phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-12-12 20:12:33 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

remove setting interops threads

Browse source
This commit is contained in:
Swapna Lekkala 2025-10-21 12:13:03 -07:00
parent a46fe682dc
commit 462cdb051c
1 changed files with 1 additions and 5 deletions
Download patch file Download diff file Expand all files Collapse all files

6
llama_stack/providers/utils/inference/embedding_mixin.py
View file

@ -93,11 +93,7 @@ class SentenceTransformerEmbeddingMixin:
# PyTorch's OpenMP kernels can segfault on macOS when spawned from background
# threads with the default parallel settings, so force a single-threaded CPU run.
log.debug(f"Constraining torch threads on {platform_name} to a single worker")
try:
torch.set_num_threads(1)
torch.set_num_interop_threads(1)
except Exception:
log.debug(f"Failed to adjust torch thread counts on {platform_name}", exc_info=True)
torch.set_num_threads(1)
return SentenceTransformer(model, trust_remote_code=True)