Update README.md

README.md

@@ -1,2 +1,2 @@
 # fairseq-rl
-Modified version of fairseq, including new implementations for criterions using reinforcement learning methods. The implementation of  two versions of RL criterions are in [Version 1](https://github.com/TianchunH97/fairseq-rl/blob/master/fairseq-rl/fairseq/criterions/v1.py) and [Version 2](https://github.com/TianchunH97/fairseq-rl/blob/master/fairseq-rl/fairseq/criterions/v2.py), respectively.
+Modified version of fairseq, including new implementations for criterions using reinforcement learning methods. The implementation of  two versions of RL criterions are in [Version 1](https://github.com/TianchunH97/fairseq-rl/blob/master/fairseq-rl/fairseq/criterions/v1.py) and [Version 2](https://github.com/TianchunH97/fairseq-rl/blob/master/fairseq-rl/fairseq/criterions/v2.py), respectively. Some experiments have been run with the new criterions. The results and conclusions are reported [here](https://github.com/TianchunH97/fairseq-rl/blob/master/A_Study_of_RL_loss.pdf)