Handle short-token edge case in BPE pair selection

[amit-chaubey]

Fixes #1002

Summary

• Return None in find_freq_pair when fewer than 2 token IDs exist (or no pairs are found).
• Keep train() from failing on short inputs by stopping gracefully.
• Add a brief notebook note + tiny sanity checks.

Why

Short inputs could produce an empty pair set, causing max()/min() to fail before train() can break on pair_id is None.

Validation

• find_freq_pair([]) -> None
• find_freq_pair([42]) -> None
• train(""), train("H"), and train("He") run without error.

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[rasbt]

Thanks for catching (and fixing) this issue @CLu1207 and @amit-chaubey

I checked the other more complete BPE notebook and this one should be fine.

Anyways, thanks for also adding the edge case tests.

[amit-chaubey]

Thanks for the review and the merge, much appreciated! 🙌