> For the complete documentation index, see [llms.txt](https://docs-v3.activeloop.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs-v3.activeloop.ai/v3.8.16/performance-features/deep-memory.md).

# Deep Memory

Overview of Deep Lake tools for increasing retrieval accuracy

## How to Use Deep Memory to Improve Retrieval Accuracy in Your LLM Apps

Deep Memory is a suite of tools that enables you to optimize your Vector Store for your use-case and achieve higher accuracy in your LLM apps.

### Embedding Transformation

Deep Memory computes a transformation that converts your embeddings into an embedding space that is tailored for your use case. This increases the accuracy of your Vector Search by up to 22%, which significantly impacts the user experience of your LLM applications. 

Furthermore, Deep Memory can also be used to decrease costs by reducing the amount of context `(k)` that must be injected into the LLM prompt to achieve a given accuracy, thereby reducing token usage.

<figure><img src="https://content.gitbook.com/content/NarYgplYsWUwWA0iaflm/blobs/83gmpH7Ft0C9ZMf3zNi9/Deep_Memory_Comparison.png" alt=""><figcaption></figcaption></figure>