Skip to content

Add z image turbo warmup script #845

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
helloyongyang merged 1 commit into from
Jan 29, 2026
Merged

Add z image turbo warmup script #845

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
1 change: 1 addition & 0 deletions lightx2v/models/runners/z_image/z_image_runner.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -340,6 +340,7 @@ def run_vae_decoder(self, latents):
gc.collect()
return images

@ProfilingContext4DebugL1("RUN pipeline")
def run_pipeline(self, input_info):
self.input_info = input_info

Expand Down
2 changes: 0 additions & 2 deletions scripts/bench/run_lightx2v_qwen_2512_with_warmup.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,12 +1,10 @@
import os
import sys
from datetime import datetime

lightx2v_path = "/path/to/LightX2V"
model_path = "/path/to/Qwen/Qwen-Image-2512"

sys.path.append(lightx2v_path)
os.environ["PROFILING_DEBUG_LEVEL"] = "2"

from lightx2v import LightX2VPipeline # noqa: E402

Expand Down
2 changes: 0 additions & 2 deletions scripts/bench/run_lightx2v_qwen_edit_2511_with_warmup.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,12 +1,10 @@
import os
import sys
from datetime import datetime

lightx2v_path = "/path/to/LightX2V"
model_path = "/path/to/Qwen/Qwen-Image-Edit-2511"

sys.path.append(lightx2v_path)
os.environ["PROFILING_DEBUG_LEVEL"] = "2"

from lightx2v import LightX2VPipeline # noqa: E402

Expand Down
2 changes: 0 additions & 2 deletions scripts/bench/run_lightx2v_wan22_t2v_8gpu_with_warmup.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,14 +1,12 @@
# torchrun --nproc_per_node=8 run_lightx2v_wan22_t2v_8gpu_with_warmup.py

import os
import sys
from datetime import datetime

lightx2v_path = "/path/to/LightX2V"
model_path = "/path/to/Wan-AI/Wan2.2-T2V-A14B"

sys.path.append(lightx2v_path)
os.environ["PROFILING_DEBUG_LEVEL"] = "2"

from lightx2v import LightX2VPipeline # noqa: E402

Expand Down
54 changes: 54 additions & 0 deletions scripts/bench/run_lightx2v_z_image_turbo_with_warmup.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,54 @@
import sys
from datetime import datetime

lightx2v_path = "/path/to/LightX2V"
model_path = "/path/to/Tongyi-MAI/Z-Image-Turbo"

sys.path.append(lightx2v_path)

from lightx2v import LightX2VPipeline # noqa: E402

ts = datetime.now().strftime("%y%m%d%H%M")

model_cls = "z_image"

task = "t2i"


pipe = LightX2VPipeline(
model_path=model_path,
model_cls=model_cls,
task=task,
)

pipe.create_generator(config_json=f"{lightx2v_path}/configs/z_image/z_image_turbo_t2i.json")


# Generation parameters
seed = 42
prompt = "A fantasy landscape with mountains and a river, detailed, vibrant colors"

negative_prompt = " "

target_shape = [1024, 1024]

save_result_path = f"{lightx2v_path}/save_results/{model_cls}_{task}_{ts}.png"

# warmup
pipe.generate(
seed=seed,
prompt=prompt,
target_shape=target_shape,
negative_prompt=negative_prompt,
save_result_path=save_result_path,
)

# Generate video
pipe.generate(
seed=seed,
prompt=prompt,
target_shape=target_shape,
negative_prompt=negative_prompt,
save_result_path=save_result_path,
return_result_tensor=True,
)
Comment on lines +37 to +54
Copy link
Copy Markdown
Contributor

@gemini-code-assist gemini-code-assist Bot Jan 29, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

There are a couple of improvements that can be made here:

  1. The arguments passed to pipe.generate are almost identical for the warmup and the actual generation call, leading to code duplication. To improve maintainability, you can store the common arguments in a dictionary and unpack them during the calls.
  2. The comment on line 46, # Generate video, is misleading since this is a text-to-image (t2i) task that generates a .png file. It should be changed to # Generate image.

Here's a suggestion that addresses both points:

generate_args = {
    "seed": seed,
    "prompt": prompt,
    "target_shape": target_shape,
    "negative_prompt": negative_prompt,
    "save_result_path": save_result_path,
}

# warmup
pipe.generate(**generate_args)

# Generate image
pipe.generate(**generate_args, return_result_tensor=True)