[Misc] Fix Current vLLM config is not set. warnings, assert to avoid issues in the future (#31747)
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com> Signed-off-by: Lucas Wilkinson <LucasWilkinson@users.noreply.github.com> Co-authored-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
This commit is contained in:
co-authored by
Luka Govedič
parent
5d3b6097ad
commit
6cdf015c3c
@@ -14,7 +14,7 @@ import torch.distributed
|
||||
import torch.nn as nn
|
||||
|
||||
import vllm.envs as envs
|
||||
from vllm.config import CUDAGraphMode, VllmConfig
|
||||
from vllm.config import CUDAGraphMode, VllmConfig, set_current_vllm_config
|
||||
from vllm.config.compilation import CompilationMode
|
||||
from vllm.distributed import (
|
||||
ensure_model_parallel_initialized,
|
||||
@@ -268,7 +268,9 @@ class Worker(WorkerBase):
|
||||
# to hijack tensor allocation.
|
||||
def load_model(self) -> None:
|
||||
eep_scale_up = os.environ.get("VLLM_ELASTIC_EP_SCALE_UP_LAUNCH") == "1"
|
||||
with self._maybe_get_memory_pool_context(tag="weights"):
|
||||
with self._maybe_get_memory_pool_context(
|
||||
tag="weights"
|
||||
) and set_current_vllm_config(self.vllm_config):
|
||||
self.model_runner.load_model(eep_scale_up=eep_scale_up)
|
||||
|
||||
def update_config(self, overrides: dict[str, Any]) -> None:
|
||||
|
||||
Reference in New Issue
Block a user