train.sh

#!/bin/bash
DATA_PATH=/disk/gjw/msr-vtt

# Initialize from a pretrained checkpoint
INIT_MODEL=/disk/gjw/CLIP4Clip/ckpts/ckpt_msrvtt_retrieval_looseType/pytorch_model.bin.0
RESUME_MODEL=/disk/gjw/CLIP4Clip/ckpts/ckpt_msrvtt_retrieval_looseType_0909/pytorch_opt.bin.0
CUDA_VISIBLE_DEVICES=0 torchrun --standalone --nproc_per_node=1 --master_port=29502 \
cache_main_task_retrieval.py --do_train --num_thread_reader=2 \
--epochs=5 --batch_size=256 --n_display=50 \
--train_csv ${DATA_PATH}/MSRVTT_train.9k.csv \
--val_csv ${DATA_PATH}/MSRVTT_JSFUSION_test.csv \
--data_path ${DATA_PATH}/MSRVTT_data.json \
--features_path ${DATA_PATH}/compressed_videos \
--output_dir ckpts/ckpt_msrvtt_retrieval_looseType_0909/ \
--lr 1e-4 --max_words 32 --max_frames 12 --batch_size_val 16 \
--datatype msrvtt --expand_msrvtt_sentences  \
--feature_framerate 1 --coef_lr 1e-3 \
--freeze_layer_num 12 --slice_framepos 2 \
--loose_type --linear_patch 2d --sim_header meanP \
--pretrained_clip_name ViT-B/32 \
--proj 3008 \
--init_model ${INIT_MODEL} \
--resume_model ${RESUME_MODEL}