ExAws.Bedrock

The easiest way to build and scale generative AI applications with foundation models -- https://aws.amazon.com/bedrock/

Service module for Elixir AWS.

NOTE this is a work in progress as the operations are uploaded over the next few days and requires a minimum ex_aws version of 2.5.1.

Installation

The package can be installed by adding :ex_aws_bedrock to your list of dependencies in mix.exs along with :ex_aws, and your preferred HTTP client and JSON codec.

def deps do
  [
    {:ex_aws, ">= 2.5.1"},
    {:ex_aws_bedrock, "~> 1.5"},
    {:hackney, "~> 1.9"},
    {:jason, "~> 1.1"},
    {:poison, "~> 3.0"}
  ]
end

Unit tests

The default suite of unit tests verify the requests generated by this library. If you wish to test against AWS with live requests then you need to configure ExAws in the standard way as described in it's README.

For example you could create an .env file:

export AWS_ACCESS_KEY_ID="AK..."
export AWS_SECRET_ACCESS_KEY="..."
export AWS_REGION="us-east-1"

Then set the variables and run the tests with:

source .env && mix test --only aws

Usage

The AWS bedrock actions and operations are defined in ExAws.Bedrock module.

Each model takes their inference parameters as JSON documents which are defined in the model parameters section of the AWS user guide. You can pass plain Elixir maps which are then JSON encoded, or define structs which implement the encoder protocol of your chosen JSON codec. This library has an example under ExAws.Bedrock.Titan.TextModel.

License

The MIT License (MIT).