Skip to content

Latest commit

 

History

History
10 lines (9 loc) · 384 Bytes

README.md

File metadata and controls

10 lines (9 loc) · 384 Bytes

LSTM-NLP-Vietnamese-Tensorflow-

Tenrsorflow version: 1.10.0

1.Word Embedding(Word2Vec): 300 dims, no char-base embeddings , Batch-size = 20, learning rate = 0.001 , CRFs on top

With 22 epochs,
F1 score: 74.43%
Accuracy: 69.71%

2. BiLSTM - Char_embeddings - word_embeddings - CRFs on top

F1 score: 88.06%
Accuracy: 86.65%