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 14, 23:45 UTC
add clear messages for deprecated models (#21424)
Signed-off-by: youkaichao <youkaichao@gmail.com>
youkaichao committed
Jul 23, 2025 at 15:03 UTC
2f5c14de6a7e281a86ab5e2376de95f7021dff0a
2 files changed
+12
-1
vllm/model_executor/model_loader/utils.py
+10
-1
@@ -25,7 +25,8 @@ from vllm.model_executor.models.adapters import (as_embedding_model,
25
as_reward_model,
26
as_seq_cls_model)
27
from vllm.model_executor.models.interfaces import SupportsQuant
28
-from vllm.model_executor.models.registry import _TRANSFORMERS_MODELS
28
+from vllm.model_executor.models.registry import (_PREVIOUSLY_SUPPORTED_MODELS,
29
+ _TRANSFORMERS_MODELS)
30
from vllm.utils import is_pin_memory_available
31
32
logger = init_logger(__name__)
@@ -261,6 +262,14 @@ def get_model_architecture(
262
vllm_not_supported = False
263
break
264
265
+ if any(arch in _PREVIOUSLY_SUPPORTED_MODELS for arch in architectures):
266
+ previous_version = _PREVIOUSLY_SUPPORTED_MODELS[architectures[0]]
267
+ raise ValueError(
268
+ f"Model architecture {architectures[0]} was supported"
269
+ f" in vLLM until version {previous_version}, and is "
270
+ "not supported anymore. Please use an older version"
271
+ " of vLLM if you want to use this model architecture.")
272
+
273
if (model_config.model_impl == ModelImpl.TRANSFORMERS or
274
model_config.model_impl == ModelImpl.AUTO and vllm_not_supported):
275
architectures = resolve_transformers_arch(model_config, architectures)
vllm/model_executor/models/registry.py
+2
@@ -276,6 +276,8 @@ _SUBPROCESS_COMMAND = [
276
sys.executable, "-m", "vllm.model_executor.models.registry"
277
]
278
279
+_PREVIOUSLY_SUPPORTED_MODELS = {"Phi3SmallForCausalLM": "0.9.2"}
280
+
281
282
@dataclass(frozen=True)
283
class _ModelInfo: