Compare CHAI vs. StableVicuna

We are developing a premier platform for conversational AI, having initiated our journey with a unique dataset containing billions of chat interactions, investing over $3 million into training language models that captivate users. Today, millions engage with our platform daily, as we tirelessly enhance our models to ensure they remain increasingly entertaining. Explore chat AIs from various corners of the world and interact with them to uncover their diverse capabilities. With a community of millions involved in chatting, creating, and sharing unique chat AI personalities, we are dedicated to empowering users to experience the most enjoyable chat AI available. Our models are built on billions of tokens, supplemented by countless reward signals provided by our user base. Through conducting AB tests with real users, we've achieved a new model that outperforms OpenAI ChatGPT in terms of session duration. We not only develop and refine our own language models, but we are also in a constant cycle of training using our exclusive chat message dataset, making sure our platform evolves with the needs and interests of our community. This relentless pursuit of excellence ensures that our chat AI remains at the forefront of innovation in conversational technology.

StableVicuna represents the inaugural large-scale open-source chatbot developed through reinforced learning from human feedback (RLHF). It is an advanced version of the Vicuna v0 13b model, which has undergone further instruction fine-tuning and RLHF training. To attain the impressive capabilities of StableVicuna, we use Vicuna as the foundational model and adhere to the established three-stage RLHF framework proposed by Steinnon et al. and Ouyang et al. Specifically, we perform additional training on the base Vicuna model with supervised fine-tuning (SFT), utilizing a blend of three distinct datasets. The first is the OpenAssistant Conversations Dataset (OASST1), which consists of 161,443 human-generated messages across 66,497 conversation trees in 35 languages. The second dataset is GPT4All Prompt Generations, encompassing 437,605 prompts paired with responses created by GPT-3.5 Turbo. Lastly, the Alpaca dataset features 52,000 instructions and demonstrations that were produced using OpenAI's text-davinci-003 model. This collective approach to training enhances the chatbot's ability to engage effectively in diverse conversational contexts.