Concept_Norm/run_model_margin.sh at main · Machine-Learning-for-Medical-Language/Concept_Norm · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
#!/bin/bash
#SBATCH --partition=chip-gpu             # queue to be used
#SBATCH --account=chip
#SBATCH --time=12:00:00             # Running time (in hours-minutes-seconds)
#SBATCH --job-name=conorm             # Job name
#SBATCH --mail-type=BEGIN,END,FAIL      # send and email when the job begins, ends or fails
#SBATCH --mail-user=dongfang.xu@childrens.harvard.edu      # Email address to send the job status
#SBATCH --output=log/exp_margin%j.txt    # Name of the output file
#SBATCH --error=log/exp_margin%j.err
#SBATCH --nodes=1               # Number of gpu nodes
#SBATCH --gres=gpu:Titan_RTX:1                # Number of gpu devices on one gpu node


pwd; hostname; date

module load singularity

margins='0.0 0.05 0.1 0.15 0.2 0.25 0.3'
for margin in $margins
do

    OUTPUT_DIR=/temp_work/ch223150/outputs/joint_model/e15_b16_seq128_5e5_sc30_m$margin

    singularity exec -B $TEMP_WORK --nv /temp_work/ch223150/image/hpc-ml_centos7-python3.7-transformers4.4.1.sif  python3.7 train_system_joint.py \
        --model_name_or_path /home/ch223150/projects/models/0_Transformer \
        --data_dir /home/ch223150/projects/Concept_Norm/data/n2c2/joint_input/st/ \
        --output_dir $OUTPUT_DIR \
        --task_name st_joint \
        --do_train \
        --do_eval \
        --do_predict \
        --train_batch_size 16 \
        --num_train_epochs 15 \
        --overwrite_output_dir true \
        --overwrite_cache true \
        --max_seq_length 128 \
        --token true \
        --label_names st_labels \
        --pad_to_max_length true \
        --learning_rate 5e-5 \
        --margin $margin \
        --scale 30 \
        --evals_per_epoch 1

done