Hackers' Guide: Gen AI for Gaming, Social, or Digital Entertainment

Welcome, Hackers!

As you embark on the challenge to create the best AI use-case for Gaming, Social, or Digital Entertainment, this guide will provide you with a roadmap to navigate through the process of ideation to execution. Here’s how you can work towards developing a possible use-case:

Ideation and Conceptualization

Begin by brainstorming ideas that align with the challenge directions. Whether it’s an AI companion for gamers, an AI-driven RPG, or an innovative social interaction tool, define the core functionality and unique selling proposition of your concept.

Choose Your AI Model

You have two paths to choose from when it comes to AI models. We’d recommend using an open-source model for this challenge. Do take note of the model licenses - if you’re looking to further develop this idea for commercial use beyond this hackathon, you should check whether the model you’re using allows for that.

Resources:

https://huggingface.co/models

Recommended model to start finetuning

Mistral 7B - this is the base model created by Mistral, you can take any instruct dataset, pair with any prompt template to finetune.

https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.1

Llama 2 7b chat - this is the chat model finetuned by Llama 2, it is fine tuned with Llama2 chat template, so you would have to format your dataset into this prompt template before finetuning.

https://huggingface.co/meta-llama/Llama-2-7b-chat-hf

Apart from the said models, you can choose from a variety of community made models, like the popular silicon-maid, toppy, mythomax, spicyboros, dolphin, pygmalion etc. These models are either finetuned or merged, and you should take note of the base model and the prompt template used to finetune it. That way, you can format your dataset to the same prompt template to ensure consistency in performance.