float123 opened a new issue #17892: How to get the output of the specified layer？ URL: https://github.com/apache/incubator-mxnet/issues/17892 Hi， I need to get the output of a certain layer, I did the following： ``` sym, arg_params, aux_params = mx.model.load_checkpoint('./data/retina', 0) print('sym', sym) # [face_rpn_cls_prob_reshape_stride32, face_rpn_bbox_pred_stride32, face_rpn_landmark_pred_stride32,.......] data = mx.sym.Variable('data') sym1 = mx.sym.Variable('face_rpn_cls_prob_reshape_stride32') group = data + sym1 group= group.get_internals() mod = mx.mod.Module(symbol=group, context=mx.gpu(0), label_names=None) mod.bind(data_shapes=[('data', (1, 3, 600, 600))], for_training=False) mod.set_params(arg_params, aux_params) mod.forward(img, is_train=False) net_out = mod.get_outputs() ``` RuntimeError: face_rpn_cls_prob_reshape_stride32 is not presented or ``` data = mx.sym.Variable('data') sym1 = mx.sym.Variable('face_rpn_cls_prob_reshape_stride32') sym1 = sym1.get_internals() group = mx.symbol.Group([data, sym1]) mod = mx.mod.Module(symbol=group, context=mx.gpu(0), label_names=None) ``` RuntimeError: simple_bind error. Arguments: data: (1, 3, 600, 600) [23:49:46] src/executor/../common/exec_utils.h:392: InferShape pass cannot decide shapes for the following arguments (-1 means unknown dimensions). Please consider providing them as inputs: face_rpn_cls_prob_reshape_stride32: None or ``` sym1 = mx.sym.Variable('face_rpn_cls_prob_reshape_stride32') group= sym1.get_internals() mod = mx.mod.Module(symbol=group, context=mx.gpu(0), label_names=None) ``` ValueError: You created Module with Module(..., data_names=['data']) but input with name 'data' is not found in symbol.list_arguments(). Did you mean one of: face_rpn_cls_prob_reshape_stride32 I always encounter these problems when using .get_internals () and mx.symbol.Group (),Using the output face_rpn_cls_prob_reshape_stride32_output will also have an error. how do I need to do it? thank you very much.
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to 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
