Enable GPU fit and CPU transform in UMAP
#6032
Conversation
divyegala
added
feature request
python/cuml/cuml/manifold/umap.pyx
Outdated
| state['_knn_dists'], state['_knn_indices'], state['_knn_search_index'] = \ | ||
| nearest_neighbors(host_raw_data, self.n_neighbors, self.metric, | ||
| self.metric_kwds, False, self.random_state) |
There was a problem hiding this comment.
I think it should be good, but if there's a possibility to avoid running the KNN, that would be ideal.
python/cuml/cuml/manifold/umap.pyx
Outdated
| @property | ||
| def _n_neighbors(self): | ||
| return self.n_neighbors | ||
|
|
||
| @property | ||
| def _a(self): | ||
| return self.a | ||
|
|
||
| @property | ||
| def _b(self): | ||
| return self.b | ||
|
|
||
| @property | ||
| def _initial_alpha(self): | ||
| return self.learning_rate | ||
|
|
||
| @property | ||
| def _disconnection_distance(self): | ||
| return DISCONNECTION_DISTANCES.get(self.metric, np.inf) |
There was a problem hiding this comment.
Aren't setters required to transfer attributes from umaplib to cuML?
There was a problem hiding this comment.
Hmm, I'm not sure how it was working before then. We never had matching name conventions with them, look at _a and a for example. Is it possible we did not support cpu train/gpu infer before?
There was a problem hiding this comment.
Actually, I think they are transfered during instantiation through hyperparameters listed in get_param_names. I do not know if we want to make these attributes writable.
There was a problem hiding this comment.
Okay, so we'll skip the setters then?
There was a problem hiding this comment.
Actually I'm facing test errors now that these properties are part of get_attr_names. I'll need to add setters
|
/merge |
No description provided.