thunlp/ERNIE

tried run_tacred.py, but training progress stuck at 0% for three days

Coopercoppers opened this issue · 1 comments

Hi, I tried to finetune the model for tacred, and i used the code python3 code/run_tacred.py --do_train --do_lower_case --data_dir data/tacred --ernie_model ernie_base --max_seq_length 256 --train_batch_size 32 --learning_rate 2e-5 --num_train_epochs 4.0 --output_dir output_tacred --fp16 --loss_scale 128 --threshold 0.4. However, the interface says that the progress stuck at 0% for three days,
QQ截图20210711235241. I am at loss about what to do? Any suggestions?

zzy14 commented

Maybe your server has run out of memory. Did you try other tasks?