DocumentationAPI ReferenceTutorialsGitHub Code ExamplesDiscord Community

Guides Overview

Here are links to pages which give you step-by-step instructions on getting the most out of Haystack. They are designed to help you through tasks that require interaction with more than one part of the repository.

Creating Custom ComponentsLearn how to create custom components and use them in Pipelines.
Vector-Based vs Keyword-Based RetrieversLearn about the types of Retrievers, how you can choose the best one for your use case, or how you can combine different approaches for hybrid retrieval.
Language ModelsLearn about the different types of language models and when to use them.
Languages Other Than EnglishAdapt a Haystack system to work with languages other than English.
Domain AdaptationImprove your system's performance on data from a specific domain.
OptimizationSpeed up your system or improve its general performance.
EvaluationMeasure how well your system is performing.
Chatbot IntegrationBuild a chatbot that knows how to take advantage of Haystack's NLP power.
Table Question AnsweringPerform retrieval and question answering on tabular data.
Model DistillationCreate a version of a pretrained model that sacrifices some accuracy for speed.
Metadata FilteringNarrow down the dataset that you're working with based on metadata attached to your Documents.
Generative Pseudo LabellingAdapt your dense Retriever to a new domain without human annotation.
Enabling GPU AccelerationUse GPU to speed up your Haystack system.
Log LevelConfigure what messages you want to display in the console output.
TelemetryLearn about the information Haystack uses to bring improvements to its users.