LAVT-RIS/donghwa/scripts/submit_train_refcocog_mosaic.sh

#!/bin/bash

#SBATCH --job-name=lavt_ccggr      # Submit a job named "example"
#SBATCH [email protected]
#SBATCH --mail-type=BEGIN,END,FAIL
#SBATCH --partition=a3000         # a6000 or a100
#SBATCH --gres=gpu:2
#SBATCH --time=7-00:00:00        # d-hh:mm:ss, max time limit
#SBATCH --mem=84000              # cpu memory size
#SBATCH --cpus-per-task=8        # cpu num
#SBATCH --output=log_refcocog_google_random_460_0up_resume.txt         # std output filename

ml cuda/11.0                # 필요한 쿠다 버전 로드
eval "$(conda shell.bash hook)"  # Initialize Conda Environment
conda activate lavt             # Activate your conda environment



# train
# mkdir ./models

# mkdir ./models/gref_umd/
# srun python -m torch.distributed.launch --nproc_per_node 2 --master_port 12345 train_mosaic.py --model lavt --dataset refcocog --splitBy umd --model_id gref_umd --batch-size 14 --lr 0.00005 --wd 1e-2 --swin_type base --pretrained_swin_weights ./pretrained_weights/swin_base_patch4_window12_384_22k.pth --epochs 40 --img_size 480 2>&1 | tee ./models/gref_umd/output

# mkdir ./models/mosaic_gref_umd_lavt_one/
# srun python -m torch.distributed.launch --nproc_per_node 2 --master_port 12345 train_mosaic.py --model lavt_one --dataset refcocog --splitBy umd --model_id mosaic_gref_umd_lavt_one --batch-size 14 --lr 0.00005 --wd 1e-2 --swin_type base --pretrained_swin_weights ./pretrained_weights/swin_base_patch4_window12_384_22k.pth --epochs 40 --img_size 480 2>&1 | tee ./models/mosaic_gref_umd_lavt_one/output

# mkdir ./models/gref_google
# srun python -m torch.distributed.launch --nproc_per_node 2 --master_port 12345 train_mosaic.py --model lavt_one --dataset refcocog --splitBy google --model_id gref_google_lavt_one --batch-size 8 --lr 0.00005 --wd 1e-2 --swin_type base --pretrained_swin_weights ./pretrained_weights/swin_base_patch4_window12_384_22k.pth --epochs 40 --img_size 480 2>&1 | tee ./models/mosaic_gref_google_lavt_one/output

# mkdir ./models/mosaic_gref_google_lavt_one
# srun python -m torch.distributed.launch --nproc_per_node 2 --master_port 13347 train_mosaic.py --model lavt_one --dataset refcocog --splitBy google --model_id mosaic_gref_google_lavt_one --batch-size 14 --lr 0.00005 --wd 1e-2 --swin_type base --pretrained_swin_weights ./pretrained_weights/swin_base_patch4_window12_384_22k.pth --epochs 50 --img_size 480 2>&1 | tee ./models/mosaic_gref_google_lavt_one/output

# tensorboard X
# srun python -m torch.distributed.launch --nproc_per_node 1 --master_port 14567 train_mosaic.py --model lavt_one --dataset refcocog --splitBy umd --model_id lmdb_test --batch-size 5 --lr 0.00005 --wd 1e-2 --swin_type base --pretrained_swin_weights ./pretrained_weights/swin_base_patch4_window12_384_22k.pth --epochs 50 --img_size 480 2>&1 | tee ./models/lmdb_test/output

# tensorboard O
# mkdir ./experiments/refcocog_google/gref_google_random_460_0up
srun python -m torch.distributed.launch --nproc_per_node 2 --master_port 35327 train_mosaic.py \
    --model lavt_one --dataset refcocog --splitBy google --model_id gref_google_random_460_0up \
    --batch-size 16 --lr 0.00005 --wd 1e-2 --swin_type base --pretrained_swin_weights ./pretrained_weights/swin_base_patch4_window12_384_22k.pth \
    --epochs 40 --img_size 480 --config config/random_460.yaml \
    --resume experiments/refcocog_google/gref_google_random_460_0up/model_best_gref_google_random_460_0up.pth 2>&1 | tee ./experiments/refcocog_google/gref_google_random_460_0up/log_resume.txt