[Misc] Clean up fast loader daemon quant method verification (#55656)

Signed-off-by: Isotr0py <Isotr0py@outlook.com>
This commit is contained in:
Isotr0py
2026-09-12 20:16:39 -07:00
committed by GitHub
parent e52be1a62d
commit 533dbe424b
5 changed files with 75 additions and 117 deletions
@@ -39,6 +39,19 @@ class BaseModelLoader(ABC):
inplace weights loading for an already-initialized model"""
raise NotImplementedError
def create_model(
self, vllm_config: VllmConfig, model_config: ModelConfig, prefix: str = ""
) -> nn.Module:
"""Create a model with the given configurations."""
model = initialize_model(
vllm_config=vllm_config,
model_config=model_config,
prefix=prefix,
)
log_online_quantization(vllm_config)
log_model_inspection(model)
return model
@instrument(span_name="Load model")
def load_model(
self, vllm_config: VllmConfig, model_config: ModelConfig, prefix: str = ""
@@ -52,15 +65,12 @@ class BaseModelLoader(ABC):
target_device = torch.device(load_device)
with set_default_torch_dtype(model_config.dtype):
with target_device:
model = initialize_model(
model = self.create_model(
vllm_config=vllm_config,
model_config=model_config,
prefix=prefix,
)
log_online_quantization(vllm_config)
log_model_inspection(model)
logger.debug("Loading weights on %s ...", load_device)
self.load_weights(model, model_config)