Pushing the docs to dev/ for branch: master, commit 125a54d9440567afe77ed45f57be0fa7edaa74c3

Author: sklearn-ci

dev/_downloads/3409d9766d352cc9f9b169d4a799a87a/auto_examples_python.zip

@@ -4,22 +4,23 @@
 =====================================
 
 This example presents how to chain KNeighborsTransformer and TSNE in a
-pipeline, and how to wrap the packages `annoy` and `nmslib` to replace
-KNeighborsTransformer and perform approximate nearest neighbors.
-These package can be installed with `pip install annoy nmslib`.
+pipeline. It also shows how to wrap the packages `annoy` and `nmslib` to
+replace KNeighborsTransformer and perform approximate nearest neighbors.
+These packages can be installed with `pip install annoy nmslib`.
 
-Note: Currently TSNE(metric='precomputed') does not modify the precomputed
+Note: Currently `TSNE(metric='precomputed')` does not modify the precomputed
 distances, and thus assumes that precomputed euclidean distances are squared.
 In future versions, a parameter in TSNE will control the optional squaring of
 precomputed distances (see #12401).
 
-Note: In :class:`KNeighborsTransformer` we use the definition which includes
-each training point as its own neighbor in the count of `n_neighbors`, and for
+Note: In KNeighborsTransformer we use the definition which includes each
+training point as its own neighbor in the count of `n_neighbors`, and for
 compatibility reasons, one extra neighbor is computed when
 `mode == 'distance'`. Please note that we do the same in the proposed wrappers.
 
 Sample output:
 
+```
 Benchmarking on MNIST_2000:
 ---------------------------
 AnnoyTransformer:                    0.583 sec
@@ -39,6 +40,7 @@
 TSNE with NMSlibTransformer:         43.295 sec
 TSNE with KNeighborsTransformer:     64.845 sec
 TSNE with internal NearestNeighbors: 64.984 sec
+```
 """
 # Author: Tom Dupre la Tour
 #

dev/_downloads/76064e174f651530bf4f9c3043b45bad/approximate_nearest_neighbors.py

@@ -15,7 +15,7 @@
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "\n# Approximate nearest neighbors in TSNE\n\n\nThis example presents how to chain KNeighborsTransformer and TSNE in a\npipeline, and how to wrap the packages `annoy` and `nmslib` to replace\nKNeighborsTransformer and perform approximate nearest neighbors.\nThese package can be installed with `pip install annoy nmslib`.\n\nNote: Currently TSNE(metric='precomputed') does not modify the precomputed\ndistances, and thus assumes that precomputed euclidean distances are squared.\nIn future versions, a parameter in TSNE will control the optional squaring of\nprecomputed distances (see #12401).\n\nNote: In :class:`KNeighborsTransformer` we use the definition which includes\neach training point as its own neighbor in the count of `n_neighbors`, and for\ncompatibility reasons, one extra neighbor is computed when\n`mode == 'distance'`. Please note that we do the same in the proposed wrappers.\n\nSample output:\n\nBenchmarking on MNIST_2000:\n---------------------------\nAnnoyTransformer:                    0.583 sec\nNMSlibTransformer:                   0.321 sec\nKNeighborsTransformer:               1.225 sec\nTSNE with AnnoyTransformer:          4.903 sec\nTSNE with NMSlibTransformer:         5.009 sec\nTSNE with KNeighborsTransformer:     6.210 sec\nTSNE with internal NearestNeighbors: 6.365 sec\n\nBenchmarking on MNIST_10000:\n----------------------------\nAnnoyTransformer:                    4.457 sec\nNMSlibTransformer:                   2.080 sec\nKNeighborsTransformer:               30.680 sec\nTSNE with AnnoyTransformer:          30.225 sec\nTSNE with NMSlibTransformer:         43.295 sec\nTSNE with KNeighborsTransformer:     64.845 sec\nTSNE with internal NearestNeighbors: 64.984 sec\n"
+        "\n# Approximate nearest neighbors in TSNE\n\n\nThis example presents how to chain KNeighborsTransformer and TSNE in a\npipeline. It also shows how to wrap the packages `annoy` and `nmslib` to\nreplace KNeighborsTransformer and perform approximate nearest neighbors.\nThese packages can be installed with `pip install annoy nmslib`.\n\nNote: Currently `TSNE(metric='precomputed')` does not modify the precomputed\ndistances, and thus assumes that precomputed euclidean distances are squared.\nIn future versions, a parameter in TSNE will control the optional squaring of\nprecomputed distances (see #12401).\n\nNote: In KNeighborsTransformer we use the definition which includes each\ntraining point as its own neighbor in the count of `n_neighbors`, and for\ncompatibility reasons, one extra neighbor is computed when\n`mode == 'distance'`. Please note that we do the same in the proposed wrappers.\n\nSample output:\n\n```\nBenchmarking on MNIST_2000:\n---------------------------\nAnnoyTransformer:                    0.583 sec\nNMSlibTransformer:                   0.321 sec\nKNeighborsTransformer:               1.225 sec\nTSNE with AnnoyTransformer:          4.903 sec\nTSNE with NMSlibTransformer:         5.009 sec\nTSNE with KNeighborsTransformer:     6.210 sec\nTSNE with internal NearestNeighbors: 6.365 sec\n\nBenchmarking on MNIST_10000:\n----------------------------\nAnnoyTransformer:                    4.457 sec\nNMSlibTransformer:                   2.080 sec\nKNeighborsTransformer:               30.680 sec\nTSNE with AnnoyTransformer:          30.225 sec\nTSNE with NMSlibTransformer:         43.295 sec\nTSNE with KNeighborsTransformer:     64.845 sec\nTSNE with internal NearestNeighbors: 64.984 sec\n```\n"
       ]
     },
     {