put the unsqueeze() in code in dense_alter_op.py file before the breaking part. Nothing happened and the problem is the same just like in the beginning (also tried the squeeze() for inversly solving). 
   The variable 'tinfos' has the two tensor arrays of dimensions [3, ] and [1,3]. Debugging step by step concluded that variable 'tinfos' doesn't assign any value to the destination variables. Tried to separate arrays and assign them with multiple approaches without success. Also attempt to transform the arrays without success.
   I have the same approach to the different models (ONNX and PyTorch) and the both of two models have the same results. I think that the problem is much deeper then simple problem of unsqueezing in dense_alter_op.py file. 
   
   Do you have some ideas how to solve this in another way?
   
   CC: @masahi @ganler 


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org