The paddle version of the top three in each track of the IJCAI 2024 competition.
Inference codes only now.
Please refer to the .ipynb files in each directory to download the data and set the corresponding parameters.
Donwload checkpoints:
cd PaddleScience/jointContribution/IJCAI_2024
# linux
wget -nc https://paddle-org.bj.bcebos.com/paddlescience/models/contrib/IJCAI_2024_ckpts.tar.gz
# windows
# curl https://paddle-org.bj.bcebos.com/paddlescience/models/contrib/IJCAI_2024_ckpts.tar.gz
Unzip the checkpoints and move them to the corresponding directory:
tar -xvzf IJCAI_2024_ckpts.tar.gz
# aminos
mkdir -p ./aminos/Logger/states/
mv ./ckpts/aminos/90.pdparams ./aminos/Logger/states/90.pdparams
# tenfeng
mkdir -p ./results/
mv ./ckpts/tenfeng/checkpoint.pdparams ./tenfeng/results/checkpoint.pdparams
# leejt
mv ./ckpts/leejt/model.pdparams ./leejt/model.pdparams
# bju
mv ./ckpts/bju/geom/ckpt ./bju/geom/
mv ./ckpts/bju/pretrained_checkpoint.pdparams ./bju/pretrained_checkpoint.pdparams
# zhongzaicanyu
mv ./ckpts/zhongzaicanyu/pretrained_checkpoint.pdparams ./zhongzaicanyu/pretrained_checkpoint.pdparams
First enter the corresponding directory. For example "aminos":
cd aminos
Install requirements:
pip install -r requirements.txt
Run Inference:
### aminos
python infer.py --dataset_dir "./Datasets" --load_index="90"
### tenfeng
python infer.py --epochs 69 --milestones 40 50 60 65 68 --gpu_id 0 --depth 5 --hidden_dim 256 --num_slices 32 --batch_size 4 --loss_type 'rl2' --submit --log_dir "./results" --training_data_dir "./Dataset/train_track_B_e" --testing_data_dir "./Dataset/Testset_track_B_e"
### leejt
python infer.py
### bju
python infer.py --train_data_dir "./Dataset/Trainset_track_B" --test_data_dir "./Dataset/Testset_track_B/Inference" --info_dir "./Dataset/Testset_track_B/Auxiliary" --ulip_ckpt "./geom/ckpt/checkpoint_pointbert.pdparams"
### zhongzaicanyu
python infer.py --data_dir "./Dataset/data_centroid_track_B_vtk" --test_data_dir "./Dataset/track_B_vtk" --save_dir "./Dataset/data_centroid_track_B_vtk_preprocessed_data"