Generative AI Google Brain

gemini-prompts-examples-text-summarization

3 min read

Text Summarization with Generative Models on Vertex AI

Source notebook

Repo path: gemini/prompts/examples/text_summarization.ipynb · Open on GitHub · intro

Demonstrates Gemini text summarization prompts on Vertex AI and evaluates summaries with ROUGE.

Summary

This notebook teaches how to use a generative model on Vertex AI to summarize text in several formats, including short summaries, TL;DRs, bullet points, dialogue summaries with to-dos, hashtags, and title options. It initializes the Vertex AI SDK, loads gemini-2.0-flash, varies GenerationConfig parameters, and calls generate_content with task-specific prompts. It also compares a model-generated summary with a human reference summary using ROUGE scores.

Key code patterns

Initialize Vertex AI

import vertexai
 
vertexai.init(project=PROJECT_ID, location=LOCATION)

Sets the Google Cloud project and region before using Vertex AI generative models.

Load Gemini model

from vertexai.generative_models import GenerationConfig, GenerativeModel
 
generation_model = GenerativeModel("gemini-2.0-flash")

Creates a reusable Gemini model client for all prompt examples.

Generate summary

generation_config = GenerationConfig(temperature=0.1, max_output_tokens=256)
 
response = generation_model.generate_content(
    contents=prompt,
    generation_config=generation_config,
).text

Shows the core pattern for sending a summarization prompt with controlled generation parameters.

Tune generation settings

generation_config = GenerationConfig(
    temperature=0.2,
    max_output_tokens=256,
    top_k=1,
    top_p=0.8,
)

Demonstrates changing temperature, top_k, top_p, and output length for different summarization tasks.

Evaluate with ROUGE

from rouge import Rouge
 
ROUGE = Rouge()
ROUGE.get_scores(candidate, reference)

Compares a model-generated summary against a human-created reference summary.

Models & APIs used

  • Models: gemini-2.0-flash
  • APIs / services: Vertex AI
  • SDKs / libraries: google-cloud-aiplatform, vertexai, rouge

When to use this

Use this pattern when building prompt-based text summarization workflows with Gemini on Vertex AI and basic ROUGE evaluation.

Gotchas & caveats

  • Vertex AI is a billable Google Cloud component.
  • A Google Cloud project is required and the Vertex AI API must be enabled.
  • Colab users must authenticate with google.colab.auth.authenticate_user().
  • The notebook uses LOCATION = “us-central1” for Vertex AI initialization.
  • GenerationConfig values can change outputs, so the notebook recommends experimenting with parameters.

Best practices

  • Initialize Vertex AI with an explicit project and location before model calls.
  • Use low temperature for concise summary generation when consistency is desired.
  • Adjust temperature, max_output_tokens, top_k, and top_p for different output styles.
  • Write prompts that specify the desired format, such as bullet points, TL;DR, to-dos, or title options.
  • Evaluate generated summaries against human-created summaries with ROUGE metrics.

Graph View

Backlinks