bomonike

google-ai.png How to useGoogle AI services.

Overview

Google has several divisions that are focused on different AI topics.

Glossary

LLMs vs Foundation Models

From https://arxiv.org/abs/2108.07258 “On the Opportunities and Risks of Foundation Models”

“Vertex AI” is an assortment of AI services:

Generative AI can generate text, images, audio, and video.

Predictive AI can predict the outcome of events.

Services

Bard, Gemini, AutoML, VertexAI

Google NotebookLM asks questions and returns answers in natural language.

  1. Google Bard Large Language Model (25 minutes)

    1. Introduction to Bard and Gemini
    2. Prompt patterns for Bard and Gemini to generate natural language text

    3. Refining natural language text

    4. Generating Bash command line scripts
    5. Generating Python code

  2. Generating Images with Generative AI Studio (15 minutes)

    1. Introduction to Generative AI Studio
    2. Overview of modalities: text, image, audio, and video
    3. Example use cases (text to image, image to text,image to image, image to video)
    4. Designing prompts for text generation
    5. Exercise: Generating images with Generative AI Studio (10 minutes)

  3. Segment 5: Foundation Models and Model Garden in Vertex AI (30 minutes)

    1. Introduction to Foundation Models and Modalities
    2. Overview of Model Garden in Vertex AI
    3. General and Special Purpose Language Models
    4. Image and Vision Models
    5. Multi-modal Image and Text
    6. Exercises: Working with Vertex AI Model Garden (10 minutes)

  4. Segment 6: Overview of Vertex AI (15 minutes)

    1. Managed ML Models
    2. Model monitoring and metrics
    3. Pipelines automate and orchestrate ML workflows structured as a directed acyclic graphs (DAGs) of containerized tasks. Compiled ML pipelines defined in YAML files.
    4. Workflows: Each task in a pipeline performs a specific step in the workflow, such as data preprocessing, model training, or model deployment
    5. Feature Store = Centralized repository for organizing, storing, and serving ML features. Supports offline execution engine feature data in BigQuery, Dataflow, or Dataproc Serverless.
    6. Batch prediction for batch usecases: model endpoint to serve predictions and a Dataflow job to fetch and process the data. Supports custom container images, debugging, and feature filtering
    7. Exercise: Exploring Vertex AI (5 minutes)

  5. Segment 7: Building ML Models with AutoML to automate the development of ML models without the need for machine learning expertise.

    1. Benefits of AutoML: • PyTorch • Tensorflow • scikit-learn • XGBoost
    2. Use Cases of AutoML: • Identify objects in images, • Content classification, sentiment analysis, entity extraction, • Predictions based on tabular data
    3. Limitations of AutoML: • Depends on high-quality, labeled data • Limited ability to customize models compared to building custom models in Vertex AI • May not be sufficient for more complex ML modeling • Expensive with large datasets or extensive usage

    4. Components: AutoML Vision, AutoML Natural Language, AutoML Tables

    5. AutoML Demontstration
    6. Exercise: Building a Model with AutoML (10 minutes)

https://cloud.google.com/vertex-ai/docs/start/introduction-unified-platform

Setup

  1. Sign up for a Google Cloud account:

    https://cloud.google.com/docs/get-started

    https://cloud.google.com/free

  2. Read

    https://ai.google.dev/aistudio

  3. Prompt Gallery

    https://ai.google.dev/gemini-api/prompts

    https://ai.google.dev/aistudio/prompt-gallery

  4. Get API key

    aistudio.google.com

Tutorials

OReilly: AI Google Cloud Services by Dan Sullivan, Lead Solutions Architect, Hydrolix and author of Google Cloud Essentials and Google Cloud AI Services books.

GCP

To request a quota increase: https://cloud.google.com/docs/quotas/view-manage

References

https://cloudonair.withgoogle.com/events/startup-school-ai-q4

Gabi: please send us and email at startupschool-external@google.com, and we’ll try to help you after the session

https://aistudio.google.com/prompts/new_chat

What is the difference between “Google Vertex AI studio” (https://console.cloud.google.com/vertex-ai/studio/) and “Google AI studio” (https://aistudio.google.com/app/prompts/new_data)

https://idx.dev/

idx.google.com

https://console.cloud.google.com/vertex-ai

Instructions for Cloud Skills Boost and activating credits: https://services.google.com/fh/files/emails/ssgenai_cloudskillsboost_instructions.pdf

Nikita Andreyev https://riqli.com/main/search/67230274a7bbc5dea4d00a43 We are using Generative AI and Vertex for test preparation, and get main information from topics

Jay Veal For the folks asking about gen AI examples in education: - Design customized language lessons and exercises tailored to learner needs - Develop tailored tutoring materials and improvement guidance for students

https://www.cloudskillsboost.google/focuses/86502?parent=catalog Get Started with Vertex AI Studio In this lab, you will learn how to:

Vertex AI is a comprehensive machine learning development platform that provides both predictive and generative AI capabilities. It allows you to train, evaluate, and deploy predictive machine learning models for forecasting purposes. Additionally, you can utilize the platform to discover, tune, and serve generative AI models to produce content.

Vertex AI Studio lets you quickly test and customize generative AI models so you can leverage their capabilities in your applications. It provides a variety of tools and resources including both UI (user interface) and coding examples that make it easy to start with generative AI, even if you don’t have a background in machine learning.

This lab guides you through Vertex AI Studio, where you’ll unlock the potential of cutting-edge generative AI models. You’ll explore Gemini and use it to analyze images, design prompts, and generate conversations directly on the Google Cloud console. No need for API or Python SDKs - it’s all accessible through the intuitive user interface.