[Misc] Split model loader (#17712)

Signed-off-by: Jee Jee Li <pandaleefree@gmail.com>
This commit is contained in:
Jee Jee Li
2025-05-07 12:42:26 +08:00
committed by GitHub
parent 8d84d836d1
commit 822de7fb94
21 changed files with 1681 additions and 1566 deletions
+2 -2
View File
@@ -37,7 +37,7 @@ from vllm.model_executor.layers.linear import (ColumnParallelLinear,
RowParallelLinear)
from vllm.model_executor.layers.quantization import QuantizationConfig
from vllm.model_executor.layers.rotary_embedding import get_rope
from vllm.model_executor.model_loader.loader import _initialize_model
from vllm.model_executor.model_loader.utils import initialize_model
from vllm.model_executor.model_loader.weight_utils import default_weight_loader
from vllm.model_executor.sampling_metadata import SamplingMetadata
from vllm.multimodal import MULTIMODAL_REGISTRY
@@ -670,7 +670,7 @@ class Llama4ForConditionalGeneration(nn.Module, SupportsMultiModal,
self.config,
None,
prefix=maybe_prefix(prefix, "multi_modal_projector"))
self.language_model = _initialize_model(
self.language_model = initialize_model(
vllm_config=vllm_config.with_hf_config(config.text_config,
["LlamaForCausalLM"]),
prefix=maybe_prefix(prefix, "language_model"),
+1 -1
View File
@@ -17,7 +17,7 @@ from vllm.config import VllmConfig
from vllm.forward_context import get_forward_context
from vllm.model_executor.layers.activation import MulAndSilu, get_act_fn
from vllm.model_executor.layers.layernorm import RMSNorm
from vllm.model_executor.model_loader.loader import DefaultModelLoader
from vllm.model_executor.model_loader import DefaultModelLoader
from vllm.model_executor.models.module_mapping import MultiModelKeys
from vllm.model_executor.sampling_metadata import SamplingMetadata
from vllm.multimodal import MULTIMODAL_REGISTRY
+2 -2
View File
@@ -273,7 +273,7 @@ def init_vllm_registered_model(
Helper function to initialize an inner model registered to vLLM,
based on the arguments passed to the outer vLLM model.
"""
from vllm.model_executor.model_loader.loader import _initialize_model
from vllm.model_executor.model_loader.utils import initialize_model
if hf_config is None and architectures is not None:
# So that the architectures field is overridden
@@ -283,7 +283,7 @@ def init_vllm_registered_model(
vllm_config = vllm_config.with_hf_config(hf_config,
architectures=architectures)
return _initialize_model(vllm_config=vllm_config, prefix=prefix)
return initialize_model(vllm_config=vllm_config, prefix=prefix)
@overload