Generative AI represents a branch of artificial intelligence that enables computers to create content such as images, text, code, and synthetic data. These applications are underpinned by foundational technologies including large language models (LLMs).
-
LLMs are deep learning models trained on extensive datasets to perform language processing tasks. They generate new text that mimics human language based on their training data.
-
Foundation models are pre-trained machine learning (ML) models designed for further fine-tuning to achieve specific tasks in language understanding and generation. These models identify patterns in input data that form the basis for generating statistically probable outputs when prompted.
After these models have completed their learning processes, you can employ them to accomplish various tasks, including:
-
Speech and natural language tasks such as chatbots, transcription, translation, question and answer generation, and interpretation of the intent or meaning of text.
-
Image generation based on existing images or utilizing the style of one image to modify or create a new one.
Domino provides several generative AI design patterns that have business impact in production, including:
-
Prompt engineering: Develop specialized prompts to direct LLM behavior.
-
Retrieval Augmented Generation (RAG): Integrate an LLM with external knowledge sources.
-
Fine-tuning: Adapt a pre-trained LLM to specialized datasets or domains.
Domino Data Lab facilitates the entire AI lifecycle from data collection and preparation, through model development and operations, to deployment and monitoring.
You can use the following capabilities to develop generative AI applications in Domino:
| Capability | Description |
|---|---|
Accelerate project development by providing turnkey sample Projects directly in Domino, showcasing best practices for generative AI and classical ML. | |
Vector database connectors for Pinecone and Qdrant | Enable managed access to high-dimension vectorized data for generative AI algorithms like RAG. |
Provides users a safe and streamlined way to access external LLMs hosted by service providers like OpenAI, Anthropic, and more. | |
Give you on-demand code generation, data analysis, notebook generation, and more to help you analyze and develop code more efficiently. | |
Use your datasets to bring context-specific knowledge to pre-trained models. |
In addition to generative AI-specific features, Domino’s core platform manages the complex challenges of generative AI:
| Data access layer | Access vast quantities of disparate data required for generative AI from a central interface. |
|---|---|
Register models and customize the model review and validation process with complete audit records and reproducibility to ensure responsible practices. | |
Train and deploy highly compute-intensive generative AI models. | |
Run AI workloads in any cloud or on-premise environment to reduce costs, simplify scaling, and protect data. | |
Deploy and monitor GenAI LLMs and apps, on-premise or in the cloud. | |
Monitor and reduce AI costs through budgeting, alerts, and efficient cost allocation. |
Domino’s open and interoperable design handles any generative AI scenario. The following are examples of generative AI projects from the AI Hub and the Domino User Guide:
-
Enterprise Q&A applications with RAG using a Pinecone database.
-
Summarize product feedback and respond with emails with LangChain and AWS Bedrock.
-
Build LLM chatbots with Streamlit and OpenAI.
Explore more generative AI projects in the AI Hub.