cap model cache size using bytes, not # models

This commit is contained in:
Lincoln Stein
2023-05-07 18:07:28 -04:00
parent 647ffb2a0f
commit 667171ed90
4 changed files with 99 additions and 38 deletions

View File

@ -59,7 +59,7 @@ def get_model_manager(config: Args, logger: types.ModuleType) -> ModelManager:
config.conf,
precision=dtype,
device_type=device,
max_loaded_models=config.max_loaded_models,
max_cache_size=config.max_cache_size,
# temporarily disabled until model manager stabilizes
# embedding_path = Path(embedding_path),
logger = logger,