diff --git a/vllm/config/scheduler.py b/vllm/config/scheduler.py index fb6951ea7d..7900c94848 100644 --- a/vllm/config/scheduler.py +++ b/vllm/config/scheduler.py @@ -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 diff --git a/vllm/v1/core/sched/scheduler.py b/vllm/v1/core/sched/scheduler.py index 8edad4c3ec..c39e80c24e 100644 --- a/vllm/v1/core/sched/scheduler.py +++ b/vllm/v1/core/sched/scheduler.py @@ -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