[POST] ⚡pytorch lightning 으로 koBERT Fine-tuning 하기 - NSMC
$ python download_pretrained.py --save_path pretrained- It will create
'./pretrained'directory. - It will download kobert pretrained model and tokenizer files below
'./pretrained'directory.
# run example
python train.py- arguments
- seed: random seed number
- data_path: where to prepare data
- max_epoch: maximum number of epochs to train
- num_gpus: number of available gpus (-1: all avauilable)
- mode: train only korean data or not
- save_path: where to save checkpoints files
- valid_size: size of validation file
- max_seq_len: number of available gpus
- batch_size: batch size
# run example
python test.py --ckpt_path checkpoints/epoch=05-val_acc=0.897.ckpt- arguments
- ckpt_path: checkpoint file path which is execute test with
- The rest is same with train.py
# run example
python export_to_onnx.py --ckpt_path checkpoints/epoch=05-val_acc=0.897.ckpt- It will create
model.onnx.