# NousResearch/hermes-agent · Sherpa KWS tokenizer type not auto-detected

## Open now

- 1 issue, 2 PRs open on GitHub.
- #110003: \[Bug\]: Wake word: sherpa KWS hard-codes tokens_type="bpe" — any model without bpe.model crashes the engine (Chinese wenetspeech model never starts)
- PR #110087: fix(tools): support non-BPE Sherpa wake vocabularies
- PR #124849: fix(wake-word): pick the sherpa KWS tokenizer layout from the model directory

A bug where the Sherpa keyword spotting engine hard-codes BPE tokenization causes non-BPE models (e.g. Chinese WenetSpeech) to fail, with accompanying fixes to detect the tokenizer layout from the model directory.

## TL;DR

- 3 member(s): 1 open issue(s), 0 closed issue(s), 2 open PR(s), 0 merged PR(s), 0 PR(s) closed without merging.
- No fix has been merged to main; one or more PRs were attempted (see below).
- **Recommended:** review the prior/competing PR(s) below. A fix was attempted but not merged; your case may extend or revive that work.

## Maintainer resolution verdict

- Internal-tracker verdict over this cluster's issue members: `unresolved`. This is distinct from the fix status above. Weigh it yourself.

## How this cluster was grouped

- Link evidence (`STRUCTURAL_SPANNING`): every member is connected by structural evidence: shared changed files, or a closing/superseded reference.
- Pairwise link density `0.67`: the share of member pairs that carry a link, counting a structural link in full and a relevance-only link at 0.36. `1.00` means every pair is linked.

## Open related issues

- #110003: \[Bug\]: Wake word: sherpa KWS hard-codes tokens_type="bpe" — any model without bpe.model crashes the engine (Chinese wenetspeech model never starts)

## Prior & competing PRs

- **#110087: fix(tools): support non-BPE Sherpa wake vocabularies**. open
  - maintainer direction: category `uncategorized`; implication `needs_llm`; refs `5e9d9ad6b8`, `81c627e01f`, `#110057`
  - files: `tests/tools/test_wake_word.py`, `tools/wake_word_engines.py`
- **#124849: fix(wake-word): pick the sherpa KWS tokenizer layout from the model directory**. open
  - maintainer direction: category `uncategorized`; implication `needs_llm`
  - files: `tests/tools/test_wake_word_engines_layout.py`, `tools/wake_word_engines.py`

## Fix status

- No merged fix on record.
- Subsystem: `agent`