Abhirupa Mitra

Prompt Engineering is a process that empowers us to effectively guide generative AI models and control their output to produce desired results. In this specialized field, we learn techniques, approaches, and best practices for crafting effective prompts. These prompts serve as instructions to large language models (LLMs), allowing us to tap into their powerful capabilities.

Workshop Overview

1. Zero Shot Prompting

Discover how to achieve remarkable results with minimal input. We’ll explore techniques for leveraging pre-trained models without fine-tuning, enabling your AI to perform magic across a wide range of tasks.

2. Persona Patterns

Transform your AI into a chameleon! Learn how to inject personality and context into your prompts, tailoring responses to specific personas or scenarios.

3. Forecasting Patterns

Predictive powers at your fingertips! Dive into forecasting patterns that allow your AI to anticipate future events or outcomes, even when faced with incomplete information.

4. Cognitive Verifier Patterns

Equip your AI with common sense. We’ll explore methods to ensure self-consistency and logical coherence in generated responses, preventing your model from spouting nonsense.

5. Flipped Interaction Patterns

Turn the tables on your model! Understand how to prompt your AI to ask questions, seek clarification, or engage in interactive dialogues with users.

6. Few Shot Prompting

The ultimate cheat code for rapid learning! Learn how to fine-tune your model with just a few examples, making it adaptable to new tasks and domains.

7. Chain of Thought Prompting

Link ideas seamlessly. We’ll cover techniques for maintaining context and continuity across multiple prompts, creating coherent narratives.

8. Generated Knowledge Prompting

Tap into the wisdom of the digital oracle. Explore methods to prompt your AI to generate factual information, summaries, or detailed explanations.

9. Self-Consistency Prompting

Keep existential crises at bay! Ensure that your AI remains consistent with its own knowledge base, avoiding contradictory statements.