# Navigating in Latent Space and Retrieval-Augmented Generation

This page contains training materials for exploring vector embeddings and using them for enrichting prompts with additional information. This training session is part of the [ScaDS.AI summer school on Neuro-Symbolic AI 2026](https://scads.ai/event/summer-schools/summer-school-2026/).

## Target audience and teaching goal

These  exercises are written for computer scientists who want to use this technique together with large language models in their software applications. By the end of this session, trainees will know how to create vector embeddings of text and images, how to navigate them interactively and use them to improve outputs of large language models.

## Read more

Used software:
* [LLama-Index](https://github.com/run-llama/llama_index)
* [Stackview](https://github.com/haesleinhuepf/stackview)
* [transformers](https://github.com/huggingface/transformers)
* [Vision Embedding Space Travelling](https://github.com/ScaDS/vest)

Used AI models:
* [openai/clip-vit-base-patch32](https://huggingface.co/openai/clip-vit-base-patch32)
* [intfloat/multilingual-e5-large-instruct](https://huggingface.co/intfloat/multilingual-e5-large-instruct)

Recommended literature:
* [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks, Lewis et al (2021)](https://arxiv.org/abs/2005.11401)

## Trainers

* [Dr. Robert Haase](https://haesleinhuepf.github.io/), [ScaDS.AI Dresden/Leipzig](http://scads.ai/)

## Acknowledgements

We acknowledge the financial support by the Federal Ministry of Education and Research of Germany and by Sächsische Staatsministerium für Wissenschaft, Kultur und Tourismus in the programme Center of Excellence for AI-research „Center for Scalable Data Analytics and Artificial Intelligence Dresden/Leipzig“, project identification number: ScaDS.AI