Unleash the Ultimate AI Insult Generator

Table of Contents

1. Introduction
2. The Inspiration behind Humble
3. Meet the Team
4. Language and Technologies Used
5. Data Collection Challenges
6. Training the AI Models
7. Website Design and Development
8. The Models Page
9. The Search Bar
10. The History Page
11. Meet the Team (with Pictures)

Introduction

In this article, we will explore the world of Humble, a unique product that brings together the realms of technology and entertainment. Humble takes the concept of conversational AI to a whole new level by allowing users to interact with virtual versions of famous personalities such as Shakespeare and Gordon Ramsay. These virtual celebrities use their wit and insult prowess to engage users, creating a one-of-a-kind experience. In the following sections, we will delve into the inspiration behind Humble, introduce the team behind its creation, discuss the technologies used, explore the challenges faced during data collection, dive into the AI model training process, examine the website design and development, and showcase the unique features of the Humble website. So buckle up and get ready to be humbled!

The Inspiration behind Humble

At the heart of every great product lies an idea that sparks curiosity and excitement. For the founders of Humble, this idea came to them in a moment of self-reflection. Jun, the main founder, realized that he needed a dose of humility in his life. Inspired by the concept of celebrity roasts, the idea of creating a platform where virtual celebrities could insult users in a lighthearted manner was born. The aim was to create an entertaining and interactive experience that would engage users while also providing a humbling experience. With this vision in mind, Jun and his team set out to bring Humble to life.

Meet the Team

Behind every successful product, there is a dedicated and passionate team. The team behind Humble consists of Jun (the founder), Jason, Sahasra, Kennedy, Jack, Devin, and Mahir. Jun, the mastermind behind the concept, came up with the initial idea and played a crucial role in shaping the vision of Humble. Jason, a key member of the team, provided invaluable assistance in the project's development. Sahasra and Devin worked on the model page, showcasing their expertise in both front-end and back-end development. Kennedy focused primarily on the back-end development of the search engine. Jack contributed to both front-end design and a portion of the back-end work. Finally, Mahir brought his love for Java programming and his experience to the team. Together, this talented group of individuals worked tirelessly to make Humble a reality.

Language and Technologies Used

To bring Humble to life, the team utilized a combination of programming languages and technologies. Python played a significant role in generating the models and implementing AI capabilities. CSS, HTML, and JavaScript were utilized to design and develop the front-end of the Humble website, ensuring a visually appealing and addictive user experience. Additionally, the team made use of libraries and tools such as AI-TEXAN-PIE, Beautiful Soup, NLTK, and regular expressions for tasks ranging from web scraping to data cleaning. The website itself was built using HTML, CSS, JavaScript, and the Bootstrap framework, with W3Schools used to create a slideshow for the team page.

Data Collection Challenges

Collecting data for the Humble AI models presented its fair share of challenges. While most of the data was collected using Cavo, certain celebrities like Nicki Minaj and Shakespeare's insults proved difficult to find. To overcome this hurdle, the team turned to web scraping, scouring various websites to obtain the necessary data. However, they encountered obstacles along the way. For instance, during their attempt to scrape Gordon Ramsay's site, the HTML elements did not load completely, making it nearly impossible to extract the desired content. Furthermore, scraping the Nicki Minaj model's data proved to be a challenge due to the confusing layout of the HTML, resulting in the team needing to search for alternative websites to scrape. Despite these challenges, the final dataset included insults from Gordon Ramsay, Trump, Nicki Minaj, and Shakespeare.

Training the AI Models

The heart and soul of Humble lies in its AI models, specifically the GPT2 unsupervised deep learning language model. With multiple sizes available, the team found that the 124 million model worked perfectly with the amount of data they had. The AI models were trained using AI-TEXAN-PIE, and each celebrity had their own range of epochs, typically ranging from three thousand to five thousand. Training times varied but were generally around 18 minutes and 30 seconds, with the exception of Trump's model, which took 36 minutes due to the large amount of data. This meticulous training process ensured that the virtual celebrities could generate insults that were both entertaining and engaging.

Website Design and Development

The Humble website was designed and developed from scratch, with the team opting to create a unique user interface instead of using pre-existing templates. As they explored various templates, the team found the navigation and understanding of complex template structures to be a challenge. To simplify the process and stay true to their vision, the team decided to start from scratch. The result was a visually appealing and user-friendly website that showcased the humor and wit of the virtual celebrities.

The Models Page

One of the core features of the Humble website is the models page. Here, users can interact with virtual celebrities such as Gordon Ramsay, Donald Trump, Shakespeare, Nicki Minaj, and more. The team opted for a simple user interface that allows users to input a sentence or word, with the virtual celebrity generating a corresponding insult. Though the accuracy may vary due to limited training data, the insults generated by the AI models still offer a unique and enjoyable experience for users.

The Search Bar

To enhance the user experience, Humble features a search bar that allows users to search for insults containing specific words. This search functionality extends to all the collected data, enabling users to discover insults related to their chosen keywords. The search results are displayed randomly, adding an element of surprise and amusement to the user's journey through the world of Humble.

The History Page

Humble includes a history page that records the insults generated by the virtual celebrities during each session. The history page showcases the most recent insults at the top, allowing users to revisit and enjoy the insults they encountered. From the oldest to the latest, the history page provides a comprehensive overview of the insults users have experienced while using Humble.

Meet the Team (with Pictures)

The team behind Humble is a diverse and talented group of individuals. Each member brings their unique skills and interests to the table, contributing to the project's success. From Jun, the visionary founder, to Kennedy, who specializes in back-end development, the team is a cohesive unit, working together to create an exceptional product. To add a personal touch, the team members have included pictures and brief descriptions to introduce themselves. Through their dedication and hard work, the team has crafted an unforgettable experience with Humble.

Highlights

• Humble is an innovative product that allows users to engage with virtual celebrities who insult them in a lighthearted manner.
• The team behind Humble consists of talented individuals who worked tirelessly to bring the product to life.
• Python, CSS, HTML, and JavaScript were used to create the AI models and design the Humble website.
• Data collection presented challenges, including difficulties in scraping certain websites for insults.
• AI models were trained using AI-TEXAN-PIE, with each celebrity having their own specific range of epochs.
• The Humble website was designed from scratch to provide a unique and user-friendly experience.
• The models page allows users to interact with virtual celebrities and receive personalized insults.
• The search bar enables users to search for insults containing specific words, adding to the overall user experience.
• The history page records the insults generated during each session, allowing users to revisit and enjoy their past experiences.
• The team members have shared personal photos and descriptions, adding a personal touch to the Humble experience.