Generative AI has transcended the realm of potential and has firmly established itself in the mainstream. Enterprises across all industries are rapidly embracing its transformative powers. In this comprehensive session, attendees will gain valuable insights on constructing enterprise-grade, end-to-end generative AI solutions applicable across various modalities including text-to-text, text-to-image, text-to-video, text-to-3D, image-to-text, and image-to-image.
The session will delve into how these solutions can be tailored, built, and deployed anywhere (on-cloud, on-premises, hybrid) using real-time, accurate data. Emphasis will be given on preventing ‘hallucination’ in generative A.I. chatbots, a critical step in making Generative LLMs enterprise-ready.
Participants will explore various types of foundation models and garner knowledge on deciding whether to use generic foundation models or craft domain-specific ones. The session will also encompass training and inferencing LLMs with a focus on fine-tuning and model parallelism techniques. Enterprise-grade LLM considerations, including guardrails models and the integration of LLMs into enterprise databases, will also be discussed. This enlightening session will conclude with practical demonstrations of text-to-text and text-to-video/image/3D applications.
Key Takeaways:
- Knowledge of building and customizing enterprise-grade, end-to-end generative AI solutions.
- Insights into training and inferring LLMs, with emphasis on model parallelism techniques.
- Techniques for preventing ‘hallucination’ in generative A.I. chatbots.
- Understanding of foundation models and their selection for specific domains.
- Exposure to practical applications of generative AI through demonstrations.
