xwhan / extremely-fine-grained-entity-typing Goto Github PK

Hello XWHan, the work you contribute is gorgeous and I am sorry to disturb you, my problem is as follow:
I reran your model on the OntoNotes dataset, but couldn't perceive the result you proposed(which I can only get 65~67 ma_f1 and you submit 77.8 in your paper), I wonder if there is any trick or dataset preprocess I need to do before I rerun?

Can someone please tell how much time was required to complete the training?

I am basically using the following command -
CUDA_VISIBLE_DEVICES=1 python main.py $RUN_ID$ -lstm_type single -model_debug -enhanced_mention -data_setup joint -add_crowd -multitask -gcn

Thanks!

Hi, thanks for sharing code ! I notice that in the forward function of class GCNMultiDecoder in model_util.py

def forward(self, inputs, output_type):
    connection_matrix = self.label_matrix + self.weight * self.affinity
    label_vectors = self.transform(
        connection_matrix.mm(self.linear.weight) / connection_matrix.sum(
            1, keepdim=True))

    if output_type == "open":
        return self.linear(inputs)
    elif output_type == 'wiki':
        return F.linear(
            inputs, label_vectors[:constant.ANSWER_NUM_DICT['wiki'], :],
            self.linear.bias)
    elif output_type == 'kb':
        return F.linear(inputs,
                        label_vectors[:constant.ANSWER_NUM_DICT['kb'], :],
                        self.linear.bias)
    else:
        raise ValueError('Decoder error: output type not one of the valid')

when output_type is "open", the model simply do a linear transform of the input hidden state which means the model doesn't utilize the label correlation information when the label space is "open", is there something wrong here?

Thanks for uploading code.
Which version of gluonnlp is used?
Because when importing gluonnlp, I came across segmentation fault.

-> import gluonnlp
(Pdb) unt

Segmentation fault: 11

zsh: segmentation fault (core dumped)  python3 -m pdb main.py