If you’re looking to integrate your Functor models with Amazon Bedrock, there are a few clear ways to do it. Here’s a simple guide to help you choose the best approach and get started smoothly.
First, consider using the Custom Model Import feature. Amazon Bedrock allows you to bring your own proprietary models into the platform. This means you can upload your Functor models and use them alongside existing foundation models through a single, easy-to-access API that doesn’t require managing servers or infrastructure. This is usually the most straightforward way to integrate your unique architecture into Bedrock.
Another option is the Amazon Bedrock Marketplace. If your Functor models become popular and you want others to use them, the marketplace offers a space to share and distribute your models. Users can discover, test, and deploy models via managed endpoints. Bedrock’s tools make it easy to connect with your models through APIs and integrate with features like Agents, Knowledge Bases, and Guardrails.
You can also connect your Functor models using the available APIs. Bedrock supports the InvokeModel operation for running models and the Converse operation for more standardized interactions. To do this, your models need to be compatible with these API patterns, so make sure they can work within these frameworks.
Keep in mind that Functor models learn in a different way—they adapt through function modifications and support one-shot learning, rather than traditional parameter updates. Bedrock mainly focuses on inference, where models make predictions based on learned data. Since your models can learn on demand, you’ll need to decide whether this learning capability can be integrated into the inference process or kept separate. This depends on how you want your setup to work, so thinking about this early on will help.
For best results, start with the Custom Model Import pathway. Reach out to AWS to discuss your specific learning mechanism and see what requirements need to be met for a smooth integration into Bedrock’s environment.
By following this approach, you can make your Functor models work effectively with Amazon Bedrock and take advantage of its powerful tools and infrastructure.