point paper links to arxiv

Author: Moshe Looks

README.md

@@ -7,7 +7,7 @@ the input data. For example, [this model](tensorflow_fold/g3doc/sentiment.ipynb)
 implements [TreeLSTMs](https://arxiv.org/abs/1503.00075) for sentiment analysis
 on parse trees of arbitrary shape/size/depth.
 
-Fold implements [*dynamic batching*](https://openreview.net/pdf?id=ryrGawqex).
+Fold implements [*dynamic batching*](https://arxiv.org/abs/1702.02181).
 Batches of arbitrarily shaped computation graphs are transformed to produce a
 static computation graph. This graph has the same structure regardless of what
 input it receives, and can be executed efficiently by TensorFlow.

tensorflow_fold/g3doc/index.md

@@ -21,7 +21,7 @@ typically used to run models like tree-RNNs. When the input consists of trees
 size and shape. A standard TensorFlow model consists of a fixed graph of
 operations, which cannot accommodate variable-shaped data. Fold overcomes this
 limitation by using
-the [dynamic batching algorithm](https://openreview.net/pdf?id=ryrGawqex).
+the [dynamic batching algorithm](https://arxiv.org/abs/1702.02181).
 
 Fold consists of a high-level API called Blocks, and a low-level API called
 Loom. Blocks are pure Python, whereas Loom is a mixture of Python and

tensorflow_fold/g3doc/quick.ipynb

@@ -499,7 +499,7 @@
  "source": [
  "The `reduce_net_block` function creates a block (`net_block` that contains a two-layer fully connected (FC) network that takes a pair of scalar tensors as input and produces a scalar tensor as output. This network gets applied in a binary tree to reduce a sequence of scalar tensors to a single scalar tensor.\n",
  "\n",
- "One thing to notice here is that we are calling [`tf.squeeze`](https://www.tensorflow.org/versions/r1.0/api_docs/python/array_ops/shapes_and_shaping#squeeze) with `axis=1`, even though the Fold output type of `td.FC(1, activation=None)` (and hence the input type of the enclosing `Function` block) is a `TensorType` with shape `(1)`). This is because all Fold blocks actually run on TF tensors with an implicit leading batch dimension, which enables execution via [*dynamic batching*](https://openreview.net/pdf?id=ryrGawqex). It is important to bear this in mind when creating `Function` blocks that wrap functions that are not applied elementwise."
+ "One thing to notice here is that we are calling [`tf.squeeze`](https://www.tensorflow.org/versions/r1.0/api_docs/python/array_ops/shapes_and_shaping#squeeze) with `axis=1`, even though the Fold output type of `td.FC(1, activation=None)` (and hence the input type of the enclosing `Function` block) is a `TensorType` with shape `(1)`). This is because all Fold blocks actually run on TF tensors with an implicit leading batch dimension, which enables execution via [*dynamic batching*](https://arxiv.org/abs/1702.02181). It is important to bear this in mind when creating `Function` blocks that wrap functions that are not applied elementwise."
  ]
  },
  {

tensorflow_fold/g3doc/sentiment.ipynb

@@ -1160,7 +1160,7 @@
  "cell_type": "markdown",
  "metadata": {},
  "source": [
- "Not bad! See section 3.5.1 of [our paper](https://openreview.net/pdf?id=ryrGawqex) for discussion and a comparison of these results to the state of the art."
+ "Not bad! See section 3.5.1 of [our paper](https://arxiv.org/abs/1702.02181) for discussion and a comparison of these results to the state of the art."
  ]
  }
  ],