p04798526
/
LLaMA-Factory-310P3
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

chore: sort run_once

This commit is contained in:
wql 2024-09-05 14:41:48 +08:00
parent 36840f0310
commit ba90bf1625
1 changed files with 20 additions and 23 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

43
run_once.sh
View File

@ -23,35 +23,32 @@ else
echo "output_dir exists: $output_dir"
fi
# echo "${run_type} ${model} ${gpu_cnt} ${max_steps} ${run_name} ${output_dir}"
python prepare_yaml_file.py ${run_type} ${model} ${max_steps} ${run_name} ${output_dir}
export USE_MODELSCOPE_HUB=1
echo "Start recording gpu status "
# 0 means not printing gpu status
python gpu_status.py ${output_dir} 1 10 &
gpu_status_pid=$!
echo "${gpu_status_pid}"
# echo "Start recording gpu status "
# # 0 means not printing gpu status
# python gpu_status.py ${output_dir} 1 10 &
# gpu_status_pid=$!
# echo "${gpu_status_pid}"
sleep 60
if [ "${gpu_cnt}"="1" ]; then
ASCEND_RT_VISIBLE_DEVICES=0 llamafactory-cli train ${output_dir}/${run_name}.yml \
| tee ${output_dir}/log.txt" &
train_pid=$!
echo "Start train"
else
FORCE_TORCHRUN=1 llamafactory-cli train ${output_dir}/${run_name}.yml \
| tee ${output_dir}/log.txt" &
train_pid=$!
echo "Start train"
fi
# if [ "${gpu_cnt}"="1" ]; then
# ASCEND_RT_VISIBLE_DEVICES=0 llamafactory-cli train ${output_dir}/${run_name}.yml \
# | tee ${output_dir}/log.txt" &
# train_pid=$!
# echo "Start train"
# else
# FORCE_TORCHRUN=1 llamafactory-cli train ${output_dir}/${run_name}.yml \
# | tee ${output_dir}/log.txt" &
# train_pid=$!
# echo "Start train"
# fi
wait $train_pid
echo "Train ended"
# wait $train_pid
# echo "Train ended"
# sleep 90
kill $gpu_status_pid
echo "Gpu status ended"
# kill $gpu_status_pid
# echo "Gpu status ended"