Reranking for Neural Semantic Parsing

Open paper
Benchmark Model Rank Results
code-generation-on-conalaReranker–BLEU: 30.11Exact Match Accuracy: 2.8
code-generation-on-conala-extReranker–BLEU: 19.85
code-generation-on-djangoReranker–Accuracy: 80.2

Rank counts only results with a code link.