pocheyeniu/summarunner

Tensorflow implementation of SummaRuNNer

summarunner's Issues

How do I generate the labeled data?

I get this error when I run "train.py":

Traceback (most recent call last): File "train.py", line 133, in <module> train("model.0201") File "train.py", line 55, in train dr.load_data(FLAGS.train_data_file, max_doc_length, max_sen_length) File "/scratch/user/phan3006/MATH689/SummaRuNNer/data_reader.py", line 83, in load_data for line in f.readlines(): File "/scratch/user/phan3006/MATH689/dl_proj_summarun_py2.7/lib/python2.7/codecs.py", line 696, in readlines return self.reader.readlines(sizehint) File "/scratch/user/phan3006/MATH689/dl_proj_summarun_py2.7/lib/python2.7/codecs.py", line 605, in readlines data = self.read() File "/scratch/user/phan3006/MATH689/dl_proj_summarun_py2.7/lib/python2.7/codecs.py", line 494, in read newchars, decodedbytes = self.decode(data, self.errors) UnicodeDecodeError: 'utf8' codec can't decode byte 0x80 in position 0: invalid start byte

Training

I use the data of CNN/Daily Mail and the hyper parameter of the paper to train the model based on your project. However, I found that the loss keeps almost constant when I set the batch size to 32 or 64 and if I set the batch size to 8, it will be convergence gradually. if you know the reasons, could you tell me about it , many thanks.

数据标签中的0,1,2是什么意思？

请问，数据标签中的0,1,2是什么意思？
在预测时，预测出的y_应该怎么用那，对应的值越大成为摘要的可能越大吗？

interpret summaRunner test output

After testing outputs folder is created which contains two more folder hyp and ref. these two folders contain text files. What is difference in hyp and ref folder.
Does 1.txt tells sentence 1 summary and so on for other files?

datasets files

The links do not work

Data & w2v model: 链接: https://pan.baidu.com/s/1AVwjhBTkkbpUMgo7DAg73w 提取码: ywf5
Google: https://drive.google.com/drive/folders/1Ym2BcE2MBkxfPYWOfzwbrhpJo51OgFlJ?usp=sharing

Can you please provide me the working links of these datasets.

缺少helper.py 文件

eval.py 文件中，8到12行 的 from helper import Config 这个helper在哪里？