未验证 提交 18c15ef6 编写于 作者: P pkpk 提交者: GitHub

Update run_classifier_multi_gpu.sh (#4281)

上级 1b4bafba
...@@ -6,8 +6,6 @@ DATA_PATH="./data/glue_data/MNLI/" ...@@ -6,8 +6,6 @@ DATA_PATH="./data/glue_data/MNLI/"
CKPT_PATH="./data/saved_model/mnli_models" CKPT_PATH="./data/saved_model/mnli_models"
GPU_TO_USE=0,1,2,3 GPU_TO_USE=0,1,2,3
export CUDA_VISIBLE_DEVICES=$GPU_TO_USE
# start fine-tuning # start fine-tuning
python -m paddle.distributed.launch --selected_gpus=$GPU_TO_USE --log_dir ./cls_log run_classifier.py \ python -m paddle.distributed.launch --selected_gpus=$GPU_TO_USE --log_dir ./cls_log run_classifier.py \
--task_name ${TASK_NAME} \ --task_name ${TASK_NAME} \
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册