wei commited on
Commit
370e650
1 Parent(s): 4cbe9b0

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +67 -0
README.md CHANGED
@@ -5,3 +5,70 @@ widget:
5
  - text: "public static < T , U > Function < T , U > castFunction ( Class < U > target ) { return new CastToClass < T , U > ( target ) ; }"
6
 
7
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5
  - text: "public static < T , U > Function < T , U > castFunction ( Class < U > target ) { return new CastToClass < T , U > ( target ) ; }"
6
 
7
  ---
8
+
9
+ # CodeTrans model for code documentation generation java
10
+ Pretrained model on programming language java using the t5 base model architecture. It was first released in
11
+ [this repository](https://github.com/agemagician/CodeTrans). This model is trained on tokenized java code functions: it works best with tokenized java functions.
12
+
13
+
14
+ ## Model description
15
+
16
+ This CodeTrans model is based on the `t5-base` model. It has its own SentencePiece vocabulary model. It used multi-task training on 13 supervised tasks in the software development domain and 7 unsupervised datasets.
17
+
18
+ ## Intended uses & limitations
19
+
20
+ The model could be used to generate the description for the java function or be fine-tuned on other java code tasks. It can be used on unparsed and untokenized java code. However, if the java code is tokenized, the performance should be better.
21
+
22
+ ### How to use
23
+
24
+ Here is how to use this model to generate java function documentation using Transformers SummarizationPipeline:
25
+
26
+ ```python
27
+ from transformers import AutoTokenizer, AutoModelWithLMHead, SummarizationPipeline
28
+
29
+ pipeline = SummarizationPipeline(
30
+ model=AutoModelWithLMHead.from_pretrained("SEBIS/code_trans_t5_base_code_documentation_generation_java_multitask"),
31
+ tokenizer=AutoTokenizer.from_pretrained("SEBIS/code_trans_t5_base_code_documentation_generation_java_multitask", skip_special_tokens=True),
32
+ device=0
33
+ )
34
+
35
+ tokenized_code = "public static < T , U > Function < T , U > castFunction ( Class < U > target ) { return new CastToClass < T , U > ( target ) ; }"
36
+ pipeline([tokenized_code])
37
+ ```
38
+ Run this example in [colab notebook](https://github.com/agemagician/CodeTrans/blob/main/prediction/multitask/pre-training/function%20documentation%20generation/java/base_model.ipynb).
39
+ ## Training data
40
+
41
+ The supervised training tasks datasets can be downloaded on [Link](https://www.dropbox.com/sh/488bq2of10r4wvw/AACs5CGIQuwtsD7j_Ls_JAORa/finetuning_dataset?dl=0&subfolder_nav_tracking=1)
42
+
43
+ ## Training procedure
44
+
45
+ ### Multi-task Pretraining
46
+
47
+ The model was trained on a single TPU Pod V3-8 for 480,000 steps in total, using sequence length 512 (batch size 4096).
48
+ It has a total of approximately 220M parameters and was trained using the encoder-decoder architecture.
49
+ The optimizer used is AdaFactor with inverse square root learning rate schedule for pre-training.
50
+
51
+
52
+ ## Evaluation results
53
+
54
+ For the code documentation tasks, different models achieves the following results on different programming languages (in BLEU score):
55
+
56
+ Test results :
57
+
58
+ | Language / Model | Python | Java | Go | Php | Ruby | JavaScript |
59
+ | -------------------- | :------------: | :------------: | :------------: | :------------: | :------------: | :------------: |
60
+ | CodeTrans-ST-Small | 17.31 | 16.65 | 16.89 | 23.05 | 9.19 | 13.7 |
61
+ | CodeTrans-ST-Base | 16.86 | 17.17 | 17.16 | 22.98 | 8.23 | 13.17 |
62
+ | CodeTrans-TF-Small | 19.93 | 19.48 | 18.88 | 25.35 | 13.15 | 17.23 |
63
+ | CodeTrans-TF-Base | 20.26 | 20.19 | 19.50 | 25.84 | 14.07 | 18.25 |
64
+ | CodeTrans-TF-Large | 20.35 | 20.06 | **19.54** | 26.18 | 14.94 | **18.98** |
65
+ | CodeTrans-MT-Small | 19.64 | 19.00 | 19.15 | 24.68 | 14.91 | 15.26 |
66
+ | CodeTrans-MT-Base | **20.39** | 21.22 | 19.43 | **26.23** | **15.26** | 16.11 |
67
+ | CodeTrans-MT-Large | 20.18 | **21.87** | 19.38 | 26.08 | 15.00 | 16.23 |
68
+ | CodeTrans-MT-TF-Small | 19.77 | 20.04 | 19.36 | 25.55 | 13.70 | 17.24 |
69
+ | CodeTrans-MT-TF-Base | 19.77 | 21.12 | 18.86 | 25.79 | 14.24 | 18.62 |
70
+ | CodeTrans-MT-TF-Large | 18.94 | 21.42 | 18.77 | 26.20 | 14.19 | 18.83 |
71
+ | State of the art | 19.06 | 17.65 | 18.07 | 25.16 | 12.16 | 14.90 |
72
+
73
+
74
+ > Created by [Ahmed Elnaggar](https://twitter.com/Elnaggar_AI) | [LinkedIn](https://www.linkedin.com/in/prof-ahmed-elnaggar/) and Wei Ding | [LinkedIn](https://www.linkedin.com/in/wei-ding-92561270/)