Some problems in topi.nn.matmul() code

[YuChenHAHAHA]

First of all I want to thank you for opening this course.
When I did assignment 1, I found a problem in "tvm.topi.nn.matmul(tensor_a, tensor_b)".
The script defined the shape of tensor_b is [out_dim, in_dim].

However, in the function "matmul()" in "dense.py" in TVM (line 83), the definition of tensor_b's shape is contrary.
This will result in an error on line 86.

Thank you again for everything you’ve done.

[jinhongyii]

There may be something wrong with the document. If you take a look at the python implementation of this function, you will find that transpose_b = True corresponds to tensor_b's input shape [out_dim, in_dim], while transpose_b = False corresponds to tensor_b's input shape [in_dim, out_dim]. So, in your case, you should pass transpose_b = True as argument.

By the way, there's another API tvm.topi.nn.dense which is a wrapper for tvm.topi.nn.matmul. It is equivalent to tvm.topi.nn.matmul with transpose_a = False and transpose_b = True. So you can also use it here.