20 lines
568 B
Bash
20 lines
568 B
Bash
|
nohup python -u main.py --gpus "0," --max_epochs=20 --num_workers=32 \
|
||
|
|
--model_name_or_path bert-base-uncased \
|
||
|
|
--accumulate_grad_batches 1 \
|
||
|
|
--model_class BertKGC \
|
||
|
|
--batch_size 128 \
|
||
|
|
--checkpoint /where/you/put/the/checkpoint/file/of/the/pretrain/process \
|
||
|
|
--pretrain 0 \
|
||
|
|
--bce 0 \
|
||
|
|
--check_val_every_n_epoch 1 \
|
||
|
|
--overwrite_cache \
|
||
|
|
--data_dir dataset/umls \
|
||
|
|
--eval_batch_size 256 \
|
||
|
|
--max_seq_length 128 \
|
||
|
|
--lr 3e-5 \
|
||
|
|
--max_triplet 64 \
|
||
|
|
--add_attn_bias True \
|
||
|
|
--use_global_node True \
|
||
|
|
>logs/train_umls.log 2>&1 &
|
||
|
|
|