Dom/rpa_vision_v3
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix: SomEngine sur CPU pour cohabiter avec Qwen2.5-VL GPU

Browse Source 
Qwen2.5-VL occupe 9.8 GB de VRAM → plus de place pour YOLO.
SomEngine passe en CPU (1.4s au lieu de 0.1s, acceptable car
utilisé uniquement pendant le build_replay, pas le replay).

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
Dom
2026-04-01 09:30:00 +02:00
parent 394342be7e
commit 0bd0fbb8c5
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
core/detection/som_engine.py
View File

@@ -300,7 +300,7 @@ _shared_engine: Optional[SomEngine] = None
_shared_lock = __import__("threading").Lock()
def get_shared_engine(device: str = "cuda") -> Optional[SomEngine]:
def get_shared_engine(device: str = "cpu") -> Optional[SomEngine]:
"""Singleton SomEngine partagé entre tous les modules."""
global _shared_engine
if _shared_engine is None: