why decoder produce same generated summary ?

[PH-github95]

The model had a loss of 5 and there have always the same generated summary (for all articles ) :

[UNK] [UNK] , 28 , has been charged with two counts of first-degree murder . he has been charged with two counts of attempted murder . he was sentenced to 15 years in prison and sentenced to 18 months in prison .

[maham-collab]

I have same problem with worst results at each time step my generated summary is : [UNK] [UNK] , , , , to to to . the . the . the . the . the . the . the . the . the . the . the . the . the

[tohidarehman1988]

I have same problem with worst results at each time step my generated summary is : [UNK] [UNK] , , , , to to to . the . the . the . the . the . the . the . the . the . the . the . the . the

can you help me how you solved this problem?because i facing the same issue.

[maham-collab]

I have same problem with worst results at each time step my generated summary is : [UNK] [UNK] , , , , to to to . the . the . the . the . the . the . the . the . the . the . the . the . the

can you help me how you solved this problem?because i facing the same issue.

I tried training the model longer. by default it is set to a while(true) loop. so I let it train for more than 4 days in a row. and validated for four days too. it helped to generate better summaries. I am using a single GPU. if you are not able to generate good results try training for longer time.

[yuzexin05]

I have same problem with worst results at each time step my generated summary is : [UNK] [UNK] , , , , to to to . the . the . the . the . the . the . the . the . the . the . the . the . the

can you help me how you solved this problem?because i facing the same issue.

could you solved this problem?