Update incorrect data processing in DataCollatorForChatML

[ruijunfeng]

What does this PR do?

Fix the extra BOS token and the absence of an EOS token in the returned input_ids, and potentially the absence of a target string in the returned labels.

Fixes #2169

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you read the contributor guideline,
  Pull Request section?
• Was this discussed/approved via a GitHub issue? Please add a link
  to it if that's the case. (Incorrect data processing in DataCollatorForChatML #2169)
• Did you make sure to update the documentation with your changes? Here are the
  documentation guidelines.
• Did you write any new necessary tests?

Who can review?

Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.

[kashif]

awesome @ruijunfeng can we also have a test for this?

[ruijunfeng]

awesome @ruijunfeng can we also have a test for this?

Sure thing, I have tested it on the instruct-tuned version of Llama2 series and gemma1 series with my own dataset, and it seems to work well. Let me know if you need me to provide anything.