Skip to content

representation 미실행 버전 #4

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
sejin-sim wants to merge 1 commit into
base: main
Choose a base branch
from
Open

representation 미실행 버전 #4

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
127 changes: 0 additions & 127 deletions Train.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -444,133 +444,6 @@
"source": [
"--------------------------------------------------"
]
},
{
"cell_type": "code",
"execution_count": 8,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Save MinMaxScaler in path: ./scaler/minmax_scaler_x_repr.pkl\n"
]
}
],
"source": [
"# load representation data\n",
"data_root_dir = './data/'\n",
"train_x, train_y, test_x, test_y = utils.load_data(data_root_dir, model_name='FC') # shape=(num_of_instance, embedding_dim)\n",
"\n",
"# split train data into train/valiation data\n",
"# train data를 랜덤으로 test_size=split_ratio에 대하여 train/validation set으로 분할\n",
"split_ratio = 0.2\n",
"train_x, valid_x, train_y, valid_y = train_test_split(train_x, train_y, test_size=split_ratio, shuffle=True)\n",
"\n",
"# normalization\n",
"scaler_x_path = './scaler/minmax_scaler_x_repr.pkl'\n",
"train_x, valid_x = utils.get_train_val_data(train_x, valid_x, scaler_x_path)"
]
},
{
"cell_type": "code",
"execution_count": 9,
"metadata": {
"scrolled": true
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Start training model: FC\n",
"\n",
"Epoch 1/150\n",
"train Loss: 1.7845 Acc: 0.1682\n",
"val Loss: 1.7542 Acc: 0.2345\n",
"\n",
"Epoch 10/150\n",
"train Loss: 1.2704 Acc: 0.6596\n",
"val Loss: 1.2287 Acc: 0.6737\n",
"\n",
"Epoch 20/150\n",
"train Loss: 0.8235 Acc: 0.7495\n",
"val Loss: 0.7868 Acc: 0.8389\n",
"\n",
"Epoch 30/150\n",
"train Loss: 0.6276 Acc: 0.8051\n",
"val Loss: 0.5940 Acc: 0.8742\n",
"\n",
"Epoch 40/150\n",
"train Loss: 0.5202 Acc: 0.8373\n",
"val Loss: 0.4832 Acc: 0.8926\n",
"\n",
"Epoch 50/150\n",
"train Loss: 0.4367 Acc: 0.8708\n",
"val Loss: 0.4038 Acc: 0.9028\n",
"\n",
"Epoch 60/150\n",
"train Loss: 0.3723 Acc: 0.8912\n",
"val Loss: 0.3447 Acc: 0.9164\n",
"\n",
"Epoch 70/150\n",
"train Loss: 0.3287 Acc: 0.8978\n",
"val Loss: 0.2992 Acc: 0.9211\n",
"\n",
"Epoch 80/150\n",
"train Loss: 0.2944 Acc: 0.9112\n",
"val Loss: 0.2645 Acc: 0.9259\n",
"\n",
"Epoch 90/150\n",
"train Loss: 0.2663 Acc: 0.9170\n",
"val Loss: 0.2383 Acc: 0.9320\n",
"\n",
"Epoch 100/150\n",
"train Loss: 0.2400 Acc: 0.9267\n",
"val Loss: 0.2173 Acc: 0.9334\n",
"\n",
"Epoch 110/150\n",
"train Loss: 0.2243 Acc: 0.9286\n",
"val Loss: 0.2009 Acc: 0.9368\n",
"\n",
"Epoch 120/150\n",
"train Loss: 0.2131 Acc: 0.9289\n",
"val Loss: 0.1885 Acc: 0.9402\n",
"\n",
"Epoch 130/150\n",
"train Loss: 0.2020 Acc: 0.9303\n",
"val Loss: 0.1785 Acc: 0.9443\n",
"\n",
"Epoch 140/150\n",
"train Loss: 0.1910 Acc: 0.9335\n",
"val Loss: 0.1705 Acc: 0.9456\n",
"\n",
"Epoch 150/150\n",
"train Loss: 0.1821 Acc: 0.9362\n",
"val Loss: 0.1633 Acc: 0.9456\n",
"\n",
"Training complete in 0m 17s\n",
"Best val Acc: 0.947655\n"
]
}
],
"source": [
"# Case 5. fully-connected layers (w/ data representation)\n",
"model_name = 'FC'\n",
"model_params = config.model_config[model_name]\n",
"\n",
"data_cls = mc.Classification(model_params)\n",
"best_model = data_cls.train_model(train_x, train_y, valid_x, valid_y) # 모델 학습\n",
"data_cls.save_model(best_model, best_model_path=model_params[\"best_model_path\"]) # 모델 저장"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
Expand Down