[Quantization] Migrate GGUF diffusion model support to OOT plugin (#4769)
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn> Signed-off-by: Isotr0py <Isotr0py@outlook.com>
This commit is contained in:
1 parent
ff6f0da271
commit
b95905deb1
35 files changed
+272
-1614
No files matched your search
@@ -12,10 +12,10 @@ Adds per-component quantization for multi-stage models.
|
||||
"""
|
||||
|
||||
from .component_config import ComponentQuantizationConfig
|
||||
from .factory import SUPPORTED_QUANTIZATION_METHODS, build_quant_config
|
||||
from .factory import SUPPORTED_QUANTIZATION_METHODS, build_quant_config, register_quantization_override
|
||||
from .inc_config import OmniINCConfig
|
||||
|
||||
# Heavy configs (GGUF, MXFP8) are NOT imported here to avoid pulling in
|
||||
# Heavy configs are NOT imported here to avoid pulling in
|
||||
# optional dependencies (pynvml, torch_npu) at module load time.
|
||||
# Import them directly when needed:
|
||||
# from vllm_omni.quantization.mxfp8_config import DiffusionMXFP8Config
|
||||
@@ -25,4 +25,5 @@ __all__ = [
|
||||
"ComponentQuantizationConfig",
|
||||
"OmniINCConfig",
|
||||
"SUPPORTED_QUANTIZATION_METHODS",
|
||||
"register_quantization_override",
|
||||
]
|
||||
Reference in new issue
Block a user