Merge remote-tracking branch 'upstream/master' into import-fixes

Author: otiliastr

g3doc/install.md

@@ -11,7 +11,6 @@ Learning (NSL) in TensorFlow:
     `pip` package manager.
 *   If you have a unique machine configuration,
     [build NSL](#build-the-neural-structured-learning-pip-package) from source.
-*   You can also use [Docker for installing NSL](#using-docker).
 
 ## Install Neural Structured Learning using pip
 
@@ -54,7 +53,7 @@ Note: To exit the virtual environment, run `deactivate`.
 #### 4. (Optional) Test Neural Structured Learning.
 
 <pre class="prettyprint lang-bsh">
-<code class="devsite-terminal tfo-terminal-venv">python -c "import neural_structured_learning as nsl</code>
+<code class="devsite-terminal tfo-terminal-venv">python -c "import neural_structured_learning as nsl"</code>
 </pre>
 
 Success: Neural Structured Learning is now installed.
@@ -105,8 +104,8 @@ Note: To exit the virtual environment, run `deactivate`.
 ### 5. Install Neural Structured Learning dependencies.
 
 <pre class="prettyprint lang-bsh">
-<code class="devsite-terminal">cd neural_structured_learning</code>
-<code class="devsite-terminal tfo-terminal-venv">pip install --requirement "requirements.txt"</code>
+<code class="devsite-terminal">cd neural-structured-learning</code>
+<code class="devsite-terminal tfo-terminal-venv">pip install --requirement neural_structured_learning/requirements.txt</code>
 </pre>
 
 ### 6. (Optional) Unit Test Neural Structured Learning.
@@ -115,51 +114,22 @@ Note: To exit the virtual environment, run `deactivate`.
 <code class="devsite-terminal tfo-terminal-venv">bazel test //neural_structured_learning/...</code>
 </pre>
 
-#### 7. Test Neural Structured Learning.
+### 7. Build the pip package.
 
 <pre class="prettyprint lang-bsh">
-<code class="devsite-terminal tfo-terminal-venv">python -c "import neural_structured_learning as nsl</code>
+<code class="devsite-terminal tfo-terminal-venv">python setup.py bdist_wheel --universal --dist-dir="./wheel"</code>
 </pre>
 
-Success: The Neural Structured Learning package is built.
-
-## Using Docker
-
-Create a Neural Structured Learning development environment using Docker on
-Ubuntu or macOS.
-
-### 1. Install Docker.
-
-[Install Docker](https://docs.docker.com/install/) on your local machine.
-
-### 2. Clone the latest Neural Structured Learning source.
-
-<pre class="prettyprint lang-bsh">
-<code class="devsite-terminal">git clone https://github.com/tensorflow/neural-structured-learning.git</code>
-<code class="devsite-terminal">cd neural_structured_learning</code>
-</pre>
-
-### 3. Build a Docker image.
-
-<pre class="prettyprint lang-bsh">
-<code class="devsite-terminal">docker build . \
-    --tag neural-structured-learning:latest</code>
-</pre>
-
-### 4. Start a Docker container.
+### 8. Install the pip package.
 
 <pre class="prettyprint lang-bsh">
-<code class="devsite-terminal">docker run -it \
-    --workdir /neural_structured_learning \
-    --volume $(pwd):/neural_structured_learning \
-    neural-structured-learning:latest \
-    bash</code>
+<code class="devsite-terminal tfo-terminal-venv">pip install --upgrade ./wheel/neural_structured_learning*.whl</code>
 </pre>
 
-### 5. (Optional) Test Neural Structured Learning.
+### 9. Test Neural Structured Learning.
 
 <pre class="prettyprint lang-bsh">
-<code class="devsite-terminal">bazel test //neural_structured_learning/...</code>
+<code class="devsite-terminal tfo-terminal-venv">python -c "import neural_structured_learning as nsl"</code>
 </pre>
 
-Success: The Neural Structured Learning development environment is ready.
+Success: The Neural Structured Learning package is built.

g3doc/tutorials/graph_keras_lstm_imdb.ipynb

@@ -645,7 +645,7 @@
       "outputs": [],
       "source": [
         "!python -m neural_structured_learning.tools.pack_nbrs \\\n",
-        "--max_nbrs=1 --add_undirected_edges=True \\\n",
+        "--max_nbrs=3 --add_undirected_edges=True \\\n",
         "/tmp/imdb/train_data.tfr '' /tmp/imdb/graph_99.tsv \\\n",
         "/tmp/imdb/nsl_train_data.tfr"
       ]
@@ -743,13 +743,13 @@
         "    ### neural graph learning parameters\n",
         "    self.distance_type = nsl.configs.DistanceType.L2\n",
         "    self.graph_regularization_multiplier = 0.1\n",
-        "    self.num_neighbors = 1\n",
+        "    self.num_neighbors = 2\n",
         "    ### model architecture\n",
         "    self.num_embedding_dims = 16\n",
         "    self.num_lstm_dims = 64\n",
         "    self.num_fc_units = 64\n",
         "    ### training parameters\n",
-        "    self.train_epochs = 4\n",
+        "    self.train_epochs = 10\n",
         "    self.batch_size = 128\n",
         "    ### eval parameters\n",
         "    self.eval_steps = None  # All instances in the test set are evaluated.\n",
@@ -1459,11 +1459,12 @@
         "# Accuracy values for both the Bi-LSTM model and the feed forward NN model have\n",
         "# been precomputed for the following supervision ratios.\n",
         "\n",
-        "supervision_ratios = [0.3, 0.15, 0.05, 0.03, 0.01]\n",
+        "supervision_ratios = [0.3, 0.15, 0.05, 0.03, 0.02, 0.01, 0.005]\n",
         "\n",
         "model_tags = ['Bi-LSTM model', 'Feed Forward NN model']\n",
-        "base_model_accs = [[85, 85, 62, 58, 50], [85, 79, 61, 53, 50]]\n",
-        "graph_reg_model_accs = [[85, 84, 76, 63, 51], [85, 79, 73, 62, 50]]\n",
+        "base_model_accs = [[84, 84, 83, 80, 65, 52, 50], [87, 86, 76, 74, 67, 52, 51]]\n",
+        "graph_reg_model_accs = [[84, 84, 83, 83, 65, 63, 50],\n",
+        "                        [87, 86, 80, 75, 67, 52, 50]]\n",
         "\n",
         "plt.clf()  # clear figure\n",
         "\n",
@@ -1498,12 +1499,12 @@
         "It can be observed that as the superivision ratio decreases, model accuracy also\n",
         "decreases. This is true for both the base model and for the graph-regularized\n",
         "model, regardless of the model architecture used. However, notice that the\n",
-        "graph-regularized model is consistenly better than the base model -- sometimes\n",
-        "by as much as 15% -- and further, as the supervision ratio decreases, the\n",
-        "decrease in accuracy is much less for the graph-regularized model than the base\n",
-        "model. This is primarily because of semi-supervised learning for the\n",
-        "graph-regularized model, where structural similarity among training samples is\n",
-        "used in addition to the training samples themselves."
+        "graph-regularized model performs better than the base model for both the\n",
+        "architectures. In particular, for the Bi-LSTM model, when the supervision ratio\n",
+        "is 0.01, the accuracy of the graph-regularized model is **~20%** higher than\n",
+        "that of the base model. This is primarily because of semi-supervised learning\n",
+        "for the graph-regularized model, where structural similarity among training\n",
+        "samples is used in addition to the training samples themselves."
       ]
     },
     {

neural_structured_learning/tools/build_docs.py

@@ -25,7 +25,7 @@
 
 ```shell
 python build_docs.py \
---out_dir=/tmp/neural_structured_learning_api
+--output_dir=/tmp/neural_structured_learning_api
 ```
 
 Note:
@@ -47,7 +47,7 @@
 from tensorflow_docs.api_generator import generate_lib
 from tensorflow_docs.api_generator import public_api
 
-flags.DEFINE_string("out_dir", "/tmp/neural_structured_learning_api",
+flags.DEFINE_string("output_dir", "/tmp/neural_structured_learning_api",
                     "Where to output the docs")
 flags.DEFINE_string(
     "code_url_prefix",
@@ -79,7 +79,7 @@ def main(_):
       # documented in the location that defines them, instead of every location
       # that imports them.
       callbacks=[public_api.local_definitions_filter])
-  doc_generator.build(output_dir=FLAGS.out_dir)
+  doc_generator.build(output_dir=FLAGS.output_dir)
 
 
 if __name__ == "__main__":