#!/bin/bash source ~/anaconda3/etc/profile.d/conda.sh conda activate pytorch_p38 # Hierarchical GVP experiments on PDBBind/CASF2016 dataset # version 29 python train.py --accelerator gpu \ --max_epochs 500 \ --precision 16 \ --num_layers 2 \ --dataset_name PDBBind \ --input_type complex \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019/scoring \ --residual \ --num_workers 8 python evaluate.py --model_name gvp \ --dataset_name PDBBind \ --input_type complex \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/casf2016/scoring \ --dataset_alias casf2016_scoring \ --checkpoint_path /home/ec2-user/SageMaker/ppi-model/lightning_logs/version_29 # version 31 python train.py --accelerator gpu \ --max_epochs 500 \ --precision 16 \ --num_layers 2 \ --dataset_name PDBBind \ --input_type complex \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019/scoring \ --residual \ --num_workers 8 \ --lr 1e-3 \ --bs 128 \ --early_stopping_patience 10 python evaluate.py --model_name gvp \ --dataset_name PDBBind \ --input_type complex \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/casf2016/scoring \ --dataset_alias casf2016_scoring \ --checkpoint_path /home/ec2-user/SageMaker/ppi-model/lightning_logs/version_31 python evaluate_casf2016.py --model_name gvp \ --num_workers 16 \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/casf2016_processed \ --checkpoint_path /home/ec2-user/SageMaker/ppi-model/lightning_logs/version_36 # version 32 python train.py --accelerator gpu \ --max_epochs 500 \ --precision 16 \ --num_layers 2 \ --dataset_name PDBBind \ --input_type complex \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019/scoring \ --residual \ --num_workers 8 \ --lr 3e-3 \ --bs 128 \ --early_stopping_patience 10 python evaluate.py --model_name gvp \ --dataset_name PDBBind \ --input_type complex \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/casf2016/scoring \ --dataset_alias casf2016_scoring \ --checkpoint_path /home/ec2-user/SageMaker/ppi-model/lightning_logs/version_32 python train.py --accelerator gpu \ --max_epochs 500 \ --precision 16 \ --num_layers 4 \ --dataset_name PDBBind \ --input_type complex \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019/scoring \ --residual \ --num_workers 8 \ --lr 1e-3 \ --bs 128 \ --early_stopping_patience 10 # a wider GVP: python train.py --accelerator gpu \ --max_epochs 500 \ --precision 16 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --dataset_name PDBBind \ --input_type complex \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019_processed/scoring \ --residual \ --num_workers 8 \ --lr 1e-3 \ --bs 128 \ --early_stopping_patience 10 # MolT5 python train.py --accelerator gpu \ --devices 1 \ --max_epochs 500 \ --precision 16 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --dataset_name PDBBind \ --input_type complex \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019_processed/scoring \ --residual \ --num_workers 0 \ --lr 1e-4 \ --bs 128 \ --early_stopping_patience 10 \ --residue_featurizer_name MolT5-small \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_MolT5 python evaluate_casf2016.py --model_name gvp \ --num_workers 8 \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/casf2016 \ --checkpoint_path /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_MolT5/lightning_logs/version_1 \ --residue_featurizer_name MolT5-small for lr in 1e-5 1e-4 1e-3; do python train.py --accelerator gpu \ --devices 1 \ --max_epochs 500 \ --precision 16 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --dataset_name PDBBind \ --input_type complex \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019_processed/scoring \ --residual \ --num_workers 0 \ --lr $lr \ --bs 128 \ --early_stopping_patience 10 \ --residue_featurizer_name MolT5-base \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_MolT5_base done python train.py --accelerator gpu \ --devices 1 \ --max_epochs 500 \ --precision 16 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --dataset_name PDBBind \ --input_type complex \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019_processed/scoring \ --residual \ --num_workers 0 \ --lr 1e-4 \ --bs 128 \ --early_stopping_patience 10 \ --residue_featurizer_name MolT5-large \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_MolT5_large python evaluate_casf2016.py --model_name gvp \ --num_workers 8 \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/casf2016_processed \ --checkpoint_path /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_MolT5_large/lightning_logs/version_1 \ --residue_featurizer_name MolT5-large # Morgan fingerprint python train.py --accelerator gpu \ --devices 1 \ --max_epochs 500 \ --precision 16 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --dataset_name PDBBind \ --input_type complex \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019_processed/scoring \ --residual \ --num_workers 0 \ --lr 1e-3 \ --bs 128 \ --early_stopping_patience 10 \ --residue_featurizer_name Morgan \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_Morgan python train.py --accelerator gpu \ --devices 1 \ --max_epochs 500 \ --precision 16 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --dataset_name PDBBind \ --input_type complex \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019_processed/scoring \ --residual \ --num_workers 0 \ --lr 1e-4 \ --bs 128 \ --early_stopping_patience 10 \ --residue_featurizer_name Morgan \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_Morgan python evaluate_casf2016.py --model_name gvp \ --num_workers 8 \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/casf2016 \ --checkpoint_path /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_Morgan/lightning_logs/version_2 \ --residue_featurizer_name Morgan # Repeating MACCS exp on 4-GPU python train.py --accelerator gpu \ --devices 4 \ --max_epochs 500 \ --precision 16 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --dataset_name PDBBind \ --input_type complex \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019_processed/scoring \ --residual \ --num_workers 8 \ --lr 1e-3 \ --bs 128 \ --early_stopping_patience 10 \ --residue_featurizer_name MACCS \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_MACCS python train.py --accelerator gpu \ --devices 1 \ --max_epochs 500 \ --precision 16 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --dataset_name PDBBind \ --input_type complex \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019_processed/scoring \ --residual \ --num_workers 8 \ --lr 1e-3 \ --bs 128 \ --early_stopping_patience 10 \ --residue_featurizer_name MACCS \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_MACCS # small batch_size: python train.py --accelerator gpu \ --devices 1 \ --max_epochs 500 \ --precision 16 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --dataset_name PDBBind \ --input_type complex \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019_processed/scoring \ --residual \ --num_workers 8 \ --lr 1e-4 \ --bs 8 \ --early_stopping_patience 10 \ --residue_featurizer_name MACCS \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_MACCS # on 1-GPU machine CUDA_VISIBLE_DEVICES=3 python train.py --accelerator gpu \ --devices 1 \ --max_epochs 500 \ --precision 16 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --dataset_name PDBBind \ --input_type complex \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019/scoring \ --residual \ --num_workers 8 \ --lr 1e-4 \ --bs 128 \ --early_stopping_patience 501 \ --residue_featurizer_name MACCS \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_MACCS python train.py --accelerator gpu \ --devices 1 \ --max_epochs 500 \ --precision 16 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --dataset_name PDBBind \ --input_type complex \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019/scoring \ --residual \ --num_workers 8 \ --lr 1e-4 \ --bs 128 \ --early_stopping_patience 10 \ --residue_featurizer_name MACCS python evaluate_casf2016.py --model_name gvp \ --num_workers 8 \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/casf2016_processed \ --checkpoint_path /home/ec2-user/SageMaker/ppi-model/lightning_logs/version_6 \ --residue_featurizer_name MACCS python evaluate_casf2016.py --model_name gvp \ --num_workers 8 \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/casf2016_processed \ --checkpoint_path /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_MACCS/lightning_logs/version_7 \ --residue_featurizer_name MACCS # GVP with energy python train.py --accelerator gpu \ --devices 4 \ --max_epochs 1000 \ --precision 16 \ --dataset_name PDBBind \ --input_type complex \ --model_name gvp \ --residue_featurizer_name MolT5-small \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019/scoring \ --use_energy_decoder \ --is_hetero \ --num_workers 4 \ --lr 1e-4 \ --bs 16 \ --early_stopping_patience 50 \ --loss_der1_ratio=10.0 \ --loss_der2_ratio=10.0 \ --min_loss_der2=-20.0 \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_energy_MolT5 python evaluate_casf2016.py --model_name gvp \ --num_workers 8 \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/casf2016_processed \ --checkpoint_path /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_energy_MolT5/lightning_logs/version_5 \ --residue_featurizer_name MolT5-small \ --use_energy_decoder \ --is_hetero python train.py --accelerator gpu \ --devices 4 \ --max_epochs 1000 \ --precision 16 \ --dataset_name PDBBind \ --input_type complex \ --model_name gvp \ --residue_featurizer_name MolT5-small \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019/scoring \ --use_energy_decoder \ --is_hetero \ --num_workers 8 \ --lr 1e-4 \ --bs 16 \ --early_stopping_patience 50 \ --loss_der1_ratio=10.0 \ --loss_der2_ratio=10.0 \ --min_loss_der2=-20.0 \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_energy_MolT5 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --persistent_workers True ## ssGVP with 3 energies python train.py --accelerator gpu \ --devices 4 \ --max_epochs 1000 \ --precision 16 \ --dataset_name PDBBind \ --input_type complex \ --model_name gvp \ --residue_featurizer_name MolT5-small \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019/scoring \ --use_energy_decoder \ --is_hetero \ --intra_mol_energy \ --num_workers 8 \ --lr 1e-4 \ --bs 16 \ --early_stopping_patience 50 \ --loss_der1_ratio=10.0 \ --loss_der2_ratio=10.0 \ --min_loss_der2=-20.0 \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_3energies_MolT5 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --persistent_workers True python evaluate_casf2016.py --model_name gvp \ --num_workers 8 \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/casf2016_processed \ --checkpoint_path /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_3energies_w_MolT5/lightning_logs/version_19 \ --residue_featurizer_name MolT5-small \ --use_energy_decoder \ --intra_mol_energy \ --is_hetero \ --bs 32 python train.py --accelerator gpu \ --devices 4 \ --max_epochs 1000 \ --precision 16 \ --dataset_name PDBBind \ --input_type complex \ --model_name gvp \ --residue_featurizer_name MolT5-small \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019/scoring \ --use_energy_decoder \ --is_hetero \ --intra_mol_energy \ --num_workers 8 \ --lr 1e-4 \ --bs 16 \ --early_stopping_patience 50 \ --loss_der1_ratio=1.0 \ --loss_der2_ratio=1.0 \ --min_loss_der2=-20.0 \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_3energies_MolT5 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --persistent_workers True ## ssGVP with 3 energies (weighted) python train.py --accelerator gpu \ --devices 4 \ --max_epochs 1000 \ --precision 16 \ --dataset_name PDBBind \ --input_type complex \ --model_name gvp \ --residue_featurizer_name MolT5-small \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019/scoring \ --use_energy_decoder \ --is_hetero \ --intra_mol_energy \ --num_workers 8 \ --lr 1e-4 \ --bs 16 \ --early_stopping_patience 50 \ --loss_der1_ratio=10.0 \ --loss_der2_ratio=10.0 \ --min_loss_der2=-20.0 \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_3energies_w_MolT5 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --persistent_workers True python train.py --accelerator gpu \ --devices 4 \ --max_epochs 1000 \ --precision 16 \ --dataset_name PDBBind \ --input_type complex \ --model_name gvp \ --residue_featurizer_name MolT5-small \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019/scoring \ --use_energy_decoder \ --is_hetero \ --intra_mol_energy \ --num_workers 8 \ --lr 1e-4 \ --bs 16 \ --early_stopping_patience 50 \ --loss_der1_ratio=10.0 \ --loss_der2_ratio=10.0 \ --min_loss_der2=-20.0 \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_3energies_w_MolT5 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --persistent_workers True \ --energy_agg_type 3_1 # v14 python train.py --accelerator gpu \ --devices 4 \ --max_epochs 1000 \ --precision 16 \ --dataset_name PDBBind \ --input_type complex \ --model_name gvp \ --residue_featurizer_name MolT5-small \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019/scoring \ --use_energy_decoder \ --is_hetero \ --intra_mol_energy \ --num_workers 8 \ --lr 1e-4 \ --bs 16 \ --early_stopping_patience 50 \ --loss_der1_ratio=10.0 \ --loss_der2_ratio=10.0 \ --min_loss_der2=-20.0 \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_3energies_w_MolT5 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --persistent_workers True \ --energy_agg_type 12_1 # v15 python train.py --accelerator gpu \ --devices 4 \ --max_epochs 1000 \ --precision 16 \ --dataset_name PDBBind \ --input_type complex \ --model_name gvp \ --residue_featurizer_name MolT5-small \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019/scoring \ --use_energy_decoder \ --is_hetero \ --intra_mol_energy \ --num_workers 8 \ --lr 1e-4 \ --bs 16 \ --early_stopping_patience 50 \ --loss_der1_ratio=10.0 \ --loss_der2_ratio=10.0 \ --min_loss_der2=-20.0 \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_3energies_w_MolT5 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --persistent_workers True \ --energy_agg_type 12_0 # v16 CUDA_VISIBLE_DEVICES=4,5,6,7 python train.py --accelerator gpu \ --devices 4 \ --max_epochs 1000 \ --precision 16 \ --dataset_name PDBBind \ --input_type complex \ --model_name gvp \ --residue_featurizer_name MolT5-small \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019/scoring \ --use_energy_decoder \ --is_hetero \ --intra_mol_energy \ --num_workers 8 \ --lr 1e-4 \ --bs 16 \ --early_stopping_patience 50 \ --loss_der1_ratio=10.0 \ --loss_der2_ratio=10.0 \ --min_loss_der2=-20.0 \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_3energies_w_MolT5 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --persistent_workers True \ --energy_agg_type 3_0 # v17 CUDA_VISIBLE_DEVICES=4,5,6,7 python train.py --accelerator gpu \ --devices 4 \ --max_epochs 1000 \ --precision 16 \ --dataset_name PDBBind \ --input_type complex \ --model_name gvp \ --residue_featurizer_name MolT5-small \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019/scoring \ --use_energy_decoder \ --is_hetero \ --intra_mol_energy \ --num_workers 8 \ --lr 1e-4 \ --bs 16 \ --early_stopping_patience 50 \ --loss_der1_ratio=0.0 \ --loss_der2_ratio=0.0 \ --min_loss_der2=-20.0 \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_3energies_w_MolT5 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --persistent_workers True \ --energy_agg_type 3_0 # v18 python train.py --accelerator gpu \ --devices 4 \ --max_epochs 1000 \ --precision 16 \ --dataset_name PDBBind \ --input_type complex \ --model_name gvp \ --residue_featurizer_name MolT5-small \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019/scoring \ --use_energy_decoder \ --is_hetero \ --intra_mol_energy \ --num_workers 8 \ --lr 1e-4 \ --bs 16 \ --early_stopping_patience 50 \ --loss_der1_ratio=0.0 \ --loss_der2_ratio=0.0 \ --min_loss_der2=-20.0 \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_3energies_w_MolT5 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --persistent_workers True \ --energy_agg_type 3_1 # v19 load weights from the best 1 energy model python train.py --accelerator gpu \ --devices 4 \ --max_epochs 1000 \ --precision 16 \ --dataset_name PDBBind \ --input_type complex \ --model_name gvp \ --residue_featurizer_name MolT5-small \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019/scoring \ --use_energy_decoder \ --is_hetero \ --intra_mol_energy \ --num_workers 8 \ --lr 1e-4 \ --bs 16 \ --early_stopping_patience 50 \ --loss_der1_ratio=10.0 \ --loss_der2_ratio=10.0 \ --min_loss_der2=-20.0 \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_3energies_w_MolT5 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --persistent_workers True \ --energy_agg_type 3_1 \ --pretrained_weights /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_energy_MolT5/lightning_logs/version_5/checkpoints/epoch=647-step=36936.ckpt # v20 load weights from the best 1 energy model, 0's on der losses python train.py --accelerator gpu \ --devices 4 \ --max_epochs 1000 \ --precision 16 \ --dataset_name PDBBind \ --input_type complex \ --model_name gvp \ --residue_featurizer_name MolT5-small \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019/scoring \ --use_energy_decoder \ --is_hetero \ --intra_mol_energy \ --num_workers 8 \ --lr 1e-4 \ --bs 16 \ --early_stopping_patience 50 \ --loss_der1_ratio=0.0 \ --loss_der2_ratio=0.0 \ --min_loss_der2=-20.0 \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_3energies_w_MolT5 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --persistent_workers True \ --energy_agg_type 3_1 \ --pretrained_weights /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_energy_MolT5/lightning_logs/version_5/checkpoints/epoch=647-step=36936.ckpt ## ssGVP with 3 energies (weighted) on full PDBBind complexes inputs python train.py --accelerator gpu \ --devices 4 \ --max_epochs 1000 \ --precision 16 \ --dataset_name PDBBind \ --input_type complex \ --model_name gvp \ --residue_featurizer_name MolT5-small \ --data_dir /home/ec2-user/SageMaker/efs/data/PDBBind/pdbbind_v2019/scoring \ --use_energy_decoder \ --is_hetero \ --intra_mol_energy \ --num_workers 8 \ --lr 1e-4 \ --bs 16 \ --early_stopping_patience 50 \ --loss_der1_ratio=10.0 \ --loss_der2_ratio=10.0 \ --min_loss_der2=-20.0 \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_intact_GVP_3energies_w_MolT5 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --persistent_workers True python evaluate_casf2016.py --model_name gvp \ --num_workers 8 \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/casf2016_processed \ --checkpoint_path /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_3energies_w_MolT5/lightning_logs/version_17 \ --residue_featurizer_name MolT5-small \ --use_energy_decoder \ --intra_mol_energy \ --is_hetero \ --bs 32 # Continued training from checkpoint python train.py --accelerator gpu \ --devices 4 \ --max_epochs 500 \ --precision 16 \ --dataset_name PDBBind \ --input_type complex \ --model_name gvp \ --residue_featurizer_name MolT5-small \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019/scoring \ --use_energy_decoder \ --is_hetero \ --intra_mol_energy \ --num_workers 8 \ --lr 1e-4 \ --bs 16 \ --early_stopping_patience 500 \ --loss_der1_ratio=10.0 \ --loss_der2_ratio=10.0 \ --min_loss_der2=-20.0 \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_3energies_w_MolT5 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --persistent_workers True \ --pretrained_weights /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_3energies_w_MolT5/lightning_logs/version_5/checkpoints/epoch=337-step=19266.ckpt python evaluate_casf2016.py --model_name gvp \ --num_workers 8 \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/casf2016_processed \ --checkpoint_path /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_3energies_w_MolT5/lightning_logs/version_6/checkpoints/last.ckpt \ --residue_featurizer_name MolT5-small \ --use_energy_decoder \ --intra_mol_energy \ --is_hetero \ --bs 32 ## ssGVP with energy and end-to-end training python train.py --accelerator gpu \ --devices 4 \ --max_epochs 500 \ --precision 32 \ --dataset_name PDBBind \ --input_type complex \ --model_name hgvp \ --residue_featurizer_name MolT5-small-grad \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/pdbbind_v2019/scoring \ --use_energy_decoder \ --is_hetero \ --num_workers 8 \ --lr 1e-4 \ --bs 4 \ --early_stopping_patience 50 \ --loss_der1_ratio=10.0 \ --loss_der2_ratio=10.0 \ --min_loss_der2=-20.0 \ --num_layers 3 \ --node_h_dim 200 32 \ --edge_h_dim 64 2 \ --default_root_dir /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_energy_MolT5_grad python evaluate_casf2016.py --model_name hgvp \ --num_workers 8 \ --data_dir /home/ec2-user/SageMaker/efs/data/PIGNet/data/casf2016_processed \ --checkpoint_path /home/ec2-user/SageMaker/efs/model_logs/zichen/PDBBind_GVP_energy_MolT5_grad/lightning_logs/version_0 \ --residue_featurizer_name MolT5-small-grad \ --use_energy_decoder \ --is_hetero \ --bs 32