anodyine/vllm-omni
1
0 Fork 0
Code Issues 4 Pull Requests 2 Actions Packages Projects Releases Wiki Activity

[CI] Expand E2E source_file_dependencies for shared model code (#5994)

Browse Source 
Signed-off-by: wangyu <410167048@qq.com>
Signed-off-by: wangyu <53896905+yenuo26@users.noreply.github.com>
This commit is contained in:
wangyu authored and GitHub committed 2026-08-15 12:01:20 +08:00
1 parent b95c73742d
commit d671e6b0db
14 files changed
+135 -228
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

.buildkite/cuda/test-merge.yml
+26 -91
View File
@@ -76,6 +76,9 @@ steps:
- tests/e2e/online_serving/test_qwen3_tts_customvoice.py
- tests/e2e/offline_inference/test_qwen3_tts_customvoice.py
- vllm_omni/model_executor/models/qwen3_tts/
- vllm_omni/model_executor/models/common/qwen3_code_predictor.py
- vllm_omni/model_executor/models/common/snake_activation.py
- vllm_omni/model_executor/models/common/alias_free_activation.py
- vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
- vllm_omni/deploy/qwen3_tts.yaml
commands:
@@ -92,6 +95,9 @@ steps:
- tests/e2e/online_serving/test_qwen3_tts_base.py
- tests/e2e/offline_inference/test_qwen3_tts_base.py
- vllm_omni/model_executor/models/qwen3_tts/
- vllm_omni/model_executor/models/common/qwen3_code_predictor.py
- vllm_omni/model_executor/models/common/snake_activation.py
- vllm_omni/model_executor/models/common/alias_free_activation.py
- vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
- vllm_omni/deploy/qwen3_tts.yaml
- .buildkite/common/scripts/run_cov_split.sh
@@ -109,27 +115,15 @@ steps:
-- -m 'advanced_model and cuda' --run-level 'advanced_model'
mirror_hardwares: l4_1
- label: "TTS · Higgs-Audio-V3 Test"
source_file_dependencies:
- tests/e2e/online_serving/test_higgs_audio_v3.py
- vllm_omni/model_executor/models/higgs_audio_v3/
- vllm_omni/model_executor/stage_input_processors/higgs_audio_v3.py
- vllm_omni/deploy/higgs_multimodal_qwen3.yaml
commands:
- |
timeout 20m bash -c "
export VLLM_WORKER_MULTIPROC_METHOD=spawn
export VLLM_USE_DEEP_GEMM=0
export VLLM_MOE_USE_DEEP_GEMM=0
pytest -s -v tests/e2e/online_serving/test_higgs_audio_v3.py -m 'advanced_model' --run-level 'advanced_model'
"
mirror_hardwares: l4_1
- label: "Omni · Qwen3-Omni Test"
source_file_dependencies:
- tests/e2e/offline_inference/test_qwen3_omni.py
- tests/e2e/online_serving/test_qwen3_omni.py
- vllm_omni/model_executor/models/qwen3_omni/
- vllm_omni/model_executor/models/common/qwen3_code_predictor.py
- vllm_omni/model_executor/models/common/snake_activation.py
- vllm_omni/model_executor/models/qwen2_5_omni/qwen2_5_omni_thinker.py
- vllm_omni/model_executor/models/qwen3_tts/cuda_graph_decoder_wrapper.py
- vllm_omni/model_executor/stage_input_processors/qwen3_omni.py
- vllm_omni/deploy/qwen3_omni_moe.yaml
timeout_in_minutes: 50
@@ -141,6 +135,11 @@ steps:
source_file_dependencies:
- tests/e2e/offline_inference/test_minicpmo_4_5.py
- tests/e2e/online_serving/test_minicpmo_4_5.py
- vllm_omni/model_executor/models/minicpmo_4_5/
- vllm_omni/model_executor/models/step_audio2/
- vllm_omni/model_executor/models/cosyvoice3/code2wav_core/
- vllm_omni/model_executor/models/cosyvoice3/utils.py
- vllm_omni/model_executor/stage_input_processors/minicpmo_4_5_omni.py
- vllm_omni/deploy/minicpmo_4_5.yaml
timeout_in_minutes: 50
commands:
@@ -162,70 +161,35 @@ steps:
- tests/e2e/online_serving/helpers/minicpmo_realtime_duplex_scenarios.py
- tests/e2e/online_serving/run_minicpmo_realtime_duplex_multi_session.py
- tests/e2e/online_serving/run_minicpmo_realtime_duplex_soft_interrupt.py
- vllm_omni/deploy/minicpmo_4_5.yaml
- tests/e2e/online_serving/helpers/minicpmo_4_5_duplex.py
- vllm_omni/deploy/minicpmo_4_5_duplex.yaml
- vllm_omni/experimental/fullduplex/
- vllm_omni/model_executor/models/minicpmo_4_5/
- vllm_omni/model_executor/models/step_audio2/
- vllm_omni/model_executor/models/cosyvoice3/code2wav_core/
- vllm_omni/model_executor/models/cosyvoice3/utils.py
- vllm_omni/model_executor/stage_input_processors/minicpmo_4_5_omni.py
- vllm_omni/core/sched/
- vllm_omni/outputs/multimodal_accumulation.py
- vllm_omni/engine/stage_init_utils.py
- vllm_omni/engine/async_omni_engine.py
- vllm_omni/engine/orchestrator.py
- vllm_omni/entrypoints/cli/serve.py
- vllm_omni/entrypoints/openai/api_server.py
- vllm_omni/entrypoints/openai/duplex_capability.py
- vllm_omni/config/stage_config.py
timeout_in_minutes: 50
commands:
- pytest -s -v tests/e2e/online_serving/test_minicpmo_4_5_duplex.py -m 'advanced_model and cuda' --run-level 'advanced_model'
mirror_hardwares: h100_1
- label: "TTS · VoxCPM2 Test"
source_file_dependencies:
- tests/e2e/offline_inference/test_voxcpm2_tts.py
- tests/e2e/online_serving/test_voxcpm2_tts.py
- vllm_omni/model_executor/models/voxcpm2/
- vllm_omni/deploy/voxcpm2.yaml
timeout_in_minutes: 30
commands:
- |
timeout 30m bash -c "
export VLLM_LOGGING_LEVEL=DEBUG
export VLLM_WORKER_MULTIPROC_METHOD=spawn
pytest -s -v tests/e2e/offline_inference/test_voxcpm2_tts.py tests/e2e/online_serving/test_voxcpm2_tts.py -m 'advanced_model and cuda' --run-level 'advanced_model'
"
mirror_hardwares: l4_1
- label: "Diffusion · Qwen Image Edit Test"
source_file_dependencies:
- tests/e2e/online_serving/test_qwen_image_edit.py
- vllm_omni/diffusion/models/qwen_image/
timeout_in_minutes: 20
commands:
- pytest -s -v tests/e2e/online_serving/test_qwen_image_edit.py -m 'advanced_model and cuda' --run-level 'advanced_model'
mirror_hardwares: h100_1
- label: "Diffusion · Bagel Test"
source_file_dependencies:
- tests/e2e/offline_inference/test_bagel.py
- tests/e2e/online_serving/test_bagel.py
- vllm_omni/model_executor/models/bagel/
- vllm_omni/diffusion/models/bagel/
- vllm_omni/model_executor/stage_input_processors/bagel.py
- vllm_omni/deploy/bagel.yaml
- .buildkite/common/scripts/run_cov_split.sh
- pyproject.toml
# 20 -> 40: the coverage split runs the model twice.
timeout_in_minutes: 40
commands:
- |
.buildkite/common/scripts/run_cov_split.sh \
--model-id bagel \
--offline tests/e2e/offline_inference/test_bagel.py \
--online tests/e2e/online_serving/test_bagel.py \
-- -m 'advanced_model and cuda' --run-level 'advanced_model'
mirror_hardwares: h100_1
- label: "Diffusion · Qwen Image Test"
source_file_dependencies:
- tests/e2e/online_serving/test_qwen_image.py
- vllm_omni/diffusion/models/qwen_image/
- vllm_omni/diffusion/models/dmd2/
- vllm_omni/diffusion/distributed/autoencoders/autoencoder_kl_qwenimage.py
commands:
- |
timeout 40m bash -c '
@@ -234,23 +198,12 @@ steps:
'
mirror_hardwares: h100_1
- label: "Diffusion · Qwen Image Layered Test"
source_file_dependencies:
- tests/e2e/online_serving/test_qwen_image_layered.py
- vllm_omni/diffusion/models/qwen_image/
commands:
- |
timeout 40m bash -c "
export VLLM_IMAGE_FETCH_TIMEOUT=60
pytest -s -v tests/e2e/online_serving/test_qwen_image_layered.py -m 'advanced_model and cuda' --run-level 'advanced_model'
"
mirror_hardwares: h100_1
- label: "Diffusion · Wan22 Test"
source_file_dependencies:
- tests/e2e/offline_inference/test_wan22_t2v.py
- tests/e2e/online_serving/test_wan22_t2v.py
- vllm_omni/diffusion/models/wan2_2/
- vllm_omni/diffusion/models/dmd2/
commands:
- |
timeout 40m bash -c "
@@ -258,21 +211,3 @@ steps:
pytest -s -v tests/e2e/offline_inference/test_wan22_t2v.py tests/e2e/online_serving/test_wan22_t2v.py -m 'advanced_model and cuda and diffusion' --run-level 'advanced_model'
"
mirror_hardwares: h100_1
- label: "Diffusion · Z Image Test"
source_file_dependencies:
- tests/e2e/offline_inference/test_z_image.py
- tests/e2e/online_serving/test_z_image.py
- vllm_omni/diffusion/models/z_image/
timeout_in_minutes: 30
commands:
- pytest -s -v tests/e2e/offline_inference/test_z_image.py tests/e2e/online_serving/test_z_image.py -m 'advanced_model and cuda and diffusion' --run-level 'advanced_model'
mirror_hardwares: l4_1
- label: "Diffusion · FLUX.2 Klein Test"
source_file_dependencies:
- tests/e2e/offline_inference/test_flux2_klein.py
- vllm_omni/diffusion/models/flux2_klein/
timeout_in_minutes: 30
commands:
- pytest -s -v tests/e2e/offline_inference/test_flux2_klein.py -m 'advanced_model and cuda and diffusion' --run-level 'advanced_model'
mirror_hardwares: l4_1
.buildkite/cuda/test-ready.yml
+34 -69
View File
@@ -169,6 +169,10 @@ steps:
source_file_dependencies:
- tests/e2e/online_serving/test_qwen3_omni.py
- vllm_omni/model_executor/models/qwen3_omni/
- vllm_omni/model_executor/models/common/qwen3_code_predictor.py
- vllm_omni/model_executor/models/common/snake_activation.py
- vllm_omni/model_executor/models/qwen2_5_omni/qwen2_5_omni_thinker.py
- vllm_omni/model_executor/models/qwen3_tts/cuda_graph_decoder_wrapper.py
- vllm_omni/model_executor/stage_input_processors/qwen3_omni.py
- vllm_omni/deploy/qwen3_omni_moe.yaml
commands:
@@ -181,6 +185,11 @@ steps:
- label: "Omni · MiniCPM-o 4.5 Offline Test"
source_file_dependencies:
- tests/e2e/offline_inference/test_minicpmo_4_5.py
- vllm_omni/model_executor/models/minicpmo_4_5/
- vllm_omni/model_executor/models/step_audio2/
- vllm_omni/model_executor/models/cosyvoice3/code2wav_core/
- vllm_omni/model_executor/models/cosyvoice3/utils.py
- vllm_omni/model_executor/stage_input_processors/minicpmo_4_5_omni.py
- vllm_omni/deploy/minicpmo_4_5.yaml
commands:
- |
@@ -192,6 +201,11 @@ steps:
- label: "Omni · MiniCPM-o 4.5 Online Test"
source_file_dependencies:
- tests/e2e/online_serving/test_minicpmo_4_5.py
- vllm_omni/model_executor/models/minicpmo_4_5/
- vllm_omni/model_executor/models/step_audio2/
- vllm_omni/model_executor/models/cosyvoice3/code2wav_core/
- vllm_omni/model_executor/models/cosyvoice3/utils.py
- vllm_omni/model_executor/stage_input_processors/minicpmo_4_5_omni.py
- vllm_omni/deploy/minicpmo_4_5.yaml
commands:
- |
@@ -210,16 +224,24 @@ steps:
- tests/e2e/online_serving/helpers/minicpmo_realtime_duplex_scenarios.py
- tests/e2e/online_serving/run_minicpmo_realtime_duplex_multi_session.py
- tests/e2e/online_serving/run_minicpmo_realtime_duplex_soft_interrupt.py
- vllm_omni/deploy/minicpmo_4_5.yaml
- tests/e2e/online_serving/helpers/minicpmo_4_5_duplex.py
- vllm_omni/deploy/minicpmo_4_5_duplex.yaml
- vllm_omni/experimental/fullduplex/
- vllm_omni/model_executor/models/minicpmo_4_5/
- vllm_omni/model_executor/models/step_audio2/
- vllm_omni/model_executor/models/cosyvoice3/code2wav_core/
- vllm_omni/model_executor/models/cosyvoice3/utils.py
- vllm_omni/model_executor/stage_input_processors/minicpmo_4_5_omni.py
- vllm_omni/core/sched/
- vllm_omni/outputs/multimodal_accumulation.py
- vllm_omni/engine/stage_init_utils.py
- vllm_omni/engine/async_omni_engine.py
- vllm_omni/engine/orchestrator.py
- vllm_omni/entrypoints/cli/serve.py
- vllm_omni/entrypoints/openai/api_server.py
- vllm_omni/entrypoints/openai/duplex_capability.py
- vllm_omni/config/stage_config.py
commands:
- |
timeout 20m bash -c "
@@ -239,25 +261,13 @@ steps:
- pytest -s -v tests/dfx/perf/scripts/run_benchmark.py --test-config-file tests/dfx/perf/tests/test_minicpmo_4_5_ready.json -m 'core_model and cuda'
mirror_hardwares: h100_1
- label: "TTS · VoxCPM2 Test"
source_file_dependencies:
- tests/e2e/online_serving/test_voxcpm2_tts.py
- vllm_omni/model_executor/models/voxcpm2/
- vllm_omni/deploy/voxcpm2.yaml
timeout_in_minutes: 30
commands:
- |
timeout 30m bash -c "
export VLLM_LOGGING_LEVEL=DEBUG
export VLLM_WORKER_MULTIPROC_METHOD=spawn
pytest -s -v tests/e2e/online_serving/test_voxcpm2_tts.py -m 'core_model' --run-level 'core_model'
"
mirror_hardwares: l4_1
- label: "TTS · Qwen3-TTS CustomVoice Test"
source_file_dependencies:
- tests/e2e/online_serving/test_qwen3_tts_customvoice.py
- vllm_omni/model_executor/models/qwen3_tts/
- vllm_omni/model_executor/models/common/qwen3_code_predictor.py
- vllm_omni/model_executor/models/common/snake_activation.py
- vllm_omni/model_executor/models/common/alias_free_activation.py
- vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
- vllm_omni/deploy/qwen3_tts.yaml
commands:
@@ -272,6 +282,9 @@ steps:
source_file_dependencies:
- tests/e2e/online_serving/test_qwen3_tts_base.py
- vllm_omni/model_executor/models/qwen3_tts/
- vllm_omni/model_executor/models/common/qwen3_code_predictor.py
- vllm_omni/model_executor/models/common/snake_activation.py
- vllm_omni/model_executor/models/common/alias_free_activation.py
- vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
- vllm_omni/deploy/qwen3_tts.yaml
commands:
@@ -284,52 +297,11 @@ steps:
"
mirror_hardwares: l4_1
- label: "TTS · Higgs-Audio-V3 Test"
source_file_dependencies:
- tests/e2e/online_serving/test_higgs_audio_v3.py
- vllm_omni/model_executor/models/higgs_audio_v3/
- vllm_omni/model_executor/stage_input_processors/higgs_audio_v3.py
- vllm_omni/deploy/higgs_multimodal_qwen3.yaml
commands:
- |
timeout 20m bash -c "
export VLLM_WORKER_MULTIPROC_METHOD=spawn
export VLLM_USE_DEEP_GEMM=0
export VLLM_MOE_USE_DEEP_GEMM=0
pytest -s -v tests/e2e/online_serving/test_higgs_audio_v3.py -m 'core_model' --run-level 'core_model'
"
mirror_hardwares: l4_1
- label: "Diffusion · Qwen Image Edit Test"
source_file_dependencies:
- tests/e2e/online_serving/test_qwen_image_edit.py
- vllm_omni/diffusion/models/qwen_image/
commands:
- |
timeout 20m bash -c "
pytest -s -v tests/e2e/online_serving/test_qwen_image_edit.py -m 'core_model' --run-level 'core_model'
"
mirror_hardwares: h100_1
- label: "Diffusion · Bagel Test"
source_file_dependencies:
- tests/e2e/online_serving/test_bagel.py
- vllm_omni/model_executor/models/bagel/
- vllm_omni/diffusion/models/bagel/
- vllm_omni/model_executor/stage_input_processors/bagel.py
- vllm_omni/deploy/bagel.yaml
commands:
- |
timeout 40m bash -c "
export VLLM_IMAGE_FETCH_TIMEOUT=60
pytest -s -v tests/e2e/online_serving/test_bagel.py -m 'core_model' --run-level 'core_model'
"
mirror_hardwares: h100_1
- label: "Diffusion · Wan22 Test"
source_file_dependencies:
- tests/e2e/online_serving/test_wan22_t2v.py
- vllm_omni/diffusion/models/wan2_2/
- vllm_omni/diffusion/models/dmd2/
commands:
- |
timeout 40m bash -c "
@@ -342,6 +314,9 @@ steps:
source_file_dependencies:
- tests/e2e/online_serving/test_cosmos3.py
- vllm_omni/diffusion/models/cosmos3/
- vllm_omni/diffusion/distributed/autoencoders/autoencoder_kl_wan.py
- vllm_omni/experimental/world_models/
- vllm_omni/model_extras/cosmos3.py
commands:
- |
timeout 40m bash -c "
@@ -354,6 +329,8 @@ steps:
source_file_dependencies:
- tests/e2e/online_serving/test_qwen_image.py
- vllm_omni/diffusion/models/qwen_image/
- vllm_omni/diffusion/models/dmd2/
- vllm_omni/diffusion/distributed/autoencoders/autoencoder_kl_qwenimage.py
commands:
- |
timeout 40m bash -c "
@@ -361,15 +338,3 @@ steps:
pytest -s -v tests/e2e/online_serving/test_qwen_image.py -m 'core_model' --run-level 'core_model'
"
mirror_hardwares: h100_1
- label: "Diffusion · Qwen Image Layered Test"
source_file_dependencies:
- tests/e2e/online_serving/test_qwen_image_layered.py
- vllm_omni/diffusion/models/qwen_image/
commands:
- |
timeout 40m bash -c "
export VLLM_IMAGE_FETCH_TIMEOUT=60
pytest -s -v tests/e2e/online_serving/test_qwen_image_layered.py -m 'core_model' --run-level 'core_model'
"
mirror_hardwares: h100_1
docs/contributing/ci/ci_settings.md
+9 -6
View File
@@ -329,11 +329,13 @@ On PR builds, `upload_pipeline.py` logs `skip '…' (no changes under …)` for
### Per-model coverage
Pilot (v1): two Merge-tier (L3) jobs in `.buildkite/cuda/test-merge.yml` upload
per-model, per-entry-mode coverage as Buildkite artifacts — "Diffusion · Bagel
Test" and "TTS · Qwen3-TTS Base Test". Both run on a single GPU so the pilot is
cheap to reproduce. This is a pilot, not a rollout: no dashboard/visualization
lives in this repo, and nothing is gated on coverage.
Pilot (v1): one Merge-tier (L3) job in `.buildkite/cuda/test-merge.yml` uploads
per-model, per-entry-mode coverage as Buildkite artifacts — "TTS · Qwen3-TTS Base
Test". It runs on a single GPU so the pilot is cheap to reproduce. (A second
diffusion pilot, "Diffusion · Bagel Test", was retired from merge; reintroduce
another diffusion coverage job if a second pilot is needed again.) This is a
pilot, not a rollout: no dashboard/visualization lives in this repo, and nothing
is gated on coverage.
**Naming convention**: `coverage-<model_id>-<mode>-<step_id>.xml.gz`, where
`<model_id>` is the model's directory name under
@@ -410,7 +412,8 @@ runs a coverage job needs the same.
List both `run_cov_split.sh` and `pyproject.toml` in every opted-in job's
`source_file_dependencies` — both change what the job measures, so without them a
change there is filtered out of normal PR builds and only surfaces in a later
nightly. `tests/buildkite/test_upload_pipeline.py` asserts this for the pilots.
nightly. `tests/buildkite/test_upload_pipeline.py` covers the filter behavior with
a synthetic job (it does not pin real merge labels).
Editing only the surrounding CI YAML still does not schedule them, so a PR that
touches just the wiring needs a full E2E run (or the commands run on a GPU host)
to produce artifacts. When checking a new model's
tests/buildkite/test_upload_pipeline.py
+36 -21
View File
@@ -39,8 +39,6 @@ def _render(changed_files: list[str]) -> str:
def test_bootstrap_if_injected_by_step_key() -> None:
import yaml
rendered = _render_bootstrap_pipeline(
BOOTSTRAP_STEPS_TEMPLATE,
decision=resolve_ci_decision([]),
@@ -142,12 +140,7 @@ def test_mirror_hardwares_a2b3_npu_4_expands_agents_image_and_plugins() -> None:
]
COVERAGE_PILOT_LABELS = ("Diffusion · Bagel Test", "TTS · Qwen3-TTS Base Test")
MERGE_PIPELINE = Path(".buildkite/cuda/test-merge.yml")
def _surviving_labels(changed_files: list[str]) -> set[str]:
doc = yaml.safe_load(MERGE_PIPELINE.read_text(encoding="utf-8"))
def _surviving_labels(doc: dict, changed_files: list[str]) -> set[str]:
rendered = _render_test_pipeline(doc, changed_files=changed_files)
labels: set[str] = set()
@@ -163,6 +156,29 @@ def _surviving_labels(changed_files: list[str]) -> set[str]:
return labels
# Synthetic coverage-style job: shared inputs that change what the split measures.
_COVERAGE_SHARED_INPUTS_DOC = {
"steps": [
{
"label": "Coverage Pilot",
"source_file_dependencies": [
"tests/e2e/online_serving/test_example.py",
".buildkite/common/scripts/run_cov_split.sh",
"pyproject.toml",
],
"commands": [".buildkite/common/scripts/run_cov_split.sh --model-id example"],
},
{
"label": "Unrelated Model Test",
"source_file_dependencies": [
"tests/e2e/online_serving/test_other.py",
],
"commands": ["pytest -sv tests/e2e/online_serving/test_other.py"],
},
],
}
@pytest.mark.parametrize(
"changed_file",
[
@@ -170,18 +186,17 @@ def _surviving_labels(changed_files: list[str]) -> set[str]:
"pyproject.toml",
],
)
def test_coverage_pilots_selected_by_their_shared_inputs(changed_file: str) -> None:
"""The split helper and the coverage config both change what these jobs measure.
Without them in source_file_dependencies the jobs are filtered out of normal
PR builds, so a regression would only surface in a later nightly.
"""
labels = _surviving_labels([changed_file])
for pilot in COVERAGE_PILOT_LABELS:
assert pilot in labels, f"{pilot} dropped when {changed_file} changed"
def test_coverage_shared_inputs_select_dependent_job(changed_file: str) -> None:
"""Jobs that list coverage shared inputs must stay selected when those files change."""
labels = _surviving_labels(_COVERAGE_SHARED_INPUTS_DOC, [changed_file])
assert "Coverage Pilot" in labels
assert "Unrelated Model Test" not in labels
def test_coverage_pilots_not_selected_by_unrelated_change() -> None:
labels = _surviving_labels(["vllm_omni/entrypoints/openai/serving_chat.py"])
for pilot in COVERAGE_PILOT_LABELS:
assert pilot not in labels
def test_coverage_shared_inputs_ignored_for_unrelated_change() -> None:
labels = _surviving_labels(
_COVERAGE_SHARED_INPUTS_DOC,
["vllm_omni/entrypoints/openai/serving_chat.py"],
)
assert "Coverage Pilot" not in labels
assert "Unrelated Model Test" not in labels
tests/e2e/offline_inference/test_bagel.py
+1 -1
View File
@@ -46,7 +46,7 @@ REFERENCE_TEXT_IMG2TEXT = (
_OMNI_RUNNER_PARAM = (MODEL_NAME, STAGE_CONFIG)
pytestmark = [
pytest.mark.advanced_model,
pytest.mark.slow,
pytest.mark.diffusion,
pytest.mark.parametrize("omni_runner", [_OMNI_RUNNER_PARAM], indirect=True),
]
tests/e2e/offline_inference/test_flux2_klein.py
+2 -2
View File
@@ -72,7 +72,7 @@ def _send_inpaint_with_generator(
)
@pytest.mark.advanced_model
@pytest.mark.slow
@pytest.mark.diffusion
@hardware_test(res={"cuda": "L4"}, num_cards=1)
def test_flux2_klein_inpaint_deterministic(omni_runner_handler: OmniRunnerHandler):
@@ -93,7 +93,7 @@ def test_flux2_klein_inpaint_deterministic(omni_runner_handler: OmniRunnerHandle
)
@pytest.mark.advanced_model
@pytest.mark.slow
@pytest.mark.diffusion
@hardware_test(res={"cuda": "L4"}, num_cards=1)
def test_flux2_klein_inpaint_different_seeds_different_output(omni_runner_handler: OmniRunnerHandler):
tests/e2e/offline_inference/test_voxcpm2_tts.py
+3 -6
View File
@@ -39,8 +39,7 @@ def _extract_audio(multimodal_output: dict) -> torch.Tensor:
return audio
@pytest.mark.core_model
@pytest.mark.advanced_model
@pytest.mark.slow
@pytest.mark.tts
@hardware_test(res={"cuda": "L4"}, num_cards=1)
def test_voxcpm2_zero_shot_001(omni_runner: OmniRunner) -> None:
@@ -53,8 +52,7 @@ def test_voxcpm2_zero_shot_001(omni_runner: OmniRunner) -> None:
assert 0.5 < duration_s < 30.0, f"Audio duration out of range: {duration_s:.2f}s"
@pytest.mark.core_model
@pytest.mark.advanced_model
@pytest.mark.slow
@pytest.mark.tts
@hardware_test(res={"cuda": "L4"}, num_cards=1)
def test_voxcpm2_voice_clone_002(omni_runner: OmniRunner) -> None:
@@ -95,8 +93,7 @@ def test_voxcpm2_voice_clone_002(omni_runner: OmniRunner) -> None:
assert 0.5 < duration_s < 30.0, f"Audio duration out of range: {duration_s:.2f}s"
@pytest.mark.core_model
@pytest.mark.advanced_model
@pytest.mark.slow
@pytest.mark.tts
@hardware_test(res={"cuda": "L4"}, num_cards=1)
def test_voxcpm2_prefill_decode_mixed_batch_003(omni_runner: OmniRunner) -> None:
tests/e2e/offline_inference/test_z_image.py
+1 -1
View File
@@ -11,7 +11,7 @@ MODEL = "Tongyi-MAI/Z-Image-Turbo"
PROMPT = "A high-detail studio photo of an orange tabby cat sitting on a laptop keyboard."
@pytest.mark.advanced_model
@pytest.mark.slow
@pytest.mark.diffusion
@hardware_test(res={"cuda": "L4", "rocm": "MI325", "xpu": "B60"}, num_cards={"cuda": 1, "rocm": 1, "xpu": 2})
@pytest.mark.parametrize("omni_runner", [(MODEL, None)], indirect=True)
tests/e2e/online_serving/test_bagel.py
+2 -4
View File
@@ -75,8 +75,7 @@ def _build_img2img_messages(prompt: str, image_b64: str) -> list[dict]:
]
@pytest.mark.core_model
@pytest.mark.advanced_model
@pytest.mark.slow
@pytest.mark.diffusion
@hardware_test(res={"cuda": "H100"})
@pytest.mark.parametrize("omni_server", test_params, indirect=True)
@@ -98,8 +97,7 @@ def test_bagel_text2img_online(omni_server, openai_client) -> None:
openai_client.send_diffusion_request(request_config)
@pytest.mark.core_model
@pytest.mark.advanced_model
@pytest.mark.slow
@pytest.mark.diffusion
@hardware_test(res={"cuda": "H100", "rocm": "MI325"})
@pytest.mark.parametrize("omni_server", test_params, indirect=True)
tests/e2e/online_serving/test_higgs_audio_v3.py
+12 -15
View File
@@ -59,8 +59,7 @@ _REF_TEXT = "Okay. Yeah. I resent you. I love you. I respect you. But you know w
class TestHiggsAudioV3OnlineHappyPath:
"""Plain-text -> audio happy paths against the live HTTP server."""
@pytest.mark.core_model
@pytest.mark.advanced_model
@pytest.mark.slow
@pytest.mark.tts
@hardware_test(res={"cuda": "L4"}, num_cards=1)
def test_plain_text_wav(self, omni_server, openai_client) -> None:
@@ -79,7 +78,7 @@ class TestHiggsAudioV3OnlineHappyPath:
}
)
@pytest.mark.core_model
@pytest.mark.slow
@pytest.mark.tts
@hardware_test(res={"cuda": "L4"}, num_cards=1)
def test_plain_text_with_max_new_tokens(self, omni_server, openai_client) -> None:
@@ -97,8 +96,7 @@ class TestHiggsAudioV3OnlineHappyPath:
)
@pytest.mark.skip(reason="issue#4411")
@pytest.mark.core_model
@pytest.mark.advanced_model
@pytest.mark.slow
@pytest.mark.tts
@hardware_test(res={"cuda": "L4"}, num_cards=1)
def test_concurrent_plain_text(self, omni_server, openai_client) -> None:
@@ -115,7 +113,7 @@ class TestHiggsAudioV3OnlineHappyPath:
)
@pytest.mark.skip(reason="issue#4411")
@pytest.mark.advanced_model
@pytest.mark.slow
@pytest.mark.tts
@hardware_test(res={"cuda": "L4"}, num_cards=1)
def test_plain_text_pcm_streaming(self, omni_server, openai_client) -> None:
@@ -147,7 +145,7 @@ class TestHiggsAudioV3OnlineHappyPath:
)
@pytest.mark.skip(reason="issue#4411")
@pytest.mark.advanced_model
@pytest.mark.slow
@pytest.mark.tts
@hardware_test(res={"cuda": "L4"}, num_cards=1)
def test_concurrent_pcm_streaming(self, omni_server, openai_client) -> None:
@@ -189,7 +187,7 @@ class TestHiggsAudioV3OnlineInlineControlTokens:
They do not assert audio quality (out of scope for CI).
"""
@pytest.mark.core_model
@pytest.mark.slow
@pytest.mark.tts
@hardware_test(res={"cuda": "L4"}, num_cards=1)
def test_inline_emotion_and_expressive(self, omni_server, openai_client) -> None:
@@ -205,7 +203,7 @@ class TestHiggsAudioV3OnlineInlineControlTokens:
}
)
@pytest.mark.core_model
@pytest.mark.slow
@pytest.mark.tts
@hardware_test(res={"cuda": "L4"}, num_cards=1)
def test_inline_style_whispering(self, omni_server, openai_client) -> None:
@@ -221,7 +219,7 @@ class TestHiggsAudioV3OnlineInlineControlTokens:
}
)
@pytest.mark.core_model
@pytest.mark.slow
@pytest.mark.tts
@hardware_test(res={"cuda": "L4"}, num_cards=1)
def test_inline_prosody_speed_and_pitch(self, omni_server, openai_client) -> None:
@@ -239,7 +237,7 @@ class TestHiggsAudioV3OnlineInlineControlTokens:
}
)
@pytest.mark.core_model
@pytest.mark.slow
@pytest.mark.tts
@hardware_test(res={"cuda": "L4"}, num_cards=1)
def test_inline_pause_mid_text(self, omni_server, openai_client) -> None:
@@ -255,7 +253,7 @@ class TestHiggsAudioV3OnlineInlineControlTokens:
}
)
@pytest.mark.core_model
@pytest.mark.slow
@pytest.mark.tts
@hardware_test(res={"cuda": "L4"}, num_cards=1)
def test_inline_sfx_with_onomatopoeia(self, omni_server, openai_client) -> None:
@@ -279,8 +277,7 @@ class TestHiggsAudioV3OnlineInlineControlTokens:
class TestHiggsAudioV3OnlineVoiceClone:
"""Voice clone via the two payload shapes the model serves."""
@pytest.mark.core_model
@pytest.mark.advanced_model
@pytest.mark.slow
@pytest.mark.tts
@hardware_test(res={"cuda": "L4"}, num_cards=1)
def test_voice_clone_ref_audio_ref_text(self, omni_server, openai_client) -> None:
@@ -297,7 +294,7 @@ class TestHiggsAudioV3OnlineVoiceClone:
}
)
@pytest.mark.core_model
@pytest.mark.slow
@pytest.mark.tts
@hardware_test(res={"cuda": "L4"}, num_cards=1)
def test_voice_clone_references_alias(self, omni_server, openai_client) -> None:
tests/e2e/online_serving/test_qwen_image_edit.py
+4 -5
View File
@@ -6,7 +6,7 @@ Online serving tests for the Qwen-Image-Edit family (image-to-image via chat com
- ``test_single_image_to_image_001``: ``Qwen/Qwen-Image-Edit`` — one reference image, fixed 512×512.
- ``test_multi_images_to_image_001``: ``Qwen/Qwen-Image-Edit-2511`` — two reference images, fixed 512×512.
- ``test_different_sizes_001``: ``Qwen/Qwen-Image-Edit-2511`` only, ``advanced_model`` — mixed input
- ``test_different_sizes_001``: ``Qwen/Qwen-Image-Edit-2511`` only, ``slow`` — mixed input
resolutions; ``extra_body`` uses per-output ``width``/``height`` lists; the test client sends one
scalar-size request per list index in parallel and merges images (see ``OpenAIClientHandler.send_diffusion_request``).
@@ -52,7 +52,7 @@ def _get_diffusion_feature_cases(model: str):
]
@pytest.mark.advanced_model
@pytest.mark.slow
@pytest.mark.diffusion
@pytest.mark.parametrize(
"omni_server",
@@ -82,8 +82,7 @@ def test_single_image_to_image_001(omni_server: OmniServer, openai_client: OpenA
openai_client.send_diffusion_request(request_config)
@pytest.mark.core_model
@pytest.mark.advanced_model
@pytest.mark.slow
@pytest.mark.diffusion
@pytest.mark.parametrize(
"omni_server",
@@ -114,7 +113,7 @@ def test_multi_images_to_image_001(omni_server: OmniServer, openai_client: OpenA
openai_client.send_diffusion_request(request_config)
@pytest.mark.advanced_model
@pytest.mark.slow
@pytest.mark.diffusion
@pytest.mark.parametrize(
"omni_server",
tests/e2e/online_serving/test_qwen_image_layered.py
+1 -2
View File
@@ -44,8 +44,7 @@ def _get_diffusion_feature_cases(model: str):
]
@pytest.mark.core_model
@pytest.mark.advanced_model
@pytest.mark.slow
@pytest.mark.diffusion
@pytest.mark.parametrize("omni_server", _get_diffusion_feature_cases(MODEL), indirect=True)
def test_single_image_to_image_001(omni_server: OmniServer, openai_client: OpenAIClientHandler):
tests/e2e/online_serving/test_voxcpm2_tts.py
+2 -3
View File
@@ -46,8 +46,7 @@ tts_server_params = [
]
@pytest.mark.advanced_model
@pytest.mark.core_model
@pytest.mark.slow
@pytest.mark.tts
@hardware_test(res={"cuda": "L4"}, num_cards=1)
@pytest.mark.parametrize("omni_server", tts_server_params, indirect=True)
@@ -72,7 +71,7 @@ def test_text_to_audio_001(omni_server, openai_client) -> None:
openai_client.send_audio_speech_request(request_config, request_num=MAX_CONCURRENT)
@pytest.mark.advanced_model
@pytest.mark.slow
@pytest.mark.tts
@hardware_test(res={"cuda": "L4"}, num_cards=1)
@pytest.mark.parametrize("omni_server", tts_server_params, indirect=True)
tests/e2e/online_serving/test_z_image.py
+2 -2
View File
@@ -32,7 +32,7 @@ def _get_diffusion_feature_cases():
]
@pytest.mark.advanced_model
@pytest.mark.slow
@pytest.mark.diffusion
@pytest.mark.parametrize(
"omni_server",
@@ -60,7 +60,7 @@ def test_basic_001(
openai_client.send_diffusion_request(request_config)
@pytest.mark.advanced_model
@pytest.mark.slow
@pytest.mark.diffusion
@pytest.mark.parametrize(
"omni_server",