Generative AI Service Product Tour

 

Transform your business with generative AI

Generative AI overview view, details below
The overview view provides a menu on the left with options:
  • Overview
  • Playground
  • Dedicated AI clusters
  • Custom models
  • Endpoints
Under Scope a dropdown menu allows the user to select their compartment.

Power your apps with large language models and generative AI

OCI Generative AI is a fully managed service that provides a set of state-of-the-art, customizable LLMs that cover a wide range of use cases for text generation. Use the playground to try out the models out-of-the-box or create and host your own fine-tuned custom models based on your own data on dedicated AI clusters.

The following Metrics are provided for the current component:
  • Dedicated AI clusters
  • Custom models
  • Endpoints
Sections for Resources and Get started are cropped on the edge of the image.

A fully managed generative AI service built for business

Leverage customizable large language models (LLMs) that are pretrained and ready to use. Use these models across a large set of generative AI use cases, such as text summarization, copy generation, search, chat, and more.

Read more about Oracle’s generative AI strategy

Learn more about generative AI use cases

Generative AI Playground section view, details below
The Generative AI Playground view allows the user to select the model and example. AI parameters available for the user to change in the right column include:
  • Maximum output tokens
  • Temperature
  • Top p
  • Top k
  • Stop sequence
  • Frequency penalty
Additional parameters are cropped at the bottom of the image. Sample input and output text is shown. In this case a paragraph of instructions on updating a computer operating system is provides as input, with instructions to provide output as a numbered list, which is shown.

Use the Playground for out-of-the-box models and sample library

The Playground provides a chat-like exploratory experience to easily work with the out-of-the-box LLMs and fine-tuned custom models. Experiment with parameter settings to change the model output. Automatically generated Python and Java code snippets can be used in your applications.

See the Playground in action (7:33)

Data governance and security

Enterprise-grade generative AI with data governance and security

Only you have access to your data and custom trained models. Only you can use a custom model trained on your data. No other customers can see your data or custom models.

Learn more about data handling in OCI Generative AI

AI infrastructure

Best-in-class AI infrastructure

Use prebuilt models on demand by paying only for tokens sent and received or use dedicated AI clusters available for predictable performance and pricing. Both use OCI, which provides industry-leading performance and cost efficiency with NVIDIA H100 Tensor Core GPUs.

Discover OCI AI infrastructure

Available pretrained models and custom model endpoints view

Start fast with pretrained, out-of-the-box LLMs

Three generative models are available out of the box.

  • Cohere Command and Meta’s Llama 2 for text generation
  • Cohere Command for text summarization
  • Cohere Embed for text embedding

Learn more about the pretrained models

Text generation section view, details below
Input, output and parameters are shown from the Generative AI Playground view. AI parameters available for the user to change in the right column include:
  • Maximum output tokens
  • Temperature
  • Top p
  • Top k
  • Stop sequence
  • Frequency penalty
  • Presence penalty
Additional parameters are cropped at the bottom of the image. Sample input and output text is shown. In this case the LLM is asked to create a text job description for a data visualization expert with a list of three qualifications, and the output text is shown below (and cropped).

Text generation

Rapidly create job descriptions, blogs, emails, marketing copy, and more. Perform text translations. Ask questions conversationally in natural language.

Learn more about generation model usage and parameters

Summarization view, details below
Input, output and parameters are shown from the Summarization view. AI parameters available for the user to change in the right column include:
  • Length
  • Format
  • Extractiveness
  • Temperature
  • Additional command
Sample input and output text is shown. In this case two paragraphs about Oracle's AI strategy are provides as input, while the output contains four bullet points summarizing the paragraphs.

Summarization

Create short, accurate, and fluent summaries of articles, blogs, emails, and other bodies of text using the Cohere Summarize model. Save time and effort to comprehend and share the core ideas within a longer text document.

Learn more about summarization models

Embedding export view, details below
Cropped at the top is a list of 11 items showing statements
  • 9. We are committed to being transparent with our customers about our...
  • 10. Billing can be a complex process, and we are here to help our custom...
  • 11. We value our customers and want to ensure that they are happy with...
Below are buttons for Run and Clear, followed by an output vector projection chart with points labelled one through eleven. A button to Export embeddings to JSON is provided.

Embedding

Create vector representations of paragraphs, sentences, and phrases. Visualize the output vector to identify outliers and similarly grouped phrases. Use these vector representations for semantic search, text classification, and many other use cases.

Fine-tuning configuration section view, details below
Create model interface is shown. Three steps are listed as links:
  • Model definition
  • Fine-tuning configuration (current step)
  • Data selection
Fine-tuning configuration options are available:
  • Base model
  • Fine-tuning method
  • Compartment / Dedicated AI cluster

Refine models with your own business knowledge

Fine-tune Cohere’s Command and Command Light models with your domain knowledge. Use the custom model endpoints in your applications. Use dedicated GPU clusters for predictable performance and pricing.

See the model endpoint documentation

Watch details on fine-tuning (4:04)

Generative AI overview view, details below
The overview view provides a menu on the left with options:
  • Overview
  • Playground
  • Dedicated AI clusters
  • Custom models
  • Endpoints
Under Scope a dropdown menu allows the user to select their compartment.

Power your apps with large language models and generative AI

OCI Generative AI is a fully managed service that provides a set of state-of-the-art, customizable LLMs that cover a wide range of use cases for text generation. Use the playground to try out the models out-of-the-box or create and host your own fine-tuned custom models based on your own data on dedicated AI clusters.

The following Metrics are provided for the current component:
  • Dedicated AI clusters
  • Custom models
  • Endpoints
Sections for Resources and Get started are cropped on the edge of the image.

注:为免疑义,本网页所用以下术语专指以下含义:

  1. Oracle专指Oracle境外公司而非甲骨文中国。
  2. 相关Cloud或云术语均指代Oracle境外公司提供的云技术或其解决方案。