This is an AI chatbot powered by a transformer model, designed to interact with users in Vietnamese. The chatbot can handle various tasks, such as answering questions, providing recommendations, and analyzing users based on their age, gender, and interests. It also integrates with speech-to-text and text-to-speech models for seamless communication.
The goal of this project is to create an intelligent and interactive chatbot capable of understanding and generating human-like responses in Vietnamese. We leverage the transformer model, a cutting-edge deep learning architecture in the field of Natural Language Processing (NLP), to achieve this objective. The chatbot is trained on a diverse dataset to ensure accurate and contextually relevant responses.
To use the Transformer chatbot:
clone https://github.com/blak-tran/vietnamese_chatbot_research.git
User interaction Multi-task handling, including answering questions, providing recommendations, and more Context-aware responses, resulting in more natural interactions User analysis Integration with speech-to-text and text-to-speech models
- Python: The core programming language used for building the chatbot.
- Tensorflow: Deep learning framework utilized for implementing the transformer model.
- Transformers Library: Leveraged for accessing pre-trained transformer models.
- FastAPI: Web framework used for building the chatbot's user interface.
To use the chatbot, follow these steps:
-
Setup We used Python 3.9.9 and PyTorch 1.10.1 to train and test our models, but the codebase is expected to be compatible with Python 3.8-3.11 and recent PyTorch versions. The codebase also depends on a few Python packages, most notably OpenAI's tiktoken for their fast tokenizer implementation. pip install -r requirement.txt
-
Inference
uvicorn t2s:app --host 172.17.12.221 --port 8000 uvicorn s2t:app --host 172.17.12.221 --port 8000 uvicorn user_analyst_api:app --host 172.17.12.221 --port 8000 python main_flow_inference.py