temporal file for testing purposes

35b8c03b · jrzaurin · ba65b112 · 35b8c03b
隐藏空白更改
内联并排

Showing with 90 addition and 75 deletion

main.py main.py +90 -75

未找到文件。
--- a/main.py
+++ b/main.py
@@ -5,89 +5,104 @@ import numpy as np
 import pandas as pd
 from pathlib import Path
 from torchvision import transforms
-from widedeep.models.wide_deep import WideDeep, WideDeepLoader
+from pytorch_widedeep.models.wide_deep import WideDeep, WideDeepLoader
 from sklearn.metrics import mean_squared_error
+import pdb
 if __name__ == '__main__':
    use_cuda = torch.cuda.is_available()
-    wd_dataset = pickle.load(open("data/airbnb/wide_deep_data/wd_dataset.p", "rb"))
+    wd = pickle.load(open('data/wd_dataset.p', 'rb'))
-    params = dict()
-    params['wide'] = dict(
-        wide_dim = wd_dataset['train']['wide'].shape[1]
-        )
-    params['deep_dense'] = dict(
-        embeddings_input = wd_dataset['cat_embeddings_input'],
-        embeddings_encoding_dict = wd_dataset['cat_embeddings_encoding_dict'],
-        continuous_cols = wd_dataset['continuous_cols'],
-        deep_column_idx = wd_dataset['deep_column_idx'],
-        hidden_layers = [64,32],
-        dropout = [0.5]
-        )
-    params['deep_text'] = dict(
-        vocab_size = len(wd_dataset['vocab'].itos),
-        embedding_dim = wd_dataset['word_embeddings_matrix'].shape[1],
-        hidden_dim = 64,
-        n_layers = 2,
-        rnn_dropout = 0.5,
-        spatial_dropout = 0.1,
-        padding_idx = 1,
-        attention = False,
-        bidirectional = True,
-        embedding_matrix = wd_dataset['word_embeddings_matrix']
-        )
-    params['deep_img'] = dict(
-        pretrained = True,
-        freeze='all',
-        )
-    model = WideDeep(output_dim=1, **params)
+    model = WideDeep(output_dim=1, wide_dim=wd.wide.shape[1],
-    # optimizer={'widedeep': ['Adam', 0.1]}
+        embeddings_input = wd.cat_embeddings_input,
-    # lr_scheduler = {'widedeep': ['MultiStepLR', [3,5,7], 0.1]}
+        embeddings_encoding_dict=wd.cat_embeddings_encoding_dict,
-    optimizer=dict(
+        continuous_cols=wd.continuous_cols,
-        wide=['Adam', 0.1],
+        deep_column_idx=wd.deep_column_idx, vocab_size=len(wd.vocab.itos),
-        deep_dense=['Adam', 0.01],
+        pretrained=False)
-        deep_text=['RMSprop', 0.01,0.1],
+    model.compile(method='regression', optimizer='Adam')
-        deep_img= ['Adam', 0.01]
+    print(model.optimizer)
-        )
+    print(model.lr_scheduler)
-    lr_scheduler=dict(
+    print(model)
-        wide=['StepLR', 3, 0.1],
-        deep_dense=['StepLR', 3, 0.1],
-        deep_text=['MultiStepLR', [3,5,7], 0.1],
+    # wd_dataset = pickle.load(open("data/airbnb/wide_deep_data/wd_dataset.p", "rb"))
-        deep_img=['MultiStepLR', [3,5,7], 0.1]
+    # params = dict()
-        )
+    # params['wide'] = dict(
-    model.compile(method='regression', optimizer=optimizer, lr_scheduler=lr_scheduler)
+    #     wide_dim = wd_dataset['train']['wide'].shape[1]
-    if use_cuda:
+    #     )
-        model = model.cuda()
+    # params['deep_dense'] = dict(
-    # # ImageNet metrics
+    #     embeddings_input = wd_dataset['cat_embeddings_input'],
-    # mean=[0.485, 0.456, 0.406] #RGB
+    #     embeddings_encoding_dict = wd_dataset['cat_embeddings_encoding_dict'],
-    # std=[0.229, 0.224, 0.225]  #RGB
+    #     continuous_cols = wd_dataset['continuous_cols'],
-    # cv2 reads BGR
+    #     deep_column_idx = wd_dataset['deep_column_idx'],
-    mean=[0.406, 0.456, 0.485] #BGR
+    #     hidden_layers = [64,32],
-    std=[0.225, 0.224, 0.229]  #BGR
+    #     dropout = [0.5]
-    transform  = transforms.Compose([
+    #     )
-        transforms.ToTensor(),
+    # params['deep_text'] = dict(
-        transforms.Normalize(mean=mean, std=std)
+    #     vocab_size = len(wd_dataset['vocab'].itos),
-    ])
+    #     embedding_dim = wd_dataset['word_embeddings_matrix'].shape[1],
-    train_set = WideDeepLoader(wd_dataset['train'], transform, mode='train')
+    #     hidden_dim = 64,
-    valid_set = WideDeepLoader(wd_dataset['valid'], transform, mode='train')
+    #     n_layers = 2,
-    test_set = WideDeepLoader(wd_dataset['test'], transform, mode='test')
+    #     rnn_dropout = 0.5,
-    train_loader = torch.utils.data.DataLoader(dataset=train_set,
+    #     spatial_dropout = 0.1,
-        batch_size=64, num_workers=4, shuffle=True)
+    #     padding_idx = 1,
-    valid_loader = torch.utils.data.DataLoader(dataset=valid_set,
+    #     attention = False,
-        batch_size=64, num_workers=4, shuffle=True)
+    #     bidirectional = True,
-    test_loader = torch.utils.data.DataLoader(dataset=test_set,
+    #     embedding_matrix = wd_dataset['word_embeddings_matrix']
-        batch_size=32,shuffle=False)
+    #     )
-    model.fit(n_epochs=10, train_loader=train_loader, eval_loader=valid_loader)
+    # params['deep_img'] = dict(
-    preds = model.predict(test_loader)
+    #     pretrained = True,
-    y = wd_dataset['test']['target']
+    #     freeze='all',
-    print(np.sqrt(mean_squared_error(y, preds)))
+    #     )
-    # save
-    MODEL_DIR = Path('data/models')
+    # model = WideDeep(output_dim=1, **params)
-    if not MODEL_DIR.exists(): os.makedirs(MODEL_DIR)
+    # # optimizer={'widedeep': ['Adam', 0.1]}
-    torch.save(model.state_dict(), MODEL_DIR/'widedeep.pkl')
+    # # lr_scheduler = {'widedeep': ['MultiStepLR', [3,5,7], 0.1]}
+    # optimizer=dict(
+    #     wide=['Adam', 0.1],
+    #     deep_dense=['Adam', 0.01],
+    #     deep_text=['RMSprop', 0.01,0.1],
+    #     deep_img= ['Adam', 0.01]
+    #     )
+    # lr_scheduler=dict(
+    #     wide=['StepLR', 3, 0.1],
+    #     deep_dense=['StepLR', 3, 0.1],
+    #     deep_text=['MultiStepLR', [3,5,7], 0.1],
+    #     deep_img=['MultiStepLR', [3,5,7], 0.1]
+    #     )
+    # model.compile(method='regression', optimizer=optimizer, lr_scheduler=lr_scheduler)
+    # if use_cuda:
+    #     model = model.cuda()
+    # # # ImageNet metrics
+    # # mean=[0.485, 0.456, 0.406] #RGB
+    # # std=[0.229, 0.224, 0.225]  #RGB
+    # # cv2 reads BGR
+    # mean=[0.406, 0.456, 0.485] #BGR
+    # std=[0.225, 0.224, 0.229]  #BGR
+    # transform  = transforms.Compose([
+    #     transforms.ToTensor(),
+    #     transforms.Normalize(mean=mean, std=std)
+    # ])
+    # train_set = WideDeepLoader(wd_dataset['train'], transform, mode='train')
+    # valid_set = WideDeepLoader(wd_dataset['valid'], transform, mode='train')
+    # test_set = WideDeepLoader(wd_dataset['test'], transform, mode='test')
+    # train_loader = torch.utils.data.DataLoader(dataset=train_set,
+    #     batch_size=64, num_workers=4, shuffle=True)
+    # valid_loader = torch.utils.data.DataLoader(dataset=valid_set,
+    #     batch_size=64, num_workers=4, shuffle=True)
+    # test_loader = torch.utils.data.DataLoader(dataset=test_set,
+    #     batch_size=32,shuffle=False)
+    # model.fit(n_epochs=10, train_loader=train_loader, eval_loader=valid_loader)
+    # preds = model.predict(test_loader)
+    # y = wd_dataset['test']['target']
+    # print(np.sqrt(mean_squared_error(y, preds)))
+    # # save
+    # MODEL_DIR = Path('data/models')
+    # if not MODEL_DIR.exists(): os.makedirs(MODEL_DIR)
+    # torch.save(model.state_dict(), MODEL_DIR/'widedeep.pkl')
    # load
    # model = WideDeep(1, **params)