Frequently Asked Questions

The main difference between an SLM and an LLM is the size. Arcee’s views SLM’s ranging in 7B - 70B parameters in size, with our sweet spot sitting between 7B-13B parameters.

Closed-source models have several drawbacks for companies that want to build their own use-case specific models.

• Even though closed-source general purpose LLMs are very powerful, hitting them at scale can be very costly.
• You are at the mercy of closed-source model providers. When their API goes down, or has an increase in traffic, your performance is negatively affected.
• Closed-source model training and training datasets are typically a black box. This means you have very little insight into how they were trained or what data they were trained on.
• Training LLMs with closed-course models exposes your data to third-party platforms, which compromises data security.
• You never have ownership of a closed-source model. Your model should be your asset. Just as data was the new gold, LLMs should be your new gold. These assets should be owned and controlled by you.

In most cases, smaller, more specialized language models running on your own cloud is a more suitable business case. This is what Arcee is built for.

The short answer is yes. We are open-source model agnostic for the most part so you can utilize your model in our SLM adaptation system. Arcee’s SLM adaptation has been built with a modular focus.

Arcee has built a way to take the power of LLMs and bring them down into smaller, specialized and scalable models. We are able to do this thanks to our unique SLM Adaptation System, where we offer customers pre-training, alignment, and continuous retrieval-augmented generation all in one place. Furthermore, our focus on doing everything 100% in your own cloud via virtual private cloud (VPC) deployment means both your data and your models are 100% owned by your from start to finish.

Yes, your model will be 100% yours. It runs inside your cloud, inside your VPC and you can wield it anyway you see fit.

We do not go all the way back to pretraining from scratch, but take the domain adaptive continual pre-training approach. This means we start with a great open-source model, such as Mistral or Llama 2, and extend the pretraining of the existing model by injecting new data into it. The end result is a model that holds all the great general capability of the base model, and all the domain knowledge of your injected corpus. This makes for a much more powerful model than pre-training from scratch, at a fraction of the price.

If you go to such depths of domain adaptation as we do, then you can be smaller with your model size. We also believe the data and knowledge in large general models are largely unnecessary in many business use cases, and are overkill for the ROI that you are trying to achieve.

Arcee has different tiers for API calls in production with a sliding scale for consumption and support tailored to your use case and needs. Please contact sales@arcee.ai for more details.

Yes. Arcee is all about the company owning the model. We have different packages for pre-training only as well as end-to-end deployment. To get a sense of what you can get from Arcee throughout your journey with us, click here for a visualization of our end-to-end deployment and maintenance lifecycle.

One of the core pillars of our product is 100% in-VPC (virtual private cloud) deployment. This means that when you build your SLMs with Arcee, all the way from pre-training to deployment, everything stays in your own cloud. This is the case for both your data AND your SLMs.

Currently we are focused on serving customers who use AWS. In the near future, we plan on expanding to other cloud providers as well. As we expand we are looking for test cases in other clouds so please reach out if you have a use case outside of AWS.

For now, we do not offer multi-modal models. However, this is an area that we are actively exploring, so stay tuned for more information in the coming months!

We do offer multilingual model training, and have done so with some of our customers already. Get in touch with our team with more information about your needs to get a quote for multilingual models.