[Bug] [FRONTEND][ONNX] fails to convert a valid onnx model since the Less operator cannot broadcast for valid inputs #18030
Description
Expected behavior
The onnx frontend should import the model correctly.
Actual behavior
For the following model, the onnx frontend cannot import it.
tvm_model = from_onnx(model, keep_params_in_input=True)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/carla/Documents/tvm/python/tvm/relax/frontend/onnx/onnx_frontend.py", line 3695, in from_onnx
return g.from_onnx(graph, opset)
^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/carla/Documents/tvm/python/tvm/relax/frontend/onnx/onnx_frontend.py", line 3326, in from_onnx
self._construct_nodes(graph)
File "/home/carla/Documents/tvm/python/tvm/relax/frontend/onnx/onnx_frontend.py", line 3506, in _construct_nodes
raise err
File "/home/carla/Documents/tvm/python/tvm/relax/frontend/onnx/onnx_frontend.py", line 3503, in _construct_nodes
op = self.bb.normalize(op)
^^^^^^^^^^^^^^^^^^^^^
File "/home/carla/Documents/tvm/python/tvm/relax/block_builder.py", line 667, in normalize
return _ffi_api.BlockBuilderNormalize(self, expr) # type: ignore
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "tvm/ffi/cython/./function.pxi", line 228, in tvm.ffi.core.Function.__call__
tvm.error.InternalError: In Op(relax.less), the first input shape at dim 0 is T.int64(3) and the second input shape at dim 3 is T.int64(5), which are not broadcastable.
[17:27:21] /home/carla/Documents/tvm/src/relax/ir/block_builder.cc:64: Warning: BlockBuilder destroyed with remaining blocks!The Less operator supports multidirectional (i.e., Numpy-style) broadcasting. Hence, the inputs A and B should be broadcastable.
I also using numpy to verify that A and B is broadcastable for the less operation, the results are as follows, which has the same shapes in the model:
Environment
OS: Ubuntu 20.04
TVM: 0.21.dev0(eca92bd)
Steps to reproduce
This bug can be reproduced by the following code with the model in the attachment. As shown in the code, the model can be executed by onnxruntime, which indicates that the model is valid.
import sys
import numpy as np
import onnx
import onnxruntime
import tvm
from tvm import relax
from tvm.relax.frontend.onnx import from_onnx
import pickle
def test():
onnx_model = onnx.load("11.onnx")
with open("inputs.pkl", "rb") as fp:
inputs = pickle.load(fp)
try:
ort_session = onnxruntime.InferenceSession(
onnx_model.SerializeToString(), providers=["CPUExecutionProvider"]
)
ort_output = ort_session.run([], inputs)
except Exception as e:
print(e)
sys.exit(1)
print("ONNXRuntime:\n", ort_output)
tvm_model = from_onnx(onnx_model)
if __name__ == "__main__":
test()Triage
- needs-triage