SEBIS/code_trans_t5_base_program_synthese_multitask_finetune

1---2tags:3- summarization4widget:5- text: "you are given an array of numbers a and a number b , compute the difference of elements in a and b"6 7---8 9 10# CodeTrans model for program synthesis11Pretrained model on programming language lisp inspired DSL using the t5 base model architecture. It was first released in12[this repository](https://github.com/agemagician/CodeTrans). 13 14 15## Model description16 17This 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. It is then fine-tuned on the program synthesis task for the lisp inspired DSL code.18 19## Intended uses & limitations20 21The model could be used to generate lisp inspired DSL code given the human language description tasks.22 23### How to use24 25Here is how to use this model to generate lisp inspired DSL code using Transformers SummarizationPipeline:26 27```python28from transformers import AutoTokenizer, AutoModelWithLMHead, SummarizationPipeline29 30pipeline = SummarizationPipeline(31    model=AutoModelWithLMHead.from_pretrained("SEBIS/code_trans_t5_base_program_synthese_multitask_finetune"),32    tokenizer=AutoTokenizer.from_pretrained("SEBIS/code_trans_t5_base_program_synthese_multitask_finetune", skip_special_tokens=True),33    device=034)35 36tokenized_code = "you are given an array of numbers a and a number b , compute the difference of elements in a and b"37pipeline([tokenized_code])38```39Run this example in [colab notebook](https://github.com/agemagician/CodeTrans/blob/main/prediction/multitask/fine-tuning/program%20synthesis/base_model.ipynb).40## Training data41 42The 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)43 44 45## Training procedure46 47### Multi-task Pretraining48 49The model was trained on a single TPU Pod V3-8 for 500,000 steps in total, using sequence length 512 (batch size 4096).50It has a total of approximately 220M parameters and was trained using the encoder-decoder architecture.51The optimizer used is AdaFactor with inverse square root learning rate schedule for pre-training.52 53### Fine-tuning54 55This model was then fine-tuned on a single TPU Pod V2-8 for 30,000 steps in total, using sequence length 512 (batch size 256), using only the dataset only containing lisp inspired DSL data.56 57 58## Evaluation results59 60For the code documentation tasks, different models achieves the following results on different programming languages (in BLEU score):61 62Test results :63 64|   Language / Model   |      LISP      |65| -------------------- | :------------: |66|   CodeTrans-ST-Small    |     89.43      |67|   CodeTrans-ST-Base     |     89.65      |68|   CodeTrans-TF-Small    |     90.30      |69|   CodeTrans-TF-Base     |     90.24      |70|   CodeTrans-TF-Large    |     90.21      |71|   CodeTrans-MT-Small    |     82.88      |72|   CodeTrans-MT-Base     |     86.99      |73|   CodeTrans-MT-Large    |     90.27      |74|   CodeTrans-MT-TF-Small |   **90.31**    |75|   CodeTrans-MT-TF-Base  |     90.30      |76|   CodeTrans-MT-TF-Large |     90.17      |77|   State of the art   |     85.80      |78 79 80 81> 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/)82 83