fixed training script (#19)

This commit is contained in:
Frank Lee 2024-02-29 16:07:34 +08:00 committed by GitHub
parent c1767416bf
commit 3bea560af3
No known key found for this signature in database
GPG key ID: B5690EEEBB952194

View file

@ -11,18 +11,18 @@ ROOT_DIR=$FOLDER_DIR/..
cd $ROOT_DIR
# define dataset shards
COLLATED_VIDEO_DIR=./dataset/MSRVTT-collated/val/videos
COLLATED_VIDEO_DIR=./dataset/MSRVTT-collated/train/videos
PROCESSED_DATASET=(
./dataset/MSRVTT-processed/val/part-00000
./dataset/MSRVTT-processed/val/part-00001
./dataset/MSRVTT-processed/val/part-00002
./dataset/MSRVTT-processed/val/part-00003
./dataset/MSRVTT-processed/val/part-00004
./dataset/MSRVTT-processed/val/part-00005
./dataset/MSRVTT-processed/val/part-00006
./dataset/MSRVTT-processed/val/part-00007
./dataset/MSRVTT-processed/val/part-00008
./dataset/MSRVTT-processed/val/part-00009
./dataset/MSRVTT-processed/train/part-00000
./dataset/MSRVTT-processed/train/part-00001
./dataset/MSRVTT-processed/train/part-00002
./dataset/MSRVTT-processed/train/part-00003
./dataset/MSRVTT-processed/train/part-00004
./dataset/MSRVTT-processed/train/part-00005
./dataset/MSRVTT-processed/train/part-00006
./dataset/MSRVTT-processed/train/part-00007
./dataset/MSRVTT-processed/train/part-00008
./dataset/MSRVTT-processed/train/part-00009
)
# create timestamp to differentiate between runs
@ -37,7 +37,7 @@ torchrun --standalone \
--lr 1e-4 \
--accumulation_steps 32 \
--grad_checkpoint \
--dataset $PROCESSED_DATASET \
--dataset "${PROCESSED_DATASET[@]}" \
--video_dir $COLLATED_VIDEO_DIR \
--save_interval 224 \
--checkpoint_dir ./checkpoints/$timestamp \