[英]How to trigger a python function inside a tf.keras custom loss function?
在我的自定义丢失函数中,我需要调用一个传递计算出的TD错误和一些索引的纯python函数。 该函数不需要返回任何内容或进行区分。 这是我想要调用的函数:
def update_priorities(self, traces_idxs, td_errors):
"""Updates the priorities of the traces with specified indexes."""
self.priorities[traces_idxs] = td_errors + eps
我已经尝试使用tf.py_function
来调用包装器函数,但只有在它嵌入图形中时才会被调用,即如果它有输入和输出并且使用了输出。 因此,我试图通过一些张量而不对它们执行任何操作,现在函数被调用。 这是我的整个自定义丢失函数:
def masked_q_loss(data, y_pred):
"""Computes the MSE between the Q-values of the actions that were taken and the cumulative
discounted rewards obtained after taking those actions. Updates trace priorities.
"""
action_batch, target_qvals, traces_idxs = data[:,0], data[:,1], data[:,2]
seq = tf.cast(tf.range(0, tf.shape(action_batch)[0]), tf.int32)
action_idxs = tf.transpose(tf.stack([seq, tf.cast(action_batch, tf.int32)]))
qvals = tf.gather_nd(y_pred, action_idxs)
def update_priorities(_qvals, _target_qvals, _traces_idxs):
"""Computes the TD error and updates memory priorities."""
td_error = _target_qvals - _qvals
_traces_idxs = tf.cast(_traces_idxs, tf.int32)
mem.update_priorities(_traces_idxs, td_error)
return _qvals
qvals = tf.py_function(func=update_priorities, inp=[qvals, target_qvals, traces_idxs], Tout=[tf.float32])
return tf.keras.losses.mse(qvals, target_qvals)
但是由于调用mem.update_priorities(_traces_idxs, td_error)
我收到以下错误
ValueError: An operation has `None` for gradient. Please make sure that all of your ops have a gradient defined (i.e. are differentiable). Common ops without gradient: K.argmax, K.round, K.eval.
我不需要为update_priorities
计算渐变,我只想在图形计算中的特定点调用它而忘记它。 我怎样才能做到这一点?
在包装函数内的张量上使用.numpy()
修复了问题:
def update_priorities(_qvals, _target_qvals, _traces_idxs):
"""Computes the TD error and updates memory priorities."""
td_error = np.abs((_target_qvals - _qvals).numpy())
_traces_idxs = (tf.cast(_traces_idxs, tf.int32)).numpy()
mem.update_priorities(_traces_idxs, td_error)
return _qvals
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.