Updates to README.md and framework.md.

PiperOrigin-RevId: 266432540

Author: Neural-Link Team

README.md

@@ -8,15 +8,14 @@ Structure can be explicit as represented by a graph [1,2,5] or implicit as
 induced by adversarial perturbation [3,4].
 
 Structured signals are commonly used to represent relations or similarity
-among samples that may be labeled or unlabeled. Therefore, leveraging these
-signals during neural network training harnesses both labeled and unlabeled
-data, which can improve model accuracy, particularly when **the amount of labeled
-data is relatively small**. Additionally, models trained with samples that are
-generated by adding adversarial perturbation have been shown to be
-**robust against malicious attacks**, which are designed to mislead a model's
-prediction or classification.
-
-NSL generalizes to Neural Graph Learning [1] as well as Adversarial
+among samples that may be labeled or unlabeled. Leveraging these signals during
+neural network training harnesses both labeled and unlabeled data, which can
+improve model accuracy, particularly when **the amount of labeled data is
+relatively small**. Additionally, models trained with samples that are generated
+by adversarial perturbation have been shown to be **robust against malicious
+attacks**, which are designed to mislead a model's prediction or classification.
+
+NSL generalizes to Neural Graph Learning [1] as well as to Adversarial
 Learning [3]. The NSL framework in TensorFlow provides the following easy-to-use
 APIs and tools for developers to train models with structured signals:
 
@@ -29,12 +28,11 @@ APIs and tools for developers to train models with structured signals:
 The NSL framework is designed to be flexible and can be used to train any kind
 of neural network. For example, feed-forward, convolution, and recurrent neural
 networks can all be trained using the NSL framework. In addition to supervised
-and semi-supervised learning (low amount of supervision), NSL can also be
-generalized to unsupervised learning. Furthermore, incorporating structure
-is done only during training; there is no change to the serving/inference
-workflow. As a result, no additional cost (latency, memory consumption, etc)
-because of neural structured learning is incurred during serving. Please visit
-our tutorials for a practical introduction to NSL.
+and semi-supervised learning (a low amount of supervision), NSL can in theory be
+generalized to unsupervised learning. Incorporating structured signals is done
+only during training, so the performance of the serving/inference workflow
+remains unchanged. Please check out our tutorials for a practical introduction
+to NSL.
 
 ## Getting started
 
@@ -80,7 +78,7 @@ for tracking requests and bugs. For questions, please direct them to [Stack Over
 ["nsl"](https://stackoverflow.com/questions/tagged/nsl)
 tag.
 
-## Reference
+## References
 
 [[1] T. Bui, S. Ravi and V. Ramavajjala. "Neural Graph Learning: Training Neural Networks Using Graphs." WSDM 2018](https://ai.google/research/pubs/pub46568.pdf)
 

g3doc/tutorials/framework.md

@@ -80,15 +80,15 @@ To obtain hands-on experience with Neural Structured Learning, we have three
 tutorials that cover various scenarios where structured signals may be
 explicitly given, induced or constructed:
 
-*   [Graph regularization for document classification using natural graphs](graph_keras_mlp_cora#top_of_page).
+*   [Graph regularization for document classification using natural graphs](graph_keras_mlp_cora.ipynb).
     In this tutorial, we explore the use of graph regularization to classify
     documents that form a natural (organic) graph.
 
-*   [Graph regularization for sentiment classification using synthesized graphs](graph_keras_lstm_imdb#top_of_page).
+*   [Graph regularization for sentiment classification using synthesized graphs](graph_keras_lstm_imdb.ipynb).
     In this tutorial, we demonstrate the use of graph regularization to classify
     movie review sentiments by constructing (synthesizing) structured signals.
 
-*   [Adversarial learning for image classification](adversarial_keras_cnn_mnist#top_of_page).
+*   [Adversarial learning for image classification](adversarial_keras_cnn_mnist.ipynb).
     In this tutorial, we explore the use of adversarial learning (where
     structured signals are induced) to classify images containing numeric
     digits.