/Ads-with-AI

A mega project that includes multiple features to enhance Ads and their Click-Through Rate. The llama-2 model will be fine-tuned for generating query captions. Stable Diffusion to generate Ad Creatives. Datasets like the Meta Ads dataset will help the product determine its target audience.

Primary LanguageJupyter Notebook

Ads-with-AI

Project Overview

Ads-with-AI is a comprehensive project aimed at enhancing advertisements and improving their Click-Through Rate (CTR) through the integration of various features. The project incorporates fine-tuning the Mistral-7B model on puns for generating quirky captions, utilizing Stable Diffusion to create Ad Creatives, and leveraging datasets such as Meta Ads, Puns, and Ads datasets for analysis.

Features

  1. Mistral-7B Model for Query Captions:

    • Fine-tuning the Mistral-7B model for generating query captions that enhance the overall ad content.

  2. Stable Diffusion for Ad Creatives:

    • Implementing Stable Diffusion to generate visually appealing and creative advertisements.

  3. Datasets for Training:

    • Meta Ads dataset for understanding target audience behavior.
    • Puns dataset for incorporating humor and creativity into ad captions.
    • Ads datasets for comprehensive analysis and insights.

Step-by-Step Guide

1. Clone the Repository

git clone https://github.com/your-username/Ads-with-AI.git
cd Ads-with-AI

2. Install Dependencies

pip install -r requirements.txt

3. Download Datasets

4. Model Fine-Tuning

Fine-tune the Llama-2 model using the provided Llama-2 Model.

5. Implement Stable Diffusion

Follow the guidelines in the Stable Diffusion Resource to integrate Stable Diffusion for generating Ad Creatives.

6. Data Analysis

Utilize the Ads datasets for in-depth analysis and insights into ad performance.