---
license: apache-2.0
language:
- en
pipeline_tag: text2text-generation
tags:
- bpmn
- Business Process
---
# T5-Small Finetuned for Purchase Order Workflow Business Processes

## Model Description
This is a fine-tuned version of the T5-Small model, designed specifically for the extraction of BPMN (Business Process Model and Notation) diagrams from textual descriptions related to Purchase Order Workflow Business Processes. This AI-driven approach leverages advanced language modeling techniques to transform natural language descriptions into BPMN models, facilitating the modernization and automation of business processes in this specific area. **This model serves as a proof of concept and is not yet ready for real-life applications.**

## Key Features
- **Language Model Base**: T5-Small, known for its efficiency and efficacy in understanding and generating text.
- **Specialization**: Fine-tuned specifically for BPMN generation in Purchase Order Workflows, improving accuracy and relevancy in business process modeling.
- **Dataset**: Trained on the "MaD: A Dataset for Interview-based BPM in Business Process Management" dataset, cited from the research article available at [IEEE Xplore](https://ieeexplore.ieee.org/document/10191898).

## Applications
- **Business Process Management**: Automates the generation of BPMN diagrams, which are crucial for documenting and improving Purchase Order Workflow Business Processes.
- **AI Research and Development**: Provides a research basis for further exploration into the integration of NLP and business process management.
- **Educational Tool**: Assists in teaching the concepts of BPMN and AI's role in business process automation, particularly in the context of Purchase Order Workflows.

## Configuration
- **Pre-trained Model**: Google's T5-Small
- **Training Environment**: Utilized a dataset from "MaD: A Dataset for Interview-based BPM in Business Process Management" for training and validation.
- **Hardware Used**: 
  - **CPU**: Apple M1 MAX with 10 cores (8 performance + 2 efficiency), 3.20 GHz
  - **GPU**: Integrated, 32 cores
  - **RAM**: 64 GB LPDDR5
  - **Storage**: 2 TB SSD
  - **OS**: macOS 12.7.1 (Monterey)
- **Training Script**: [Finetuning T5-Small BPMN](https://github.com/ofachati/Finetuning-T5small-BPMN)

## Installation and Requirements
The model can be accessed and installed via the Hugging Face model hub. Requirements for using this model include Python 3.6 or newer and access to a machine with adequate computational capabilities to run inference with the T5 architecture.

## Contributors
- **Pascal Poizat**: [Hugging Face Profile](https://huggingface.co/pascalpoizat) - Training hardware
- **Omar El Fachati**: [Hugging Face Profile](https://huggingface.co/fachati) - Training script