Update README.md

README.md

@@ -46,3 +46,15 @@ warp = WarpLJP(args=args)
 for data in dataset:
   x,_,y = warp.processing_single(data)
 ```
+
+## code
+We have provided a class of Monte Carlo tree sampling-based inference methods as baselines in the code, which can be directly invoked via scripts:
+```python
+sh test.sh
+```
+If fine-tuning a reward model is required, execute:
+```python
+sh train.sh
+```
+The reward model in inference can be loaded by adding --reward_model_path to the command.
+