From 4b5e00166ac67d00824a1e247129c8e5ca4cda8a Mon Sep 17 00:00:00 2001 From: Anton Kiselev Date: Sat, 18 May 2019 14:53:33 +0300 Subject: [PATCH] Changes in configuration. --- requirements.txt | 4 +++- training.py | 12 +++++++----- 2 files changed, 10 insertions(+), 6 deletions(-) diff --git a/requirements.txt b/requirements.txt index 913428f..eab399d 100644 --- a/requirements.txt +++ b/requirements.txt @@ -1 +1,3 @@ -tensorboardX \ No newline at end of file +tensorboardX +pytorch_pretrained_bert +tqdm \ No newline at end of file diff --git a/training.py b/training.py index 24fb76b..59d50a4 100644 --- a/training.py +++ b/training.py @@ -16,12 +16,14 @@ if __name__ == '__main__': parser = argparse.ArgumentParser(description='bert_adapter') parser.add_argument('--num_epochs', type=int, default=5, metavar='NI', help='num epochs (default: 5)') - parser.add_argument('--batch-size', type=int, default=50, metavar='S') - parser.add_argument('--n_workers', type=int, default=4, metavar='S') - parser.add_argument('--num-threads', type=int, default=4, metavar='BS', + parser.add_argument('--batch-size', type=int, default=10, metavar='S') + parser.add_argument('--n_workers', type=int, default=2, metavar='S') + parser.add_argument('--num-threads', type=int, default=2, help='num threads (default: 4)') parser.add_argument('--dropout', type=float, default=0.4, metavar='D', 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', help='Name for tensorboard model') parser.add_argument('--train_file', type=str, @@ -40,7 +42,7 @@ if __name__ == '__main__': 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) 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) # Load pre-trained model (weights) - model = BertModel.from_pretrained('bert-base-uncased') + model = BertModel.from_pretrained(args.bert_model_name) config = AdapterConfig( hidden_size=768, adapter_size=5,