R语言进行LSTM递归预测时报错

对未来数据进行递归预测时运行以下代码：

library(keras)
library(tensorflow)
atestPredict<-vector()
for(i in 1:length(atest))
{predict<-model_keras %>% predict(atestXY$dataX[i,],verbose=2)
atestPredict<-cbind(atestPredict,predict)
if(i<=look_back){atestXY$dataX[i+1,(look_back-i+1):look_back]<-atestPredict}
if(i>look_back){atestXY$dataX[i+1,]<-atestPredict[(length(atestPredict)-look_back+1):length(atestPredict)]}}

报错：
Error in py_call_impl(callable, dots$args, dots$keywords) :
ValueError: Error when checking input: expected lstm_6_input to have 3 dimensions, but got array with shape (10, 1)

请问出现维度问题应该怎么修改呢？
更新：
刚才发现是前面模型训练就有问题
library(keras)
model_keras<-keras_model_sequential()
model_keras %>%
layer_lstm(units=20,input_shape=c(1,look_back))%>%
layer_dense(units=10)%>%
compile(loss="mean_squared_error",optimizer="adam")%>%
fit(atrainXY$dataX,atrainXY$dataY,epochs=30,batch_size=10,verbose=2)
这里三维数组输进去就变成了二维😭求帮助解决