Skip to content
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

Added TRT config for inference #1907

Open
wants to merge 2 commits into
base: main
Choose a base branch
from
Open

Added TRT config for inference #1907

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
04a9326
Added TRT config for inference
borisfom Dec 20, 2024
0b4545b
cleanup
borisfom Dec 20, 2024
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
7 changes: 6 additions & 1 deletion generation/maisi/configs/config_infer.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,5 +18,10 @@
2.0
],
"autoencoder_sliding_window_infer_size": [48,48,48],
"autoencoder_sliding_window_infer_overlap": 0.25
"autoencoder_sliding_window_infer_overlap": 0.25,
"controlnet": "$@controlnet_def",
"diffusion_unet": "$@diffusion_unet_def",
"autoencoder": "$@autoencoder_def",
"mask_generation_autoencoder": "$@mask_generation_autoencoder_def",
"mask_generation_diffusion": "$@mask_generation_diffusion_def"
}
22 changes: 22 additions & 0 deletions generation/maisi/configs/config_trt.json
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,22 @@
{
"+imports": [
"$from monai.networks import trt_compile"
],
"c_trt_args": {
"export_args": {
"dynamo": "$False",
"report": "$True"
},
"output_lists": [
[
-1
],
[
]
]
},
"device": "cuda",
"controlnet": "$trt_compile(@controlnet_def.to(@device), @trained_controlnet_path, @c_trt_args)",
"diffusion_unet": "$trt_compile(@diffusion_unet_def.to(@device), @trained_diffusion_path)",
"mask_generation_diffusion": "$trt_compile(@mask_generation_diffusion_def.to(@device), @trained_mask_generation_diffusion_path)"
Copy link
Contributor

Choose a reason for hiding this comment

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

Hi @borisfom , thanks for the changes.
Can I confirm if mask_generation_diffusion supports TRT? Should we also modify https://github.com/Project-MONAI/model-zoo/blob/dev/models/maisi_ct_generative/configs/inference_trt.json ?

Copy link
Author

Choose a reason for hiding this comment

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

Well mask_generation_diffusion was never called - it probably is TRT compliant, but never tested.
No need to modify current model-zoo config until we have a config that engages this component.

}
26 changes: 21 additions & 5 deletions generation/maisi/scripts/inference.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -48,6 +48,12 @@ def main():
default="./configs/config_infer.json",
help="config json file that stores inference hyper-parameters",
)
parser.add_argument(
"-x",
"--extra-config-file",
default=None,
help="config json file that stores inference extra parameters",
)
parser.add_argument(
"-s",
"--random-seed",
Expand Down Expand Up @@ -140,6 +146,16 @@ def main():
setattr(args, k, v)
print(f"{k}: {v}")

#
# ## Read in optional extra configuration setting - typically acceleration options (TRT)
#
#
if args.extra_config_file is not None:
extra_config_dict = json.load(open(args.extra_config_file, "r"))
for k, v in extra_config_dict.items():
setattr(args, k, v)
print(f"{k}: {v}")

check_input(
args.body_region,
args.anatomy_list,
Expand All @@ -158,25 +174,25 @@ def main():

device = torch.device("cuda")

autoencoder = define_instance(args, "autoencoder_def").to(device)
autoencoder = define_instance(args, "autoencoder").to(device)
checkpoint_autoencoder = torch.load(args.trained_autoencoder_path)
autoencoder.load_state_dict(checkpoint_autoencoder)

diffusion_unet = define_instance(args, "diffusion_unet_def").to(device)
diffusion_unet = define_instance(args, "diffusion_unet").to(device)
checkpoint_diffusion_unet = torch.load(args.trained_diffusion_path)
diffusion_unet.load_state_dict(checkpoint_diffusion_unet["unet_state_dict"], strict=True)
scale_factor = checkpoint_diffusion_unet["scale_factor"].to(device)

controlnet = define_instance(args, "controlnet_def").to(device)
controlnet = define_instance(args, "controlnet").to(device)
checkpoint_controlnet = torch.load(args.trained_controlnet_path)
monai.networks.utils.copy_model_state(controlnet, diffusion_unet.state_dict())
controlnet.load_state_dict(checkpoint_controlnet["controlnet_state_dict"], strict=True)

mask_generation_autoencoder = define_instance(args, "mask_generation_autoencoder_def").to(device)
mask_generation_autoencoder = define_instance(args, "mask_generation_autoencoder").to(device)
checkpoint_mask_generation_autoencoder = torch.load(args.trained_mask_generation_autoencoder_path)
mask_generation_autoencoder.load_state_dict(checkpoint_mask_generation_autoencoder)

mask_generation_diffusion_unet = define_instance(args, "mask_generation_diffusion_def").to(device)
mask_generation_diffusion_unet = define_instance(args, "mask_generation_diffusion").to(device)
checkpoint_mask_generation_diffusion_unet = torch.load(args.trained_mask_generation_diffusion_path)
mask_generation_diffusion_unet.load_state_dict(checkpoint_mask_generation_diffusion_unet["unet_state_dict"])
mask_generation_scale_factor = checkpoint_mask_generation_diffusion_unet["scale_factor"]
Expand Down
Loading