[Help Wanted] Problems encountered when converting the onnx model to tvm.relay

[Zeus1116]

Excuse me. I encountered a series of errors while converting the onnx model generated by nnsmith to tvm.relay. I would like to ask if this is a bug caused by an internal error in TVM or an error in the model generated by NNSmith.



The command I am using is "tvm.relay.frontend.onxx.from_onnx(onnx_model)".

[Zeus1116]

Meanwhile, when running these models using nnsmith's exec module, the same error will also be recorded.

[ganler]

Together with #136, and similar to #126, -- if you are posting an issue here, I'd assume you are assuming this bug is caused by NNSmith. Yet, because the error happens in the deep call stack of TVM (e.g., codegen) -- I am actually not sure if this is anywhere caused by NNSmith. :)

I can look into this sometime later but it'd be good for you to first localize the who the bug belongs to. As it can also be very likely a TVM bug, your installation error (e.g., inadequate dependency built for TVM to support this instruction), etc. This pipeline used to work very well with NNSmith, so I doubt if this bug comes from NNSmith.

[Zeus1116]

Together with #136, and similar to #126, -- if you are posting an issue here, I'd assume you are assuming this bug is caused by NNSmith. Yet, because the error happens in the deep call stack of TVM (e.g., codegen) -- I am actually not sure if this is anywhere caused by NNSmith. :)

I can look into this sometime later but it'd be good for you to first localize the who the bug belongs to. As it can also be very likely a TVM bug, your installation error (e.g., inadequate dependency built for TVM to support this instruction), etc. This pipeline used to work very well with NNSmith, so I doubt if this bug comes from NNSmith.

Thank you for your reply, I will try to locate the error.