The impact of lexical and grammatical processing on generating code from natural language
Open paper
Benchmark
Model
Rank
Results
code-generation-on-conala
TranX + BERT w/mined
#3
BLEU: 34.2
Exact Match Accuracy: 5.8
code-generation-on-django
TranX + BERT w/mined
#3
Accuracy: 81.03
BLEU Score: 79.86
Rank counts only results with a code link.