Question

我有以下lstm模型：

class LSTM_model():

def __init__(self):
    w2v_model = gensim.models.Word2Vec(sentences, size=150, window=10, min_count=2, workers=10)
    pretrained_weights = w2v_model.wv.syn0
    vocab_size, emdedding_size = pretrained_weights.shape
    self.w2v_model = w2v_model
    self.keras_lstm_model = Sequential()
    self.keras_lstm_model.add(Embedding(input_dim = vocab_size, output_dim = emdedding_size, weights = [pretrained_weights]))
    self.keras_lstm_model.add(LSTM(units = emdedding_size))
    self.keras_lstm_model.add(Dense(units = vocab_size))
    self.keras_lstm_model.add(Activation('sigmoid'))
    self.keras_lstm_model.compile(optimizer = 'adam', loss = 'sparse_categorical_crossentropy', metrics = ['mae','acc'])

我的目的是预测上下文中给定单词的概率。我有句子列表，我想训练这个模型：

   def train_lstm_model(self, sentences):

        sentences_as_indexes = []


        filtered_sentences = list(filter(lambda x : all([w in self.w2v_model.wv.vocab for w in x]) , sentences)) #filter to take only sentences with no OOV words
        for sentence in filtered_sentences:
            if all([w in self.w2v_model.wv.vocab for w in sentence]): #todo use filter
                indexes_row = []
                for word in sentence:

                    idx = self.w2v_model.wv.vocab.get(word).index
                    indexes_row.append(idx)

                sentences_as_indexes.append(indexes_row)
        X = pd.DataFrame([sentence[:-1] for sentence in sentences_as_indexes])
        y = pd.DataFrame([sentence[-1] for sentence in sentences_as_indexes])
        print(datetime.datetime.now(), ": Fitting LSTM model , size of X is ", X.shape)

        self.keras_lstm_model.fit(X, y) #HERE the error

这里的句子是大约一百万个句子的列表（我的训练集）。我在fit（）中收到以下错误：

tensorflow.python.framework.errors_impl.InvalidArgumentError：索引[7,38] = -2147483648不在[0，694415）中 [[节点：embedding_1 / embedding_lookup = GatherV2 [Taxis = DT_INT32，Tindices = DT_INT32，Tparams = DT_FLOAT，_class = [“ loc：@ training / Adam / Assign_2”]，_ device =“ / job：localhost / replica：0 / task ：0 /设备：CPU：0“]（嵌入_1 /嵌入/读取，嵌入_1 /投放，训练/亚当/渐变/嵌入/嵌入_1 /嵌入_lookup_grad / concat /轴）]]

我虽然可能是因为我拥有的数据是OOV，但是如您所见-我过滤了句子以仅包含词汇中所有单词的句子。

什么可能导致此错误？

谢谢！

tensorflow.python.framework.errors_impl.InvalidArgumentError：在KERAS LSTM模型中

0 个答案: