Skip to content

[diffusion] Add cache-dit CI tests #19213

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
mickqian merged 11 commits into from
May 10, 2026
+60 −0
Merged

[diffusion] Add cache-dit CI tests #19213

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
11 commits
Select commit Hold shift + click to select a range
0593ced
add
qimcis Feb 24, 2026
5c45e52
fix
qimcis Feb 24, 2026
10ffede
add
qimcis Feb 24, 2026
2ff693b
fix
qimcis Feb 24, 2026
018ba0b
scm
qimcis Feb 25, 2026
57bbc8b
fix baseline
qimcis Feb 25, 2026
6bad7e2
configs
qimcis Mar 10, 2026
0c272f8
add
qimcis May 9, 2026
1819297
fix
qimcis May 9, 2026
17b1be2
fix
qimcis May 9, 2026
c4cbdae
Merge branch 'main' into cachedit-test
qimcis May 10, 2026
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
13 changes: 13 additions & 0 deletions python/sglang/multimodal_gen/test/server/configs/cache_dit_scm_config.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,13 @@
cache_config:
max_warmup_steps: 2
warmup_interval: 2
max_cached_steps: -1
max_continuous_cached_steps: 2
Fn_compute_blocks: 1
Bn_compute_blocks: 0
residual_diff_threshold: 0.12
enable_taylorseer: true
taylorseer_order: 1
num_inference_steps: 8
steps_computation_mask: "medium"
steps_computation_policy: dynamic
47 changes: 47 additions & 0 deletions python/sglang/multimodal_gen/test/server/gpu_cases.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,3 +1,6 @@
from dataclasses import replace
from pathlib import Path

from sglang.multimodal_gen.runtime.platforms import current_platform
from sglang.multimodal_gen.test.server.testcase_configs import (
MODELOPT_FLUX1_FP8_TRANSFORMER,
Expand Down Expand Up @@ -49,6 +52,8 @@
DEFAULT_WAN_2_2_TI2V_5B_MODEL_NAME_FOR_TEST,
)

_CACHE_DIT_CONFIG_DIR = Path(__file__).parent / "configs"

# All test cases with clean default values
# To test different models, simply add more DiffusionCase entries
ONE_GPU_CASES: list[DiffusionTestCase] = [
Expand All @@ -68,6 +73,28 @@
),
T2I_sampling_params,
),
DiffusionTestCase(
"qwen_image_t2i_cache_dit_scm_config_diffusers_1gpu",
DiffusionServerArgs(
model_path=DEFAULT_QWEN_IMAGE_MODEL_NAME_FOR_TEST,
extras=[
"--backend",
"diffusers",
"--cache-dit-config",
str(_CACHE_DIT_CONFIG_DIR / "cache_dit_scm_config.yaml"),
],
),
replace(
T2I_sampling_params,
output_size="512x512",
extras={"num_inference_steps": 8, "seed": 0},
),
run_perf_check=False,
run_consistency_check=False,
run_component_accuracy_check=False,
run_models_api_check=False,
run_t2v_input_reference_check=False,
),
DiffusionTestCase(
"flux_image_t2i",
DiffusionServerArgs(model_path=DEFAULT_FLUX_1_DEV_MODEL_NAME_FOR_TEST),
Expand Down Expand Up @@ -521,6 +548,26 @@
),
T2V_sampling_params,
),
DiffusionTestCase(
"wan2_1_t2v_1_3b_cache_dit_sp_only_2gpu",
DiffusionServerArgs(
model_path=DEFAULT_WAN_2_1_T2V_1_3B_MODEL_NAME_FOR_TEST,
ulysses_degree=2,
enable_cache_dit=True,
env_vars={"SGLANG_CACHE_DIT_WARMUP": "2"},
),
replace(
T2V_sampling_params,
output_size="832x480",
num_frames=5,
extras={"num_inference_steps": 8, "seed": 0},
),
run_perf_check=False,
run_consistency_check=False,
run_component_accuracy_check=False,
run_models_api_check=False,
run_t2v_input_reference_check=False,
),
DiffusionTestCase(
"fsdp-inference",
DiffusionServerArgs(
Expand Down
Loading