why loss=dh*h？

As the scalablegcn code shows：

```
def _optimize_store(self, node, node_embeddings):
if not self.gradient_stores:
return tf.zeros([]), tf.no_op()

losses = []
clear_ops = []
for gradient_store, node_embedding in zip(
self.gradient_stores, node_embeddings):
embedding_gradient = tf.nn.embedding_lookup(gradient_store, node)
with tf.control_dependencies([embedding_gradient]):
clear_ops.append(
utils_embedding.embedding_update(gradient_store, node,
tf.zeros_like(embedding_gradient)))
losses.append(tf.reduce_sum(node_embedding * embedding_gradient))

store_loss = tf.add_n(losses)
with tf.control_dependencies(clear_ops):
return store_loss, self.store_optimizer.minimize(store_loss)
```

Start by reading the _optimize_store method in the scalablegcn code shown and trace how gradient_store and node_embeddings are produced and consumed. Done means providing a clear, evidence-based explanation of the loss expression and documenting it if the project’s conventions support that; no test or file path is identified in the issue.

Written by the indexing model from the issue text.