Update broken docs.tiledb.com links (#90)

Author: gsakkis

examples/models/pytorch_tiledb_models_example.ipynb

@@ -248,7 +248,7 @@
    "metadata": {},
    "source": [
     "The above step will create a TileDB array in your working directory. For information about the structure of a dense\n",
-    "TileDB array in terms of files on disk please take a look [here](https://docs.tiledb.com/main/basic-concepts/data-format).\n",
+    "TileDB array in terms of files on disk please take a look [here](https://docs.tiledb.com/main/concepts/data-format).\n",
     "Let's open our TileDB array model and check metadata. Metadata that are of type list, dict or tuple have been JSON\n",
     "serialized while saving, i.e., we need json.loads to deserialize them."
    ]
@@ -315,7 +315,7 @@
    "metadata": {},
    "source": [
     "For the case of PyTorch models, internally, we save model's state_dict and optimizer's state_dict,\n",
-    "as [variable sized attributes)](https://docs.tiledb.com/main/solutions/tiledb-embedded/api-usage/writing-arrays/var-length-attributes)\n",
+    "as [variable sized attributes)](https://docs.tiledb.com/main/how-to/arrays/writing-arrays/var-length-attributes)\n",
     "(pickled), i.e., we can open the TileDB and get only the state_dict of the model or optimizer,\n",
     "without bringing the whole model in memory. For example, we can load model's and optimizer's state_dict\n",
     "for model tiledb-pytorch-mnist-1 as follows."
@@ -382,7 +382,7 @@
    "metadata": {},
    "source": [
     "What is really nice with saving models as TileDB array, is native versioning based on fragments as described\n",
-    "[here](https://docs.tiledb.com/main/basic-concepts/data-format#immutable-fragments). We can load a model, retrain it\n",
+    "[here](https://docs.tiledb.com/main/concepts/data-format#immutable-fragments). We can load a model, retrain it\n",
     "with new data and update the already existing TileDB model array with new model parameters and metadata. All information, old\n",
     "and new will be there and accessible. This is extremely useful when you retrain with new data or trying different architectures for the same\n",
     "problem, and you want to keep track of all your experiments without having to store different model instances. In our case,\n",
@@ -452,7 +452,7 @@
    "metadata": {},
    "source": [
     "Finally, a very interesting and useful, for machine learning models, TileDB feature that is described\n",
-    "[here](https://docs.tiledb.com/main/basic-concepts/data-format#groups) and [here](https://docs.tiledb.com/main/solutions/tiledb-embedded/api-usage/object-management#creating-tiledb-groups)\n",
+    "[here](https://docs.tiledb.com/main/concepts/data-format#groups) and [here](https://docs.tiledb.com/main/how-to/object-management#creating-tiledb-groups)\n",
     "are groups. Assuming we want to solve the MNIST problem, and we want to try several architectures. We can save each architecture\n",
     "as a separate TileDB array with native versioning each time it is re-trained, and then organise all models that solve the same problem (MNIST)\n",
     "as a TileDB array group with any kind of hierarchy. Let's firstly define a new model architecture."

examples/models/sklearn_tiledb_models_example.ipynb

@@ -129,9 +129,9 @@
    },
    "source": [
     "We can now save the trained model as a TileDB array. For information about the structure of a dense\n",
-    "TileDB array in terms of files on disk please take a look [here](https://docs.tiledb.com/main/basic-concepts/data-format).\n",
+    "TileDB array in terms of files on disk please take a look [here](https://docs.tiledb.com/main/concepts/data-format).\n",
     "At the moment (will change in the future) we use pickle, which is one of the [most common scenarios for sklearn models](https://scikit-learn.org/stable/modules/model_persistence.html),\n",
-    "to serialize the whole model and save it as a [variable sized attribute](https://docs.tiledb.com/main/solutions/tiledb-embedded/api-usage/writing-arrays/var-length-attributes)\n",
+    "to serialize the whole model and save it as a [variable sized attribute](https://docs.tiledb.com/main/how-to/arrays/writing-arrays/var-length-attributes)\n",
     "in a TileDB array.  We first declare a SklearnTileDBModel object (with the corresponding uri and model attributes) and then save the model as a TileDB array.\n",
     "Finally, we can save any kind of metadata (in any structure, i.e., list, tuple or dictionary) by passing a dictionary to the meta attribute."
    ]
@@ -222,7 +222,7 @@
    "source": [
     "Moving on, we can load the trained models for evaluation or retraining, as usual with Sklearn models.\n",
     "What is really nice with saving models as TileDB array, is native versioning based on fragments as described\n",
-    "[here](https://docs.tiledb.com/main/basic-concepts/data-format#immutable-fragments). We can load a model, retrain it\n",
+    "[here](https://docs.tiledb.com/main/concepts/data-format#immutable-fragments). We can load a model, retrain it\n",
     "with new data and update the already existing TileDB model array with new model parameters and metadata. All information, old\n",
     "and new will be there and accessible. This is extremely useful when you retrain with new data or trying different architectures\n",
     "for the same problem, and you want to keep track of all your experiments without having to store different model instances. In our case,\n",
@@ -307,7 +307,7 @@
    "metadata": {},
    "source": [
     "Finally, a very interesting and useful, for machine learning models, TileDB feature that is described\n",
-    "[here](https://docs.tiledb.com/main/basic-concepts/data-format#groups) and [here](https://docs.tiledb.com/main/solutions/tiledb-embedded/api-usage/object-management#creating-tiledb-groups)\n",
+    "[here](https://docs.tiledb.com/main/concepts/data-format#groups) and [here](https://docs.tiledb.com/main/how-to/object-management#creating-tiledb-groups)\n",
     "are groups. Assuming we want to solve the MNIST problem, and we want to try several architectures. We can save each architecture\n",
     "as a separate TileDB array with native versioning each time it is re-trained, and then organise all models that solve the same problem (MNIST)\n",
     "as a TileDB array group with any kind of hierarchy. Let's firstly define a new model architecture, then train a model and save\n",

examples/models/tensorflow_keras_tiledb_models_example.ipynb

@@ -140,7 +140,7 @@
    "metadata": {},
    "source": [
     "The above step will create a TileDB array in your working directory. For information about the structure of a dense\n",
-    "TileDB array in terms of files on disk please take a look [here](https://docs.tiledb.com/main/basic-concepts/data-format).\n",
+    "TileDB array in terms of files on disk please take a look [here](https://docs.tiledb.com/main/concepts/data-format).\n",
     "Let's open our TileDB array model and check metadata. Metadata that are of type list, dict or tuple have been JSON\n",
     "serialized while saving, i.e., we need json.loads to deserialize them."
    ]
@@ -342,7 +342,7 @@
    "metadata": {},
    "source": [
     "What is really nice with saving models as TileDB array, is native versioning based on fragments as described\n",
-    "[here](https://docs.tiledb.com/main/basic-concepts/data-format#immutable-fragments). We can load a model, retrain it\n",
+    "[here](https://docs.tiledb.com/main/concepts/data-format#immutable-fragments). We can load a model, retrain it\n",
     "with new data and update the already existing TileDB model array with new model parameters and metadata. All information, old\n",
     "and new will be there and accessible. This is extremely useful when you retrain with new data or trying different architectures\n",
     "for the same problem, and you want to keep track of all your experiments without having to store different model instances. In our case,\n",
@@ -410,7 +410,7 @@
    "metadata": {},
    "source": [
     "Finally, a very interesting and useful, for machine learning models, TileDB feature that is described\n",
-    "[here](https://docs.tiledb.com/main/basic-concepts/data-format#groups) and [here](https://docs.tiledb.com/main/solutions/tiledb-embedded/api-usage/object-management#creating-tiledb-groups)\n",
+    "[here](https://docs.tiledb.com/main/concepts/data-format#groups) and [here](https://docs.tiledb.com/main/how-to/object-management#creating-tiledb-groups)\n",
     "are groups. Assuming we want to solve the MNIST problem, and we want to try several architectures. We can save each architecture\n",
     "as a separate TileDB array with native versioning each time it is re-trained, and then organise all models that solve the same problem (MNIST)\n",
     "as a TileDB array group with any kind of hierarchy. Let's firstly define a new model architecture."

tiledb/ml/readers/pytorch.py

@@ -41,7 +41,7 @@ def __init__(
         will work, but in case it's not equal the tile extent of the first dimension of
         the TileDB array, you won't achieve highest read speed. For more details on tiles,
         tile extent and indices in TileDB, please check here:
-        https://docs.tiledb.com/main/solutions/tiledb-embedded/performance-tips/choosing-tiling-and-cell-layout#dense-arrays
+        https://docs.tiledb.com/main/how-to/performance/performance-tips/choosing-tiling-and-cell-layout#dense-arrays
 
         :param x_array: Array that contains features.
         :param y_array: Array that contains labels.

tiledb/ml/readers/tensorflow.py

@@ -43,7 +43,7 @@ def __new__(
         will work, but in case it's not equal the tile extent of the first dimension of
         the TileDB array, you won't achieve highest read speed. For more details on tiles,
         tile extent and indices in TileDB, please check here:
-        https://docs.tiledb.com/main/solutions/tiledb-embedded/performance-tips/choosing-tiling-and-cell-layout#dense-arrays
+        https://docs.tiledb.com/main/how-to/performance/performance-tips/choosing-tiling-and-cell-layout#dense-arrays
 
         :param x_array: Array that contains features.
         :param y_array: Array that contains labels.