Introduction to the Amazon Bedrock Workshop
The Amazon Bedrock Workshop is a hands-on experience designed to equip developers and solution builders with the skills to leverage foundation models (FMs) using Amazon Bedrock. It aligns with a self-paced or instructor-led workshop, where participants can dive deep into practical labs that demonstrate the application of Generative AI in various scenarios.
About Amazon Bedrock
Amazon Bedrock is a fully managed service that facilitates access to foundation models from third-party providers and Amazon through an API. It allows users to select the most appropriate model for their specific use case. This enables the creation of text, images, and other media, enhancing organizational productivity by improving operations like text composition, summarization, question answering, and chatbot creation.
Workshop Structure
The workshop is structured as a series of labs that illustrate common use cases for Generative AI. Some of the main activities include:
-
Text Generation: This lab teaches text generation with Bedrock, text summarization with models like Titan and Claude, question answering with Titan, and entity extraction. Estimated to take about 45 minutes.
-
Knowledge Bases and RAG: Participants learn to manage RAG (retrieve and generate) functions using tools like Langchain. This lab also takes approximately 45 minutes.
-
Model Customization: Focuses on fine-tuning models such as Titan Lite and Llama2. Note that this requires running on a personal AWS account. The expected duration for this lab is 30 minutes.
-
Image and Multimodal Generation: This section includes generating images with the Bedrock Titan image generator and working with multimodal embeddings. Again, this lab should take about 30 minutes.
-
Agents: Participants learn to build customer service and insurance claims agents using the Bedrock platform, with each lab session estimated to last 30 minutes.
-
Open Source Examples (Optional): These additional sessions, taking about 30 minutes each, cover Langchain text generation, knowledge base RAG examples, chatbot examples, NVIDIA NeMo Guardrails examples, and NodeJS Bedrock examples.
Getting Started
To begin, participants need to set up a notebook environment. The workshop is presented with Python notebooks, which can be executed in environments such as:
- SageMaker Studio: This is a recommended choice for those desiring a robust AI/ML experience.
- SageMaker Notebook Instances: Ideal for a more straightforward, fully-managed experience.
- Local or Other Notebook Environments: Participants must ensure AWS credentials are configured properly.
AWS IAM permissions must be enabled for the identity used in the notebook environment. This involves setting appropriate permissions in the AWS IAM Console.
Cloning the Workshop
Participants can clone the workshop repository to their chosen environment. Instructions include installing necessary tools, retrieving the repository via git clone, and accessing the notebooks for exploration. Starting with the prerequisite notebook will guide users in installing SDKs, creating a client, and beginning API calls with Python.
Additional Resources
The workshop provides step-by-step guided instructions for a smoother experience. As participants progress, they can expect to gain significant hands-on experience with how Amazon Bedrock can be utilized for innovative AI solutions.