Nigel
/
mindspore
forked from mindspore-Ecosystem/mindspore
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

fix bert export bug

This commit is contained in:
yuzhenhua 2020-12-02 15:46:49 +08:00
parent 3eb4c14d86
commit 0d1f209d69
1 changed files with 10 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
model_zoo/official/nlp/bert/export.py
View File

@ -21,23 +21,26 @@ import mindspore.common.dtype as mstype
from mindspore.train.serialization import load_checkpoint, export from mindspore.train.serialization import load_checkpoint, export
from src.finetune_eval_model import BertCLSModel, BertSquadModel, BertNERModel from src.finetune_eval_model import BertCLSModel, BertSquadModel, BertNERModel
from src.finetune_eval_config import optimizer_cfg, bert_net_cfg from src.finetune_eval_config import bert_net_cfg
from src.bert_for_finetune import BertNER from src.bert_for_finetune import BertNER
from src.utils import convert_labels_to_index from src.utils import convert_labels_to_index
context.set_context(mode=context.GRAPH_MODE, device_target="Ascend")
parser = argparse.ArgumentParser(description='Bert export') parser = argparse.ArgumentParser(description='Bert export')
parser.add_argument("--device_id", type=int, default=0, help="Device id")
parser.add_argument('--use_crf', type=str, default="false", help='Use cfg, default is false.') parser.add_argument('--use_crf', type=str, default="false", help='Use cfg, default is false.')
parser.add_argument('--downstream_task', type=str, choices=["NER", "CLS", "SQUAD"], default="NER", parser.add_argument('--downstream_task', type=str, choices=["NER", "CLS", "SQUAD"], default="NER",
help='at present，support NER only') help='at present，support NER only')
parser.add_argument('--num_class', type=int, default=41, help='The number of class, default is 41.') parser.add_argument('--num_class', type=int, default=41, help='The number of class, default is 41.')
parser.add_argument("--batch_size", type=int, default=16, help="batch size")
parser.add_argument('--label_file_path', type=str, default="", help='label file path, used in clue benchmark.') parser.add_argument('--label_file_path', type=str, default="", help='label file path, used in clue benchmark.')
parser.add_argument('--ckpt_file', type=str, required=True, help='Bert ckpt file.') parser.add_argument('--ckpt_file', type=str, required=True, help='Bert ckpt file.')
parser.add_argument('--output_file', type=str, default='Bert.air', help='bert output air name.') parser.add_argument('--output_file', type=str, default='Bert.air', help='bert output air name.')
parser.add_argument('--file_format', type=str, choices=["AIR", "ONNX", "MINDIR"], default='AIR', help='file format') parser.add_argument('--file_format', type=str, choices=["AIR", "ONNX", "MINDIR"], default='AIR', help='file format')
args = parser.parse_args() args = parser.parse_args()
context.set_context(mode=context.GRAPH_MODE, device_target="Ascend", device_id=args.device_id)
label_list = [] label_list = []
with open(args.label_file_path) as f: with open(args.label_file_path) as f:
for label in f: for label in f:
@ -57,7 +60,7 @@ else:
if __name__ == '__main__': if __name__ == '__main__':
if args.downstream_task == "NER": if args.downstream_task == "NER":
if args.use_crf.lower() == "true": if args.use_crf.lower() == "true":
net = BertNER(bert_net_cfg, optimizer_cfg.batch_size, False, num_labels=number_labels, net = BertNER(bert_net_cfg, args.batch_size, False, num_labels=number_labels,
use_crf=True, tag_to_index=tag_to_index) use_crf=True, tag_to_index=tag_to_index)
else: else:
net = BertNERModel(bert_net_cfg, False, number_labels, use_crf=(args.use_crf.lower() == "true")) net = BertNERModel(bert_net_cfg, False, number_labels, use_crf=(args.use_crf.lower() == "true"))
@ -71,10 +74,10 @@ if __name__ == '__main__':
load_checkpoint(args.ckpt_file, net=net) load_checkpoint(args.ckpt_file, net=net)
net.set_train(False) net.set_train(False)
input_ids = Tensor(np.zeros([optimizer_cfg.batch_size, bert_net_cfg.seq_length]), mstype.int32) input_ids = Tensor(np.zeros([args.batch_size, bert_net_cfg.seq_length]), mstype.int32)
input_mask = Tensor(np.zeros([optimizer_cfg.batch_size, bert_net_cfg.seq_length]), mstype.int32) input_mask = Tensor(np.zeros([args.batch_size, bert_net_cfg.seq_length]), mstype.int32)
token_type_id = Tensor(np.zeros([optimizer_cfg.batch_size, bert_net_cfg.seq_length]), mstype.int32) token_type_id = Tensor(np.zeros([args.batch_size, bert_net_cfg.seq_length]), mstype.int32)
label_ids = Tensor(np.zeros([optimizer_cfg.batch_size, bert_net_cfg.seq_length]), mstype.int32) label_ids = Tensor(np.zeros([args.batch_size, bert_net_cfg.seq_length]), mstype.int32)
if args.downstream_task == "NER" and args.use_crf.lower() == "true": if args.downstream_task == "NER" and args.use_crf.lower() == "true":
input_data = [input_ids, input_mask, token_type_id, label_ids] input_data = [input_ids, input_mask, token_type_id, label_ids]