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 15, 04:45 UTC
[Minor] Allow redirecting model path for HfRunner in test (#19795)
Signed-off-by: Isotr0py <2037008807@qq.com>
Isotr0py committed
Jun 19, 2025 at 14:04 UTC
6021999573910709e25e7ca838ec8b647959a07d
1 file changed
+2
tests/conftest.py
+2
@@ -33,6 +33,7 @@ from vllm.inputs import (ExplicitEncoderDecoderPrompt, TextPrompt,
33
from vllm.logger import init_logger
34
from vllm.outputs import RequestOutput
35
from vllm.sampling_params import BeamSearchParams
36
+from vllm.transformers_utils.utils import maybe_model_redirect
37
from vllm.utils import cuda_device_count_stateless
38
39
logger = init_logger(__name__)
@@ -321,6 +322,7 @@ class HfRunner:
322
skip_tokenizer_init: bool = False,
323
auto_cls: type[_BaseAutoModelClass] = AutoModelForCausalLM,
324
) -> None:
325
+ model_name = maybe_model_redirect(model_name)
326
self.model_name = model_name
327
328
self.config = AutoConfig.from_pretrained(