Update to proper EOS ids for Qwen2 and Qwen2.5

[joecummings]

Context

What is the purpose of this PR? Is it to

• add a new feature
• fix a bug
• update tests and/or documentation
• other (please add here)

As detailed in #2337, the ending token for Qwen2 and Qwen2.5 should actually be "<|im_end|>", not "<|endoftext|>". This fixes that and updates tests accordingly.

Qwen2 tokenizer source
Qwen2.5 tokenizer source

Changelog

What are the changes made in this PR?

• Change eos token id to be the token id for "<|im_end|>"
• Set UNK id to null b/c that's the same way it is in HF code
• Update tests

Test plan

Please make sure to do each of the following if applicable to your PR. If you're unsure about any one of these just ask and we will happily help. We also have a contributing page for some guidance on contributing.

• run pre-commit hooks and linters (make sure you've first installed via pre-commit install)
• add unit tests for any new functionality
• update docstrings for any new or updated methods or classes
• run unit tests via pytest tests
• run recipe tests via pytest tests -m integration_test
• manually run any new or modified recipes with sufficient proof of correctness
• include relevant commands and any other artifacts in this summary (pastes of loss curves, eval results, etc.)

UX

If your function changed a public API, please add a dummy example of what the user experience will look like when calling it.
Here is a docstring example
and a tutorial example

• I did not change any public API
• I have added an example to docs or docstrings

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/torchtune/2342

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❗ 1 Active SEVs

There are 1 currently active SEVs. If your PR is affected, please view them below:

• ROCM Infra failures during checkout of PyTorch

✅ No Failures

As of commit 6f01c17 with merge base 059cad9 ():
💚 Looks good so far! There are no failures yet. 💚

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[felipemello1]

why did we need this before?

[joecummings]

I don't think we did. It was just incorrectly set.

[felipemello1]

was "self.pad_id" not here also a bug?

[joecummings]

No b/c eos_id was set to the pad_id. So now it's the same values, but proper references.

[felipemello1]

did you have a chance to reproduce the python code detailed in the ticket and make sure the output is correct now? I can approve after that

[joecummings]

did you have a chance to reproduce the python code detailed in the ticket and make sure the output is correct now? I can approve after that

I confirm that when I run the code on the ticket, I get the following output, which is correct according to the code on Hugging Face:

<|im_start|>user
user content<|im_end|>
<|im_start|>assistant
assistant content<|im_end|>