Skip to content

Adding capability to start a training from model checkpoint instead of doing it from scratch #297

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

Open
wants to merge 4 commits into
base: main
Choose a base branch
from
Open

Adding capability to start a training from model checkpoint instead of doing it from scratch #297

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
bcf9c27
Update train.py
karolzak Feb 1, 2024
630948e
Update places_example.yaml
karolzak Feb 1, 2024
448ea11
Update docker.yaml
karolzak Feb 1, 2024
167c316
Update celeba_example.yaml
karolzak Feb 1, 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
12 changes: 10 additions & 2 deletions bin/train.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,7 @@
from pytorch_lightning.loggers import TensorBoardLogger
from pytorch_lightning.plugins import DDPPlugin

from saicinpainting.training.trainers import make_training_model
from saicinpainting.training.trainers import make_training_model, load_checkpoint
from saicinpainting.utils import register_debug_signal_handlers, handle_ddp_subprocess, handle_ddp_parent_process, \
handle_deterministic_config

Expand Down Expand Up @@ -48,7 +48,15 @@ def main(config: OmegaConf):
metrics_logger = TensorBoardLogger(config.location.tb_dir, name=os.path.basename(os.getcwd()))
metrics_logger.log_hyperparams(config)

training_model = make_training_model(config)
if "load_checkpoint_path" in config.location:
print("Loading model checkpoint from path:", config.location.load_checkpoint_path)
training_model = load_checkpoint(
train_config=config,
path=config.location.load_checkpoint_path,
strict=False
)
else:
training_model = make_training_model(config)

trainer_kwargs = OmegaConf.to_container(config.trainer.kwargs, resolve=True)
if need_set_deterministic:
Expand Down
2 changes: 2 additions & 0 deletions configs/training/location/celeba_example.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,3 +3,5 @@ data_root_dir: /home/user/lama/celeba-hq-dataset/
out_root_dir: /home/user/lama/experiments/
tb_dir: /home/user/lama/tb_logs/
pretrained_models: /home/user/lama/
# path to model checkpoint that will be loaded before start of training
# load_checkpoint_path: /home/user/lama/big-lama/models/best.ckpt
2 changes: 2 additions & 0 deletions configs/training/location/docker.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,3 +3,5 @@ data_root_dir: /data/data
out_root_dir: /data/experiments
tb_dir: /data/tb_logs
pretrained_models: /some_path
# path to model checkpoint that will be loaded before start of training
# load_checkpoint_path: /home/user/lama/big-lama/models/best.ckpt
2 changes: 2 additions & 0 deletions configs/training/location/places_example.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,3 +3,5 @@ data_root_dir: /home/user/inpainting-lama/places_standard_dataset/
out_root_dir: /home/user/inpainting-lama/experiments
tb_dir: /home/user/inpainting-lama/tb_logs
pretrained_models: /home/user/inpainting-lama/
# path to model checkpoint that will be loaded before start of training
# load_checkpoint_path: /home/user/lama/big-lama/models/best.ckpt