vit_g_k710_it_ucf101_ft.sh

#!/usr/bin/env bash
set -x

export MASTER_PORT=$((12000 + $RANDOM % 20000))
export OMP_NUM_THREADS=1

# official train/test splits. valid numbers: 1, 2, 3
SPLIT=${SPLIT:-1}

OUTPUT_DIR='YOUR_PATH/work_dir/vit_g_hybrid_pt_1200e_k710_it_ucf101_'${SPLIT}'_ft'
DATA_PATH='YOUR_PATH/data/ucf101_'${SPLIT}
MODEL_PATH='YOUR_PATH/model_zoo/vit_g_hybrid_pt_1200e_k710_ft.pth'

JOB_NAME=$1
PARTITION=${PARTITION:-"video"}
# 8 for 1 node, 16 for 2 node, etc.
GPUS=${GPUS:-8}
GPUS_PER_NODE=${GPUS_PER_NODE:-8}
CPUS_PER_TASK=${CPUS_PER_TASK:-10}
SRUN_ARGS=${SRUN_ARGS:-""}
PY_ARGS=${@:2}

# batch_size can be adjusted according to the graphics card
srun -p $PARTITION \
        --job-name=${JOB_NAME} \
        --gres=gpu:${GPUS_PER_NODE} \
        --ntasks=${GPUS} \
        --ntasks-per-node=${GPUS_PER_NODE} \
        --cpus-per-task=${CPUS_PER_TASK} \
        --kill-on-bad-exit=1 \
        --async \
        ${SRUN_ARGS} \
        python run_class_finetuning.py \
        --model vit_giant_patch14_224 \
        --data_set UCF101 \
        --nb_classes 101 \
        --data_path ${DATA_PATH} \
        --finetune ${MODEL_PATH} \
        --log_dir ${OUTPUT_DIR} \
        --output_dir ${OUTPUT_DIR} \
        --batch_size 3 \
        --num_sample 2 \
        --input_size 224 \
        --short_side_size 224 \
        --save_ckpt_freq 10 \
        --num_frames 16 \
        --sampling_rate 4 \
        --opt adamw \
        --lr 1e-3 \
        --layer_decay 0.90 \
        --num_workers 10 \
        --opt_betas 0.9 0.999 \
        --weight_decay 0.05 \
        --epochs 50 \
        --drop_path 0.35 \
        --head_drop_rate 0.5 \
        --test_num_segment 5 \
        --test_num_crop 3 \
        --dist_eval --enable_deepspeed \
        ${PY_ARGS}