adding comments

python/mxnet/contrib/onnx/mx2onnx/_op_translations.py

@@ -1845,6 +1845,10 @@ def convert_cast(node, **kwargs):
     proc_nodes = kwargs["proc_nodes"]
     inputs = node["inputs"]
     dtype = node["attrs"]["dtype"]
+
+    # dtype can be mapped only with types from TensorProto
+    # float32 is mapped to float and float64 to double in onnx
+    # following tensorproto mapping https://github.com/onnx/onnx/blob/master/onnx/mapping.py
     if dtype == 'float32':
         dtype = 'float'
     elif dtype == 'float64':

python/mxnet/contrib/onnx/onnx2mx/_op_translations.py

@@ -121,6 +121,7 @@ def argmax(attrs, inputs, proto_obj):
     axis = attrs.get('axis', 0)
     keepdims = attrs.get('keepdims', 1)
     argmax_op = symbol.argmax(inputs[0], axis=axis, keepdims=keepdims)
+    # onnx argmax operator always expects int64 as output type
     cast_attrs = {'dtype': 'int64'}
     return 'cast', cast_attrs, argmax_op
 
@@ -129,6 +130,7 @@ def argmin(attrs, inputs, proto_obj):
     axis = attrs.get('axis', 0)
     keepdims = attrs.get('keepdims', 1)
     argmin_op = symbol.argmin(inputs[0], axis=axis, keepdims=keepdims)
+    # onnx argmax operator always expects int64 as output type
     cast_attrs = {'dtype': 'int64'}
     return 'cast', cast_attrs, argmin_op