[Core] Cleanup KVConnector handling with PP + fix MRV2 (#43732)

Signed-off-by: Nick Hill <nickhill123@gmail.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
This commit is contained in:
Nick Hill
2026-05-28 13:12:03 -07:00
committed by GitHub
co-authored by mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
parent 9090368b65
commit 7e53283b1c
9 changed files with 38 additions and 48 deletions
+5 -2
View File
@@ -39,6 +39,7 @@ from vllm.v1.kv_cache_interface import (
KVCacheGroupSpec,
KVCacheTensor,
)
from vllm.v1.outputs import EMPTY_MODEL_RUNNER_OUTPUT
from vllm.v1.sample.metadata import SamplingMetadata
from vllm.v1.spec_decode.metadata import SpecDecodeMetadata
from vllm.v1.worker.gpu_input_batch import InputBatch
@@ -279,7 +280,8 @@ def test_sample_tokens_receives_pp_sampled_ids_only_on_non_last_rank(
lambda: SimpleNamespace(world_size=world_size, is_last_rank=is_last_rank),
)
assert GPUModelRunner.sample_tokens(runner, None) is None
output = GPUModelRunner.sample_tokens(runner, None)
assert output in (EMPTY_MODEL_RUNNER_OUTPUT, None)
assert receive_calls == expected_calls
@@ -298,7 +300,8 @@ def test_sample_tokens_skips_pp_group_lookup_without_async_scheduling(
pytest.fail,
)
assert GPUModelRunner.sample_tokens(runner, None) is None
output = GPUModelRunner.sample_tokens(runner, None)
assert output in (EMPTY_MODEL_RUNNER_OUTPUT, None)
def test_select_common_block_size_no_valid_option():