OpenAssistant has been launched to provide “Conversational AI for everyone.” It is reportedly the first model trained on human data that is open-source and optimized for instructions.
An open-source AI chatbot has been released as a free option to ChatGPT, made possible by the LAION e.V (Large-Scale Artificial Intelligence Network). This association has also provided data sets that have allowed the creation of Stable Diffusion, an open AI system for image synthesis, which has been available since August 2022.
The publishers generously make the dataset used for training and pre-trained models open source. This code and data, accessed on Hugging Face, are offered to the public at no cost. Furthermore, a chatbot has been created to assist.
Unlock The Power Of AI Development With Free Data Sets
Andreas Köpf, a developer, and Yannic Kilcher, a tech influencer well-known on YouTube, have spearheaded the project. Working with the LAION community, they have gathered text-based input and feedback over the past few months to generate an impeccable training data set. Their newly released tool is said to be capable of producing State-of-the-Art (SOTA) models for developers.
A large variety of topics and writing styles are represented in the data. Kilcher reports that over 600,000 human-generated samples were included in the dataset and used to train the model.
The conversationally calibrated dataset, similar to LAION’s large image datasets, is available for developers to use as a foundation to train language models and AI applications. These models are accessible under the Apache 2.0 license (apart from those based on LLaMA, which has not yet been published due to licensing issues).
OpenAssistant represents a major milestone in the democratization of conversational AI technology. With this open-source tool, anyone can create powerful chatbots and virtual assistants, bringing the benefits of conversational AI to a wider audience than ever before.
Source: Developer
