For Pooling:
pytorch has " ceil_mode" to specify which value(ceil/floor) to use while calculating output shape.
http://pytorch.org/docs/master/nn.html#maxpool2d

Similarly, MXNET has "pooling_convention" for using ceil or floor for output shape from maxpool.
https://mxnet.incubator.apache.org/api/python/symbol.html#mxnet.symbol.Pooling

But in ONNX pooling , there is no attribute to determine if the input framework used ceil or floor for pooling operations? https://github.com/onnx/onnx/blob/6270bb68c54647f6c9ca56d640a73fe2dec2ad80/docs/Operators.md#MaxPool

Is there a reason ONNX does not have this attribute? How can we determine this ?

@bddppq