search

kundajelab / deeplift

Public facing deeplift repo
MIT License
830 stars 164 forks source link

KeyError: 'zeropadding2d' #119

Open hoda213 opened 3 years ago

hoda213 commented 3 years ago

Hello, I am trying to apply deeplift on my image dataset with densenet121 model but I get error: my code is:

import deeplift from deeplift.layers import NonlinearMxtsMode from deeplift.conversion import kerasapi_conversion as kc from tensorflow.keras.layers import ZeroPadding2D,BatchNormalization,Conv2D,Concatenate saved_model_file='model.h5'

Three different models, one each for RevealCancel, Gradient and GuidedBackprop revealcancel_model = kc.convert_model_from_saved_files( h5_file=saved_model_file, nonlinear_mxts_mode=NonlinearMxtsMode.RevealCancel) grad_model = kc.convert_model_from_saved_files( h5_file=saved_model_file, nonlinear_mxts_mode=NonlinearMxtsMode.Gradient) guided_backprop_model = kc.convert_model_from_saved_files( h5_file=saved_model_file, nonlinear_mxts_mode=NonlinearMxtsMode.GuidedBackprop)

and my error is: KeyError: 'zeropadding2d'

KeyError Traceback (most recent call last)

in () 14 guided_backprop_model = kc.convert_model_from_saved_files( 15 h5_file=saved_model_file, ---> 16 nonlinear_mxts_mode=NonlinearMxtsMode.GuidedBackprop) 3 frames /usr/local/lib/python3.6/dist-packages/deeplift/conversion/kerasapi_conversion.py in convert_model_from_saved_files(h5_file, json_file, yaml_file, **kwargs) 411 layer_config["config"]["weights"] = layer_weights 412 --> 413 return model_conversion_function(model_config=model_config, **kwargs) 414 415 /usr/local/lib/python3.6/dist-packages/deeplift/conversion/kerasapi_conversion.py in convert_functional_model(model_config, nonlinear_mxts_mode, verbose, dense_mxts_mode, conv_mxts_mode, maxpool_deeplift_mode, layer_overrides, custom_conversion_funcs) 835 maxpool_deeplift_mode=maxpool_deeplift_mode, 836 layer_overrides=layer_overrides, --> 837 custom_conversion_funcs=custom_conversion_funcs) 838 839 for output_layer in converted_model_container.output_layers: /usr/local/lib/python3.6/dist-packages/deeplift/conversion/kerasapi_conversion.py in functional_container_conversion(config, name, verbose, nonlinear_mxts_mode, dense_mxts_mode, conv_mxts_mode, maxpool_deeplift_mode, layer_overrides, custom_conversion_funcs, outer_inbound_node_infos, node_id_to_deeplift_layers, node_id_to_input_node_info, name_to_deeplift_layer) 588 else: 589 conversion_function = layer_name_to_conversion_function( --> 590 layer_config["class_name"]) 591 592 #We need to deal with the case of shared layers, i.e. the same /usr/local/lib/python3.6/dist-packages/deeplift/conversion/kerasapi_conversion.py in layer_name_to_conversion_function(layer_name) 345 # lowercase to create resistance to capitalization changes 346 # was a problem with previous Keras versions --> 347 return name_dict[layer_name.lower()] 348 349 I would be grateful if you have any idea to solve it
NGD1L commented 7 months ago

Hello，i meet this trouble too！Do you solve it?