zhreshold commented on a change in pull request #12812: [MXNET-886] ONNX
export: HardSigmoid, Less, Greater, Equal, InstanceNorm
URL: https://github.com/apache/incubator-mxnet/pull/12812#discussion_r225744224
##########
File path: python/mxnet/contrib/onnx/mx2onnx/export_onnx.py
##########
@@ -118,31 +118,39 @@ def forward_pass(inputs, sym, arg_params, aux_params,
output_label):
if graph_input not in arg_params and graph_input not in
aux_params
and graph_input != output_label]
+ data_forward = []
data_shapes = []
# Adding extra dimension of batch_size 1 if the batch_size is
different for multiple inputs.
for idx, input_name in enumerate(data_names):
- data_shapes.append((input_name, inputs[idx].shape))
+ val = inputs[idx]
+ data_shapes.append((input_name, val.shape))
+ data_forward.append(nd.array(val))
# create module, passing cpu context
ctx = context.cpu()
- test_mod = mod.Module(symbol=sym, data_names=data_names, context=ctx,
label_names=None)
- test_mod.bind(for_training=False, data_shapes=data_shapes,
label_shapes=None)
- # initializing parameters for calculating result of each individual
node
- if arg_params is None and aux_params is None:
- test_mod.init_params()
- else:
- test_mod.set_params(arg_params=arg_params, aux_params=aux_params,
allow_missing=True)
+ # module bind method requires all data to have same batch size,
Review comment:
why not use sym.bind whatsoever
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
