ChatGPT’s Rival has Arrived: Hugging Face Introduces Open-Source Version of ChatGPT

The Generative AI race is becoming more explosive and intense. Most big tech giants want to be a part of this massive race.

Hugging Face, the machine learning community and AI tools platform, launched HuggingChat, an open-source ChatGPT replica that anyone can use or download for themselves. The Open Assistant Conversational AI Model serves as the foundation for the HuggingChat ChatGPT clone.

I believe we need open-source alternatives to ChatGPT for more transparency, inclusivity, accountability and distribution of power.

Excited to introduce HuggingChat, an open-source early prototype interface, powered by OpenAssistant, a model that was released a few weeks ago. pic.twitter.com/8U1OY0jnzP

— clem ? (@ClementDelangue) April 25, 2023

The Large-scale Artificial Intelligence Open Network (LAION), a non-profit organization, is the organization behind Open Assistant. LAION believes that machine learning research and its applications should be made more accessible because they have the potential to significantly improve our world. Their main principles are to release code, open datasets, and machine learning models.

They encourage the effective utilization of both energy and computing power to meet the difficulties posed by climate change by making models, datasets, and software recyclable without the requirement to train completely from scratch every time.

Recommended: 4 Top ChatGPT Alternatives That Are Refurbishing the AI Conversational World

According to the Open Assistant conversation model’s GitHub page:

• Open Assistant is a project that aims to make a fantastic chat-based broad language model available to everyone.
• It believes it can aid the world by enhancing language itself, just like stable diffusion did for the creation of new forms of art and visuals.

How was HuggingChat Trained?

The OpenAssistant Conversations Dataset (OASST1), which has been rather recent and contains data that was gathered up until April 12th, 2023, was used to train HuggingChat.

This model employs the same reinforcement learning from human feedback (RLHF) training approach developed by OpenAI.

RLHF is a method for building an excellent data set of queries and responses that have been annotated and graded by humans and may be used to teach an AI to comply with instructions. They succeeded in their mission to make the RLHF approach accessible to everyone wishing to train an AI with this release.

The paper explained,

“In an effort to democratize research on large-scale alignment, we release OpenAssistant Conversations, a human-generated, human-annotated assistant-style conversation corpus consisting of 161,443 messages distributed across 66,497 conversation trees, in 35 different languages, annotated with 461,292 quality ratings.”

• Over 13,000 volunteers from around the world participated in a crowdsourcing project to create the dataset.
• Crowdsourcing was an effective method for producing multilingual training data that helped to create a high-quality dataset.

The crowdsourcing strategy, in the opinion of the researchers, also established limits in the standard of the dataset in the form of cultural and individual biases of those who produced and scored the training data.

Recommended: 5 Ways You Can Use These ChatGPT Alternatives to Fulfill your Business Goals

They cautioned that more active individuals had a tendency to contribute more, leading to an imbalance in the distribution of their beliefs and prejudices.

The researchers come to the conclusion that the dataset might not accurately reflect the variety of perspectives among all the contributors.

Researchers believe that they have developed strict contribution standards that every user must abide by in order to safeguard the efficacy of our dataset. These rules are intended to deter the addition of unsafe material to our dataset and to motivate contributors to produce high-quality responses.

Disclaimer: Expectations of ChatGPT-level productivity are unwarranted at this time. Its version number, 0.0, on the app page should give you an idea of how developed it is right now.