diff --git a/README.md b/README.md
index 3306c2345ea6c59e89820cdc15eb6dd011bf1cf1..acb8d279da45bed8307ae19d6fabb301e0ad2df2 100755
--- a/README.md
+++ b/README.md
@@ -9,7 +9,7 @@ The code of TUS is in `convlab2/policy/tus` and a rule-based DST of user is also
 ### Train the user simulator
 `python3 convlab2/policy/tus/multiwoz/train.py --user_config convlab2/policy/tus/multiwoz/exp/default.json`
 
-One default configuration is placed in `convlab2/policy/tus/multiwoz/exp/default.json`
+One default configuration is placed in `convlab2/policy/tus/multiwoz/exp/default.json`. They can be modified based on your requirements. For example, the output directory can be specified in the configuration (`model_dir`).
 
 ### Train a dialogue policy with TUS
 You can use it as a normal user simulator by `PipelineAgent`. For example,
@@ -19,8 +19,9 @@ from convlab2.dialog_agent.agent import PipelineAgent
 from convlab2.dst.rule.multiwoz.usr_dst import UserRuleDST
 from convlab2.policy.tus.multiwoz.TUS import UserPolicy
 
+user_config_file = "convlab2/policy/tus/multiwoz/exp/default.json"
 dst_usr = UserRuleDST()
-user_config = json.load(open(args.user_config))
+user_config = json.load(open(user_config_file))
 policy_usr = UserPolicy(user_config)
 simulator = PipelineAgent(None, dst_usr, policy_usr, None, 'user')
 ```