GH-61: used sklearn

flair/visual/__init__.py

@@ -0,0 +1 @@
+from .tsne import tSNE

flair/visual/tsne.py

@@ -0,0 +1,31 @@
+from sklearn.manifold import TSNE
+import tqdm
+import numpy
+
+
+class tSNE:
+    def __init__(self, embeddings):
+
+        self.embeddings = embeddings
+
+        self.transform = \
+            TSNE(n_components=2, verbose=1, perplexity=40, n_iter=300)
+
+    def _prepare(self, sentences):
+
+        X = []
+
+        print('computing embeddings')
+        for sentence in tqdm.tqdm(sentences):
+            self.embeddings.embed(sentence)
+
+            for token in sentence:
+                X.append(token.embedding.detach().numpy()[None, :])
+
+        X = numpy.concatenate(X, 0)
+
+        return X
+
+    def fit(self, sentences):
+
+        return self.transform.fit_transform(self._prepare(sentences))