Subnet 11
Dippy
Impel
Dippy Roleplay creates the best open-source roleplay LLM to combat loneliness and enhance companionship
SN11 : Dippy Roleplay
| Subnet | Description | Category | Company |
|---|---|---|---|
| SN11 : Dippy Roleplay | Roleplays LLMs | Model development | Dippy |
The Dippy Roleplay subnet on Bittensor aims to create the world’s best open-source roleplay LLM by leveraging the collective efforts of the open-source community. This subnet addresses the critical issue of loneliness, which affects a significant portion of the population and is linked to various mental and physical health problems.
Current state-of-the-art LLMs (Claude, OpenAI, etc.) are designed for the assistant use case and lack the empathetic qualities necessary for companionship. While some companies (c.ai, Chai, Inflection, etc.) have developed closed-source roleplay LLMs, the open-source alternatives lag significantly behind in performance.
Their team at Impel. understands this issue intimately through building Dippy, a proactive AI companion app for iOS. In this subnet, they will bring together the entire open-source ecosystem to build the world’s best roleplay LLM.
The concept of leveraging AI characters for companionship purposes, such as interaction and support, has gained popularity, emphasizing emotional and social engagement. Integrating AI companions into daily activities like providing mental health prompts based on user behaviors mirrors a shift towards personalized and proactive AI applications. The core thesis revolves around the idea that advanced AI will evolve into companionship and proactive behavior is favored over reactive responses.
Role of Empathetic AI in App Development
The idea of incorporating a hero’s journey concept into the functionality of an app, tailored to the user’s state of mind, highlights the potential impact of AI companions on user experience and productivity. Having an AI companion that enhances efficiency while providing engaging experiences akin to what has historically captivated users through media and storytelling can revolutionize interactions with technology.
Overview of Miner and Validator Functions
Miners utilize existing frameworks, fine-tuning techniques, or MergeKit to train, fine-tune, or merge models, creating unique roleplay Large Language Models (LLMs). These models are then submitted to a shared Hugging Face pool.
Validators assess model performance and evaluate submissions based on various metrics such as empathy and conciseness using our protocol. We provide a suite of testing and benchmarking protocols with state-of-the-art datasets.
The team from Impel includes members with backgrounds from prestigious companies like Microsoft, IBM, and Twitter, who have collectively created apps with over 100 million downloads that were groundbreaking in the AI space.
The Impel team is focused on creating AI models that not only offer utility but also incorporate emotional intelligence elements like compassion, empathy, and humor to enhance user engagement.
After experiencing success with the viral app “Wombo,” the Impel founders decided to shift their focus to creating AI products with a proactive and context-aware approach, aiming to cater to billions of consumers.
Despite being a young company founded in August 2023, impel quickly secured a significant 2.1 million preseed funding round within just a month of incorporation, showcasing rapid growth and potential.
Akshat Jagga – CEO
Angad Arneja – COO