fix(offline): patch transformers mistral-regex check to survive HF failures (#530)

* fix(offline): patch transformers mistral-regex check to survive HF failures

transformers 4.57.x's `PreTrainedTokenizerBase._patch_mistral_regex` calls
`huggingface_hub.model_info(repo_id)` unconditionally during any non-local
tokenizer load to probe for Mistral-family models. The call raises on
`HF_HUB_OFFLINE=1`, on network outages, and on slow/blocked HF endpoints,
and transformers doesn't catch any of it — the exception bubbles out of
`from_pretrained` and kills the load for unrelated engines (Qwen TTS,
Qwen CustomVoice, TADA, etc.).

0.4.2's load-time `force_offline_if_cached` guard walked straight into
this trap: on cached online users it flipped `HF_HUB_OFFLINE=1` and
converted a healthy load into a hard crash. 0.4.3's inference-path guard
masked it; #524 removed the inference guard in 0.4.4, and users updating
to 0.4.4 started hitting the same error on the load path instead
(#526).

Fix:
- Wrap `_patch_mistral_regex` so any exception from the inner HF
  metadata check is swallowed and the tokenizer is returned unchanged.
  Voicebox never loads Mistral models, so the regex rewrite this check
  gates is a no-op for us; matches the success-path behavior for
  non-Mistral repos (tokenization_utils_base.py:2503).
- Drop the `force_offline_if_cached` wraps from every load path
  (pytorch_backend Qwen + Whisper, qwen_custom_voice_backend,
  mlx_backend Qwen + Whisper). With the mistral patch in place they
  provide zero value and only risk re-introducing the same class of
  bug. Helper and its unit tests stay — still correct for targeted
  future use.
- Add `backend/tests/test_offline_patch.py` covering
  OfflineModeIsEnabled / ConnectionError suppression, success
  pass-through, idempotence, and the missing-method no-op path.

Fixes #526.

* fix(offline): install mistral-regex patch for non-MLX backends

The previous commit left the patch wired only through ``mlx_backend.py``'s
existing import of ``hf_offline_patch``. On Windows/Linux/CUDA users who
never load the MLX backend (everyone who hit #526), the patch module was
never imported, so ``patch_transformers_mistral_regex`` never ran and the
crash persisted.

Hoist the import into ``backends/__init__.py``. Every backend imports from
this package, so the module-level patch install runs before any
``from_pretrained`` call regardless of which engine the user picks.

Caught by CodeRabbit and Cursor Bugbot on #530.
This commit is contained in:
Jamie Pine
2026-04-21 22:01:29 -07:00
committed by GitHub
parent 74e004400f
commit d61e884104
6 changed files with 202 additions and 38 deletions
+2 -5
View File
@@ -20,7 +20,6 @@ ensure_original_qwen_config_cached()
from . import TTSBackend, STTBackend, LANGUAGE_CODE_TO_NAME, WHISPER_HF_REPOS
from .base import is_model_cached, combine_voice_prompts as _combine_voice_prompts, model_load_progress
from ..utils.cache import get_cache_key, get_cached_voice_prompt, cache_voice_prompt
from ..utils.hf_offline_patch import force_offline_if_cached
class MLXTTSBackend:
@@ -99,8 +98,7 @@ class MLXTTSBackend:
logger.info("Loading MLX TTS model %s...", model_size)
with force_offline_if_cached(is_cached, model_name):
self.model = load(model_path)
self.model = load(model_path)
self._current_model_size = model_size
self.model_size = model_size
@@ -311,8 +309,7 @@ class MLXSTTBackend:
model_name = WHISPER_HF_REPOS.get(model_size, f"openai/whisper-{model_size}")
logger.info("Loading MLX Whisper model %s...", model_size)
with force_offline_if_cached(is_cached, progress_model_name):
self.model = load(model_name)
self.model = load(model_name)
self.model_size = model_size
logger.info("MLX Whisper model %s loaded successfully", model_size)