fix(config): validate max_num_scheduled_tokens >= 0 on all paths (#44207)

Signed-off-by: Oxygen56 <1391083091@qq.com>
This commit is contained in:
Willow Lopez
2026-06-03 16:06:45 +00:00
committed by GitHub
parent 0c6631f02a
commit 51e0c579b0
2 changed files with 2 additions and 2 deletions
+1 -1
View File
@@ -53,7 +53,7 @@ class SchedulerConfig:
In real usage, this should be set in `EngineArgs.create_engine_config`.
"""
max_num_scheduled_tokens: int | None = None
max_num_scheduled_tokens: int | None = Field(default=None, ge=0)
"""Maximum number of tokens that the scheduler may issue in a single iteration.
This is usually equal to max_num_batched_tokens, but can be smaller in cases
+1 -1
View File
@@ -104,7 +104,7 @@ class Scheduler(SchedulerInterface):
self.max_num_running_reqs = self.scheduler_config.max_num_seqs
self.max_num_scheduled_tokens = (
self.scheduler_config.max_num_scheduled_tokens
if self.scheduler_config.max_num_scheduled_tokens
if self.scheduler_config.max_num_scheduled_tokens is not None
else self.scheduler_config.max_num_batched_tokens
)
self.max_model_len = vllm_config.model_config.max_model_len