run_glue_gazesup_roberta_low_resource.sh

#!/bin/bash

#rte:accuracy, epoch
#mrpc: f1, epoch
#stsb: spearmanr, epoch
#cola: matthews_correlation, epoch
#sst2: accuracy, steps
#qnli: accuracy, steps
#qqp: f1, steps
#mnli: accuracy, steps

# task_to_lr = {'rte': 3e-5,
#         'mrpc': 3e-5,
#         'stsb': 2e-5,
#         'sst2': 1e-5,
#         'cola': 2e-5,
#         'qqp': 2e-5,
#         'mnli': 1e-5,
#         'qnli': 2e-5,
#         }

export TASK_NAME=rte
export LR=3e-05

for MAX_TRAIN_SAMPLES in 200 500 1000
do
  for AUG_WEIGHT in 1.0 0.7 0.5 0.3 0.1 0.01 0.001
  do
    for DATA_SEED in 111 222 333 444 555
    do
      CUDA_VISIBLE_DEVICES=3 python train_glue_gazesup_roberta_low_resource.py \
          --model_name_or_path RoBERTa-base \
          --task_name $TASK_NAME \
          --output_dir result/gazesup_RoBERTa/$TASK_NAME/$MAX_TRAIN_SAMPLES/$AUG_WEIGHT/$DATA_SEED/ \
          --num_train_epochs 20 \
          --learning_rate $LR \
          --per_device_train_batch_size 32 \
          --max_seq_length 128 \
          --augweight $AUG_WEIGHT \
          --evaluation_strategy epoch \
          --save_strategy epoch \
          --metric_for_best_model accuracy \
          --train_as_val True \
          --max_train_samples $MAX_TRAIN_SAMPLES\
          --low_resource_data_seed $DATA_SEED \
          --load_best_model_at_end \
          --overwrite_output_dir \
          --do_train \
          --do_eval \
          --do_predict False\
          --fp16 \
          "$@"
    done
  done
done