update model card

README.md

@@ -1,6 +1,12 @@
 ---
 license: apache-2.0
 pipeline_tag: reinforcement-learning
+datasets:
+- cheryyunl/Make-An-Agent
 ---
 
-Dataset and pretrained models for Make-An-Agent
+Pretrained models for Make-An-Agent. See https://github.com/cheryyunl/Make-An-Agent for usage instructions.
+
+`autoencoder.pt` and `behavior_embedding.pt` are trained with parameter data and trajectory data in `train_data/data.pt` of our dataset.
+
+`model-best.pt` is the policy generator model trained by latent parameter representation data and behavior embedding data in `train_data/process_data.pt`