--- a
+++ b/README.md
@@ -0,0 +1,56 @@
+# NeurIPS 2018: AI for Prosthetics Challenge Solution
+
+A solution based on OpenAI baselines ([repo](https://github.com/openai/baselines)) for 9th place 
+[NeurIPS 2018: AI for Prosthetics Challenge](https://www.crowdai.org/challenges/neurips-2018-ai-for-prosthetics-challenge)
+
+## Table of Contents
+* [Prerequisites](#prerequisites)
+* [Training](#training)
+* [Evaluation](#evaluation)
+* [Contributing](#contributing)
+
+## Prerequisites
+
+First of all, you should follow the challenge environment's instruction to set up the simulator, 
+details described in: 
+https://github.com/stanfordnmbl/osim-rl/  
+```bash
+source activate opensim-rl
+```
+Other dependencies is required as follow:
+- Ray
+- TensorFlow
+- mpi4py, cloudpickle, joblib (required by baselines)
+
+## Training
+
+To train the model from scratch, you need to comment the `--checkpoint-path` in `run.sh`, 
+or type commands as follow:
+```bash
+python -m nips.round2_train \
+    --num-timesteps=10000000 \
+    --num-steps=128 \
+    --num-minibatches=16 \
+    --num-cpus=24 --num-casks=8 \
+    --save-interval=5 \
+    --repeat=2
+```
+The baseline code shall be the minimal for this challenge, 
+and we select PPO ([paper](https://arxiv.org/abs/1707.06347)) as the algorithm.  
+You should specify your available CPUs and GPUs in the bash script. The training will be deployed 
+distributed by Ray ([docs](https://ray.readthedocs.io/en/latest/index.html)).  
+
+## Evaluation
+
+To evaluate your model, store your checkpoints file into `checkpoints/` and change 
+the arguments in script.  
+```bash
+# remember to place your checkpoints in your script file first
+./run.sh
+```
+As the comparison, our model finally get 9813.224 in the Round 2 of challenge.
+
+## Contributing
+
+- [wangzhengfei0730](https://github.com/wangzhengfei0730)
+- [QPHutu](https://github.com/QPHutu)