-
Notifications
You must be signed in to change notification settings - Fork 2
/
sh_eval_resume.sh
16 lines (16 loc) · 934 Bytes
/
sh_eval_resume.sh
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
# export CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7
# for step in 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000 22000 24000 26000
for step in 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000 22000 24000 26000 28000 30000 32000 34000 36000 38000 40000 42000 44000 46000 48000 50000 52000 54000 56000 58000 60000 62000 64000 66000 68000 70000 72000 74000 76000 78000 80000 82000 84000 86000 88000 90000 92000 94000 96000 98000 100000 102000 104000 106000 108000 110000 112000
do
echo "Run evaluate resume on steps $step"
python run_eval.py \
--dev_dir ./ner_datasets/resume_dev.json \
--max_source_length 512 \
--max_target_length 512 \
--beam_width 10 \
--model /mnt/lustrenew/lujinghui1/Unified_NER/my_trained_models_10000/$step/ \
--dataset_name resume \
--eval_result_dir ./unified_ner_v1.5/results_$step/ \
--batch_size 8
done
# spring.submit arun --gpu -n1 -s --job-name=R-SC210077.00110 "bash cluster_run.sh"