CI pipeline failed. Use the following command to fix this.

pip install black
black .

@Jian1273 checks not passed

@wangzaistone Except the black check, did the submitted code passes through the test? If yes, I think we can first merge it and run black by ourselves later on.

@wangzaistone Except the black check, did the submitted code passes through the test? If yes, I think we can first merge it and run black by ourselves later on.

Yes, But first do CR and tests. @wangzaistone + @junewgl

OK，I'm going to test the code @csunny

OK，I'm going to test the code @csunny

After my testing, the code can run successfully, but it needs to add a data set: oaast_rm_zh.json

After running, the weights will be saved in the output directory, as shown below:

cc @csunny @wangzaistone @qidanrui

Thanks for your contribution @Jian1273 and I have some suggestions：
1.You can consider adding an effect evaluation code for the RM model, such as Microsoft's RLHF code
2.You can take the time to submit the RL training related code to verify the overall RLHF effect.
3.After you submit all the RLHF code, we can test it on the text2sql task together.

I've rerun the CI and the code passes all checks. @junewgl @Jian1273 @csunny @wangzaistone