def _set_doc_stats(X, kwargs):
model = s_TfidfTransformer(**kwargs)
# Below is only required if we have to set stats
if model.use_idf:
model._set_doc_stats(X)
return model
def __init__(self, *, client=None, verbose=False, **kwargs):
"""
Create new distributed TF-IDF transformer instance
Parameters
-----------
client : dask.distributed.Client optional Dask client to use
"""
super().__init__(client=client, verbose=verbose, **kwargs)
self.datatype = "cupy"
# Make any potential model args available and catch any potential
# ValueErrors before distributed training begins.
self._set_internal_model(s_TfidfTransformer(**kwargs))