Creator: John Mario Montoya Zapata 👨💻
| User | Version | date |
|---|---|---|
| John Mario Montoya Zapata | 0.1.0 | 2025-02-28 |
| John Mario Montoya Zapata | 1.0.2 | 2025-03-03 |
TuyaBot-LLM is a chatbot based on a RAG architecture, which uses an large language model (LLM), the unsloth/Llama-3.2-1B-Instruct from meta, and the web information available from the financial company TUYA S.A to answer questions related to its products and services.
- Demo: How to interact with the local LLM
- How reproduce in your local machine
- Repository structure
- Cloning this repository
- Setting up a virtual environment
The device tested on is an HP Victus with the following specifications:
- Processor 12th Gen Intel(R) Core(TM) i5-12500H 2.50 GHz
- Installed RAM 16.0 GB (15.7 GB usable)
- Installed GPU NVIDIA GFORCE RTX 3050 4.0 GB
- Clone the repository
- Create a virtual environment and activate it
- Install dependencies: In bash, use the next commands, Be careful, your GPU shoul support cuda 12.6. Here to get more information. Otherwise, the model will be installed on CPU.
pip install -r requirements.txtpip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu126- In parent directory run the next command to get the base information
python main.py make_data- Run application using the next command
python app.pyThis project structure was partially influenced by the Cookiecutter Data Science project and reproducible-model repository.
Check this post by Jeremy Jordan for get guidelines on managing ML projects.
Other resources.
├── LICENSE
|
├── README.md <- You are here
|
├── app/ <- Folder to store the API that exposes the model
|
├── credentials/ <- Folder to store credentials files
|
├── data/ <- Folder that contains data used or generated
│ ├── external/ <- Data from third parties (external to the core company of the project)
│ ├── interim/ <- Data in an intermediate state of processing
│ ├── processed/ <- Data fully processed and ready to be used in modeling
│ └── raw/ <- The original, immutable data dump
|
├── docs/ <- A default Sphinx project; see sphinx-doc.org for details
|
├── models/ <- Trained and serialized models, model predictions, or model summaries
│ └── mlruns/ <- Folder to store trained, serialized and tracked models using the MLflow library
|
├── notebooks/ <- Jupyter notebooks. Naming convention is a number (for ordering),
│ the creator's initials, and a short `-` delimited description, e.g.
│ `101-jmmz-initial-data-exploration.ipynb`.
|
├── references/ <- Data dictionaries, manuals, and all other explanatory materials
|
├── reports/ <- Generated analysis as HTML, PDF, LaTeX, etc
│ ├── figures/ <- Generated graphics and figures to be used in reporting
│ ├── pdfs/ <- PDF files for reporting
│ └── logs/ <- Store some flat file reports concerning the execution of commands by terminal mainly
|
├── tuyabot_llm/ <- Source code for use in this project
│ ├── __init__.py <- Makes src a Python module
│ ├── archive/ <- Old scripts that are removed by restructuring the code. They are kept for future reference
│ ├── data/ <- Scripts to generate, obtain, clean or load raw data
│ ├── features/ <- Scripts to turn data into features for modeling
│ ├── models/ <- Scripts to use trained models to make predictions and to retrain models
│ ├── performance/ <- Scripts to evaluate the performance of models and to calculate metrics from the trained model
│ ├── visualization/ <- Scripts to generate evaluation graphs or reports
│ └── utils/
│ ├── __init__.py
│ └── absolute_paths.py <- Module for handling absolute path
│
├── queries/ <- Folder to store .sql files used at some point in the modeling process
│ ├── develop/ <- Queries created in bulding process
│ └── production/ <- Clean queries used to production
|
├── environment.yml <- The environment file for reproducing the analysis environment, e.g.
│ generated with `conda env export --from-history --file environment.yml`
|
├── requirements.txt <- The requirements file for reproducing the analysis environment, e.g.
│ generated with `pip freeze > requirements.txt`
|
├── .gitignore <- Gitignore file
|
├── install.md <- Instructions to configure virtual environments and install the package as a distributable
|
├── app.py <- Application running with Gradio locally
|
├── main.py <- Main file to orchestrate re-trains and execution of source code stored in src folder
|
├── main.ipynb <- Like main.py but intended for testing before launching deployments
|
├── setup.py <- Defines project metadata, dependencies, and installation requirements for distribution.
|
└── run.sh <- Executable with predefined commands to run main.py file on a remote server
- To clone this repository using SSH run the next command in your git console
git clone git@github.com:johnma96/tuyabot-llm.git
- To clone this repository using HTTPS run the next command in your git console
git clone https://github.com/johnma96/tuyabot-llm.git
For more details see Clone a repository.
In order to not create conflics between your libraries and the requirements libraries for this project, we highly recomend you to create a new virtual environment to install the requirements libraries in there.
Check out the installation guide here
For more details consult:
The required libraries are listed in the file requirements.txt or environment.yml. Please read the installation guide information for greater detail.