MCPcopy
hub / github.com/dmlc/dgl / _bucketing

Function _bucketing

python/dgl/core.py:177–204  ·  view source on GitHub ↗

Internal function to create groups on the values. Parameters ---------- val : Tensor Value tensor. Returns ------- unique_val : Tensor Unique values. bucketor : callable[Tensor -> list[Tensor]] A bucketing function that splits the given tensor da

(val)

Source from the content-addressed store, hash-verified

175
176
177def _bucketing(val):
178 """Internal function to create groups on the values.
179
180 Parameters
181 ----------
182 val : Tensor
183 Value tensor.
184
185 Returns
186 -------
187 unique_val : Tensor
188 Unique values.
189 bucketor : callable[Tensor -> list[Tensor]]
190 A bucketing function that splits the given tensor data as the same
191 way of how the :attr:`val` tensor is grouped.
192 """
193 sorted_val, idx = F.sort_1d(val)
194 unique_val = F.asnumpy(F.unique(sorted_val))
195 bkt_idx = []
196 for v in unique_val:
197 eqidx = F.nonzero_1d(F.equal(sorted_val, v))
198 bkt_idx.append(F.gather_row(idx, eqidx))
199
200 def bucketor(data):
201 bkts = [F.gather_row(data, idx) for idx in bkt_idx]
202 return bkts
203
204 return unique_val, bucketor
205
206
207def data_dict_to_list(graph, data_dict, func, target):

Callers 1

invoke_udf_reduceFunction · 0.85

Calls 2

asnumpyMethod · 0.80
appendMethod · 0.80

Tested by

no test coverage detected