Is this a assert a bug?

quantization
1

I am using the following code link where it uses the convert_fx from torch.ao.quantization.quantize_fx

However, the following assertion is fired:

_lower_to_native_backend.py", line 606, in _lower_static_weighted_ref_module
    assert(len(ref_node.args) == 1)
AssertionError

When I comment out this assert everthing seems to work fine. The question is if this assert is correct?

2

Hi Ali,

Would you mind sharing two things that would help us debug?

  1. What op is “ref_node” when it fails? (print(ref_node))
  2. What are the args of “ref_node”? (print(ref_node.args))

Thanks,
-Andrew