From 9a399adba87dfcfe0128af31f4708fa1c16f036d Mon Sep 17 00:00:00 2001 From: tzhouam Date: Thu, 3 Sep 2026 13:30:46 +0800 Subject: [PATCH] chore: satisfy changed-file policy Scoped pre-commit over the 27 files this rebase touches: ruff format reflow, the vLLM-Omni SPDX copyright line on the new modules, and wording the typos hook rejected. The Python 3.10 mypy hook still reports 24 errors across five of these files. All 24 sit on pre-existing lines; none fall on a line this rebase changed, which the repository documents as the CI-skipped condition. Co-Authored-By: Claude Opus 5 Claude-Session: https://claude.ai/code/session_0146Xu5Sgj4vXxLxvLjV5YPj Signed-off-by: tzhouam --- tests/diffusion/diffusion_kv/test_layout.py | 2 +- tests/diffusion/diffusion_kv/test_manager.py | 4 ++-- .../test_paged_attention_adapter.py | 2 +- .../test_paged_attention_adapter_gpu.py | 8 +++---- tests/diffusion/diffusion_kv/test_request.py | 4 ++-- .../diffusion_kv/test_spec_discovery.py | 2 +- tests/diffusion/layers/test_adalayernorm.py | 2 +- .../models/dreamzero/test_qkv_fusion.py | 2 +- .../glm_image/test_glm_image_quantization.py | 2 +- .../test_mistral_encoder_tp.py | 2 +- .../sana_video/test_transformer_sana_video.py | 2 +- .../test_transformer_sana_video_tp.py | 2 +- .../models/t5_encoder/test_t5_encoder_tp.py | 2 +- tests/diffusion/test_diffusion_scheduler.py | 4 ++-- tests/helpers/kv_layout.py | 2 +- .../diffusion/diffusion_kv/initialization.py | 2 +- vllm_omni/diffusion/diffusion_kv/layout.py | 23 ++++++------------- .../diffusion_kv/model_runner_backend.py | 4 +--- .../ar_diffusion/kv_cache/manager.py | 1 + 19 files changed, 30 insertions(+), 42 deletions(-) diff --git a/tests/diffusion/diffusion_kv/test_layout.py b/tests/diffusion/diffusion_kv/test_layout.py index 2f410f74c..713849e95 100644 --- a/tests/diffusion/diffusion_kv/test_layout.py +++ b/tests/diffusion/diffusion_kv/test_layout.py @@ -1,5 +1,5 @@ # SPDX-License-Identifier: Apache-2.0 -# SPDX-FileCopyrightText: Copyright contributors to the vLLM project +# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project """Physical KV cache layout selection and propagation for the Diffusion cache.""" from __future__ import annotations diff --git a/tests/diffusion/diffusion_kv/test_manager.py b/tests/diffusion/diffusion_kv/test_manager.py index 8abc296f1..ed39f6523 100644 --- a/tests/diffusion/diffusion_kv/test_manager.py +++ b/tests/diffusion/diffusion_kv/test_manager.py @@ -1,14 +1,14 @@ # SPDX-License-Identifier: Apache-2.0 -# SPDX-FileCopyrightText: Copyright contributors to the vLLM project +# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project import pytest import torch import vllm.v1.core.single_type_kv_cache_manager as native_kv_managers from vllm.v1.kv_cache_interface import FullAttentionSpec, KVCacheConfig, KVCacheGroupSpec +from tests.helpers.kv_layout import build_kv_cache_tensor from vllm_omni.diffusion.diffusion_kv.manager import DiffusionKVAdmissionError, DiffusionKVCacheManager from vllm_omni.diffusion.diffusion_kv.request import DiffusionKVContext, DiffusionKVRequest -from tests.helpers.kv_layout import build_kv_cache_tensor pytestmark = [pytest.mark.core_model, pytest.mark.cpu, pytest.mark.diffusion] diff --git a/tests/diffusion/diffusion_kv/test_paged_attention_adapter.py b/tests/diffusion/diffusion_kv/test_paged_attention_adapter.py index a13e09051..535c1b411 100644 --- a/tests/diffusion/diffusion_kv/test_paged_attention_adapter.py +++ b/tests/diffusion/diffusion_kv/test_paged_attention_adapter.py @@ -15,10 +15,10 @@ import torch from torch import nn from vllm.v1.kv_cache_interface import FullAttentionSpec +from tests.helpers.kv_layout import layout_for_backend from vllm_omni.diffusion.attention.backends.flash_attn import FlashAttentionImpl from vllm_omni.diffusion.attention.layer import Attention from vllm_omni.diffusion.diffusion_kv import paged_attention_adapter as adapter_module -from tests.helpers.kv_layout import layout_for_backend from vllm_omni.diffusion.diffusion_kv.paged_attention_adapter import ( DiffusionPagedAttentionAdapter, DiffusionPagedAttentionMetadata, diff --git a/tests/diffusion/diffusion_kv/test_paged_attention_adapter_gpu.py b/tests/diffusion/diffusion_kv/test_paged_attention_adapter_gpu.py index 192e60bea..d890dcc7b 100644 --- a/tests/diffusion/diffusion_kv/test_paged_attention_adapter_gpu.py +++ b/tests/diffusion/diffusion_kv/test_paged_attention_adapter_gpu.py @@ -1,5 +1,5 @@ # SPDX-License-Identifier: Apache-2.0 -# SPDX-FileCopyrightText: Copyright contributors to the vLLM project +# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project from __future__ import annotations @@ -14,6 +14,7 @@ from vllm.v1.kv_cache_interface import FullAttentionSpec, KVCacheConfig, KVCache from vllm.v1.worker.gpu.attn_utils import init_attn_backend, init_kv_cache from vllm.v1.worker.gpu.block_table import BlockTables +from tests.helpers.kv_layout import build_kv_cache_tensor from vllm_omni.diffusion.attention.backends.abstract import AttentionMetadata from vllm_omni.diffusion.attention.backends.flash_attn import FlashAttentionImpl from vllm_omni.diffusion.diffusion_kv.layout import resolve_diffusion_kv_cache_layout @@ -24,7 +25,6 @@ from vllm_omni.diffusion.diffusion_kv.paged_attention_adapter import ( DiffusionPagedAttentionRowBinding, ) from vllm_omni.diffusion.vllm_config import _DiffusionVllmModelConfig -from tests.helpers.kv_layout import build_kv_cache_tensor pytestmark = [pytest.mark.core_model, pytest.mark.diffusion, pytest.mark.gpu] @@ -90,9 +90,7 @@ def test_adapter_executes_native_paged_attention_on_non_contiguous_blocks() -> N canonical_spec = native_layer.spec kv_cache_config = KVCacheConfig( num_blocks=4, - kv_cache_tensors=[ - build_kv_cache_tensor(canonical_spec, 4, [_LAYER_NAME]) - ], + kv_cache_tensors=[build_kv_cache_tensor(canonical_spec, 4, [_LAYER_NAME])], kv_cache_groups=[ KVCacheGroupSpec( layer_names=[_LAYER_NAME], diff --git a/tests/diffusion/diffusion_kv/test_request.py b/tests/diffusion/diffusion_kv/test_request.py index 4ca6703d8..ac9185e09 100644 --- a/tests/diffusion/diffusion_kv/test_request.py +++ b/tests/diffusion/diffusion_kv/test_request.py @@ -1,5 +1,5 @@ # SPDX-License-Identifier: Apache-2.0 -# SPDX-FileCopyrightText: Copyright contributors to the vLLM project +# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project import pytest import torch @@ -13,8 +13,8 @@ from vllm.v1.kv_cache_interface import ( ) from vllm.v1.request import RequestStatus -from vllm_omni.diffusion.diffusion_kv.request import DiffusionKVContext, DiffusionKVRequest from tests.helpers.kv_layout import build_kv_cache_tensor +from vllm_omni.diffusion.diffusion_kv.request import DiffusionKVContext, DiffusionKVRequest pytestmark = [pytest.mark.core_model, pytest.mark.cpu, pytest.mark.diffusion] diff --git a/tests/diffusion/diffusion_kv/test_spec_discovery.py b/tests/diffusion/diffusion_kv/test_spec_discovery.py index d07c41355..1ccef7a9f 100644 --- a/tests/diffusion/diffusion_kv/test_spec_discovery.py +++ b/tests/diffusion/diffusion_kv/test_spec_discovery.py @@ -11,11 +11,11 @@ from torch import nn from vllm.v1.kv_cache_interface import FullAttentionSpec, KVCacheConfig, KVCacheGroupSpec import vllm_omni.diffusion.worker.diffusion_worker as diffusion_worker_module +from tests.helpers.kv_layout import build_kv_cache_tensor, layout_for_backend from vllm_omni.diffusion.attention.layer import Attention from vllm_omni.diffusion.diffusion_kv.config import DiffusionKVCacheMode from vllm_omni.diffusion.worker.diffusion_model_runner import DiffusionModelRunner from vllm_omni.diffusion.worker.diffusion_worker import DiffusionWorker -from tests.helpers.kv_layout import build_kv_cache_tensor, layout_for_backend pytestmark = [pytest.mark.core_model, pytest.mark.cpu, pytest.mark.diffusion] diff --git a/tests/diffusion/layers/test_adalayernorm.py b/tests/diffusion/layers/test_adalayernorm.py index 51fc52f35..94cadf5ac 100644 --- a/tests/diffusion/layers/test_adalayernorm.py +++ b/tests/diffusion/layers/test_adalayernorm.py @@ -1,5 +1,5 @@ # SPDX-License-Identifier: Apache-2.0 -# SPDX-FileCopyrightText: Copyright contributors to the vLLM project +# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project """Unit tests for shared AdaLayerNorm layers used by FLUX and other models.""" import os diff --git a/tests/diffusion/models/dreamzero/test_qkv_fusion.py b/tests/diffusion/models/dreamzero/test_qkv_fusion.py index c7b2f032f..d67a01e48 100644 --- a/tests/diffusion/models/dreamzero/test_qkv_fusion.py +++ b/tests/diffusion/models/dreamzero/test_qkv_fusion.py @@ -1,5 +1,5 @@ # SPDX-License-Identifier: Apache-2.0 -# SPDX-FileCopyrightText: Copyright contributors to the vLLM project +# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project """Equivalence test for DreamZero self-attn QKV fusion. diff --git a/tests/diffusion/models/glm_image/test_glm_image_quantization.py b/tests/diffusion/models/glm_image/test_glm_image_quantization.py index c08bfb894..1a6170ca2 100644 --- a/tests/diffusion/models/glm_image/test_glm_image_quantization.py +++ b/tests/diffusion/models/glm_image/test_glm_image_quantization.py @@ -1,5 +1,5 @@ # SPDX-License-Identifier: Apache-2.0 -# SPDX-FileCopyrightText: Copyright contributors to the vLLM project +# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project """Unit tests for GLM-Image quantization support (W4A16/AutoRound). These tests verify that the GLM-Image DiT transformer correctly accepts and uses diff --git a/tests/diffusion/models/mistral_encoder/test_mistral_encoder_tp.py b/tests/diffusion/models/mistral_encoder/test_mistral_encoder_tp.py index 17fc1540e..62bcf9ce8 100644 --- a/tests/diffusion/models/mistral_encoder/test_mistral_encoder_tp.py +++ b/tests/diffusion/models/mistral_encoder/test_mistral_encoder_tp.py @@ -1,5 +1,5 @@ # SPDX-License-Identifier: Apache-2.0 -# SPDX-FileCopyrightText: Copyright contributors to the vLLM project +# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project from types import SimpleNamespace import pytest diff --git a/tests/diffusion/models/sana_video/test_transformer_sana_video.py b/tests/diffusion/models/sana_video/test_transformer_sana_video.py index 3e1e40c2a..9e6404b21 100644 --- a/tests/diffusion/models/sana_video/test_transformer_sana_video.py +++ b/tests/diffusion/models/sana_video/test_transformer_sana_video.py @@ -1,5 +1,5 @@ # SPDX-License-Identifier: Apache-2.0 -# SPDX-FileCopyrightText: Copyright contributors to the vLLM project +# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project import os diff --git a/tests/diffusion/models/sana_video/test_transformer_sana_video_tp.py b/tests/diffusion/models/sana_video/test_transformer_sana_video_tp.py index 7105f5b4c..a44c90146 100644 --- a/tests/diffusion/models/sana_video/test_transformer_sana_video_tp.py +++ b/tests/diffusion/models/sana_video/test_transformer_sana_video_tp.py @@ -1,5 +1,5 @@ # SPDX-License-Identifier: Apache-2.0 -# SPDX-FileCopyrightText: Copyright contributors to the vLLM project +# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project """TP unit tests for SANA-Video: distributed RMSNorm, sharded attention and sharded GLUMB temporal conv. diff --git a/tests/diffusion/models/t5_encoder/test_t5_encoder_tp.py b/tests/diffusion/models/t5_encoder/test_t5_encoder_tp.py index 6eb469fac..cca73fb08 100644 --- a/tests/diffusion/models/t5_encoder/test_t5_encoder_tp.py +++ b/tests/diffusion/models/t5_encoder/test_t5_encoder_tp.py @@ -1,5 +1,5 @@ # SPDX-License-Identifier: Apache-2.0 -# SPDX-FileCopyrightText: Copyright contributors to the vLLM project +# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project import pytest import torch from transformers import T5Config diff --git a/tests/diffusion/test_diffusion_scheduler.py b/tests/diffusion/test_diffusion_scheduler.py index 6bc0bdaba..20a2f0384 100644 --- a/tests/diffusion/test_diffusion_scheduler.py +++ b/tests/diffusion/test_diffusion_scheduler.py @@ -1,5 +1,5 @@ # SPDX-License-Identifier: Apache-2.0 -# SPDX-FileCopyrightText: Copyright contributors to the vLLM project +# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project import asyncio import queue @@ -12,6 +12,7 @@ import vllm.v1.core.single_type_kv_cache_manager as native_kv_managers from pytest_mock import MockerFixture from vllm.v1.kv_cache_interface import FullAttentionSpec, KVCacheConfig, KVCacheGroupSpec +from tests.helpers.kv_layout import build_kv_cache_tensor from vllm_omni.diffusion.data import DiffusionOutput, DiffusionRequestAbortedError from vllm_omni.diffusion.diffusion_engine import DiffusionEngine, DiffusionExecutionMode from vllm_omni.diffusion.diffusion_kv.config import DiffusionKVCacheMode @@ -27,7 +28,6 @@ from vllm_omni.diffusion.sched import ( from vllm_omni.diffusion.sched.interface import CachedRequestData, NewRequestData from vllm_omni.diffusion.worker.utils import BatchRunnerOutput, RunnerOutput from vllm_omni.inputs.data import OmniDiffusionSamplingParams -from tests.helpers.kv_layout import build_kv_cache_tensor pytestmark = [pytest.mark.core_model, pytest.mark.cpu, pytest.mark.diffusion] diff --git a/tests/helpers/kv_layout.py b/tests/helpers/kv_layout.py index 0306cee42..8351de8da 100644 --- a/tests/helpers/kv_layout.py +++ b/tests/helpers/kv_layout.py @@ -1,5 +1,5 @@ # SPDX-License-Identifier: Apache-2.0 -# SPDX-FileCopyrightText: Copyright contributors to the vLLM project +# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project """Shared access to the production ``KVCacheTensor`` layout derivation. Tests deliberately go through the same code production uses diff --git a/vllm_omni/diffusion/diffusion_kv/initialization.py b/vllm_omni/diffusion/diffusion_kv/initialization.py index ffb0c2176..d1a79d934 100644 --- a/vllm_omni/diffusion/diffusion_kv/initialization.py +++ b/vllm_omni/diffusion/diffusion_kv/initialization.py @@ -1,5 +1,5 @@ # SPDX-License-Identifier: Apache-2.0 -# SPDX-FileCopyrightText: Copyright contributors to the vLLM project +# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project from __future__ import annotations diff --git a/vllm_omni/diffusion/diffusion_kv/layout.py b/vllm_omni/diffusion/diffusion_kv/layout.py index 1ba0cd326..960df395d 100644 --- a/vllm_omni/diffusion/diffusion_kv/layout.py +++ b/vllm_omni/diffusion/diffusion_kv/layout.py @@ -1,5 +1,5 @@ # SPDX-License-Identifier: Apache-2.0 -# SPDX-FileCopyrightText: Copyright contributors to the vLLM project +# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project """Physical KV cache layout selection for the native Diffusion paged cache. vLLM 0.29 removed ``AttentionSpec.indexes_kv_by_block_stride`` and replaced it @@ -44,8 +44,8 @@ def resolve_diffusion_kv_cache_layout( A layout already present on the config wins, mirroring upstream's ``resolve_kv_cache_layout``; it is validated against the requirement rather - than silently overwritten, because an inconsistent layout mis-addresses K/V - instead of raising. + than silently overwritten, because an inconsistent layout reads K/V + from the wrong offsets instead of raising. """ required_block_outermost = bool(indexes_kv_by_block_stride) @@ -53,11 +53,7 @@ def resolve_diffusion_kv_cache_layout( if cache_config is None: # A config double with nowhere to record the choice; report what the # requirement implies without pretending it was pinned. - return ( - BLOCK_STRIDE_DIFFUSION_KV_CACHE_LAYOUT - if required_block_outermost - else DEFAULT_DIFFUSION_KV_CACHE_LAYOUT - ) + return BLOCK_STRIDE_DIFFUSION_KV_CACHE_LAYOUT if required_block_outermost else DEFAULT_DIFFUSION_KV_CACHE_LAYOUT current = getattr(cache_config, "kv_cache_layout", None) if current is not None: @@ -72,11 +68,7 @@ def resolve_diffusion_kv_cache_layout( ) return layout - layout = ( - BLOCK_STRIDE_DIFFUSION_KV_CACHE_LAYOUT - if required_block_outermost - else DEFAULT_DIFFUSION_KV_CACHE_LAYOUT - ) + layout = BLOCK_STRIDE_DIFFUSION_KV_CACHE_LAYOUT if required_block_outermost else DEFAULT_DIFFUSION_KV_CACHE_LAYOUT cache_config.kv_cache_layout = layout.name return layout @@ -107,8 +99,7 @@ def adopt_kv_cache_layout(vllm_config: VllmConfig, kv_cache_config) -> KVCacheLa cache_config.kv_cache_layout = name elif current != name: raise ValueError( - f"Worker KV cache layout {current!r} disagrees with the layout " - f"resolved by the control plane ({name!r})." + f"Worker KV cache layout {current!r} disagrees with the layout resolved by the control plane ({name!r})." ) return KVCacheLayout[name] @@ -132,7 +123,7 @@ def build_kv_cache_tensor( The strides come from upstream's own ``compute_layout_strides``, called the way ``vllm.v1.core.kv_cache_utils`` calls it, because hand-computed strides - mis-address K/V silently instead of raising. + read K/V from the wrong offsets silently instead of raising. """ layout = layout or DEFAULT_DIFFUSION_KV_CACHE_LAYOUT diff --git a/vllm_omni/diffusion/diffusion_kv/model_runner_backend.py b/vllm_omni/diffusion/diffusion_kv/model_runner_backend.py index 79a0ded62..20f4db165 100644 --- a/vllm_omni/diffusion/diffusion_kv/model_runner_backend.py +++ b/vllm_omni/diffusion/diffusion_kv/model_runner_backend.py @@ -293,9 +293,7 @@ class DiffusionKVModelRunnerBackend: # and dropped attn_groups/cache_dtype from the signature. adopt_kv_cache_layout(self.vllm_config, kv_cache_config) for layer_adapter in self._kv_cache_layer_adapters.values(): - assert_backend_layout_supported( - self.vllm_config, getattr(layer_adapter, "attn_backend", None) - ) + assert_backend_layout_supported(self.vllm_config, getattr(layer_adapter, "attn_backend", None)) init_kv_cache( kv_caches, self.vllm_config.compilation_config.static_forward_context, diff --git a/vllm_omni/experimental/ar_diffusion/kv_cache/manager.py b/vllm_omni/experimental/ar_diffusion/kv_cache/manager.py index 24f09210d..f4cb416f6 100644 --- a/vllm_omni/experimental/ar_diffusion/kv_cache/manager.py +++ b/vllm_omni/experimental/ar_diffusion/kv_cache/manager.py @@ -1,4 +1,5 @@ # SPDX-License-Identifier: Apache-2.0 +# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project """Engine-level KV cache orchestration for AR-Diffusion models. This is the *body* of AR-Diffusion's KV management: it owns a vLLM ``KVCacheManager`` (a