A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 12, 17:15 UTC
Fix: Resolve circular import in model_loader/utils.py (#29189)
Signed-off-by: nandan2003 <nandan.vallamdasu@outlook.com> Signed-off-by: Nandan Vallamdasu <nandan.vallamdasu@outlook.com> Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com> Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
Nandan Vallamdasu committed
Nov 22, 2025 at 18:28 UTC
6965a392a4cd38ee65ac6a9c2730e0a7c62a658d
1 file changed
+7
-6
vllm/model_executor/model_loader/utils.py
+7
-6
@@ -19,12 +19,7 @@ from vllm.model_executor.layers.quantization.base_config import (
19
QuantizationConfig,
20
QuantizeMethodBase,
21
)
22
-from vllm.model_executor.models.adapters import (
23
- as_embedding_model,
24
- as_reward_model,
25
- as_seq_cls_model,
26
- try_create_mm_pooling_model_cls,
27
-)
22
+
23
from vllm.model_executor.models.interfaces import SupportsQuant, supports_multimodal
24
from vllm.utils.platform_utils import is_pin_memory_available
25
@@ -172,6 +167,12 @@ _MODEL_ARCH_BY_HASH = dict[int, tuple[type[nn.Module], str]]()
167
168
169
def _get_model_architecture(model_config: ModelConfig) -> tuple[type[nn.Module], str]:
170
+ from vllm.model_executor.models.adapters import (
171
+ as_embedding_model,
172
+ as_reward_model,
173
+ as_seq_cls_model,
174
+ try_create_mm_pooling_model_cls,
175
+ )
176
architectures = getattr(model_config.hf_config, "architectures", [])
177
178
model_cls, arch = model_config.registry.resolve_model_cls(