Tested on two data sizes in training model.
Epoch: 008; loss = 0.8370 cost time 7.9410 Accuracy: 0.7930 Loss in test 0.0000
precision recall f1-score support
<extra_id_0>education 0.9367 0.8706 0.9024 85
<extra_id_0>entertainment 0.7661 0.8879 0.8225 107
<extra_id_0>finance 0.6667 0.7478 0.7049 115
<extra_id_0>game 0.8333 0.7843 0.8081 102
<extra_id_0>politics 0.7119 0.8842 0.7887 95
<extra_id_0>realty 0.8942 0.8455 0.8692 110
<extra_id_0>science 0.8169 0.5800 0.6784 100
<extra_id_0>society 0.9359 0.7604 0.8391 96
<extra_id_0>sports 0.9149 0.8866 0.9005 97
<extra_id_0>stocks 0.5981 0.6882 0.6400 93
accuracy 0.7930 1000
macro avg 0.8075 0.7935 0.7954 1000
weighted avg 0.8052 0.7930 0.7940 1000
Epoch: 019; loss = 0.3964 cost time 73.6844 Accuracy: 0.8546 Loss in test 0.0000
Accuracy: 0.8546 Loss in test 0.0000
precision recall f1-score support
education 0.9221 0.9350 0.9285 1000
entertainment 0.8498 0.8490 0.8494 1000
finance 0.8697 0.8480 0.8587 1000
game 0.8690 0.8690 0.8690 1000
politics 0.8121 0.8340 0.8229 1000
realty 0.8633 0.8840 0.8735 1000
science 0.8070 0.7860 0.7964 1000
society 0.8600 0.8600 0.8600 1000
sports 0.9076 0.9140 0.9108 1000
stocks 0.7827 0.7670 0.7747 1000
accuracy 0.8546 10000
macro avg 0.8543 0.8546 0.8544 10000
weighted avg 0.8543 0.8546 0.8544 10000