fix(config): validate max_num_scheduled_tokens >= 0 on all paths (#44207)
Signed-off-by: Oxygen56 <1391083091@qq.com>
This commit is contained in:
@@ -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
|
||||
|
||||
@@ -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
|
||||
|
||||
Reference in New Issue
Block a user