Pytorch implementation of summarization with Pointer-Generator Networks.

An implementation of a paper "Get To The Point: Summarization with Pointer-Generator Networks" with a few modifications added.

Baseline scores comparison

Evaluating on ria-news dataset.

Baseline\Metric	rouge-1-f	rouge-1-p	rouge-1-r	rouge-2-f	rouge-2-p	rouge-2-r	rouge-l-f	rouge-l-p	rouge-l-r
First sentence	0.23394544	0.16467702	0.44054993	0.10301857	0.07112514	0.20716452	0.16290575	0.15132185	0.40389810
3 sentences	0.15235018	0.09042597	0.53039291	0.05835634	0.03373963	0.24088673	0.08697666	0.08455916	0.49655811
Seq2seq+attn	0.41582201	0.43718827	0.40561483	0.24086001	0.25307833	0.23581065	0.38644251	0.41508011	0.38504612
Pointer-Generator	0.41094998	0.43332178	0.39906258	0.24201008	0.25566738	0.23547465	0.38392726	0.41334106	0.38063485
Seq2seq+attn+glove	0.41925908	0.44047456	0.40921883	0.24513830	0.25763531	0.23992956	0.39001231	0.41847043	0.38874970
Pointer-Generator+Glove	0.41242970	0.43542033	0.40028182	0.24193624	0.25587737	0.23518939	0.38503439	0.41518183	0.38158816
Seq2Seq+attn+unsup	0.27785800	0.29142466	0.27177418	0.13108816	0.13720493	0.12893772	0.25623700	0.27484959	0.25632151
Seq2Seq+attn+ner	0.15539128	0.16671571	0.14977648	0.05272446	0.05646063	0.05105246	0.14240497	0.15737464	0.14127297
Seq2Seq+attn+1sent	0.38811201	0.40609767	0.37947066	0.22280892	0.23364022	0.21827619	0.36166143	0.38600110	0.36074401
Seq2Seq+attn+3sent	0.19930877	0.21369066	0.19196596	0.08043450	0.08500956	0.07857434	0.18417329	0.20316124	0.18246018