Embedding层
keras.layers.embeddings.Embedding(input_dim, output_dim, embeddings_initializer='uniform', embeddings_regularizer=None, activity_regularizer=None, embeddings_constraint=None, mask_zero=False, input_length=None)
input_dim：大或等于0的整数，字典长度，即输入数据最大下标+1
output_dim：大于0的整数，代表全连接嵌入的维度
形如（samples，sequence_length）的2D张量
形如(samples, sequence_length, output_dim)的3D张量
model = Sequential() model.add(Embedding(1000, output_dim=64, input_length=10)) # input_dim=1000是字典长度 # the model will take as input an integer matrix of size (batch, input_length). # the largest integer (i.e. word index) in the input should be no larger than 999 (vocabulary size). # model.input_shape = (None,10), where None is the batch dimension, 10 is input_length(time_step), input_dim is 1. # model.output_shape = (None, 10, 64), where None is the batch dimension, 10 is output_length, 64 is output_dim. # 32 samples, 10 time steps, input is a number between 0 and 1000 in 2-D array input_array. input_array = np.random.randint(1000, size=(32, 10)) model.compile('rmsprop', 'mse') output_array = model.predict(input_array) assert output_array.shape == (32, 10, 64)Reference
http://keras-cn.readthedocs.io/en/latest/layers/embedding_layer/
