Add gradient checkpointing papers

doc/intro2.md

@@ -29,6 +29,13 @@ useful for someone to get up-to-date with the NLP research of today.
 - ["Effective Approaches to Attention-based Neural Machine Translation" (Luong
   et al. - sep. 2015)](https://arxiv.org/abs/1508.04025) - one of the other
   Attention papers
+- ["Training Deep Nets with Sublinear Memory Cost" (Chen et al. - apr.
+  2016)](https://arxiv.org/abs/1604.06174) - introduces gradient checkpointing
+  for the first time, showing how to train larger models with little
+  computational overhead
+- ["Memory-Efficient Backpropagation Through Time" (Gruslys et al. - jun.
+  2016)](https://arxiv.org/abs/1606.03401) - improves on gradient checkpointing
+  for RNNs
 - ["Layer Normalization" (Hinton et al. - jul.
   2016)](https://arxiv.org/abs/1607.06450) - the Layer Normalization paper, used
   by the original Transformer architecture