From: MarianCG: a code generation transformer model inspired by machine translation
Rank
Model
BLEU score
Exact match accuracy
Year
1
MarianCG (Ours)
90.41
81.83
2022
2
TranX + BERT w/ mined [28]
79.86
81.03
3
BERT + TAE [26]
-
81.77
2021
4
Reranker [21]
80.2
2019
5
TranX [20]
73.7
2018
6
ipn [40]
77.6
62.3
2016
7
Phrasal Statistical MT [40]
47.6
31.5