xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-13 07:15:00 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Bugfix] Fix GPUModelRunner has no attribute lora_manager (#24762)

Browse Source 
Signed-off-by: Jee Jee Li <pandaleefree@gmail.com>
This commit is contained in:
Jee Jee Li 2025-09-13 14:55:14 +08:00 committed by GitHub
parent 89e08d6d18
commit 99bfef841f
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 10 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
vllm/v1/worker/lora_model_runner_mixin.py
View File

@ -63,8 +63,7 @@ class LoRAModelRunnerMixin:
def _set_active_loras(self, prompt_lora_mapping: tuple[int, ...], def _set_active_loras(self, prompt_lora_mapping: tuple[int, ...],
token_lora_mapping: tuple[int, ...], token_lora_mapping: tuple[int, ...],
lora_requests: set[LoRARequest]) -> None: lora_requests: set[LoRARequest]) -> None:
if not self.lora_manager: self._ensure_lora_enabled()
raise RuntimeError("LoRA is not enabled.")
# Set is_prefill to True, so we always use the SGMV kernels on # Set is_prefill to True, so we always use the SGMV kernels on
# non-cuda platforms. # non-cuda platforms.
@ -75,6 +74,11 @@ class LoRAModelRunnerMixin:
is_prefill=True) is_prefill=True)
self.lora_manager.set_active_adapters(lora_requests, lora_mapping) self.lora_manager.set_active_adapters(lora_requests, lora_mapping)
def _ensure_lora_enabled(self) -> None:
if not hasattr(self, "lora_manager"):
raise RuntimeError(
"LoRA is not enabled. Use --enable-lora to enable LoRA.")
def set_active_loras(self, input_batch: InputBatch, def set_active_loras(self, input_batch: InputBatch,
num_scheduled_tokens: np.ndarray) -> None: num_scheduled_tokens: np.ndarray) -> None:
@ -172,21 +176,17 @@ class LoRAModelRunnerMixin:
self.lora_manager.remove_all_adapters() self.lora_manager.remove_all_adapters()
def add_lora(self, lora_request: LoRARequest) -> bool: def add_lora(self, lora_request: LoRARequest) -> bool:
if not self.lora_manager: self._ensure_lora_enabled()
raise RuntimeError("LoRA is not enabled.")
return self.lora_manager.add_adapter(lora_request) return self.lora_manager.add_adapter(lora_request)
def remove_lora(self, lora_id: int) -> bool: def remove_lora(self, lora_id: int) -> bool:
if not self.lora_manager: self._ensure_lora_enabled()
raise RuntimeError("LoRA is not enabled.")
return self.lora_manager.remove_adapter(lora_id) return self.lora_manager.remove_adapter(lora_id)
def pin_lora(self, lora_id: int) -> bool: def pin_lora(self, lora_id: int) -> bool:
if not self.lora_manager: self._ensure_lora_enabled()
raise RuntimeError("LoRA is not enabled.")
return self.lora_manager.pin_adapter(lora_id) return self.lora_manager.pin_adapter(lora_id)
def list_loras(self) -> set[int]: def list_loras(self) -> set[int]:
if not self.lora_manager: self._ensure_lora_enabled()
raise RuntimeError("LoRA is not enabled.")
return self.lora_manager.list_adapters() return self.lora_manager.list_adapters()