Jump to Content
AI21 Studio DocumentationGo to Studio
HomeGuidesAPI ReferenceUpdates
Go to StudioAPI KeyAccountLog InAI21 Studio Documentation
API KeyAccountLog In
HomeGuidesAPI ReferenceUpdates

Getting Started

  • Overview
  • Quickstart
  • AI21 Model Availability Across Platforms

Key Concepts

  • Large Language Models
  • Tokenizer & Tokenization
  • When Will a Language Model Stop Generating Text?
  • How does a model predict the next word?
  • Prompt Engineering
  • Advanced
    • Special Tokens and Whitespaces
    • Repetition Penalties
    • Logit Bias

Foundation Models

  • Jurassic-2 Models
  • Custom Models

Task-Specific models

  • Task-Specific Models Overview
  • Paraphrase
  • Grammatical Error Corrections
  • Text Improvements
  • Summarize Models
    • Summarize
    • Summarize by Segment
  • Text Segmentation
  • Contextual Answers [Single Document]
  • Contextual Answers [Document Library]
  • Semantic Search
  • Embeddings

How To...

  • Build a Dataset
    • Datasets: Best Practices
    • Generating More Data Using AI21 Studio
    • Tips: Building a Dataset
  • Train a Custom Model
    • Tips: Training a Custom Model
  • Query your Custom Model
  • Generation Sets Handbook

Working With Amazon Services

  • Overview
  • Amazon SageMaker
    • Python SDK - with Amazon SageMaker
    • Choosing the Right Instance Type
    • AI21 Models on Amazon SageMaker: Throughput Guide
  • Amazon Bedrock
    • Python SDK - with Amazon Bedrock

Usage

  • Rate limits
  • Responsible Use
  • Safety Research

Amazon SageMaker

Our foundation models (Jurassic-2 series) and task-specific models are available through Amazon SageMaker! You can now deploy each of them in your own private environment. Here you'll find all the information you need, including:

  • How to use the dedicated Python SDK
  • Choosing the best instance for each model

And more.

Updated 6 months ago