本文整理汇总了Python中torch.Tensor.numpy方法的典型用法代码示例。如果您正苦于以下问题：Python Tensor.numpy方法的具体用法？Python Tensor.numpy怎么用？Python Tensor.numpy使用的例子？那么恭喜您, 这里精选的方法代码示例或许可以为您提供帮助。您也可以进一步了解该方法所在类torch.Tensor的用法示例。
在下文中一共展示了Tensor.numpy方法的1个代码示例，这些例子默认根据受欢迎程度排序。您可以为喜欢或者感觉有用的代码点赞，您的评价将有助于系统推荐出更棒的Python代码示例。
示例1: get_predicted_clusters
# 需要导入模块: from torch import Tensor [as 别名]
# 或者: from torch.Tensor import numpy [as 别名]
def get_predicted_clusters(top_spans: torch.Tensor,
antecedent_indices: torch.Tensor,
predicted_antecedents: torch.Tensor) -> Tuple[List[Tuple[Tuple[int, int], ...]],
Dict[Tuple[int, int],
Tuple[Tuple[int, int], ...]]]:
# Pytorch 0.4 introduced scalar tensors, so our calls to tuple() and such below don't
# actually give ints unless we convert to numpy first. So we do that here.
top_spans = top_spans.numpy() # (num_spans, 2)
antecedent_indices = antecedent_indices.numpy() # (num_spans, num_antecedents)
predicted_antecedents = predicted_antecedents.numpy() # (num_spans,)
predicted_clusters_to_ids: Dict[Tuple[int, int], int] = {}
clusters: List[List[Tuple[int, int]]] = []
for i, predicted_antecedent in enumerate(predicted_antecedents):
if predicted_antecedent < 0:
continue
# Find predicted index in the antecedent spans.
predicted_index = antecedent_indices[i, predicted_antecedent]
# Must be a previous span.
assert i > predicted_index
antecedent_span: Tuple[int, int] = tuple(top_spans[predicted_index]) # type: ignore
# Check if we've seen the span before.
if antecedent_span in predicted_clusters_to_ids.keys():
predicted_cluster_id: int = predicted_clusters_to_ids[antecedent_span]
else:
# We start a new cluster.
predicted_cluster_id = len(clusters)
clusters.append([antecedent_span])
predicted_clusters_to_ids[antecedent_span] = predicted_cluster_id
mention: Tuple[int, int] = tuple(top_spans[i]) # type: ignore
clusters[predicted_cluster_id].append(mention)
predicted_clusters_to_ids[mention] = predicted_cluster_id
# finalise the spans and clusters.
final_clusters = [tuple(cluster) for cluster in clusters]
# Return a mapping of each mention to the cluster containing it.
mention_to_cluster: Dict[Tuple[int, int], Tuple[Tuple[int, int], ...]] = {
mention: final_clusters[cluster_id]
for mention, cluster_id in predicted_clusters_to_ids.items()
}
return final_clusters, mention_to_cluster