scripts/cwm/3frame_no_clumping_maskvit_gpu.sh

OUTPUT_DIR='/ccn2/u/honglinc/cwm_checkpoints/ablation_3frame_no_clumping_maskvit/'
DATA_PATH="${HOME}/BBNet/bbnet/models/VideoMAE-main/video_file_lists/kinetics_400_train_list.txt"

OMP_NUM_THREADS=1 python -m torch.distributed.launch --nproc_per_node=8 \
		    --master_addr=10.102.2.153 --master_port=32240 \
		    --nnodes=2 --node_rank=1 \
		     run_cwm_pretraining.py \
		    --data_path ${DATA_PATH} \
		    --mask_type rotated_table_maskvit \
		    --mask_ratio 0.99 \
		    --mask_kwargs '{"tube_length": 1}' \
		    --model vitbase_8x8patch_3frames_1tube \
		    --context_frames 2 \
		    --target_frames 1 \
		    --temporal_units 'ms' \
		    --sampling_rate 150 \
		    --context_target_gap 150 150 \
		    --batch_size 16 \
		    --accum_iter 16 \
		    --opt adamw \
		    --opt_betas 0.9 0.95 \
		    --warmup_epochs 40 \
		    --save_ckpt_freq 10 \
		    --epochs 800 \
		    --no_normlize_target \
		    --rescale_size 224 \
		    --augmentation_type 'multiscale' \
		    --augmentation_scales 1.0 0.875 0.75 0.66 \
		    --log_dir ${OUTPUT_DIR} \
		    --output_dir ${OUTPUT_DIR} \
		    --print_freq 20 \
		    --num_workers 32 \
		    --min_lr 1e-5