File size: 5,377 Bytes
afc1f82 16bf35b 9b54e42 16bf35b 9b54e42 16bf35b 9b54e42 16bf35b 9b54e42 16bf35b 9b54e42 16bf35b 9b54e42 16bf35b |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 |
---
tags:
- summarization
widget:
- text: "you are given an array of numbers a and a number b , compute the difference of elements in a and b"
---
# CodeTrans model for program synthesis
## Table of Contents
- [Model Details](#model-details)
- [How to Get Started With the Model](#how-to-get-started-with-the-model)
- [Uses](#uses)
- [Risks, Limitations and Biases](#risks-limitations-and-biases)
- [Training](#training)
- [Evaluation](#evaluation)
- [Environmental Impact](#environmental-impact)
- [Citation Information](#citation-information)
## Model Details
- **Model Description:** This CodeTrans model is based on the `t5-small` model. It has its own SentencePiece vocabulary model. It used transfer-learning pre-training on 7 unsupervised datasets in the software development domain. It is then fine-tuned on the program synthesis task for the lisp inspired DSL code.
- **Developed by:** [Ahmed Elnaggar](https://www.linkedin.com/in/prof-ahmed-elnaggar/),[Wei Ding](https://www.linkedin.com/in/wei-ding-92561270/)
- **Model Type:** Summarization
- **Language(s):** English
- **License:** Unknown
- **Resources for more information:**
- [Research Paper](https://arxiv.org/pdf/2104.02443.pdf)
- [GitHub Repo](https://github.com/agemagician/CodeTrans)
## How to Get Started With the Model
Here is how to use this model to generate lisp inspired DSL code using Transformers SummarizationPipeline:
```python
from transformers import AutoTokenizer, AutoModelWithLMHead, SummarizationPipeline
pipeline = SummarizationPipeline(
model=AutoModelWithLMHead.from_pretrained("SEBIS/code_trans_t5_small_program_synthese_transfer_learning_finetune"),
tokenizer=AutoTokenizer.from_pretrained("SEBIS/code_trans_t5_small_program_synthese_transfer_learning_finetune", skip_special_tokens=True),
device=0
)
tokenized_code = "you are given an array of numbers a and a number b , compute the difference of elements in a and b"
pipeline([tokenized_code])
```
Run this example in [colab notebook](https://github.com/agemagician/CodeTrans/blob/main/prediction/multitask/transfer%20learning%20fine-tuning/small_model.ipynb).
## Training data
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)
## Uses
#### Direct Use
The model could be used to generate lisp inspired DSL code given the human language description tasks.
## Training
#### Training Data
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)
The authors provide additionally notes about the vocabulary used, in the [associated paper](https://arxiv.org/pdf/2104.02443.pdf):
> We used the SentencePiece model (Kudo, 2018) to construct the vocabulary for this research, as well as to decode and encode the input/output.
## Training procedure
#### Preprocessing
##### Transfer-learning Pretraining
The model was trained on a single TPU Pod V3-8 for 500,000 steps in total, using sequence length 512 (batch size 4096).
It has a total of approximately 220M parameters and was trained using the encoder-decoder architecture.
The optimizer used is AdaFactor with inverse square root learning rate schedule for pre-training.
###### Fine-tuning
This model was then fine-tuned on a single TPU Pod V2-8 for 5,000 steps in total, using sequence length 512 (batch size 256), using only the dataset only containing lisp inspired DSL data.
## Evaluation
#### Results
For the code documentation tasks, different models achieves the following results on different programming languages (in BLEU score):
Test results :
| Language / Model | LISP |
| -------------------- | :------------: |
| CodeTrans-ST-Small | 89.43 |
| CodeTrans-ST-Base | 89.65 |
| CodeTrans-TF-Small | 90.30 |
| CodeTrans-TF-Base | 90.24 |
| CodeTrans-TF-Large | 90.21 |
| CodeTrans-MT-Small | 82.88 |
| CodeTrans-MT-Base | 86.99 |
| CodeTrans-MT-Large | 90.27 |
| CodeTrans-MT-TF-Small | **90.31** |
| CodeTrans-MT-TF-Base | 90.30 |
| CodeTrans-MT-TF-Large | 90.17 |
| State of the art | 85.80 |
## Environmental Impact
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700). We present the hardware type based on the [associated paper](https://arxiv.org/pdf/2105.09680.pdf).
- **Hardware Type:** Nvidia RTX 8000 GPUs
- **Hours used:** Unknown
- **Cloud Provider:** GCC TPU v2-8 and v3-8.
- **Compute Region:** Unknown
- **Carbon Emitted:** Unknown
## Citation Information
```bibtex
@misc{elnaggar2021codetrans,
title={CodeTrans: Towards Cracking the Language of Silicon's Code Through Self-Supervised Deep Learning and High Performance Computing},
author={Ahmed Elnaggar and Wei Ding and Llion Jones and Tom Gibbs and Tamas Feher and Christoph Angerer and Silvia Severini and Florian Matthes and Burkhard Rost},
year={2021},
eprint={2104.02443},
archivePrefix={arXiv},
primaryClass={cs.SE}
}
```
|