Skip to content
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 6 additions & 6 deletions vllm_spyre/v1/core/scheduler.py
Original file line number Diff line number Diff line change
Expand Up @@ -153,19 +153,19 @@ def __init__(self, *args, **kwargs) -> None:
self.block_size = SpyrePlatform.get_block_size()
self.max_batch_tkv_limit = os.getenv("VLLM_DT_MAX_BATCH_TKV_LIMIT",
default='-1')
assert self.max_batch_tkv_limit != '-1', "Expecting the env var"
"VLLM_DT_MAX_BATCH_TKV_LIMIT to be set in platform.py"
assert self.max_batch_tkv_limit != '-1', (
"Expecting the env var VLLM_DT_MAX_BATCH_TKV_LIMIT to be set in "
"platform.py")

def update_from_output(
self,
scheduler_output: SchedulerOutput,
model_runner_output: ModelRunnerOutput,
) -> dict[int, EngineCoreOutputs]:
# Need an instance of CBSpyreModelRunnerOutput which holds the tkv value
assert isinstance(
model_runner_output, CBSpyreModelRunnerOutput
), "Expecting an instance of CBSpyreModelRunnerOutput"
"when doing continuous batching."
assert isinstance(model_runner_output, CBSpyreModelRunnerOutput), (
"Expecting an instance of CBSpyreModelRunnerOutput when doing "
"continuous batching.")
self.tkv = model_runner_output.tkv
self.n_free_blocks = model_runner_output.n_free_blocks
return super(SpyreScheduler,
Expand Down
Loading