AhmedSSoliman commited on
Commit
8d09cba
·
1 Parent(s): c8eec10

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +31 -4
README.md CHANGED
@@ -7,13 +7,19 @@ widget:
7
  - text: 'find all files in directory "/mydir" with extension ".txt"'
8
  ---
9
 
10
- # MarianCG: A TRANSFORMER MODEL FOR AUTOMATIC CODE GENERATION
11
- This model is to improve the solving of the code generation problem and implement a transformer model that can work with high accurate results. We implemented MarianCG transformer model which is a code generation model that can be able to generate code from natural language. This work declares the impact of using Marian machine translation model for solving the problem of code generation. In our implementation we prove that a machine translation model can be operated and working as a code generation model. Finally, we set the new contributors and state-of-the-art on CoNaLa reaching 10.2 Exact Match Accuracy and a BLEU score of 34.43 in the code generation problem with CoNaLa dataset.
 
 
 
 
 
 
 
12
 
13
  MarianCG model and its implemetation with the code of training and the generated output is available at this repository:
14
  https://github.com/AhmedSSoliman/MarianCG-NL-to-Code
15
 
16
-
17
  CoNaLa Dataset for Code Generation is available at
18
  https://huggingface.co/datasets/AhmedSSoliman/CoNaLa-Large
19
 
@@ -39,4 +45,25 @@ Tasks:
39
  - Code Generation
40
  - Text2Text Generation
41
  - Text Generation
42
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7
  - text: 'find all files in directory "/mydir" with extension ".txt"'
8
  ---
9
 
10
+ ```
11
+ ```
12
+ [![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/mariancg-a-code-generation-transformer-model/code-generation-on-conala)](https://paperswithcode.com/sota/code-generation-on-conala?p=mariancg-a-code-generation-transformer-model)
13
+ ```
14
+
15
+ ```
16
+
17
+ # MarianCG: a code generation transformer model inspired by machine translation
18
+ This model is to improve the solving of the code generation problem and implement a transformer model that can work with high accurate results. We implemented MarianCG transformer model which is a code generation model that can be able to generate code from natural language. This work declares the impact of using Marian machine translation model for solving the problem of code generation. In our implementation, we prove that a machine translation model can be operated and working as a code generation model. Finally, we set the new contributors and state-of-the-art on CoNaLa reaching a BLEU score of 30.92 and Exact Match Accuracy of 6.2 in the code generation problem with CoNaLa dataset.
19
 
20
  MarianCG model and its implemetation with the code of training and the generated output is available at this repository:
21
  https://github.com/AhmedSSoliman/MarianCG-NL-to-Code
22
 
 
23
  CoNaLa Dataset for Code Generation is available at
24
  https://huggingface.co/datasets/AhmedSSoliman/CoNaLa-Large
25
 
 
45
  - Code Generation
46
  - Text2Text Generation
47
  - Text Generation
48
+ ---
49
+
50
+
51
+
52
+
53
+ # Citation
54
+
55
+ We now have a paper for this work and you can cite:
56
+
57
+ ```
58
+ @article{soliman2022mariancg,
59
+ title={MarianCG: a code generation transformer model inspired by machine translation},
60
+ author={Soliman, Ahmed S and Hadhoud, Mayada M and Shaheen, Samir I},
61
+ journal={Journal of Engineering and Applied Science},
62
+ volume={69},
63
+ number={1},
64
+ pages={1--23},
65
+ year={2022},
66
+ publisher={SpringerOpen}
67
+ }
68
+
69
+ ```