Overview

Overview

Griptape provides several abstractions for working with data.

Artifacts are used for passing different types of data, such as text, lists, and blobs, between Griptape components.

Embedding Drivers are used to generate vector embeddings from text.

Loaders are used to load textual data from different sources.

Chunkers are used to split arbitrarily long text into chunks of certain token length.

Tokenizers are used to tokenize and detokenize text in order to track LLM token limits.

Audio Engines are used for working with audio.

Extraction Engines are used for extracting structured content.

Image Query Engines are used for querying images with text.

Image Generation Engines are used for generating images.

Summary Engines are used for summarizing text content.

RAG Engines are used for implementing modular RAG pipelines.

Vector Store Drivers are used to store and query vector databases.

Prompt Drivers are used to call LLM APIs.