fix: include vocab sizes in EAGLE3 vocab mapping cache key

[wiketool]

Motivation

draft_model_config.draft_vocab_size and draft_model_config.vocab_size affect vocab mapping generation. But the current vocab mapping cache key only follows the dataset cache key. This can incorrectly reuse a stale mapping when dataset/tokenizer inputs are unchanged but vocab dimensions differ.

Modifications

• Keep the existing processed dataset cache key unchanged.
• Add a vocab-mapping-specific cache key by appending draft_vocab_size and vocab_size to the existing dataset cache params string.

Checklist

• Format your code according to the Code Formatting with Pre-Commit.
• Add unit tests as outlined in the Running Unit Tests.
• Update documentation / docstrings / example tutorials as needed, according to Writing Documentation.
• Provide throughput / latency benchmark results and accuracy evaluation results as needed, according to Benchmark and Profiling and Accuracy Results.
• For reviewers: If you haven't made any contributions to this PR and are only assisting with merging the main branch, please remove yourself as a co-author when merging the PR.
• Please feel free to join our Slack channel at https://sgl-fru7574.slack.com/archives/C09784E3EN6 to discuss your PR.

[gemini-code-assist]

Code Review

This pull request refactors the cache key generation in scripts/train_eagle3.py by separating the dataset cache key from the vocabulary mapping cache key. It introduces a new vocab_cache_key that incorporates the draft and target model vocabulary sizes, ensuring that vocabulary mapping caches are correctly invalidated when vocabulary sizes change. There are no review comments to evaluate, and the changes look correct.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.