immich/machine-learning/immich_ml
todorangrg cde18e6b1b perf(ml): cut RKNN multi-core RAM via native dup_context pool
Replace RKNNLite with a pybind11 pool that shares weights across worker
contexts so MACHINE_LEARNING_RKNN_THREADS>1 no longer multiplies model memory.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-07-31 15:46:12 +02:00
..
models perf(ml): cut RKNN multi-core RAM via native dup_context pool 2026-07-31 15:46:12 +02:00
sessions perf(ml): cut RKNN multi-core RAM via native dup_context pool 2026-07-31 15:46:12 +02:00
__init__.py chore(ml): installable package (#17153) 2025-03-27 19:49:09 +00:00
__main__.py chore(deps): update machine-learning (#28239) 2026-06-02 16:44:50 +00:00
config.py fix(ml): stabilize MIGraphX inference (#28444) 2026-05-26 18:41:56 +00:00
gunicorn_conf.py chore(ml): installable package (#17153) 2025-03-27 19:49:09 +00:00
log_conf.json chore(ml): installable package (#17153) 2025-03-27 19:49:09 +00:00
main.py chore(deps): update machine-learning (#28239) 2026-06-02 16:44:50 +00:00
schemas.py chore(deps): update machine-learning (#28239) 2026-06-02 16:44:50 +00:00