DeepSeek Multi-Model Comparison

Description

This project enables the comparison of multiple language models, such as DeepSeek-V3 and Llama3.1, for tasks like text generation. It features a REST API built with FastAPI for real-time inference and Dockerized deployment for easy scalability.

Table of Contents

• Description
• Project Structure
• Installation
• Usage
• License

Project Structure

DeepSeek-MultiModel-Comparison/
├── notebooks/
│   ├── model_comparison.ipynb   # Notebook principal para pruebas y análisis.
├── api/
│   ├── app.py                   # API REST para probar modelos (usando FastAPI o Flask).
│   ├── requirements.txt         # Librerías necesarias para el API.
├── models/
│   ├── load_models.py           # Código para cargar DeepSeek-V3 y otros modelos.
├── datasets/
│   ├── input_data.json          # Datos de entrada para pruebas.
├── results/
│   ├── performance_metrics.csv  # Resultados de métricas.
├── README.md                    # Descripción completa del proyecto.
└── Dockerfile                   # Contenedorización con Docker.

Installation

1. Clone the repository:

   git clone https://github.com/your-username/DeepSeek-MultiModel-Comparison.git
   cd DeepSeek-MultiModel-Comparison

2. Build the Docker container:

   docker build -t deepseek-api .

3. Run the container:

   docker run -p 8000:8000 deepseek-api

Usage

1. Access the API at http://localhost:8000/infer.

2. Provide the following query parameters:

   • model_name: Name of the model (e.g., "deepseek" or "llama").
   • text: Input text for generation.

3. Example request using curl:

   curl -X GET "http://localhost:8000/infer?model_name=deepseek&text=The+future+of+AI+is"

License

This project is licensed under the MIT License. See the LICENSE file for details.