diff --git a/.gitignore b/.gitignore
index 4ca6f0c5f..f2ab678fb 100644
--- a/.gitignore
+++ b/.gitignore
@@ -37,6 +37,7 @@ docs/auto_examples
 docs/examples/graph*.*
 docs/examples/*.onnx
 docs/examples/pipeline*.dot*
+docs/sg_execution_times.rst
 tests/TESTDUMP
 tests/tests_dump
 tests/graph.dot*
diff --git a/docs/conf.py b/docs/conf.py
index f0f9f1f31..41613d290 100644
--- a/docs/conf.py
+++ b/docs/conf.py
@@ -117,8 +117,9 @@
     "numpy": "https://numpy.org/",
     "onnx": "https://github.com/onnx/onnx",
     "ONNX": "https://onnx.ai/",
-    "ONNX operators": "https://github.com/onnx/onnx/blob/master/docs/Operators.md",
-    "ONNX ML operators": "https://github.com/onnx/onnx/blob/master/docs/Operators-ml.md",
+    "ONNX operators": "https://onnx.ai/onnx/operators/",
+    "ONNX ML operators": "https://onnx.ai/onnx/operators/",
+    "ONNX ML Operators": "https://onnx.ai/onnx/operators/",
     "onnxmltools": "https://github.com/onnx/onnxmltools",
     "onnxruntime": "https://microsoft.github.io/onnxruntime/",
     "openmp": "https://en.wikipedia.org/wiki/OpenMP",
@@ -127,6 +128,7 @@
     "pytorch": "https://pytorch.org/",
     "scikit-learn": "https://scikit-learn.org/stable/",
     "skorch": "https://skorch.readthedocs.io/en/stable/",
+    "skl2onnx": "https://github.com/onnx/sklearn-onnx",
     "sklearn-onnx": "https://github.com/onnx/sklearn-onnx",
     "sphinx-gallery": "https://github.com/sphinx-gallery/sphinx-gallery",
     "xgboost": "https://xgboost.readthedocs.io/en/latest/",
diff --git a/docs/examples/plot_backend.py b/docs/examples/plot_backend.py
index 251399c02..3599ff5c4 100644
--- a/docs/examples/plot_backend.py
+++ b/docs/examples/plot_backend.py
@@ -44,7 +44,7 @@
 # Let's use ONNX backend API to test it.
 
 model = onnx.load(name)
-rep = backend.prepare(model, "CPU")
+rep = backend.prepare(model)
 x = np.array(
     [[-1.0, -2.0, 5.0, 6.0], [-1.0, -2.0, -3.0, -4.0], [-1.0, -2.0, 7.0, 8.0]],
     dtype=np.float32,
@@ -62,7 +62,7 @@
 # The backend can also directly load the model
 # without using *onnx*.
 
-rep = backend.prepare(name, "CPU")
+rep = backend.prepare(name)
 x = np.array(
     [[-1.0, -2.0, -3.0, -4.0], [-1.0, -2.0, -3.0, -4.0], [-1.0, -2.0, -3.0, -4.0]],
     dtype=np.float32,
diff --git a/docs/examples/plot_pipeline_xgboost.py b/docs/examples/plot_pipeline_xgboost.py
index 69e4ef4db..272369cc2 100644
--- a/docs/examples/plot_pipeline_xgboost.py
+++ b/docs/examples/plot_pipeline_xgboost.py
@@ -2,8 +2,6 @@
 
 
 """
-.. _example-xgboost:
-
 Convert a pipeline with a XGBoost model
 ========================================
 
diff --git a/docs/index.rst b/docs/index.rst
index 466185e70..e2a5906fe 100644
--- a/docs/index.rst
+++ b/docs/index.rst
@@ -56,7 +56,8 @@ onnx nodes.
 .. runpython::
     :showcode:
     
-    from skl2onnx import __max_supported_opset__
+    from skl2onnx import __max_supported_opset__, __version__
+    print("documentation for version:", __version__)
     print("Last supported opset:", __max_supported_opset__)
 
 **Backend**
@@ -125,3 +126,7 @@ Shouheng Yi, Shauheen Zahirazami, Yiwen Zhu, Du Li, Xuan Li, Wenbing Li.
 **License**
 
 It is licensed with `Apache License v2.0 <../LICENSE>`_.
+
+**Older versions**
+
+* `1.16.0 <versions/v1.16.0/>`
diff --git a/docs/parameterized.rst b/docs/parameterized.rst
index 6bf236bce..81d0cf0f1 100644
--- a/docs/parameterized.rst
+++ b/docs/parameterized.rst
@@ -27,7 +27,7 @@ produces this part of the graph but there exist two options.
 The first one is using *Scan* operator, the second one is
 using a dedicated operator called *CDist* which is not part
 of the regular ONNX operator until issue
-`2442 <https://github.com/onnx/onnx/issues/2442>`_
+`#2442 <https://github.com/onnx/onnx/issues/2442>`_
 is addressed. By default, *Scan* is used, *CDist* can be used
 by giving:
 
@@ -58,7 +58,7 @@ ZipMap
 
 The operator *ZipMap* produces a list of dictionaries.
 It repeats class names or ids but that's not necessary
-(see issue `2149 <https://github.com/onnx/onnx/issues/2149>`_).
+(see issue `#2149 <https://github.com/onnx/onnx/issues/2149>`_).
 By default, ZipMap operator is added, it can be deactivated by using:
 
 ::
diff --git a/docs/tutorial/plot_gexternal_xgboost.py b/docs/tutorial/plot_gexternal_xgboost.py
index b689c780b..adde4cf15 100644
--- a/docs/tutorial/plot_gexternal_xgboost.py
+++ b/docs/tutorial/plot_gexternal_xgboost.py
@@ -2,8 +2,6 @@
 
 
 """
-.. _example-xgboost:
-
 Convert a pipeline with a XGBoost model
 ========================================
 
diff --git a/skl2onnx/convert.py b/skl2onnx/convert.py
index db8948589..04cdeeee6 100644
--- a/skl2onnx/convert.py
+++ b/skl2onnx/convert.py
@@ -122,8 +122,6 @@ def convert_sklearn(
         in input data and the ONNX graph
         (defined by parameter *initial_types*), only integers are supported.
 
-    .. _l-conv-options:
-
     Converters options
     ++++++++++++++++++
 
diff --git a/skl2onnx/operator_converters/text_vectoriser.py b/skl2onnx/operator_converters/text_vectoriser.py
index 6c26f11fe..2601f4812 100644
--- a/skl2onnx/operator_converters/text_vectoriser.py
+++ b/skl2onnx/operator_converters/text_vectoriser.py
@@ -183,12 +183,6 @@ def convert_sklearn_text_vectorizer(
     Regular expression ``[^\\\\\\\\n]`` is used to split
     a sentance into character (and not works) if ``analyser=='char'``.
     The mode ``analyser=='char_wb'`` is not implemented.
-
-    .. versionchanged:: 1.6
-        Parameters have been renamed: *sep* into *separators*,
-        *regex* into *tokenexp*.
-    ````
-
     """  # noqa
     op = operator.raw_operator