zhanjunlang / span_oie Goto Github PK

Since the confidence score is not defined in the codes, but it is important to evaluate the performances.

and I doubt the realness of the paper

Can you provide pretrained model which can give the results reported in the paper.

您好，我现在已经训练完一轮，该怎么测试呢？模型的预测结果格式什么？有无现成测试过程样例可以参考？

Hello,
I am attempting to run evaluations on the dataset for my own version but can not understand a part of the script:
In the general reader:
`arg1 = data[3]
rel = data[2]
arg_else = data[4:]
confidence = data[1]
text = data[0]

curExtraction = Extraction(pred = rel, head_pred_index=-1, sent = text, confidence = float(confidence))
curExtraction.addArg(arg1)
for arg in arg_else:
curExtraction.addArg(arg)
d[text] = d.get(text, []) + [curExtraction]`

Why do we add the head as arg1? Or is the head passed using this head_pred_index?

您好，在训练数据中每一句话有sentence2id字段是什么意思呢？

Hi,
I trained the model with the uploaded code, but couldn't reproduce the performance on the paper.
Is the settings in the code, such as random seed, hyper-parameters, and model architecture, the same as that shown in the paper?

I revised some settings like pos_dim and syntax_flag following the paper configurations.
Is there additional things for me to consider to train the model?
or could you release the trained model for future research?

Thank you.

你好，最近读了你的论文，顺着论文的github找到这列，想继续阅读一下源码。请问可以上传代码吗。

具体指关于论文中的指标复现的代码，谢谢