|
- --------------------------------------
- some config:
- data_dir = ./data
- output_dir = ./output
- embedding_path = ./embedding/hlbl-embeddings-scaled.EMBEDDING_SIZE=50.txt
- word_dim = 50
- model_name = CNN
- mode = 1
- seed = 5782
- cuda = 0
- epoch = 20
- dropout = 0.5
- batch_size = 128
- lr = 0.001
- max_len = 100
- pos_dis = 50
- pos_dim = 5
- hidden_size = 100
- filter_num = 200
- window = 3
- L2_decay = 1e-05
- device = cuda:0
- model_dir = ./output/CNN
- --------------------------------------
- start to load data ...
- finish!
- --------------------------------------
- CNN(
- (word_embedding): Embedding(246123, 50)
- (pos1_embedding): Embedding(103, 5)
- (pos2_embedding): Embedding(103, 5)
- (conv): Conv2d(1, 200, kernel_size=(3, 60), stride=(1, 1), padding=(1, 0))
- (maxpool): MaxPool2d(kernel_size=(100, 1), stride=(100, 1), padding=0, dilation=1, ceil_mode=False)
- (tanh): Tanh()
- (dropout): Dropout(p=0.5, inplace=False)
- (linear): Linear(in_features=200, out_features=100, bias=True)
- (dense): Linear(in_features=100, out_features=19, bias=True)
- )
- traning model parameters:
- word_embedding.weight : torch.Size([246123, 50])
- pos1_embedding.weight : torch.Size([103, 5])
- pos2_embedding.weight : torch.Size([103, 5])
- conv.weight : torch.Size([200, 1, 3, 60])
- conv.bias : torch.Size([200])
- linear.weight : torch.Size([100, 200])
- linear.bias : torch.Size([100])
- dense.weight : torch.Size([19, 100])
- dense.bias : torch.Size([19])
- --------------------------------------
- start to train the model ...
- [001] train_loss: 2.446 | dev_loss: 2.449 | micro f1 on dev: 0.5220 >>> save models!
- [002] train_loss: 1.856 | dev_loss: 1.850 | micro f1 on dev: 0.4652
- [003] train_loss: 1.403 | dev_loss: 1.426 | micro f1 on dev: 0.5612 >>> save models!
- [004] train_loss: 1.186 | dev_loss: 1.265 | micro f1 on dev: 0.6535 >>> save models!
- [005] train_loss: 0.976 | dev_loss: 1.131 | micro f1 on dev: 0.6994 >>> save models!
- [006] train_loss: 0.806 | dev_loss: 1.039 | micro f1 on dev: 0.7259 >>> save models!
- [007] train_loss: 0.658 | dev_loss: 0.976 | micro f1 on dev: 0.7508 >>> save models!
- [008] train_loss: 0.571 | dev_loss: 0.972 | micro f1 on dev: 0.7557 >>> save models!
- [009] train_loss: 0.457 | dev_loss: 0.927 | micro f1 on dev: 0.7690 >>> save models!
- [010] train_loss: 0.364 | dev_loss: 0.905 | micro f1 on dev: 0.7666
- [011] train_loss: 0.289 | dev_loss: 0.896 | micro f1 on dev: 0.7796 >>> save models!
- [012] train_loss: 0.234 | dev_loss: 0.912 | micro f1 on dev: 0.7846 >>> save models!
- [013] train_loss: 0.189 | dev_loss: 0.916 | micro f1 on dev: 0.7926 >>> save models!
- [014] train_loss: 0.151 | dev_loss: 0.924 | micro f1 on dev: 0.7857
- [015] train_loss: 0.123 | dev_loss: 0.959 | micro f1 on dev: 0.7862
- [016] train_loss: 0.097 | dev_loss: 0.980 | micro f1 on dev: 0.7779
- [017] train_loss: 0.078 | dev_loss: 0.985 | micro f1 on dev: 0.7906
- [018] train_loss: 0.064 | dev_loss: 1.025 | micro f1 on dev: 0.7840
- [019] train_loss: 0.051 | dev_loss: 1.025 | micro f1 on dev: 0.7831
- [020] train_loss: 0.042 | dev_loss: 1.066 | micro f1 on dev: 0.7876
- --------------------------------------
- start test ...
- test_loss: 0.916 | micro f1 on test: 0.7926
-
-
-
-
-
-
-
- #自测A6000结果
-
- some config:
- data_dir = ./data
- output_dir = ./output
- embedding_path = ./embedding/hlbl-embeddings-scaled.EMBEDDING_SIZE=50.txt
- word_dim = 50
- model_name = CNN
- mode = 1
- seed = 5782
- cuda = 0
- epoch = 20
- dropout = 0.5
- batch_size = 128
- lr = 0.001
- max_len = 100
- pos_dis = 50
- pos_dim = 5
- hidden_size = 100
- filter_num = 200
- window = 3
- L2_decay = 1e-05
- device = cuda:0
- model_dir = ./output/CNN
- --------------------------------------
- start to load data ...
- finish!
- --------------------------------------
- CNN(
- (word_embedding): Embedding(246123, 50)
- (pos1_embedding): Embedding(103, 5)
- (pos2_embedding): Embedding(103, 5)
- (conv): Conv2d(1, 200, kernel_size=(3, 60), stride=(1, 1), padding=(1, 0))
- (maxpool): MaxPool2d(kernel_size=(100, 1), stride=(100, 1), padding=0, dilation=1, ceil_mode=False)
- (tanh): Tanh()
- (dropout): Dropout(p=0.5, inplace=False)
- (linear): Linear(in_features=200, out_features=100, bias=True)
- (dense): Linear(in_features=100, out_features=19, bias=True)
- )
- traning model parameters:
- word_embedding.weight : torch.Size([246123, 50])
- pos1_embedding.weight : torch.Size([103, 5])
- pos2_embedding.weight : torch.Size([103, 5])
- conv.weight : torch.Size([200, 1, 3, 60])
- conv.bias : torch.Size([200])
- linear.weight : torch.Size([100, 200])
- linear.bias : torch.Size([100])
- dense.weight : torch.Size([19, 100])
- dense.bias : torch.Size([19])
- --------------------------------------
- start to train the model ...
- [001] train_loss: 2.429 | dev_loss: 2.434 | micro f1 on dev: 0.5081 >>> save models!
- [002] train_loss: 1.805 | dev_loss: 1.802 | micro f1 on dev: 0.4451
- [003] train_loss: 1.391 | dev_loss: 1.418 | micro f1 on dev: 0.5715 >>> save models!
- [004] train_loss: 1.166 | dev_loss: 1.249 | micro f1 on dev: 0.6581 >>> save models!
- [005] train_loss: 0.980 | dev_loss: 1.134 | micro f1 on dev: 0.6858 >>> save models!
- [006] train_loss: 0.833 | dev_loss: 1.068 | micro f1 on dev: 0.7280 >>> save models!
- [007] train_loss: 0.681 | dev_loss: 0.993 | micro f1 on dev: 0.7303 >>> save models!
- [008] train_loss: 0.569 | dev_loss: 0.974 | micro f1 on dev: 0.7503 >>> save models!
- [009] train_loss: 0.464 | dev_loss: 0.947 | micro f1 on dev: 0.7571 >>> save models!
- [010] train_loss: 0.382 | dev_loss: 0.936 | micro f1 on dev: 0.7633 >>> save models!
- [011] train_loss: 0.306 | dev_loss: 0.917 | micro f1 on dev: 0.7732 >>> save models!
- [012] train_loss: 0.265 | dev_loss: 0.955 | micro f1 on dev: 0.7730
- [013] train_loss: 0.212 | dev_loss: 0.944 | micro f1 on dev: 0.7790 >>> save models!
- [014] train_loss: 0.168 | dev_loss: 0.946 | micro f1 on dev: 0.7865 >>> save models!
- [015] train_loss: 0.150 | dev_loss: 0.982 | micro f1 on dev: 0.7853
- [016] train_loss: 0.111 | dev_loss: 0.962 | micro f1 on dev: 0.7835
- [017] train_loss: 0.096 | dev_loss: 0.999 | micro f1 on dev: 0.7802
- [018] train_loss: 0.075 | dev_loss: 1.006 | micro f1 on dev: 0.7821
- [019] train_loss: 0.060 | dev_loss: 1.027 | micro f1 on dev: 0.7826
- [020] train_loss: 0.052 | dev_loss: 1.047 | micro f1 on dev: 0.7843
- --------------------------------------
- start test ...
- test_loss: 0.946 | micro f1 on test: 0.7865
|