Code Summarization on Python GitHub dataset (test)
0.2527BLEU-1Hybrid2Seq+Attn+DRL
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Hybrid2Seq+Attn+DRLEncoder Architecture=Hybrid (LSTM + AST-based LSTM), Attention=True, DRL=True (Actor-Critic)2018.11 | 0.2527 | 0.1033 | 0.064 | 0.0441 | 0.0929 | 0.3913 | 0.7501 | |
| Seq2Seq+Attn+DRLEncoder Architecture=Sequence-to-Sequence, Attention=True, DRL=True (Actor-Critic)2018.11 | 0.2421 | 0.0919 | 0.0513 | 0.0325 | 0.0882 | 0.3935 | 0.639 | |
| Tree2Seq+Attn+DRLEncoder Architecture=Tree-to-Sequence, Attention=True, DRL=True (Actor-Critic)2018.11 | 0.2309 | 0.0854 | 0.0499 | 0.0338 | 0.0843 | 0.3767 | 0.606 | |
| Hybrid2Seq+AttnEncoder Architecture=Hybrid (LSTM + AST-based LSTM), Attention=True, DRL=False2018.11 | 0.1965 | 0.0516 | 0.028 | 0.0189 | 0.0693 | 0.3154 | 0.3475 | |
| Seq2Seq+AttnEncoder Architecture=Sequence-to-Sequence, Attention=True, DRL=False2018.11 | 0.1897 | 0.0419 | 0.02 | 0.0133 | 0.0649 | 0.3083 | 0.2594 | |
| Tree2Seq+AttnEncoder Architecture=Tree-to-Sequence, Attention=True, DRL=False2018.11 | 0.1887 | 0.0417 | 0.0197 | 0.0129 | 0.0644 | 0.3068 | 0.2331 | |
| Hybrid2SeqEncoder Architecture=Hybrid (LSTM + AST-based LSTM), Attention=False, DRL=False2018.11 | 0.1837 | 0.0379 | 0.0183 | 0.0122 | 0.0604 | 0.302 | 0.2223 | |
| Seq2SeqEncoder Architecture=Sequence-to-Sequence, Attention=False, DRL=False2018.11 | 0.166 | 0.0251 | 0.01 | 0.0056 | 0.0535 | 0.2838 | 0.1262 | |
| Tree2SeqEncoder Architecture=Tree-to-Sequence, Attention=False, DRL=False2018.11 | 0.1649 | 0.0236 | 0.0096 | 0.0053 | 0.0501 | 0.2794 | 0.1168 |