wassname/bert_adapter
1
0
Fork 0
mirror of https://github.com/wassname/bert_adapter.git synced 2026-06-27 16:44:45 +08:00
Code Issues Packages Projects Releases Wiki Activity

Changes in configuration.

Browse Source
This commit is contained in:
Anton Kiselev
2019-05-18 14:53:33 +03:00
parent fc96b0b092
commit 4b5e00166a
2 changed files with 10 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files
requirements.txt
+3 -1
View File
@@ -1 +1,3 @@
tensorboardX tensorboardX
pytorch_pretrained_bert
tqdm
training.py
+7 -5
View File
@@ -16,12 +16,14 @@ if __name__ == '__main__':
parser = argparse.ArgumentParser(description='bert_adapter') parser = argparse.ArgumentParser(description='bert_adapter')
parser.add_argument('--num_epochs', type=int, default=5, metavar='NI', parser.add_argument('--num_epochs', type=int, default=5, metavar='NI',
help='num epochs (default: 5)') help='num epochs (default: 5)')
parser.add_argument('--batch-size', type=int, default=50, metavar='S') parser.add_argument('--batch-size', type=int, default=10, metavar='S')
parser.add_argument('--n_workers', type=int, default=4, metavar='S') parser.add_argument('--n_workers', type=int, default=2, metavar='S')
parser.add_argument('--num-threads', type=int, default=4, metavar='BS', parser.add_argument('--num-threads', type=int, default=2,
help='num threads (default: 4)') help='num threads (default: 4)')
parser.add_argument('--dropout', type=float, default=0.4, metavar='D', parser.add_argument('--dropout', type=float, default=0.4, metavar='D',
help='dropout rate (default: 0.4)') help='dropout rate (default: 0.4)')
parser.add_argument('--bert_model_name', type=str, default='bert-base-multilingual-cased',
help='Bert model name (ex. "bert-base-multilingual-cased")')
parser.add_argument('--tensorboard', type=str, default='default_tb', metavar='TB', parser.add_argument('--tensorboard', type=str, default='default_tb', metavar='TB',
help='Name for tensorboard model') help='Name for tensorboard model')
parser.add_argument('--train_file', type=str, parser.add_argument('--train_file', type=str,
@@ -40,7 +42,7 @@ if __name__ == '__main__':
torch.set_num_threads(args.num_threads) torch.set_num_threads(args.num_threads)
tokenizer = BertTokenizer.from_pretrained('bert-base-uncased') tokenizer = BertTokenizer.from_pretrained(args.bert_model_name)
train_dataset = TokenizedDataFrameDataset(tokenizer, file_path=args.train_file) train_dataset = TokenizedDataFrameDataset(tokenizer, file_path=args.train_file)
test_dataset = TokenizedDataFrameDataset(tokenizer, file_path=args.test_file) test_dataset = TokenizedDataFrameDataset(tokenizer, file_path=args.test_file)
@@ -49,7 +51,7 @@ if __name__ == '__main__':
test_loader = DataLoader(test_dataset, batch_size=args.batch_size, num_workers=args.n_workers) test_loader = DataLoader(test_dataset, batch_size=args.batch_size, num_workers=args.n_workers)
# Load pre-trained model (weights) # Load pre-trained model (weights)
model = BertModel.from_pretrained('bert-base-uncased') model = BertModel.from_pretrained(args.bert_model_name)
config = AdapterConfig( config = AdapterConfig(
hidden_size=768, adapter_size=5, hidden_size=768, adapter_size=5,