小编这次要给大家分享的是keras加载lstm+crf模型出错怎么办，文章内容丰富，感兴趣的小伙伴可以来了解一下，希望大家阅读完这篇文章之后能够有所收获。

错误展示

new_model = load_model(“model.h6”)

报错：

1、keras load_model valueError: Unknown Layer :CRF

2、keras load_model valueError: Unknown loss function:crf_loss

错误修改

1、load_model修改源码：custom_objects = None 改为 def load_model(filepath, custom_objects, compile=True)：

2、new_model = load_model(“model.h6”,custom_objects={‘CRF': CRF,‘crf_loss': crf_loss,‘crf_viterbi_accuracy': crf_viterbi_accuracy}

以上修改后，即可运行。

Code Example：

# coding: utf-8from keras.models import Sequentialfrom keras.layers import Embeddingfrom keras.layers import LSTMfrom keras.layers import Bidirectionalfrom keras.layers import Densefrom keras.layers import TimeDistributedfrom keras.layers import Dropoutfrom keras_contrib.layers.crf import CRFfrom keras_contrib.utils import save_load_utilsVOCAB_SIZE = 2500EMBEDDING_OUT_DIM = 128TIME_STAMPS = 100HIDDEN_UNITS = 200DROPOUT_RATE = 0.3NUM_CLASS = 5def build_embedding_bilstm2_crf_model(): """ 带embedding的双向LSTM + crf """ model = Sequential() model.add(Embedding(VOCAB_SIZE, output_dim=EMBEDDING_OUT_DIM, input_length=TIME_STAMPS)) model.add(Bidirectional(LSTM(HIDDEN_UNITS, return_sequences=True))) model.add(Dropout(DROPOUT_RATE)) model.add(Bidirectional(LSTM(HIDDEN_UNITS, return_sequences=True))) model.add(Dropout(DROPOUT_RATE)) model.add(TimeDistributed(Dense(NUM_CLASS))) crf_layer = CRF(NUM_CLASS) model.add(crf_layer) model.compile('rmsprop', loss=crf_layer.loss_function, metrics=[crf_layer.accuracy]) return modeldef save_embedding_bilstm2_crf_model(model, filename): save_load_utils.save_all_weights(model,filename)def load_embedding_bilstm2_crf_model(filename): model = build_embedding_bilstm2_crf_model() save_load_utils.load_all_weights(model, filename) return modelif __name__ == '__main__': model = build_embedding_bilstm2_crf_model()

注意：

如果执行build模型报错，则很可能是keras版本的问题。在keras-contrib==2.0.8且keras==2.0.8时，上面代码不会报错。

看完这篇关于keras加载lstm+crf模型出错怎么办的文章，如果觉得文章内容写得不错的话，可以把它分享出去给更多人看到。