LSTM-CRF

LSTM-CRF copied to clipboard

Feature template

1

How do you combine feature templates with neural networks?

qinya

About the data

1

thank you share your code, can you share your data with us.

chaoyueziji

some bugs

6

I think there are some bugs about FLAGS

Felix0805

The CNN-BiLSTM+CRF model has some problems with network structure. It will be fixed in the next few versions.

heshenghuan

embeddings required?

1

how can I close embeddings?

syllenz

where is the dataset? thank you

1

where is the dataset? thank you.

yingdongucas

What's the principle of the sum of embedding vector as feature template score ?

1

In `hybrid_model.py`, you add `feature template`, if I am right, the implementation is [line 116](https://github.com/heshenghuan/LSTM-CRF/blob/master/hybrid_model.py#L116), ``` scores = feat_sum + tf.matmul(outputs, self.W) + self.b ``` where `feat_sum` is defined as:...

smy69

Don't use tf.softmax before you sent the scores to TF's crf API. I found that could cause a sharp drop in performance.

heshenghuan