获取具有可变序列长度的激活时的Tensorflow GRU单元错误

时间:2016-04-29 23:33:17

标签: python python-2.7 tensorflow recurrent-neural-network gated-recurrent-unit

我想在一些时间序列数据上运行GRU单元格,以根据最后一层中的激活对它们进行聚类。我对GRU单元实现做了一个小改动

def __call__(self, inputs, state, scope=None):
"""Gated recurrent unit (GRU) with nunits cells."""
with vs.variable_scope(scope or type(self).__name__):  # "GRUCell"
  with vs.variable_scope("Gates"):  # Reset gate and update gate.
    # We start with bias of 1.0 to not reset and not update.
    r, u = array_ops.split(1, 2, linear([inputs, state], 2 * self._num_units, True, 1.0))
    r, u = sigmoid(r), sigmoid(u)
  with vs.variable_scope("Candidate"):
    c = tanh(linear([inputs, r * state], self._num_units, True))
  new_h = u * state + (1 - u) * c

  # store the activations, everything else is the same
  self.activations = [r,u,c]
return new_h, new_h

在此之后,我按照以下方式连接激活,然后在调用此GRU单元格的脚本中返回它们

@property
def activations(self):
    return self._activations


@activations.setter
def activations(self, activations_array):
    print "PRINT THIS"         
    concactivations = tf.concat(concat_dim=0, values=activations_array, name='concat_activations')
    self._activations = tf.reshape(tensor=concactivations, shape=[-1], name='flatten_activations')

我以下列方式调用GRU单元

outputs, state = rnn.rnn(cell=cell, inputs=x, initial_state=initial_state, sequence_length=s)

其中s是一个批处理长度数组,其中包含输入批处理的每个元素中的时间戳数。

最后我使用

获取
fetched = sess.run(fetches=cell.activations, feed_dict=feed_dict)

执行时我收到以下错误

追踪(最近一次通话):   文件“xxx.py”,第162行,in     fetched = sess.run(fetches = cell.activations,feed_dict = feed_dict)   文件“/xxx/local/lib/python2.7/site-packages/tensorflow/python/client/session.py”,第315行,在运行中     return self._run(None,fetches,feed_dict)   在_run中输入文件“/xxx/local/lib/python2.7/site-packages/tensorflow/python/client/session.py”,第511行     feed_dict_string)   在_do_run中输入文件“/xxx/local/lib/python2.7/site-packages/tensorflow/python/client/session.py”,第564行     target_list)   在_do_call中输入文件“/xxx/local/lib/python2.7/site-packages/tensorflow/python/client/session.py”,第588行     six.reraise(e_type,e_value,e_traceback)   在_do_call中输入文件“/xxx/local/lib/python2.7/site-packages/tensorflow/python/client/session.py”,第571行     return fn(* args)   在_run_fn中输入文件“/xxx/local/lib/python2.7/site-packages/tensorflow/python/client/session.py”,第555行

返回tf_session.TF_Run(session,feed_dict,fetch_list,target_list) tensorflow.python.pywrap_tensorflow.StatusNotOK:无效参数:为RNN / cond_396 / ClusterableGRUCell / flatten_activations返回的张量:0无效。

有人可以通过传递可变长度序列来了解如何在最后一步从GRU单元格获取激活吗?感谢。

1 个答案:

答案 0 :(得分:0)

要从最后一步获取激活,您希望激活成为您所在州的一部分,由tf.rnn返回。