Skip to content

fix: unify qwen tts cache dir for tokenizer loading on Windows #218

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
seidenbergerscott wants to merge 1 commit into
base: main
Choose a base branch
from
Open

fix: unify qwen tts cache dir for tokenizer loading on Windows #218

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 10 additions & 2 deletions backend/backends/pytorch_backend.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -178,6 +178,12 @@ def _load_model_sync(self, model_size: str):
status="downloading",
)

# Use a single cache root for both HF Hub + Transformers to avoid split-cache
# cases where speech_tokenizer files are downloaded to one location while
# AutoFeatureExtractor resolves from another.
from huggingface_hub import constants as hf_constants
tts_cache_dir = hf_constants.HF_HUB_CACHE

# Load the model (tqdm is patched, but filters out non-download progress)
try:
# Don't pass device_map on CPU: accelerate's meta-tensor mechanism
Expand All @@ -186,14 +192,16 @@ def _load_model_sync(self, model_size: str):
if self.device == "cpu":
self.model = Qwen3TTSModel.from_pretrained(
model_path,
torch_dtype=torch.float32,
cache_dir=tts_cache_dir,
dtype=torch.float32,
low_cpu_mem_usage=False,
)
else:
self.model = Qwen3TTSModel.from_pretrained(
model_path,
cache_dir=tts_cache_dir,
device_map=self.device,
torch_dtype=torch.bfloat16,
dtype=torch.bfloat16,
)
finally:
# Exit the patch context
Expand Down