nicolabs/Open-Sora
1
0
Fork 0
mirror of https://github.com/hpcaitech/Open-Sora.git synced 2026-05-21 11:59:01 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

format

Browse source
This commit is contained in:
Shen-Chenhui 2024-06-07 10:09:39 +00:00
parent 2401377f10
commit a214d9a50d
1 changed files with 37 additions and 37 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

74
parallel_eval.sh
View file

@ -15,7 +15,7 @@ VBENCH_SAMPLE_DIR=samples/samples_${MODEL_NAME}_${CKPT_BASE}_vbench
sleep_time=10m
human_eval_ready=0
human_ready_count=7
human_ready_count=8
loss_eval_ready=0
loss_ready_count=6
vbench_gen_ready=0
@ -154,9 +154,9 @@ function check_vbench_i2v_calc(){
echo "completed tasks: ${finished_list[@]}"
if [ ${#finished_list[@]}=$vbench_calc_ready_count ];
if [ ${#finished_list[@]}=$vbench_i2v_calc_ready_count ];
then
vbench_calc_ready=1
vbench_i2v_calc_ready=1
else
echo ${#finished_list[@]}
fi
@ -167,43 +167,43 @@ function check_vbench_i2v_calc(){
start=$(date +%s)
# ### human eval, ~60min
# echo "$(date): running human eval"
# bash eval/human_eval/launch.sh $CKPT $NUM_FRAMES $MODEL_NAME
# while [ $human_eval_ready -eq 0 ]
# do
# sleep ${sleep_time}
# check_human_eval
# done
### human eval, ~60min
echo "$(date): running human eval"
bash eval/human_eval/launch.sh $CKPT $NUM_FRAMES $MODEL_NAME
while [ $human_eval_ready -eq 0 ]
do
sleep ${sleep_time}
check_human_eval
done
# ### eval loss, ~160min
# echo "$(date): running eval loss"
# bash eval/eval_loss/launch.sh $CKPT $MODEL_NAME
# while [ $loss_eval_ready -eq 0 ]
# do
# sleep ${sleep_time}
# check_loss_eval
# done
# python eval/loss/tabulate_rl_loss.py --log_dir $LOG_BASE
### eval loss, ~160min
echo "$(date): running eval loss"
bash eval/eval_loss/launch.sh $CKPT $MODEL_NAME
while [ $loss_eval_ready -eq 0 ]
do
sleep ${sleep_time}
check_loss_eval
done
python eval/loss/tabulate_rl_loss.py --log_dir $LOG_BASE
# ### vbench gen, ~80min
# echo "$(date): running vbench gen"
# bash eval/vbench/launch.sh $CKPT $NUM_FRAMES $MODEL_NAME
# while [ $vbench_gen_ready -eq 0 ]
# do
# sleep ${sleep_time}
# check_vbench_gen
# done
### vbench gen, ~80min
echo "$(date): running vbench gen"
bash eval/vbench/launch.sh $CKPT $NUM_FRAMES $MODEL_NAME
while [ $vbench_gen_ready -eq 0 ]
do
sleep ${sleep_time}
check_vbench_gen
done
# ### vbench calc, ~30min
# echo "$(date): running vbench calc"
# bash eval/vbench/launch_calc.sh $VBENCH_SAMPLE_DIR $LOG_BASE
# while [ $vbench_calc_ready -eq 0 ]
# do
# sleep ${sleep_time}
# check_vbench_calc
# done
# python eval/vbench/tabulate_vbench_scores.py --score_dir ${LOG_BASE}/vbench
### vbench calc, ~30min
echo "$(date): running vbench calc"
bash eval/vbench/launch_calc.sh $VBENCH_SAMPLE_DIR $LOG_BASE
while [ $vbench_calc_ready -eq 0 ]
do
sleep ${sleep_time}
check_vbench_calc
done
python eval/vbench/tabulate_vbench_scores.py --score_dir ${LOG_BASE}/vbench
### vbench_i2v gen, ~65min
echo "$(date): running vbench_i2v gen"