#!/bin/bash export TOKENIZERS_PARALLELISM=true dt=`date '+%Y%m%d_%H%M%S'` dataset=$1 shift encoder='roberta-large' args=$@ elr="1e-5" dlr="1e-3" bs=128 mbs=8 unfreeze_epoch=4 k=5 #num of gnn layers gnndim=200 # Existing arguments but changed for GreaseLM encoder_layer=-1 max_node_num=200 seed=5 lr_schedule=fixed if [ ${dataset} = obqa ] then n_epochs=70 max_epochs_before_stop=10 ie_dim=400 else n_epochs=30 max_epochs_before_stop=10 ie_dim=400 fi max_seq_len=100 ent_emb=tzw # Added for GreaseLM info_exchange=true ie_layer_num=1 resume_checkpoint=None resume_id=None sep_ie_layers=false random_ent_emb=false echo "***** hyperparameters *****" echo "dataset: $dataset" echo "enc_name: $encoder" echo "batch_size: $bs mini_batch_size: $mbs" echo "learning_rate: elr $elr dlr $dlr" echo "gnn: dim $gnndim layer $k" echo "ie_dim: ${ie_dim}, info_exchange: ${info_exchange}" echo "******************************" save_dir_pref='runs' mkdir -p $save_dir_pref run_name=greaselm__ds_${dataset}__enc_${encoder}__k${k}__sd${seed}__iedim${ie_dim}__${dt} log=logs/train_${dataset}__${run_name}.log.txt ###### Training ###### python3 -u greaselm.py \ --dataset $dataset \ --encoder $encoder -k $k --gnn_dim $gnndim -elr $elr -dlr $dlr -bs $bs --seed $seed -mbs ${mbs} --unfreeze_epoch ${unfreeze_epoch} --encoder_layer=${encoder_layer} -sl ${max_seq_len} --max_node_num ${max_node_num} \ --n_epochs $n_epochs --max_epochs_before_stop ${max_epochs_before_stop} \ --save_dir ${save_dir_pref}/${dataset}/${run_name} \ --run_name ${run_name} \ --ie_dim ${ie_dim} --info_exchange ${info_exchange} --ie_layer_num ${ie_layer_num} --resume_checkpoint ${resume_checkpoint} --resume_id ${resume_id} --sep_ie_layers ${sep_ie_layers} --random_ent_emb ${random_ent_emb} --ent_emb ${ent_emb//,/ } --lr_schedule ${lr_schedule} \ $args \ # > ${log} 2>&1 & # echo log: ${log}