Skip to content

Open-Sora: Democratizing Efficient Video Production for All

Python

fixed training script (#19)

Frank Lee3bea560af368
diff --git a/scripts/train.sh b/scripts/train.sh
index 754fce7..b38de61 100644
--- a/scripts/train.sh
+++ b/scripts/train.sh
@@ -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 \