clean up

scripts/preprocess/preprocess_wan_data_t2v.sh

@@ -2,15 +2,15 @@
 GPU_NUM=1 # 2,4,8
 MODEL_PATH="Wan-AI/Wan2.1-T2V-1.3B-Diffusers"
 MODEL_TYPE="wan"
-DATA_MERGE_PATH="data/cats_480/merge.txt"
-OUTPUT_DIR="cats_480_2_latents_parq_neg2"
-VALIDATION_PATH="assets/2_cat_prompt.txt"
+DATA_MERGE_PATH="your/path/to/Mixkit-Src/merge.txt"
+OUTPUT_DIR="your/path"
+VALIDATION_PATH="assets/prompt.txt"
 
 torchrun --nproc_per_node=$GPU_NUM \
     fastvideo/data_preprocess/preprocess.py \
     --model_path $MODEL_PATH \
     --data_merge_path $DATA_MERGE_PATH \
-    --preprocess_video_batch_size 1 \
+    --preprocess_video_batch_size 8 \
     --max_height 480 \
     --max_width 832 \
     --num_frames 81 \
@@ -19,6 +19,6 @@ torchrun --nproc_per_node=$GPU_NUM \
     --model_type $MODEL_TYPE \
     --train_fps 16 \
     --validation_prompt_txt $VALIDATION_PATH \
-    --samples_per_file 1 \
-    --flush_frequency 1 \
+    --samples_per_file 32 \
+    --flush_frequency 128 \
     --preprocess_task "t2v"