Arcee AI today released its 70 billion parameter large-scale language model SuperNova, an AI model that can be deployed within an enterprise and is designed to address enterprise concerns about data privacy, model stability and customization. Unlike API-based services such as OpenAI and Anthropic, SuperNova allows enterprises to deploy and customize models on their own infrastructure, with complete control. The editor of Downcodes will give you a detailed explanation of SuperNova’s technological innovation, deployment methods, customization capabilities and its impact on corporate AI strategies.
Arcee AI today launched SuperNova, a 70-billion-parameter language model that offers enterprise deployment, advanced command-following capabilities, and full customization options. The model is designed to provide a powerful, affordable alternative to key issues with API-based services such as OpenAI and Anthropic for enterprise data privacy, model stability, and customization.
In an AI space dominated by cloud APIs, Arcee AI is taking a different approach with SuperNova. This large language model (LLM) can be deployed and customized within an enterprise's own infrastructure. SuperNova, released today, is based on Meta's Llama-3.1-70B-Instruct architecture and features what Arcee claims are superior instruction-following capabilities and a novel post-training process that adapts to specific business needs.
technological innovation
The development of SuperNova involves a multifaceted post-training approach
Lucas Atkins, the lead engineer on the project, revealed the secret: "We trained three models, one distilled from the essence of Llama405B, one fed large with the data set generated by our EvolKit, and one with deep DPO on Llama3instruct transformation. Finally, we combined them with a new magic that retained each model’s superpowers.”
Arcee claims that this contributes to SuperNova's command-following capabilities, especially the distillation from the 405B parametric model, which not only shows that SuperNova can capture the essence of larger models while maintaining modest hardware deployment.
Enterprise deployment and customization
SuperNova is designed to be deployed in an enterprise's own cloud environment and will be available first on the AWS Marketplace. Arcee is also working to make it available on the Google and Azure marketplaces.
Mark McQuade, co-founder of Arcee AI, highlights the benefits of this deployment approach: "The model is deployed in your AWS VPC, but it also spins up a web server, a chat interface, and a database to store your chat history. Organization Everyone can interact with it intimately.”
This deployment method solves enterprises' concerns about data privacy and model stability. Unlike API-based services that may change without notice, SuperNova gives enterprises complete control. McQuade noted that this is especially important given the recent turmoil in the AI industry: "OpenAI just dumped 3.5...A lot of companies built their businesses around the 3.5 API. So once that API changes, your application is broken. But in "In our world, nothing changes unless you want it to change, because it's your model and your way of running it."
Customization and continuous improvement
One of SuperNova’s selling points is its ability to be fine-tuned and retrained in an enterprise environment.
Atkins explains the process and its benefits: "Over time, we can retrain the model entirely in your own environment to better match your preferences. As we save these chats, if you want the model to be based on Your unique preferences as a business are improved across the board, and we have the ability to do that without having the data leave your system.”
This capability enables technical teams to adapt models to domain-specific knowledge or company-specific requirements. This is a huge advantage over cloud-based API services that typically don't allow this level of customization.
Open source components
While the full 70B model is not open source, Arcee is releasing several components for the developer community:
Free API for testing and evaluation: This allows developers to try SuperNova without committing to full deployment.
SuperNova-Lite: An open source version of the model with 8B parameters. This smaller model may be useful for developers working in resource-constrained environments or those who want to understand the architecture before deploying the full model.
EvolKit: Their dataset generation pipeline for creating complex QA pairs. This tool can be valuable for organizations looking to create custom training data for their specific use cases.
By open sourcing these components, Arcee contributes to the broader AI community while also providing potential customers with tools to evaluate and customize their products. Arcee SuperNova is also available on AWS Marketplace.
Performance Claims and Benchmarks
Arcee claims that SuperNova performs well in various areas, especially mathematical reasoning. "This one does really well on math benchmarks," Atkins noted. "However, the company encourages third-party evaluations to verify their claims.
"We will provide an API for people to use. If there are third parties who want to run trusted benchmarks to evaluate themselves, we can arrange to provide them with access to the weights. We want to have complete transparency into this model." Atkins said .
This openness to third-party evaluation is laudable because it allows for independent verification of Arcee's claims. It will be particularly interesting to see how SuperNova performs on standard benchmarks compared to models from leading AI companies such as OpenAI, Anthropic and others.
The impact of enterprise AI strategies
The release of SuperNova comes at a time when many companies are re-evaluating their AI strategies. While cloud-based API services have always dominated this space, there is growing interest in deployable, customizable models that offer more control and flexibility.
SuperNova’s approach solves several key issues:
AI sovereignty dilemma
As enterprises navigate the rapidly evolving AI landscape, the release of SuperNova sheds light on a growing tension in the industry: the trade-off between the convenience and power of cloud-based AI services versus the control and customization provided by deployable models. This dichotomy presents what we might call the “AI sovereignty dilemma.”
On the one hand, cloud-based API services like GPT-4 and Claude offer state-of-the-art performance and continuous updates, but at the cost of data privacy issues and limited customization. On the other hand, models like SuperNova promise full control and customization but require in-house expertise to deploy and maintain.
Arcee's approach with SuperNova seeks to bridge this gap, providing a model that can be deployed on-premises while still delivering capabilities designed to compete with leading cloud-based services. This hybrid approach may be particularly attractive to industries with strict regulatory requirements or those that handle highly sensitive data.
Official blog: https://blog.arcee.ai/meet-arcee-supernova-our-flagship-70b-model-alternative-to-openai/
All in all, Arcee AI's SuperNova provides enterprises with a new way to deploy and customize AI, providing significant advantages in data privacy, model stability, and customizability. Although it requires a certain amount of technical investment for deployment and maintenance, its long-term benefits and positive impact on enterprise AI strategies are worthy of attention. The emergence of SuperNova also provides companies with new options in the "AI sovereignty dilemma."