Things should have known before attending to Microsoft AI Day, Thailand (1 May 2024)
be prepared
Microsoft Fabric
Microsoft Fabric is an all-in-one analytics solution for enterprises. It covers everything from data movement to data science, real-time analytics, and business intelligence.
Components:
- Data Lake: A storage repository that holds vast amounts of raw data in its native format.
- Data Engineering: Provides a Spark platform for large-scale data transformation.
- Data Integration: Facilitates seamless data movement and integration.
- Data Science: Supports building and deploying machine learning models.
- Real-Time Analytics: Enables real-time insights from streaming data.
- Power BI Integration: Integrates with Power BI for interactive visualizations.
Advantages:
- Simplifies analytics needs by offering an integrated environment.
- Centralized administration and governance.
- Unified data lake for preferred analytics tools.
- Source
Vector Search
Vector search uses mathematical vectors to represent and efficiently search through complex, unstructured data. It focuses on discovering linked concepts rather than just keywords.
Working Principle:
- Represents data points as vectors in a high-dimensional space.
- Calculates similarity between query vectors and possible vector paths.
- Uses techniques like cosine similarity or Euclidean distance.
Benefits:
- Enhanced semantic relationships and contextual meaning.
- Efficiently finds similar items in datasets.
- Source
RAG (Retrieval-Augmented Generation)
RAG optimizes large language models (LLMs) by referencing authoritative knowledge bases outside their training data. It supplements prompts with external information.
Use Cases:
- Improves factual reliability without extensive model fine-tuning.
- Enhances LLM output for specific domains.
Components:
- Prompt Engineering: Skillful creation of text prompts.
- Fine-Tuning: Adapting pre-trained LLMs for specific use cases.
- RAG (Knowledge Augmentation): Incorporating external sources into prompts.
- Source
LLMOps (Large Language Model Operations)
LLMOps manages the lifecycle of LLMs and LLM-powered applications.
Key Concepts:
- Collaboration Environment: Provides tools for AI developers and IT administrators.
- Prompt Engineering, Fine-Tuning, and RAG are part of LLMOps.
- Central Setup and Management: Security configuration, cost control, and governance.
- Source
Azure AI Studio
Azure AI Studio is a generative AI development hub. It provides tools for building, deploying, and managing AI applications.
Capabilities:
- Collaboration Environment: Allows teams to build and manage AI applications.
- Prompt Engineering, Fine-Tuning, and RAG can be used within Azure AI Studio.
- Central Management: Security settings, project organization, and governance controls.
- Source
Prompt Flow
Prompt Flow involves skillfully creating text prompts to guide LLMs toward producing desired outputs. It’s crucial for achieving accurate and contextually relevant responses.
- Techniques: Few-shot, chain-of-thought (CoT) prompting.
- Use Cases: Interaction with LLMs via API calls or user-friendly platforms.
- Source